CN112069317A - Method for acquiring assembly time and processor - Google Patents
Method for acquiring assembly time and processor Download PDFInfo
- Publication number
- CN112069317A CN112069317A CN202010927939.8A CN202010927939A CN112069317A CN 112069317 A CN112069317 A CN 112069317A CN 202010927939 A CN202010927939 A CN 202010927939A CN 112069317 A CN112069317 A CN 112069317A
- Authority
- CN
- China
- Prior art keywords
- target
- text
- neural network
- target process
- words
- 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.)
- Pending
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/279—Recognition of textual entities
- G06F40/289—Phrasal analysis, e.g. finite state techniques or chunking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0637—Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
-
- 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
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- 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
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention provides an acquisition method of assembly time and a processor, comprising the following steps: processing the process text in the target process document by using a text mining mode, and determining the target process type of the target process text; acquiring a target neural network model corresponding to the target process type according to the target process type of the target process text; and inputting the man-hour influence factors in the target process text into the corresponding target neural network model, and acquiring the assembly man-hour output by the target neural network model and corresponding to the target process text. The embodiment of the invention carries out process-level process classification on the process documents in a text mining mode, constructs a process-level working hour prediction model by obtaining the neural network model corresponding to the process type, and obtains the working hours corresponding to the process text and the total working hours corresponding to the whole process documents based on the working hours, thereby being beneficial to improving the accuracy and efficiency of working hour prediction and further improving the estimation accuracy and management efficiency of enterprises on the working hour quota.
Description
Technical Field
The invention relates to the technical field of labor hour prediction, in particular to a method for acquiring assembly labor hour and a processor.
Background
The man-hour quota refers to the time it takes to complete a product according to a specified workflow in a manufacturing enterprise. The labor hour quota is the production guidance of enterprise planning and scheduling and personnel scheduling, is the measurement of the standard of management and examination and the production efficiency, and is an important basis for controlling the labor cost and the product quotation. In the development and production process of complex products such as satellites, the importance of the assembly process is increasingly prominent, wherein the working hours in the satellite assembly stage are about 40% of the product development cycle, and the assembly cost exceeds 50% of the complete production cost. At present, satellite manufacturing enterprises mainly rely on manual experience to carry out man-hour quota estimation, and the efficiency is low, the subjective influence is large, and the accuracy is insufficient; meanwhile, the informatization degree of the man-hour management is low, and the requirements of refining, digitalization and intellectualization of the current production management are difficult to meet. Therefore, it is necessary to develop a method for rapidly estimating the assembly man-hour of complex products such as satellites, and to design and develop a man-hour rating and management system.
Most current man-hour prediction models are mainly used for large-batch and few-variety part machining processes and part assembling processes, process information is relatively simple, and man-hour influence factors are easy to extract. For the labor hour quota prediction of the assembly of complex products such as satellites, the assembly has the problems of complex assembly relation, various assembly process information, large data volume, complex influence factors and the like, so that the labor hour quota estimation is difficult.
Disclosure of Invention
The technical purpose to be achieved by the embodiments of the present invention is to provide an acquisition method of assembly man-hour and a processor, so as to solve the problem that the estimation of the assembly man-hour of complex products such as current satellites is difficult.
In order to solve the above technical problem, an embodiment of the present invention provides an obtaining method of assembly man-hours, including:
processing the process text in the target process document by using a text mining mode, and determining a target process type of the target process text, wherein the target process text is any process text;
acquiring a target neural network model corresponding to the target process type from a pre-constructed neural network library according to the target process type of the target process text;
and inputting the working hour influence factors corresponding to the target procedure types in the target procedure texts into the corresponding target neural network models, and acquiring the assembly working hours corresponding to the target procedure texts and output by the target neural network models.
Specifically, the step of processing the process text in the target process document by using the text mining method and determining the target process type of the target process text by the above-mentioned obtaining method includes:
sequentially performing text preprocessing on the process texts in the obtained target process document to obtain a word segmentation word bank containing word segmentation results of the process texts;
and determining a target process type corresponding to the target process text according to the word segmentation word bank and a pre-constructed classification word bag model.
Preferably, the step of determining the target process type corresponding to the target process text according to the word segmentation lexicon and the pre-constructed classification bag-of-words model includes:
performing feature word extraction processing on the word segmentation word bank to obtain feature words in the word segmentation word bank and feature values corresponding to the feature words;
selecting a preset number of feature words from the target process text according to the feature values to form feature vectors, and matching the feature vectors with a classification word bag library to obtain a matching result;
and determining the target process type of the process text according to the matching result.
Specifically, before the step of determining the target process type corresponding to the target process text according to the segmented word bank and the pre-constructed classification word bag model, the obtaining method further includes:
acquiring a plurality of first test process documents in advance;
performing text preprocessing on the first test process document to obtain a test word bank corresponding to the first test process document;
extracting characteristic words from the test word bank to obtain characteristic words of the test procedure text in the test word bank and characteristic values corresponding to the characteristic words;
and determining a procedure type corresponding to the testing procedure, and obtaining a classification word bag model according to the feature words and the feature values in the testing procedure corresponding to the procedure type.
Further, in the above obtaining method, the text preprocessing step includes:
the method comprises the steps of dividing words by natural language and combining a preset professional word bank to decompose procedure text content into independent words, and removing stop words in the procedure text content according to a preset stop word bank.
Further, the step of obtaining the classification bag-of-words model according to the feature words and the feature values in the test process corresponding to the process type by the above-mentioned obtaining method includes:
and selecting the feature words with the feature values larger than a preset threshold value or a preset proportion according to the feature values of the feature words corresponding to the process types to form a classification word bag model.
Specifically, in the above-described acquiring method, before the step of acquiring the target neural network model corresponding to the target process type, the processing method further includes:
obtaining at least one neural network training model according to the working hour influence factors of the target process type;
acquiring a plurality of second test texts corresponding to the target process types, extracting working hour influence factors from the second test texts, and inputting the working hour influence factors into a neural network training model for training to obtain a training result of the neural network training model;
and determining a target neural network model corresponding to the target process type according to the training result, and constructing a neural network library.
Specifically, the obtaining method, according to the man-hour influence factor of the target process type, obtains at least one neural network training model, and includes:
determining the number of nodes of an input layer of a neural network training model according to the man-hour influence factors;
determining at least one hidden layer node number according to at least one preset hidden layer node calculation formula, the input layer node number and the output layer node number;
and obtaining at least one neural network training model according to the number of nodes of the input layer, the number of nodes of the output layer, the number of nodes of the hidden layer, a preselected activation function, initial parameters and at least one preset learning method.
Specifically, in the above-described acquisition method, the process types include:
main structure assembly, single structure assembly, other structure assembly and assembly preparation process.
Further, the acquisition method as described above,
the working hour influence factors for assembling the corresponding working procedure type as the main mechanism comprise: the method comprises the following steps of (1) precision requirement, the number of fasteners, the volume of an assembly object, a connection mode and whether debugging detection is carried out or not;
the working hour influence factors corresponding to the process type of single-machine structure assembly comprise: the number of fasteners, the number of assembly parts, the precision requirement and the volume of an assembly object;
the working hour influence factors corresponding to the process type of single-machine structure assembly comprise: using a tool, a sealing mode and clamping and positioning;
the man-hour influencing factors corresponding to the process type for the assembly preparation process include: type of action, tooling used, weight, and volume.
Another preferred embodiment of the present invention also provides a processor, including:
the first processing module is used for processing the procedure texts in the target process documents by utilizing a text mining mode and determining the target procedure types of the target procedure texts, wherein the target procedure texts are any procedure texts;
the second processing module is used for acquiring a target neural network model corresponding to the target process type from a pre-constructed neural network library according to the target process type of the target process text;
and the third processing module is used for inputting the working hour influence factors corresponding to the target process types in the target process texts into the corresponding target neural network models and acquiring the assembly working hours corresponding to the target process texts and output by the target neural network models.
Specifically, as the processor described above, the first processing module includes:
the first sub-processing module is used for sequentially carrying out text preprocessing on the process texts in the obtained target process document to obtain a word segmentation word bank containing word segmentation results of the process texts;
and the second sub-processing module is used for determining the target process type corresponding to the target process text according to the word segmentation word bank and the pre-constructed classification word bag model.
Preferably, as the processor described above, the second sub-processing module includes:
the first processing unit is used for extracting and processing the characteristic words from the participle word bank to obtain the characteristic words in the participle word bank and the characteristic values corresponding to the characteristic words;
the second processing unit is used for selecting a preset number of feature words from the target process text according to the feature values to form feature vectors, and matching the feature vectors with the classification word bag library to obtain a matching result;
and the third processing unit is used for determining the target process type of the process text according to the matching result.
Specifically, the processor as described above further includes:
the fourth processing module is used for acquiring a plurality of first test process documents in advance;
the fifth processing module is used for performing text preprocessing on the first test process document to obtain a test word bank corresponding to the first test process document;
the sixth processing module is used for extracting and processing the characteristic words of the test word bank to obtain the characteristic words of the test procedure text in the test word bank and the characteristic values corresponding to the characteristic words;
and the seventh processing module is used for determining the process type corresponding to the testing process and obtaining the classification word bag model according to the characteristic words and the characteristic values in the testing process corresponding to the process type.
Further, as for the processor, the first sub-processing module and the fifth processing module are further configured to divide the process text content into individual words by natural language word segmentation and combining with a preset professional lexicon, and reject stop words in the process text content according to the preset stop lexicon.
Further, as described above, the seventh processing module includes:
and the third sub-processing module is used for selecting the feature words with the feature values larger than a preset threshold or a preset proportion according to the feature values of the feature words corresponding to the process types to form a classification word bag model.
Specifically, the processor as described above further includes:
the eighth processing module is used for obtaining at least one neural network training model according to the working hour influence factors of the target process type;
the ninth processing module is used for acquiring a plurality of second test texts corresponding to the target process types, extracting the working hour influence factors from the second test texts, inputting the working hour influence factors into the neural network training model for training, and obtaining the training result of the neural network training model;
and the tenth processing module is used for determining a target neural network model corresponding to the target process type according to the training result and constructing a neural network library.
Specifically, as described above, the eighth processing module includes:
the fourth sub-processing module is used for determining the number of nodes of an input layer of the neural network training model according to the man-hour influence factors;
the fifth sub-processing module is used for determining at least one hidden layer node number according to at least one preset hidden layer node calculation formula, the input layer node number and the output layer node number;
and the sixth sub-processing module is used for obtaining at least one neural network training model according to the number of nodes of the input layer, the number of nodes of the output layer, the number of nodes of the hidden layer, a preselected activation function, initial parameters and at least one preset learning method.
The present invention also provides a readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the method for acquiring assembly man-hours as described above.
Compared with the prior art, the method for acquiring the assembly time and the processor provided by the embodiment of the invention at least have the following beneficial effects:
the embodiment of the invention carries out process-level process classification on the process documents in a text mining mode to obtain the process type of each process text, constructs the process-level working hour prediction model by obtaining the neural network model corresponding to the process type, and inputs the influence factors in the process text into the neural network model to obtain the working hours corresponding to the process text and the total working hours corresponding to the whole process documents, thereby being beneficial to improving the accuracy and efficiency of working hour prediction and further improving the estimation accuracy and management efficiency of enterprises on the working hour quota.
Drawings
FIG. 1 is a schematic flow chart of a method for acquiring assembly man-hours according to the present invention;
FIG. 2 is a second flowchart of the method for acquiring assembly hours according to the present invention;
FIG. 3 is a third schematic flow chart of the method for acquiring assembly time according to the present invention;
FIG. 4 is a fourth flowchart illustrating the method for acquiring assembly man-hours according to the present invention;
FIG. 5 is a fifth flowchart illustrating a method for acquiring assembly man-hours according to the present invention;
FIG. 6 is a sixth flowchart illustrating a method for acquiring assembly man-hours according to the present invention;
FIG. 7 is a block diagram of a processor according to the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments. In the following description, specific details such as specific configurations and components are provided only to help the full understanding of the embodiments of the present invention. Thus, it will be apparent to those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the invention. In addition, descriptions of well-known functions and constructions are omitted for clarity and conciseness.
It should be appreciated that reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
In various embodiments of the present invention, it should be understood that the sequence numbers of the following processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
It should be understood that the term "and/or" herein is merely one type of association relationship that describes an associated object, meaning that three relationships may exist, e.g., a and/or B may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the character "/" herein generally indicates that the former and latter related objects are in an "or" relationship.
In the embodiments provided herein, it should be understood that "B corresponding to a" means that B is associated with a from which B can be determined. It should also be understood that determining B from a does not mean determining B from a alone, but may be determined from a and/or other information.
The assembly process of complex products such as satellites belongs to typical discrete assembly, has the characteristics of multiple varieties and less batches, is single-piece production in most cases, has certain difference with the assembly of common large-batch mechanical products in the assembly process, and mainly shows that the data in the assembly process card is mainly unstructured Chinese text data; detailed records of the assembly process are recorded in the text data, and the detailed records comprise an assembly object, an assembly action, an assembly requirement, an assembly tool, an assembly step sequence and other complete assembly processes; and because the complex products such as the satellite have special structures and complex assembling objects, the complex products relate to plate products, rod products, beam products, structural components and various single-machine products; the assembly is various in action and steps, can be assembled by means of various tool equipment and materials in an auxiliary manner, and is complex to operate; meanwhile, in the assembly process, because the assembly requirement of the satellite product is strict and the assembly acceptance needs to be finished based on the previous process on the premise of carrying out part of the assembly process, various types of work can be alternately treated and assembled, such as painters, benches, heat treatment, quality detection and the like; making the assembly process complicated. Meanwhile, the text records in the process card are different from the text vocabularies used in daily life, so that the process card has certain specialty, the information extraction is difficult, and the common and traditional text data extraction method cannot be completely applicable. In addition, a large number of repeated assembly actions are provided in the assembly process, for example, assembly of instrument boards, the process basically comprises pre-fixing the nth layer of m-quadrant instrument boards, precisely measuring the nth layer of m-quadrant instrument boards, and debugging and fixing the nth layer of m-quadrant instrument boards (n is 1-4, and m is 1-4), and the repeated assembly actions have certain similarity in the embodiment of text information. Therefore, the existing man-hour prediction model cannot be directly adopted for prediction, so that the assembly man-hour prediction of a complex product is difficult, and the following technical scheme is provided for the application.
Referring to fig. 1, a preferred embodiment of the present invention provides an assembly man-hour obtaining method, including:
step S101, processing a procedure text in a target process document by using a text mining mode, and determining a target procedure type of the target procedure text, wherein the target procedure text is any procedure text;
step S102, according to the target process type of the target process text, acquiring a target neural network model corresponding to the target process type from a pre-constructed neural network library;
and step S103, inputting the working hour influence factors corresponding to the target process types in the target process texts into corresponding target neural network models, and acquiring the assembly working hours corresponding to the target process texts and output by the target neural network models.
In the embodiment of the invention, the process text of the target process text document is processed in a text mining mode, so that complex text data is simplified, the subsequent process-based processing is facilitated, and the identification difficulty caused by the text data with higher integrity, complexity, specialty and the like is avoided; meanwhile, the process types of the texts of each process are respectively determined, the assembly processes are classified at the process level, and complex process documents are simplified, so that the processing difficulty is reduced, and a good idea is provided for extracting process information. And respectively acquiring corresponding target neural network models according to the target process types of each target process text, inputting the working hour influence factors corresponding to the target process types in the target process texts into the corresponding target neural network models after the target neural network models are acquired, and calculating by the target neural network models according to the input to obtain output values, wherein the output values are the assembly working hours corresponding to the target process texts, namely the time required by the assembly according to the target process texts. The working procedure-level working hour prediction model can be established corresponding to the working procedure text, so that the prediction error is favorably reduced, and the estimation accuracy of enterprises on the working hour quota is improved; meanwhile, the working hour prediction is carried out through the neural network model, the neural network model can be optimized through forms such as training in advance, and the working hour prediction accuracy is further improved; and a neural network library is constructed in advance, and a corresponding neural network model can be directly called from the neural network library when needed, so that the time for obtaining the neural network is saved, and the working hour prediction efficiency and the working hour quota management efficiency of an enterprise are improved.
To sum up, the embodiment of the application performs process-level process classification on the process documents in a text mining mode to obtain the process type of each process text, constructs a process-level labor hour prediction model by obtaining the neural network model corresponding to the process type, and inputs the influencing factors in the process text into the neural network model to obtain the labor hour corresponding to the process text and the total labor hour corresponding to the whole process document, so that the accuracy and the efficiency of labor hour prediction are improved, and the estimation accuracy and the management efficiency of enterprises on labor hour quota are improved.
Referring to fig. 2, in particular, the step of processing the process text in the target process document by using a text mining method and determining the target process type of the target process text in the obtaining method includes:
step S201, sequentially performing text preprocessing on the process texts in the obtained target process documents to obtain a word segmentation word bank containing word segmentation results of the process texts;
and step S202, determining a target process type corresponding to the target process text according to the word segmentation word bank and a pre-constructed classification word bag model.
In a specific embodiment of the invention, when determining a target process type corresponding to a target process text, firstly, performing text preprocessing operation on each process text in a target process document in sequence to perform word segmentation on the process text to obtain a word segmentation result corresponding to each process text, and forming a word segmentation word bank by the word segmentation results corresponding to all the process texts, wherein the process texts and the word segmentation structures in the word segmentation word bank are still in one-to-one correspondence, and only text data in the original process text is simplified to reduce the extraction difficulty of main information; and then, matching the obtained word segmentation word library with a pre-constructed classification word bag library to obtain a matching result, wherein in the matching result, the process type of the classification word bag library model matched with the target process text is the target process type of the target process text, and the target process text is any one of the process texts, so that convenient process classification is realized, and the classification efficiency and the classification accuracy are effectively improved.
Referring to fig. 3, preferably, the step of determining the target process type corresponding to the target process text according to the segmented word bank and the pre-constructed classification bag-of-words model in the above-mentioned obtaining method includes:
step S301, performing feature word extraction processing on the participle word bank to obtain feature words in the participle word bank and feature values corresponding to the feature words;
step S302, selecting a preset number of feature words from the target process text according to the feature values to form feature vectors, and matching the feature vectors with a classification word bag library to obtain a matching result;
and step S303, determining the target process type of the process text according to the matching result.
In an embodiment of the present invention, the classifying includes: feature word extraction processing is performed on a word segmentation word bank corresponding to the whole procedure Document, that is, feature words corresponding to any procedure type in the word segmentation word bank are extracted and still corresponding to the procedure text are stored, and meanwhile, a feature value corresponding to the feature word in each procedure text is obtained to represent the importance degree of the feature word, wherein the feature value can be the word Frequency (TF) of the feature word, the Inverse Document Frequency (IDF) or the word Frequency-Inverse Document Frequency (TF-IDF), and in the embodiment, the word Frequency-Inverse Document Frequency is preferably used as the feature value; and then the characteristic words can be sequenced in the target process text through the characteristic values, a preset number of characteristic words are selected from the characteristic values of the characteristic words as the target words to be matched, namely, the preset number of characteristic words are selected from the characteristic values of the characteristic words as representatives and are marked as the target words, and then the target words are matched with the classification word bag library model, so that the workload during matching can be reduced, and the efficiency is improved, wherein the preset number can be a specific numerical value preset by a technician or a preset proportion. It should be noted that when the number of the feature words in the target process text is lower than a preset numerical value, all the feature words are the target words.
Preferably, when matching the target word with the classification bag library, a bayesian algorithm (Multinomial) is adopted to ensure the accuracy of the classification structure
Bayes), then, the target words form characteristic vectors according to the classification word bag library which is matched currently, matching is convenient, matching results are obtained, and according to the matching results of each classification word library, the process type corresponding to the classification word library with the highest matching degree can be determined as the target process type of the target process text.
Referring to fig. 4, specifically, before the step of determining the target process type corresponding to the target process text according to the segmented word bank and the pre-constructed classification bag-of-words model, the obtaining method further includes:
step S401, a plurality of first test process documents are obtained in advance;
step S402, performing text preprocessing on the first test process document to obtain a test word bank corresponding to the first test process document;
step S403, extracting characteristic words from the test word bank to obtain characteristic words of the test procedure text in the test word bank and characteristic values corresponding to the characteristic words;
and S404, determining a process type corresponding to the testing process, and obtaining a classification word bag model according to the feature words and the feature values in the testing process corresponding to the process type.
In a specific embodiment of the present invention, a classification bag-of-words model is pre-constructed, specifically, a plurality of first test process documents are pre-obtained, text preprocessing is performed on each first test document to obtain a corresponding test word bank, feature word extraction processing is further performed to obtain feature words and feature values of each test procedure text, and a procedure type of each test procedure is determined by manual selection or an original procedure classification method. Test procedures corresponding to the same procedure type form a corpus, wherein the corpus can be displayed in a word cloud (word cloud) form for visual observation; and constructing a test vector corresponding to the test procedure and a text feature matrix corresponding to the corpus according to the feature words and the feature values to obtain the classification word bag library providing the creep type, wherein the text feature matrix can be displayed in a form of a feature heat map for facilitating observation.
Further, in the above obtaining method, the text preprocessing step includes:
the method comprises the steps of dividing words by natural language and combining a preset professional word bank to decompose procedure text content into independent words, and removing stop words in the procedure text content according to a preset stop word bank.
In a specific embodiment of the invention, text information serving as a data source is processed to a certain extent through text preprocessing, and the text information is characterized into an intermediate data form beneficial to processing by a text mining tool, wherein process word segmentation, namely segmentation of a section of text, is firstly carried out to obtain a process of individual words, and in the embodiment, a third-party library jieba tool capable of realizing Chinese word segmentation, custom dictionary addition and stop word removal is preferably used for word segmentation in a Python language environment; secondly, because the specialty of the process document is high, a self-defined professional lexicon needs to be added for word segmentation again after the process word segmentation, which is beneficial to improving the word segmentation efficiency and accuracy, so that the professional lexicon needs to be pre-constructed according to an assembly object, an assembly process and the like before word segmentation, the professional words are prevented from being segmented or adhered to other words, the professional lexicon needs to be updated after the technology is updated, wherein the entries of the professional lexicon at least comprise: vocabulary, weight and part of speech; thirdly, because a large number of punctuation marks, mood auxiliary words and connecting words exist in the process document, the occurrence frequency of the words is high, but the words are not actually associated with the main information of the text and belong to redundant information words, such as 'this', 'that', 'in', and the like, a stop word list needs to be constructed in advance, the extracted words are removed according to the stop word list to obtain a final word segmentation result, and optionally the word segmentation result can be displayed by using a word cloud (word cloud), wherein the word number in the word cloud is positively associated with the word frequency, namely the word frequency is larger, and the word number of the word is larger.
Further, the step of obtaining the classification bag-of-words model according to the feature words and the feature values in the test process corresponding to the process type by the above-mentioned obtaining method includes:
and selecting the feature words with the feature values larger than a preset threshold value or a preset proportion according to the feature values of the feature words corresponding to the process types to form a classification word bag model.
In a specific embodiment of the present invention, after obtaining feature words and features that are tested persistently, because the number of test processes corresponding to the same process type is large, and the feature words of each test process are different, and some words appear only in some test processes, the total number of words is large through total statistics, in order to reduce the workload during subsequent matching, some feature words in the feature words need to be selected to form a classification bag-of-words model, specifically, when constructing the bag-of-words model, the feature words are sorted in descending order, a preset threshold or a preset proportion is set, and the feature words that constitute the bag-of-words model are determined according to the preset threshold or the preset proportion.
Referring to fig. 5, in particular, the acquiring method as described above, before the step of acquiring the target neural network model corresponding to the target process type, the acquiring method further includes:
step S501, obtaining at least one neural network training model according to the working hour influence factors of the target process type;
step S502, a plurality of second test texts corresponding to the target process types are obtained, the man-hour influence factors are extracted from the second test texts, and the second test texts are input into a neural network training model for training to obtain the training result of the neural network training model;
and S503, determining a target neural network model corresponding to the target process type according to the training result, and constructing a neural network library.
In another preferred embodiment of the present invention, before the step of obtaining the target neural network model corresponding to the target process type, a neural network library is constructed, which includes the neural network model corresponding to each process type, so as to call the corresponding target neural network for the target process type. When a network library is constructed, firstly, determining at least one neural network training model according to the working hour influence factors of the target process type, further obtaining a plurality of second test texts corresponding to the target process type, training the neural network training model according to the second test texts to obtain a training result, specifically, extracting the working hour influence factors from the second test texts, and inputting the working hour influence factors into the neural network training model; and selecting an optimal neural network training model as a target neural network model according to the training result, and constructing the neural network library according to the neural network model corresponding to each process type. Wherein, the above mentioned optimization includes but is not limited to the minimum relative error or the minimum time, etc.
Referring to fig. 6, in particular, the acquiring method as described above, the step of obtaining at least one neural network training model according to the man-hour influence factor of the target process type includes:
step S601, determining the number of nodes of an input layer of a neural network training model according to the man-hour influence factors;
step S602, determining at least one hidden layer node number according to at least one preset hidden layer node calculation formula, the input layer node number and the output layer node number;
step S603, obtaining at least one neural network training model according to the number of nodes of the input layer, the number of nodes of the output layer, the number of nodes of the hidden layer, a preselected activation function, initial parameters and at least one preset learning method.
In a specific embodiment of the present invention, taking a three-layer neural network model including an input layer, a hidden layer, and an output layer as an example, the step of obtaining the neural network training model includes: determining the number of input layer nodes of a neural network training model according to the public man-hour influence factors, wherein when the public man-hour influence factors have different states for the same factor, each state corresponds to one input layer node, for example, high, medium and low precision respectively correspond to one input layer node; since the purpose of the present invention is to obtain man-hours, the number of output layer nodes can be determined to be 1; and determining at least one hidden layer node number according to the input layer node number, the output layer node number and at least one preset hidden layer node calculation formula, wherein the preset hidden layer node calculation formula at least comprises:
an empirical formula based on a least square method is as follows:
II, Kolmogorov theorem:
h＝2m+1
and thirdly, the method is suitable for training of large sample size:
wherein h is the number of hidden layer nodes, m is the number of input layer nodes, n is the number of output layer nodes, p is the number of samples, and int is the rounding function.
The number of nodes of the three hidden layers can be obtained through the three calculation formulas, and then at least one neural network training model is obtained according to the number of nodes of the input layer, the number of nodes of the output layer, the number of nodes of the hidden layers, a preselected activation function, initial parameters and at least one preset learning method, wherein the activation function of the output layer uses a linear function, and the activation function of the hidden layers uses a relu function, wherein the relu function has the advantages of high convergence speed, good adaptability to regression analysis problems, capability of overcoming gradient disappearance problems and the like, and the text model has good adaptability after the relu function is trained in a later period, wherein the relu function is relu (x) MAX (x, 0). Initial parameters a random function in the tensoflow library was used to generate 1 random number seed, a random number weight with a standard deviation of 1, and a bias matrix.
In the embodiment, a secondary cost function is adopted as a loss function, and a standard degree gradient descent GD algorithm, a dynamic gradient descent algorithm, a RMSProp algorithm and an Adam algorithm are respectively selected as preset learning methods, wherein the learning rate of the standard degree gradient descent GD algorithm is preferably 0.001; the learning rate of the dynamic gradient descent algorithm is 0.01, and the dynamic parameter is 0.9; the learning rate of the Adam algorithm is 0.001. In conclusion, a plurality of neural network training models can be obtained according to different preset hidden layer node calculation formulas and different learning algorithms, and then the optimal neural network training model is determined to be the target neural network model through training analysis, so that the efficiency, accuracy and the like of the obtained target neural network model can be ensured, and the reliability of the method for obtaining the assembly time is improved.
In a preferred embodiment of the present invention, the main structural assembly is used as the process type, wherein the man-hour influencing factors include: the number of fasteners, the precision (low, medium and high), the connection mode (temporary non-screwing, gluing fixation and screwing), the assembly object, the adjustment test assembly, and 9 influence factors in total, so that the number of nodes of the input layer is 9, the number of nodes of the output layer is 1, the number of nodes of the hidden layer obtained by calculation by using the three preset hidden layer node calculation formulas is respectively 6, 9 and 38 (taking the sample number as an example 372), and the training result is obtained by training 2000 times, as shown in the following table 1:
TABLE 1 comparison of algorithms with number of hidden layer nodes
As can be seen from the data in table 1, for the main structure assembly, the relative error and the cost function of the Adam algorithm are both smaller than those of other algorithms, so that the algorithm is the optimal algorithm; however, as the number of nodes varies, the number of hidden layer nodes is 6, which is not as good as the number of hidden layer nodes 19 or 38, so the number of hidden layer nodes can be selected to be either "19" or "38".
Specifically, in the above-described acquisition method, the process types include:
main structure assembly, single structure assembly, other structure assembly and assembly preparation process.
In an embodiment of the present invention, the man-hour quota is influenced by various factors due to a large number of assembly components and large number of assembly components involved in the assembly process of a complex product.
Specifically, the man-hour influence factors extracted from the assembly object are as shown in table 1 below:
TABLE 2 influence factors on man-hours of assembly objects
In addition, the assembling action can be divided according to the assembling and matching process, the assembling working process, the assembling monitoring process and the coating and boxing process, so that the working time influence factors of the dynamic assembling process and the assembling action extraction are shown in the following table 2:
TABLE 3 man-hour influence factors of the Assembly operation
By combining the analysis of the characteristics of the assembly objects, the assembly processes and the assembly actions and the summary of the man-hour influence factors, it can be known that the man-hour influence factors of different assembly processes are greatly different, and therefore, in order to improve the neural network prediction result, the process types of the assembly process are preferably divided into: main structure assembly, single structure assembly, other structure assembly and assembly preparation process; the main structure assembly objects are instrument plates, horizontal plates, beams, vertical plates, partition plates, pull rod assemblies, process assemblies and the like; the assembly objects in the single-machine structure assembly are SADA, an engine, an antenna mounting table and the like; assembling objects in the assembly of other structures are honeycomb core splicing, embedded parts, substrate installation and the like; the assembly preparation process refers to cleaning of a cabin field and the like, hoisting and carrying of an assembly body, surface treatment and the like.
Further, the acquisition method as described above,
the working hour influence factors for assembling the corresponding working procedure type as the main mechanism comprise: the method comprises the following steps of (1) precision requirement, the number of fasteners, the volume of an assembly object, a connection mode and whether debugging detection is carried out or not;
the working hour influence factors corresponding to the process type of single-machine structure assembly comprise: the number of fasteners, the number of assembly parts, the precision requirement and the volume of an assembly object;
the working hour influence factors corresponding to the process type of single-machine structure assembly comprise: using a tool, a sealing mode and clamping and positioning;
the man-hour influencing factors corresponding to the process type for the assembly preparation process include: type of action, tooling used, weight, and volume.
Furthermore, another embodiment of the present invention specifically discloses the man-hour influence factors corresponding to different process types.
Alternatively, the above-mentioned man-hour influence factors are only preferred embodiments for easy understanding of those skilled in the art, and it is within the scope of the invention that those skilled in the art should add or delete them in the face of practical circumstances.
Referring to fig. 7, another preferred embodiment of the present invention also provides a processor including:
the first processing module 701 is configured to process a procedure text in a target process document in a text mining manner, and determine a target procedure type of the target procedure text, where the target procedure text is any procedure text;
a second processing module 702, configured to obtain, according to a target process type of the target process text, a target neural network model corresponding to the target process type from a pre-constructed neural network library;
the third processing module 703 is configured to input the labor hour influence factor corresponding to the target procedure type in the target procedure text into the corresponding target neural network model, and obtain the assembly labor hour corresponding to the target procedure text and output by the target neural network model.
Specifically, as described above, the first processing module 701 includes:
the first sub-processing module is used for sequentially carrying out text preprocessing on the process texts in the obtained target process document to obtain a word segmentation word bank containing word segmentation results of the process texts;
and the second sub-processing module is used for determining the target process type corresponding to the target process text according to the word segmentation word bank and the pre-constructed classification word bag model.
Preferably, as the processor described above, the second sub-processing module includes:
the first processing unit is used for extracting and processing the characteristic words from the participle word bank to obtain the characteristic words in the participle word bank and the characteristic values corresponding to the characteristic words;
the second processing unit is used for selecting a preset number of feature words from the target process text according to the feature values to form feature vectors, and matching the feature vectors with the classification word bag library to obtain a matching result;
and the third processing unit is used for determining the target process type of the process text according to the matching result.
Specifically, the processor as described above further includes:
the fourth processing module is used for acquiring a plurality of first test process documents in advance;
the fifth processing module is used for performing text preprocessing on the first test process document to obtain a test word bank corresponding to the first test process document;
the sixth processing module is used for extracting and processing the characteristic words of the test word bank to obtain the characteristic words of the test procedure text in the test word bank and the characteristic values corresponding to the characteristic words;
and the seventh processing module is used for determining the process type corresponding to the testing process and obtaining the classification word bag model according to the characteristic words and the characteristic values in the testing process corresponding to the process type.
Further, as for the processor, the first sub-processing module and the fifth processing module are further configured to divide the process text content into individual words by natural language word segmentation and combining with a preset professional lexicon, and reject stop words in the process text content according to the preset stop lexicon.
Further, as described above, the seventh processing module includes:
and the third sub-processing module is used for selecting the feature words with the feature values larger than a preset threshold or a preset proportion according to the feature values of the feature words corresponding to the process types to form a classification word bag model.
Specifically, the processor as described above further includes:
the eighth processing module is used for obtaining at least one neural network training model according to the working hour influence factors of the target process type;
the ninth processing module is used for acquiring a plurality of second test texts corresponding to the target process types, extracting the working hour influence factors from the second test texts, inputting the working hour influence factors into the neural network training model for training, and obtaining the training result of the neural network training model;
and the tenth processing module is used for determining a target neural network model corresponding to the target process type according to the training result and constructing a neural network library.
Specifically, as described above, the eighth processing module includes:
the fourth sub-processing module is used for determining the number of nodes of an input layer of the neural network training model according to the man-hour influence factors;
the fifth sub-processing module is used for determining at least one hidden layer node number according to at least one preset hidden layer node calculation formula, the input layer node number and the output layer node number;
and the sixth sub-processing module is used for obtaining at least one neural network training model according to the number of nodes of the input layer, the number of nodes of the output layer, the number of nodes of the hidden layer, a preselected activation function, initial parameters and at least one preset learning method.
The embodiment of the processor of the present invention is a processor corresponding to the embodiment of the acquisition method, and all implementation means in the embodiment of the acquisition method are applicable to the embodiment of the processor, so that the same technical effects can be achieved.
The present invention also provides a readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the method for acquiring assembly man-hours as described above.
Furthermore, the present invention may repeat reference numerals and/or letters in the various examples. This repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various embodiments and/or configurations discussed.
It is further noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims (12)
1. A method for acquiring assembly man-hours is characterized by comprising the following steps:
processing a process text in a target process document by using a text mining mode, and determining a target process type of the target process text, wherein the target process text is any one of the process texts;
acquiring a target neural network model corresponding to the target process type from a pre-constructed neural network library according to the target process type of the target process text;
and inputting the working hour influence factors corresponding to the target procedure types in the target procedure texts into corresponding target neural network models, and acquiring the assembly working hours corresponding to the target procedure texts and output by the target neural network models.
2. The method according to claim 1, wherein the step of processing the process text in the target process document by text mining and determining the target process type of the target process text comprises:
sequentially performing text preprocessing on the process texts in the obtained target process documents to obtain a word segmentation word bank containing word segmentation results of the process texts;
and determining a target process type corresponding to the target process text according to the word segmentation word bank and a pre-constructed classification word bag model.
3. The obtaining method according to claim 2, wherein the step of determining the target process type corresponding to the target process text according to the segmented word bank and a pre-constructed classification word bag model comprises:
extracting characteristic words from the word segmentation word bank to obtain characteristic words in the word segmentation word bank and characteristic values corresponding to the characteristic words;
selecting a preset number of feature words from the target process text according to the feature values to form feature vectors, and matching the feature vectors with the classification word bag library to obtain a matching result;
and determining the target process type of the process text according to the matching result.
4. The obtaining method according to claim 2, wherein before the step of determining the target process type corresponding to the target process text according to the segmented word bank and a pre-constructed classification bag-of-words model, the obtaining method further comprises:
acquiring a plurality of first test process documents in advance;
performing text preprocessing on the first test process document to obtain a test word bank corresponding to the first test process document;
extracting characteristic words from the test word bank to obtain characteristic words of a test procedure text in the test word bank and characteristic values corresponding to the characteristic words;
determining a procedure type corresponding to the testing procedure, and obtaining the classification bag-of-words model according to the feature words and the feature values in the testing procedure corresponding to the procedure type.
5. The acquisition method according to claim 2 or 4, wherein the text preprocessing step comprises:
the method comprises the steps of dividing words by natural language and combining a preset professional word bank to decompose procedure text content into independent words, and eliminating stop words in the procedure text content according to a preset stop word bank.
6. The method according to claim 4, wherein the step of obtaining the classification bag-of-words model according to the feature words and the feature values in the test procedure corresponding to the procedure type includes:
and selecting the feature words with the feature values larger than a preset threshold value or a preset proportion according to the feature values of the feature words corresponding to the process types to form the classification word bag model.
7. The acquisition method according to claim 1, wherein, prior to the step of acquiring the target neural network model corresponding to the target process type, the processing method further comprises:
obtaining at least one neural network training model according to the working hour influence factors of the target process type;
acquiring a plurality of second test texts corresponding to the target process types, extracting the working hour influence factors from the second test texts, and inputting the working hour influence factors into the neural network training model for training to obtain a training result of the neural network training model;
and determining a target neural network model corresponding to the target process type according to the training result, and constructing the neural network library.
8. The method of claim 7, wherein the step of deriving at least one neural network training model based on the man-hour influence factors for a target process type comprises:
determining the number of nodes of an input layer of the neural network training model according to the man-hour influence factors;
determining at least one hidden layer node number according to at least one preset hidden layer node calculation formula and the input layer node number and the output layer node number;
and obtaining at least one neural network training model according to the number of the nodes of the input layer, the number of the nodes of the output layer, the number of the nodes of the hidden layer, a preselected activation function, initial parameters and at least one preset learning method.
9. The acquisition method according to claim 1, wherein the process types include:
main structure assembly, single structure assembly, other structure assembly and assembly preparation process.
10. The acquisition method according to claim 9,
the man-hour influence factors for assembling the main mechanism corresponding to the process type include: the method comprises the following steps of (1) precision requirement, the number of fasteners, the volume of an assembly object, a connection mode and whether debugging detection is carried out or not;
the man-hour influence factors corresponding to the process type of the single-machine structure assembly comprise: the number of fasteners, the number of assembly parts, the precision requirement and the volume of an assembly object;
the man-hour influence factors corresponding to the process type of the single-machine structure assembly comprise: using a tool, a sealing mode and clamping and positioning;
the man-hour affecting factors corresponding to the process type for the assembly preparation process include: type of action, tooling used, weight, and volume.
11. A processor, comprising:
the first processing module is used for processing the procedure texts in the target process documents by utilizing a text mining mode and determining the target procedure types of the target procedure texts, wherein the target procedure texts are any one of the procedure texts;
the second processing module is used for acquiring a target neural network model corresponding to the target process type from a pre-constructed neural network library according to the target process type of the target process text;
and the third processing module is used for inputting the working hour influence factors corresponding to the target procedure types in the target procedure texts into corresponding target neural network models and acquiring the assembly working hours corresponding to the target procedure texts and output by the target neural network models.
12. A readable storage medium, characterized in that the readable storage medium stores thereon a computer program which, when executed by a processor, realizes the steps of the method for acquiring the assembly man-hour according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010927939.8A CN112069317A (en) | 2020-09-07 | 2020-09-07 | Method for acquiring assembly time and processor |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010927939.8A CN112069317A (en) | 2020-09-07 | 2020-09-07 | Method for acquiring assembly time and processor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112069317A true CN112069317A (en) | 2020-12-11 |
Family
ID=73663733
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010927939.8A Pending CN112069317A (en) | 2020-09-07 | 2020-09-07 | Method for acquiring assembly time and processor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112069317A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112819205A (en) * | 2021-01-18 | 2021-05-18 | 北京理工大学 | Method, device and system for predicting working hours |
| CN112949907A (en) * | 2021-02-04 | 2021-06-11 | 杭州品茗安控信息技术股份有限公司 | Quota matching method, device, equipment and storage medium for engineering cost |
| CN113191624A (en) * | 2021-04-28 | 2021-07-30 | 上海核工程研究设计院有限公司 | Workload prediction method for nuclear power design research and development |
| CN113570175A (en) * | 2021-09-27 | 2021-10-29 | 深圳市玄羽科技有限公司 | Method for establishing industrial refined working hour and quantity management platform |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107908716A (en) * | 2017-11-10 | 2018-04-13 | 国网山东省电力公司电力科学研究院 | 95598 work order text mining method and apparatus of word-based vector model |
| CN108733003A (en) * | 2017-04-20 | 2018-11-02 | 南京理工大学 | Slewing parts process working hour prediction technique based on kmeans clustering algorithms and system |
| CN111489032A (en) * | 2020-04-09 | 2020-08-04 | 北京理工大学 | Processing method and device for predicting assembly time of aerospace product |
| WO2020164267A1 (en) * | 2019-02-13 | 2020-08-20 | 平安科技（深圳）有限公司 | Text classification model construction method and apparatus, and terminal and storage medium |
-
2020
- 2020-09-07 CN CN202010927939.8A patent/CN112069317A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108733003A (en) * | 2017-04-20 | 2018-11-02 | 南京理工大学 | Slewing parts process working hour prediction technique based on kmeans clustering algorithms and system |
| CN107908716A (en) * | 2017-11-10 | 2018-04-13 | 国网山东省电力公司电力科学研究院 | 95598 work order text mining method and apparatus of word-based vector model |
| WO2020164267A1 (en) * | 2019-02-13 | 2020-08-20 | 平安科技（深圳）有限公司 | Text classification model construction method and apparatus, and terminal and storage medium |
| CN111489032A (en) * | 2020-04-09 | 2020-08-04 | 北京理工大学 | Processing method and device for predicting assembly time of aerospace product |
Non-Patent Citations (1)
| Title |
|---|
| 钟宏才等: "应用神经网络实现中间产品加工工时定额的快速计算", 《华东船舶工业学院学报(自然科学版)》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112819205A (en) * | 2021-01-18 | 2021-05-18 | 北京理工大学 | Method, device and system for predicting working hours |
| CN112949907A (en) * | 2021-02-04 | 2021-06-11 | 杭州品茗安控信息技术股份有限公司 | Quota matching method, device, equipment and storage medium for engineering cost |
| CN112949907B (en) * | 2021-02-04 | 2024-03-19 | 品茗科技股份有限公司 | Quota matching method, device and equipment for engineering cost and storage medium |
| CN113191624A (en) * | 2021-04-28 | 2021-07-30 | 上海核工程研究设计院有限公司 | Workload prediction method for nuclear power design research and development |
| CN113570175A (en) * | 2021-09-27 | 2021-10-29 | 深圳市玄羽科技有限公司 | Method for establishing industrial refined working hour and quantity management platform |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112069317A (en) | Method for acquiring assembly time and processor | |
| CN106407333B (en) | Spoken language query identification method and device based on artificial intelligence | |
| US10657498B2 (en) | Automated resume screening | |
| CN112184525B (en) | System and method for realizing intelligent matching recommendation through natural semantic analysis | |
| US10089581B2 (en) | Data driven classification and data quality checking system | |
| CN111489032B (en) | Processing method and device for predicting assembly time of aerospace product | |
| CN109165284A (en) | A kind of financial field human-computer dialogue intension recognizing method based on big data | |
| CN105989040A (en) | Intelligent question-answer method, device and system | |
| CN108519971B (en) | Cross-language news topic similarity comparison method based on parallel corpus | |
| US20220245356A1 (en) | Automatic discovery of business-specific terminology | |
| JP7128919B2 (en) | Skill term evaluation method and device, electronic device, computer readable medium | |
| CN113360647B (en) | 5G mobile service complaint source-tracing analysis method based on clustering | |
| CN114722833A (en) | Semantic classification method and device | |
| Tang et al. | SALKG: a semantic annotation system for building a high-quality legal knowledge graph | |
| CN113592017A (en) | Deep learning model standardization training method, management system and processing terminal | |
| CN113011156A (en) | Quality inspection method, device and medium for audit text and electronic equipment | |
| US20230418894A1 (en) | Input method and apparatus based on sample-probability quantization, and electronic device | |
| CN114330103A (en) | Spacecraft anomaly relevance model training method based on-orbit anomaly information | |
| CN110851572A (en) | Session labeling method and device, storage medium and electronic equipment | |
| Aliyanto et al. | Supervised probabilistic latent semantic analysis (sPLSA) for estimating technology readiness level | |
| CN105894410A (en) | Curriculum group support evaluation method based on knowledge automation | |
| Amalia et al. | Prediction of the waiting time period for getting a job using the naive bayes algorithm | |
| CN111814457A (en) | Power grid engineering contract text generation method | |
| Cai et al. | Risk identification of civil aviation engine control system based on particle swarm optimization-mean impact value-support vector machine | |
| KONSTANTOPOULOS | AN INTRODUCTION TO |
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 |