CN110704616A - Equipment alarm work order identification method and device - Google Patents
Equipment alarm work order identification method and device Download PDFInfo
- Publication number
- CN110704616A CN110704616A CN201910847257.3A CN201910847257A CN110704616A CN 110704616 A CN110704616 A CN 110704616A CN 201910847257 A CN201910847257 A CN 201910847257A CN 110704616 A CN110704616 A CN 110704616A
- Authority
- CN
- China
- Prior art keywords
- work order
- alarm work
- type
- training
- alarm
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/35—Clustering; Classification
- G06F16/355—Class or cluster creation or modification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/3065—Monitoring arrangements determined by the means or processing involved in reporting the monitored data
-
- 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
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Databases & Information Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method and a device for identifying an equipment alarm work order, wherein the method comprises the following steps: acquiring alarm work order content of equipment to be identified in a preset scene; inputting the content of the alarm work order of the equipment to be identified in a preset scene into a work order type identification model to obtain the type of the alarm work order of the equipment to be identified; the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types in a preset scene, and the sub-training sets with balanced types are obtained by processing sample data of a historical unbalanced type alarm work order. According to the technical scheme, the accuracy and efficiency of equipment alarm work order type identification are improved.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a method and a device for identifying an equipment alarm work order.
Background
And in the running process of the equipment, generating an alarm work order when the conditions of abnormal running and the like of the equipment are monitored. At present, alarm work orders generated particularly in group-type enterprises are increasingly huge, and in the process of classifying a large number of alarm work orders, the types of the alarm work orders are mainly recognized by common methods such as manual work, the common methods cannot correctly recognize the types of the alarm work orders, so that the alarm work orders are wrongly classified, the correctness of the type recognition of the alarm work orders needs to be checked, and the efficiency is low.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the invention provides an equipment alarm work order identification method, which is used for improving the accuracy and efficiency of equipment alarm work order identification and comprises the following steps:
acquiring an alarm work order of equipment to be identified in a preset scene;
inputting the alarm work order of the equipment to be identified in a preset scene into a work order type identification model to obtain the type of the alarm work order of the equipment to be identified; the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types in a preset scene, and the sub-training sets with balanced types are obtained by processing sample data of a historical unbalanced type alarm work order.
The embodiment of the invention also provides a device for identifying the equipment alarm work order, which is used for improving the accuracy and efficiency of identifying the equipment alarm work order and comprises the following components:
the device comprises an acquisition unit, a processing unit and a processing unit, wherein the acquisition unit is used for acquiring an alarm work order of the device to be identified in a preset scene;
the identification unit is used for inputting the alarm work order of the equipment to be identified in a preset scene into the work order type identification model to obtain the type of the alarm work order of the equipment to be identified; the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types in a preset scene, and the sub-training sets with balanced types are obtained by processing sample data of a historical unbalanced type alarm work order.
The embodiment of the invention also provides computer equipment, which comprises a memory, a processor and a computer program which is stored on the memory and can be run on the processor, wherein the processor realizes the equipment alarm work order identification method when executing the computer program.
The embodiment of the invention also provides a computer readable storage medium, and the computer readable storage medium stores a computer program for executing the equipment alarm work order identification method.
The technical scheme provided by the embodiment of the invention is as follows:
firstly, the inventor finds the technical problem that the sample data of the historical alarm work order is the sample data of the unbalanced type, and therefore proposes that: the method comprises the steps of processing sample data of a historical unbalanced type alarm work order to obtain a plurality of sub-training sets with balanced types, pre-training the sub-training sets with balanced types to generate a work order type recognition model, recognizing the alarm work order of the equipment to be recognized by using the work order type recognition model to obtain the type of the alarm work order of the equipment to be recognized, improving the accuracy of the equipment alarm work order recognition, simultaneously, omitting the step of checking the correctness of the alarm work order type recognition, and improving the efficiency of the equipment alarm work order recognition.
Secondly, the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types of a preset scene, the type of the alarm work order of the equipment to be recognized is recognized based on the work order type recognition model with the scene taken into consideration, and the accuracy of the alarm work order recognition of the equipment is also improved.
In summary, the technical scheme provided by the embodiment of the invention improves the accuracy and efficiency of equipment alarm work order identification.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a flow chart of a method for identifying an equipment alarm work order in the practice of the present invention;
FIG. 2 is a flow diagram of an equipment alarm work order in the practice of the present invention;
FIG. 3 is a schematic diagram of historical imbalance type alarm work order sample data in the practice of the present invention;
FIG. 4 is a schematic diagram of a work order type identification model in the practice of the present invention;
FIG. 5 is a schematic diagram of a classifier model included in the worksheet type recognition model in the practice of the present invention;
FIG. 6 is a schematic diagram illustrating the advantages of an XGboost recognition model in the practice of the present invention;
fig. 7 is a schematic structural diagram of an equipment alarm work order identification device in the implementation of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The inventors found out the technical problem: the existing alarm work order data is a data set with unbalanced types. The inventor finds the technical problem, provides a scheme for identifying the warning work order of the unbalanced category, and then classifies the warning work order according to the identification result. The scheme comprises the following steps: and obtaining a training set according to the classification data of the historical alarm work order, wherein the training set is obtained by the historical alarm work order, is subdivided into alarm work orders of different classes by taking the temperature class as an example, and is respectively a non-intervention class, a confirmation service state receipt class and a manual intervention class. This type of text parsing classification belongs to short text classifications in the natural language processing domain. The statistical distribution is carried out according to the category of the historical alarm work order, the training set is found to be an unbalanced data set, and the processing of the classification of the unbalanced data set is difficult. And generating a classifier aiming at the unbalanced data set by text representation and learning classification according to the training set, thereby improving the accuracy of automatic classification. The target alarm work order is classified through the classifier according to the content of the target alarm work order to obtain the alarm work order classification result, and the technical difficulties of low classification efficiency and low accuracy in the existing unbalanced classification text classification technology of natural language processing are solved.
The present invention relates to the field of data processing technologies and natural language processing, and in particular, to a method and an apparatus for classifying an unbalanced type alarm work order. The classification of the electronic work order alarm is manual classification at present, taking a temperature work order as an example, the processing of the temperature work order is classified into 3 types, namely non-intervention type, business state confirmation receipt type and manual intervention type. And the occupation ratio of the category 1 non-intervention class is far greater than that of the category 2 confirmed service state receipt class and that of the category 3 converted into the manual intervention class, and the occupation ratio of the category 1 in the randomly selected data set is 10 times that of the category 3 and is about 5 times that of the category 2. If the classification recall rate of the common classification method is low, the alarm work order cannot be correctly identified and subsequently classified, the work order data volume is more and more huge along with the increasing of work order data, and if the method suitable for machine learning of the scene (unbalanced work order type scene) can be used for realizing correct automatic classification, the efficiency can be greatly improved, and the time can be saved. The following describes the equipment alarm work order identification scheme in detail.
Fig. 1 is a schematic flow chart of an equipment alarm work order identification method in the implementation of the present invention, and as shown in fig. 1, the method includes the following steps:
step 101: acquiring an alarm work order of equipment to be identified in a preset scene;
step 102: inputting the alarm work order of the equipment to be identified in a preset scene into a work order type identification model to obtain the type of the alarm work order of the equipment to be identified; the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types in a preset scene, and the sub-training sets with balanced types are obtained by processing sample data of a historical unbalanced type alarm work order.
The technical scheme provided by the embodiment of the invention is as follows:
firstly, the inventor finds the technical problem that the sample data of the historical alarm work order is the sample data of the unbalanced type, and therefore proposes that: the method comprises the steps of processing sample data of a historical unbalanced type alarm work order to obtain a plurality of sub-training sets with balanced types, pre-training the sub-training sets with balanced types to generate a work order type recognition model, recognizing the alarm work order of the equipment to be recognized by using the work order type recognition model to obtain the type of the alarm work order of the equipment to be recognized, improving the accuracy of the equipment alarm work order recognition, simultaneously, omitting the step of checking the correctness of the alarm work order type recognition, and improving the efficiency of the equipment alarm work order recognition.
Secondly, the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types of a preset scene, the type of the alarm work order of the equipment to be recognized is recognized based on the work order type recognition model with the scene taken into consideration, and the accuracy of the alarm work order recognition of the equipment is also improved.
In summary, the technical scheme provided by the embodiment of the invention improves the accuracy and efficiency of equipment alarm work order identification.
In specific implementation, the alarm work order related to the embodiment of the invention has the following meanings: enterprises, such as corporate enterprises, may monitor devices in various regions, and if a device fails, an alarm work order may be sent to the corporate unified management platform, and the corporate may perform feedback according to specific content of the alarm work order, for example, screenshot of the content and field of the alarm work order is shown in fig. 2.
In specific implementation, the meaning of the preset scenario related to the embodiment of the present invention is: the method comprises a temperature class scene, wherein the temperature class indicates that the alarm work order is an alarm caused by temperature, and the method further comprises the following steps: other scene types, such as performance class scenes, traffic class scenes.
In specific implementation, the meaning of the work order type related to the embodiment of the invention is as follows: taking temperature classes as examples, the method comprises the following steps: no intervention class, confirmation of business state receipt class and manual intervention class; wherein: the non-intervention class representation comprises the feedback of contents in the middle and automatic invalidation of the processing; confirming whether the business state receipt indicates that whether the business state and the consultation are receipt or not is confirmed, and checking no alarm; the manual intervention type represents the condition that the work order needs to be cut, dispatched, repaired by the board card and the like, which accords with the hanging condition, the air conditioning problem and the like and can not be well processed in a short time. The work order is hung up with help of the group. Examples are as follows, e.g. without intervening class content examples: "in-process"; confirming service status receipt type content example: "Zhengzhou field checking machine room temperature is normal, equipment temperature is normal"; turning to the manual intervention class content example: "trouble group hangs up work order as appropriate".
In specific implementation, the classifier mentioned in the embodiment of the present invention may refer to a work order type recognition model. The work order classifier model may refer to a work order recognition model.
The following describes in detail each step related to the method for identifying an equipment alarm work order provided by the embodiment of the present invention with reference to fig. 2 to 7.
Firstly, a step of generating the work order type recognition model by pre-training is introduced.
In one embodiment, the work order type recognition model may be generated by pre-training as follows:
acquiring sample data of a historical unbalanced type alarm work order; the historical unbalanced type alarm work order sample data comprises alarm work order content and a corresponding type thereof; the historical unbalance type alarm work order sample data comprises a plurality of types of unbalance alarm work order sample data;
processing the alarm work order sample data with the plurality of types of unbalance to determine the number of each type of alarm work order sample;
copying the alarm work order type sample with the least number of samples into a preset number of samples; sampling and dividing the sample data of each other type of alarm work order into the preset number of parts, wherein each number of parts is the same as the number of samples of the alarm work order type with the minimum number of samples;
forming a plurality of types of balanced sub-training sets according to the following method: combining one part of the alarm work order type sample with the least number of samples with one part of the alarm work order sample data of each other type to form a sub-training set, and forming a plurality of sub-training sets with balanced types by analogy;
training each sub-training set to generate a base recognition model;
and generating the work order type identification model according to the plurality of base identification models.
The detailed steps for generating the work order type identification model will be described below.
1. Firstly, the steps of processing the alarm work order sample data with unbalanced types into a sub-training set with balanced types are introduced.
In specific implementation, a balanced training set is obtained according to historical unbalanced alarm work order sample data, the historical unbalanced alarm work order sample data comprises work order content and historical alarm work order types, and the obtained training set is a balanced data set with a specified number of copies. And performing repeated oversampling on the alarm work order data, specifically, taking the work order type with a small quantity as a reference, and performing sampling on the work orders of other types by the specified quantity of the reference quantity to finally obtain the samples of the balanced types of the specified quantity.
In specific implementation, the screenshot of the data part of the historical unbalanced alarm work order is shown in fig. 3, and includes two columns, feedback content and a label. Wherein, 1, 2 and 3 in the label column respectively represent a non-intervention class, a confirmed business state receipt class and a manual intervention class. And counting the work order type with the minimum category, and if the minimum category is the category 3, totally obtaining 2000 samples. Based on the class of the work order, if the work orders of other classes are sampled by 10 specified numbers of samples, for example, category 1 and category 2, 20000 samples are sampled respectively and then divided into 10 samples, and each sample contains the same number of category 1, 2, 3 samples to be combined to obtain a new sub-training set.
2. Next, a process of training and obtaining a worksheet type recognition model by using the obtained sub-training set is described.
In specific implementation, the alarm work order classifier (work order type recognition model) is generated by performing text representation and learning classification on each sub-training set, and specifically, a base classifier is generated by performing text representation and learning classification training on each training set. And each training set is used for adding a user dictionary, text word segmentation, stop word removal, low-frequency word removal and feature engineering, and a plurality of XGboost (Extreme Gradient Boosting) classifiers are further used for generating an alarm work order classifier in a combined mode. The detailed process is described below.
(1) First, generation of a base recognition model (a classifier) is introduced.
In one embodiment, training each of the sub-training sets to generate the base recognition model may include:
performing word segmentation on sample data of each alarm work order of each sub-training set to obtain a first word set;
removing stop words from the words in the first word set to obtain a second word set;
removing low-frequency words from the words in the second word set to obtain a third word set;
constructing feature vectors for words in the third set of words;
and training according to the feature vectors and the corresponding work order types to generate a base recognition model.
In specific implementation, the following operations are performed on each training data set (sub-training set):
① scientific research uses hidden Markov model to divide each record of training set into words, and adds it into user dictionary to obtain word set (first word set), for example, dividing "please check if recover" into "please/group/check/if/recover".
② the stop word is removed from ① (in the first word set) to get the word set (second word set), the stop word is a word without actual meaning in natural language, such as some function words, qualifiers, digital punctuation marks, etc.
③ the word set (third word set) is obtained after removing low frequency words from the word set (in the second word set) in ②, for example, words with a word frequency less than 2 can be removed, which is beneficial to extracting important words for analysis.
④, constructing feature vectors for the word sets (in the third word set) in ③ (specifically, a feature vector model as shown in fig. 4 may be generated for feature vector conversion of the subsequent alarm work order of the device to be recognized), and converting the text content of the work order into a vector format, which is used as an input of a classification algorithm (specifically, a classifier model as shown in fig. 4 may be constructed, and may also be referred to as a work order recognition model, and a structural diagram of the work order recognition model may be shown in fig. 5).
⑤, the feature vectors in ④ and the corresponding worksheet type data sets are used as the input of an XGboost (extreme gradient Boosting) classifier.
(2) Next, a process of generating the work order type recognition model from a plurality of basis recognition models, that is, generating an alarm work order classifier (work order type recognition model) by combining classifiers which are described in "(1) ⑤" above, will be described.
In one embodiment, training each of the sub-training sets to generate the base recognition model may include: training each sub-training set to generate an XGboost recognition model;
generating the work order type identification model from a plurality of base identification models may include: and generating the work order type identification model according to the plurality of XGboost identification models.
In specific implementation, a plurality of classifiers (base identification models) are trained by utilizing the feature vectors and corresponding work order type data sets thereof, and can be XGboost (Extreme Gradient Boosting/Extreme Gradient Boosting) base identification models, and the XGboost algorithm mainly aims to establish K CART trees, so that the predicted values of the tree groups are close to the true values (accuracy rates) as much as possible and have generalization capability as much as possible. The XGBoost algorithm (Extreme Gradient Boosting) has the following advantages compared with other algorithms.
XGBoost (Extreme Gradient Boosting) is a more advanced and efficient implementation of the Gradient Boosting algorithm. Advantages over other Boosting techniques: the speed is 10 times faster than the ordinary Gradient Boosting because it can implement parallel processing. One characteristic of the XGBoost (Extreme Gradient Boosting) over the GBDT (Gradient Boosting decision tree) is that it adds a regularization term to the cost function for controlling the complexity of the model, and it adds the leaf node output L2 for smoothing. Shrinkage and column subsampling were added to prevent overfitting.
The XGboost principle is as follows: for a given n samplesThe dataset of the present and m features D ═ xi,yi)(|D|＝n,xi∈Rm,yiE R) a tree ensemble model predicts the output using K cumulative functions:
wherein: f (x) wq(x)(q:R→T,w∈RT) Is the space of the CART tree (Classification and regression trees). Where q represents the structure of each tree that can map each sample into a corresponding node, and T is the number of leaf nodes in the tree. Each fkCorresponding to an independent tree structure q and leaf weights w. Unlike decision trees, each regression tree contains a continuous score value, w, at each leaf nodeiRepresenting the score of the ith node. w is aq(x)Is the score for sample x, i.e., the model prediction value. For each sample, a plurality of decision rules in tree are used to classify it into leaf nodes, and the final prediction is obtained by accumulating the scores w in the corresponding leaves (the prediction result for each sample is the sum of the prediction scores for each tree, as shown in fig. 6).
To learn the set of functions used in the model, the following regularization objectives need to be minimized:
In addition to regularizing the target, two techniques are additionally used to further prevent overfitting. The first technique is shrinkage (Shinkage), which is newly weighted by a factor η after each step of tree boosting. Similar to the learning rate in random optimization, shinkage reduces the impact of each individual tree and leaves room for future trees to optimize the model. The second technique is Column Subsampling. The use of column sampling prevents overfitting much more than conventional row sampling and also speeds up the computation of the parallel algorithm.
In conclusion, the advantage of the XGBoost can effectively process the unbalanced data set and prevent overfitting. Therefore, aiming at the particularity of the unbalanced data set, the XGboost recognition model is selected to be most suitable, and the accuracy and efficiency of the equipment alarm work order type recognition are improved.
Secondly, the above step 101 is described.
And acquiring the alarm work order content of the equipment to be identified in the preset scene, namely the target alarm work order content.
Third, next, the above step 102 is introduced.
In specific implementation, the target alarm work order is classified (identified) through an unbalanced classifier (work order type identification model) according to the content of the target alarm work order, so that classification is realized, and an alarm work order classification result is obtained.
In an embodiment, inputting the alarm work order of the device to be identified in the preset scene into the work order type identification model to obtain the type of the alarm work order of the device to be identified may include:
converting the alarm work order content of the equipment to be identified in a preset scene into a feature vector;
and obtaining the type of the alarm work order of the equipment to be identified according to the characteristic vector.
In specific implementation, as shown in fig. 4, the content of the target alarm work order is input into the classification device (the above-mentioned work order type identification model and alarm work order classifier), and the type of the target alarm work order can be obtained. The sorting apparatus may comprise several processes, as shown in fig. 4.
In specific implementation, as shown in fig. 4, in the process of training a classifier (work order type identification model), a feature vector model and a classifier model of an alarm work order of a training result are generated, that is, the work order type identification model may include the feature vector model and the classifier model shown in fig. 4. In the process of predicting the target alarm work order, the feature vector model is called first, the classifier model is further called, and finally the work order type identification model outputs the type (type) of the target alarm work order. The core module of the classifier model comprises a sampling module (acquiring the content of the work order), a word segmentation module (see the introduction in the above-mentioned "one, 2 and (1)") and a classification module (finally determining the type recognition result and further determining the category of the work order).
In summary, the scheme provided by the embodiment of the invention is as follows:
1. the machine learning method is used for replacing the traditional manual work to classify the alarm work order, so that the classification efficiency is improved.
2. The method is suitable for the unbalanced data set, and solves the technical difficulty of classifying the unbalanced data set by the short text.
Based on the same inventive concept, the embodiment of the invention also provides an equipment alarm work order identification device, as described in the following embodiments. Because the principle of solving the problems of the equipment alarm work order identification device is similar to the equipment alarm work order identification method, the implementation of the equipment alarm work order identification device can refer to the implementation of the equipment alarm work order identification method, and repeated parts are not repeated. As used hereinafter, the term "unit" or "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
Fig. 7 is a schematic structural diagram of an apparatus alarm work order identification device in the implementation of the present invention, and as shown in fig. 7, the device includes:
the acquiring unit 01 is used for acquiring the alarm work order content of the equipment to be identified in a preset scene;
the identification unit 02 is used for inputting the content of the alarm work order of the equipment to be identified in the preset scene into the work order type identification model to obtain the type of the alarm work order of the equipment to be identified; the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types in a preset scene, and the sub-training sets with balanced types are obtained by processing sample data of a historical unbalanced type alarm work order.
In an embodiment, the device alarm work order recognition apparatus may further include: the training unit is used for generating the work order type recognition model through pre-training according to the following method:
acquiring sample data of a historical unbalanced type alarm work order; the historical unbalanced type alarm work order sample data comprises alarm work order content and a corresponding type thereof; the historical unbalance type alarm work order sample data comprises a plurality of types of unbalance alarm work order sample data;
processing the alarm work order sample data with the plurality of types of unbalance to determine the number of each type of alarm work order sample;
copying the alarm work order type sample with the least number of samples into a preset number of samples; sampling and dividing the sample data of each other type of alarm work order into the preset number of copies, wherein each copy number is the same as the alarm work order type with the minimum sample number;
forming a plurality of types of balanced sub-training sets according to the following method: combining one part of the alarm work order type sample with the least number of samples with one part of the alarm work order sample data of each other type to form a sub-training set, and forming a plurality of sub-training sets with balanced types by analogy;
training each sub-training set to generate a base recognition model;
and generating the work order type identification model according to the plurality of base identification models.
In one example, training each of the sub-training sets to generate the base recognition model may include: training each sub-training set to generate an XGboost recognition model;
generating the work order type identification model according to a plurality of base identification models, comprising: and generating the work order type identification model according to the plurality of XGboost identification models.
In one example, the identification unit may be specifically configured to:
converting the alarm work order content of the equipment to be identified in the preset scene into a feature vector (namely, the feature vector model in fig. 4 can be used for realizing the conversion);
and obtaining the type of the alarm work order of the equipment to be identified according to the feature vector (namely, the type can be realized by using the classifier model in FIG. 4).
The embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the method when executing the computer program.
The embodiment of the invention also provides a computer readable storage medium, and the computer readable storage medium stores a computer program for executing the method.
The equipment alarm work order identification method and the equipment alarm work order identification device provided by the embodiment of the invention have the following beneficial technical effects: the accuracy and efficiency of equipment warning work order identification are improved.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes may be made to the embodiment of the present invention by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. An equipment alarm work order identification method is characterized by comprising the following steps:
acquiring alarm work order content of equipment to be identified in a preset scene;
inputting the content of the alarm work order of the equipment to be identified in a preset scene into a work order type identification model to obtain the type of the alarm work order of the equipment to be identified; the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types in a preset scene, and the sub-training sets with balanced types are obtained by processing sample data of a historical unbalanced type alarm work order.
2. The equipment alarm work order recognition method of claim 1, wherein the work order type recognition model is generated by pre-training according to the following method:
acquiring sample data of a historical unbalanced type alarm work order; the historical unbalanced type alarm work order sample data comprises alarm work order content and a corresponding type thereof; the historical unbalance type alarm work order sample data comprises a plurality of types of unbalance alarm work order sample data;
processing the alarm work order sample data with the plurality of types of unbalance to determine the number of each type of alarm work order sample;
copying the alarm work order type sample with the least number of samples into a preset number of samples; sampling and dividing the sample data of each other type of alarm work order into the preset number of parts, wherein each number of parts is the same as the number of samples of the alarm work order type with the minimum number of samples;
forming a plurality of types of balanced sub-training sets according to the following method: combining one part of the alarm work order type sample with the least number of samples with one part of the alarm work order sample data of each other type to form a sub-training set, and forming a plurality of sub-training sets with balanced types by analogy;
training each sub-training set to generate a base recognition model;
and generating the work order type identification model according to the plurality of base identification models.
3. The method of claim 2, wherein training each sub-training set to generate a base recognition model comprises: training each sub-training set to generate an XGboost recognition model;
generating the work order type identification model according to a plurality of base identification models, comprising: and generating the work order type identification model according to the plurality of XGboost identification models.
4. The method of claim 2, wherein training each of the sub-training sets to generate a base recognition model comprises:
performing word segmentation on sample data of each alarm work order of each sub-training set to obtain a first word set;
removing stop words from the words in the first word set to obtain a second word set;
removing low-frequency words from the words in the second word set to obtain a third word set;
constructing feature vectors for words in the third set of words;
and training according to the feature vectors and the corresponding work order types to generate a base recognition model.
5. The method for identifying the equipment alarm work order according to claim 1, wherein the step of inputting the content of the alarm work order of the equipment to be identified in the preset scene into the work order type identification model to obtain the type of the alarm work order of the equipment to be identified comprises the following steps:
converting the alarm work order content of the equipment to be identified in a preset scene into a feature vector;
and obtaining the type of the alarm work order of the equipment to be identified according to the characteristic vector.
6. An equipment alarm work order recognition device, comprising:
the acquiring unit is used for acquiring the alarm work order content of the equipment to be identified in a preset scene;
the identification unit is used for inputting the content of the alarm work order of the equipment to be identified in a preset scene into the work order type identification model to obtain the type of the alarm work order of the equipment to be identified; the work order type recognition model is generated by pre-training according to a plurality of sub-training sets with balanced types in a preset scene, and the sub-training sets with balanced types are obtained by processing sample data of a historical unbalanced type alarm work order.
7. The device alarm work order identification apparatus of claim 6, further comprising: the training unit is used for generating the work order type recognition model through pre-training according to the following method:
acquiring sample data of a historical unbalanced type alarm work order; the historical unbalanced type alarm work order sample data comprises alarm work order content and a corresponding type thereof; the historical unbalance type alarm work order sample data comprises a plurality of types of unbalance alarm work order sample data;
processing the alarm work order sample data with the plurality of types of unbalance to determine the number of each type of alarm work order sample;
copying the alarm work order type sample with the least number of samples into a preset number of samples; sampling and dividing the sample data of each other type of alarm work order into the preset number of parts, wherein each number of parts is the same as the number of samples of the alarm work order type with the minimum number of samples;
forming a plurality of types of balanced sub-training sets according to the following method: combining one part of the alarm work order type sample with the least number of samples with one part of the alarm work order sample data of each other type to form a sub-training set, and forming a plurality of sub-training sets with balanced types by analogy;
training each sub-training set to generate a base recognition model;
and generating the work order type identification model according to the plurality of base identification models.
8. The device alarm work order identification apparatus of claim 6, wherein the identification unit is specifically configured to:
converting the alarm work order content of the equipment to be identified in a preset scene into a feature vector;
and obtaining the type of the alarm work order of the equipment to be identified according to the characteristic vector.
9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 5 when executing the computer program.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program for executing the method of any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910847257.3A CN110704616B (en) | 2019-09-09 | 2019-09-09 | Equipment alarm work order identification method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910847257.3A CN110704616B (en) | 2019-09-09 | 2019-09-09 | Equipment alarm work order identification method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110704616A true CN110704616A (en) | 2020-01-17 |
| CN110704616B CN110704616B (en) | 2023-05-12 |
Family
ID=69195219
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910847257.3A Active CN110704616B (en) | 2019-09-09 | 2019-09-09 | Equipment alarm work order identification method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110704616B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112258235A (en) * | 2020-10-28 | 2021-01-22 | 国家电网有限公司客户服务中心 | Method and system for discovering new service of electric power marketing audit |
| CN112699944A (en) * | 2020-12-31 | 2021-04-23 | 中国银联股份有限公司 | Order-returning processing model training method, processing method, device, equipment and medium |
| CN113360346A (en) * | 2021-06-22 | 2021-09-07 | 北京百度网讯科技有限公司 | Method and apparatus for training a model |
| CN113497729A (en) * | 2020-04-03 | 2021-10-12 | 中移动信息技术有限公司 | Training method of alarm recognition model, alarm recognition method and device |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050114081A1 (en) * | 2003-09-22 | 2005-05-26 | Omron Corporation | Inspection method, inspection apparatus, and facility diagnosis unit |
| US20090245094A1 (en) * | 2008-03-26 | 2009-10-01 | Verizon Business Network Services Inc. | Outage analysis system |
| CN104809226A (en) * | 2015-05-07 | 2015-07-29 | 武汉大学 | Method for early classifying imbalance multi-variable time sequence data |
| JP2015225434A (en) * | 2014-05-27 | 2015-12-14 | 三菱電機株式会社 | Room entry/exit management system and room entry/exit management method |
| CN108897798A (en) * | 2018-06-12 | 2018-11-27 | 广东电网有限责任公司 | Method and device for classifying customer service work orders and electronic equipment |
| CN109491914A (en) * | 2018-11-09 | 2019-03-19 | 大连海事大学 | Defect report prediction technique is influenced based on uneven learning strategy height |
| CN109492091A (en) * | 2018-09-28 | 2019-03-19 | 科大国创软件股份有限公司 | A kind of complaint work order intelligent method for classifying based on convolutional neural networks |
| CN109635292A (en) * | 2018-12-05 | 2019-04-16 | 杭州东方通信软件技术有限公司 | Work order quality detecting method and device based on machine learning algorithm |
| CN109905269A (en) * | 2018-01-17 | 2019-06-18 | 华为技术有限公司 | The method and apparatus for determining network failure |
| CN110096410A (en) * | 2019-03-15 | 2019-08-06 | 中国平安人寿保险股份有限公司 | Alarm information processing method, system, computer installation and readable storage medium storing program for executing |
-
2019
- 2019-09-09 CN CN201910847257.3A patent/CN110704616B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050114081A1 (en) * | 2003-09-22 | 2005-05-26 | Omron Corporation | Inspection method, inspection apparatus, and facility diagnosis unit |
| US20090245094A1 (en) * | 2008-03-26 | 2009-10-01 | Verizon Business Network Services Inc. | Outage analysis system |
| JP2015225434A (en) * | 2014-05-27 | 2015-12-14 | 三菱電機株式会社 | Room entry/exit management system and room entry/exit management method |
| CN104809226A (en) * | 2015-05-07 | 2015-07-29 | 武汉大学 | Method for early classifying imbalance multi-variable time sequence data |
| CN109905269A (en) * | 2018-01-17 | 2019-06-18 | 华为技术有限公司 | The method and apparatus for determining network failure |
| CN108897798A (en) * | 2018-06-12 | 2018-11-27 | 广东电网有限责任公司 | Method and device for classifying customer service work orders and electronic equipment |
| CN109492091A (en) * | 2018-09-28 | 2019-03-19 | 科大国创软件股份有限公司 | A kind of complaint work order intelligent method for classifying based on convolutional neural networks |
| CN109491914A (en) * | 2018-11-09 | 2019-03-19 | 大连海事大学 | Defect report prediction technique is influenced based on uneven learning strategy height |
| CN109635292A (en) * | 2018-12-05 | 2019-04-16 | 杭州东方通信软件技术有限公司 | Work order quality detecting method and device based on machine learning algorithm |
| CN110096410A (en) * | 2019-03-15 | 2019-08-06 | 中国平安人寿保险股份有限公司 | Alarm information processing method, system, computer installation and readable storage medium storing program for executing |
Non-Patent Citations (2)
| Title |
|---|
| 任华 等: ""基于大数据技术的客服投诉智能分类与预警系统"", 《电信科学》 * |
| 柳加伟 等: ""告警关联规则挖掘的方法及系统"" * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113497729A (en) * | 2020-04-03 | 2021-10-12 | 中移动信息技术有限公司 | Training method of alarm recognition model, alarm recognition method and device |
| CN112258235A (en) * | 2020-10-28 | 2021-01-22 | 国家电网有限公司客户服务中心 | Method and system for discovering new service of electric power marketing audit |
| CN112699944A (en) * | 2020-12-31 | 2021-04-23 | 中国银联股份有限公司 | Order-returning processing model training method, processing method, device, equipment and medium |
| CN112699944B (en) * | 2020-12-31 | 2024-04-23 | 中国银联股份有限公司 | Training method, processing method, device, equipment and medium for returning list processing model |
| CN113360346A (en) * | 2021-06-22 | 2021-09-07 | 北京百度网讯科技有限公司 | Method and apparatus for training a model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110704616B (en) | 2023-05-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110609759B (en) | Fault root cause analysis method and device | |
| CN110704616B (en) | Equipment alarm work order identification method and device | |
| CN109191922B (en) | Large-scale four-dimensional track dynamic prediction method and device | |
| CN108416032A (en) | A kind of file classification method, device and storage medium | |
| CN111754345A (en) | Bit currency address classification method based on improved random forest | |
| CN109582963A (en) | A kind of archives automatic classification method based on extreme learning machine | |
| CN111143840B (en) | Method and system for identifying abnormity of host operation instruction | |
| CN115408525B (en) | Letters and interviews text classification method, device, equipment and medium based on multi-level label | |
| WO2024031930A1 (en) | Error log detection method and apparatus, and electronic device and storage medium | |
| KR102144010B1 (en) | Methods and apparatuses for processing data based on representation model for unbalanced data | |
| CN113704389A (en) | Data evaluation method and device, computer equipment and storage medium | |
| CN112883730A (en) | Similar text matching method and device, electronic equipment and storage medium | |
| CN115859191A (en) | Fault diagnosis method and device, computer readable storage medium and computer equipment | |
| CN117217277A (en) | Pre-training method, device, equipment, storage medium and product of language model | |
| CN115659244A (en) | Fault prediction method, device and storage medium | |
| CN109543038B (en) | Emotion analysis method applied to text data | |
| CN108536838A (en) | Very big unrelated multivariate logistic regression model based on Spark is to text sentiment classification method | |
| CN113628043A (en) | Complaint validity judgment method, device, equipment and medium based on data classification | |
| CN117574262A (en) | Underwater sound signal classification method, system and medium for small sample problem | |
| CN112579781A (en) | Text classification method and device, electronic equipment and medium | |
| CN116226747A (en) | Training method of data classification model, data classification method and electronic equipment | |
| CN113824580A (en) | Network index early warning method and system | |
| Singh et al. | Multiclass imbalanced big data classification utilizing spark cluster | |
| CN115470854A (en) | Information system fault classification method and classification system | |
| CN114881162A (en) | Method, apparatus, device and medium for predicting failure of metering automation master station |
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 | ||
| CP02 | Change in the address of a patent holder | ||
| CP02 | Change in the address of a patent holder |
Address after: Room 702-2, No. 4811, Cao'an Highway, Jiading District, Shanghai Patentee after: CHINA UNITECHS Address before: 100872 5th floor, Renmin culture building, 59 Zhongguancun Street, Haidian District, Beijing Patentee before: CHINA UNITECHS |