CN114066669B - Cloud manufacturing-oriented manufacturing service discovery method - Google Patents
Cloud manufacturing-oriented manufacturing service discovery method Download PDFInfo
- Publication number
- CN114066669B CN114066669B CN202111264538.XA CN202111264538A CN114066669B CN 114066669 B CN114066669 B CN 114066669B CN 202111264538 A CN202111264538 A CN 202111264538A CN 114066669 B CN114066669 B CN 114066669B
- Authority
- CN
- China
- Prior art keywords
- manufacturing
- service
- resources
- topic
- document
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 299
- 238000000034 method Methods 0.000 title claims abstract description 45
- 238000004806 packaging method and process Methods 0.000 claims abstract description 13
- 238000005516 engineering process Methods 0.000 claims abstract description 4
- 238000009826 distribution Methods 0.000 claims description 78
- 230000006870 function Effects 0.000 claims description 17
- 238000005070 sampling Methods 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 12
- 238000012544 monitoring process Methods 0.000 claims description 11
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 6
- 230000009467 reduction Effects 0.000 claims description 6
- 238000007726 management method Methods 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 4
- 239000000463 material Substances 0.000 claims description 4
- 238000005065 mining Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000005265 energy consumption Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000009849 deactivation Effects 0.000 claims description 2
- 238000001914 filtration Methods 0.000 claims description 2
- 238000012549 training Methods 0.000 claims description 2
- 239000000047 product Substances 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000013461 design Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 238000004088 simulation Methods 0.000 description 4
- 230000006399 behavior Effects 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 230000015654 memory Effects 0.000 description 2
- 238000013439 planning Methods 0.000 description 2
- 239000000654 additive Substances 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000005242 forging Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000000873 masking effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012858 packaging process Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012827 research and development Methods 0.000 description 1
- 239000011265 semifinished product Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/268—Morphological analysis
-
- 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
- G06F40/284—Lexical analysis, e.g. tokenisation or collocates
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- 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 a cloud manufacturing-oriented manufacturing service discovery method, which comprises the following steps: (1) According to the characteristics and model requirements of manufacturing resources in the cloud manufacturing background, a formal description model and a service packaging method for manufacturing equipment resources are provided by combining an ontology modeling technology and an extended OWL-S service description language, and a corresponding cloud manufacturing service set is constructed according to the method, so that semantic support is provided for subsequent service discovery. (2) Aiming at the demand of cloud manufacturing platform manufacturing service sharing, semantic information of corresponding cloud manufacturing services is obtained by analyzing the extended OWL-S cloud manufacturing service document in the step (1), and a cloud manufacturing service discovery method based on a subject model is researched based on the semantic information, so that the matching demand of user demands and cloud manufacturing service platform manufacturing services is met, and the efficiency of cloud manufacturing platform manufacturing resource sharing is improved.
Description
Technical Field
The invention relates to the fields of cloud manufacturing service packaging, subject mining and service discovery, in particular to a manufacturing service discovery method oriented to cloud manufacturing.
Background
Cloud manufacturing is used as a new service-oriented networked manufacturing mode, and aims to integrate physical manufacturing resources distributed in different geographic positions, so as to integrate and share the manufacturing resources. In cloud manufacturing mode, service requesters can obtain resources meeting manufacturing requirements in a transparent manner by only calling services in a service system without knowing the specific implementation details of physical manufacturing resources, and furthermore, in actual production manufacturing processes, manufacturing resources are typically distributed at each service provider discretely, and different service providers have differences in descriptions of capabilities and attributes of the manufacturing resources, which to some extent results in information being heterogeneous. Meanwhile, according to the requirements of users, manufacturing services meeting the requirements are screened from a large number of cloud manufacturing services, and building a manufacturing service candidate set matched with the requirements is a key technology for realizing a cloud manufacturing platform. The expansion method of the OWL-S service description language is proposed by Qu C and the like (Qu C,Liu F,Tao M,et al.An OWL-S based specification model of dynamic entity services for Internet of Things[J].Journal of Ambient Intelligence and Humanized Computing,2016,7(1):73-82.), a service state component part is added into a traditional OWL-S top layer ontology structure to describe the state information of the service, and ontology modeling is carried out on the service state information, however, the expansion method does not consider the quality of service attribute (QoS) and cannot describe the quality information of manufacturing equipment resources. In addition, the existing service discovery method based on grammar and keyword matching has the problems of low service description capability, poor flexibility, low service discovery matching accuracy, low retrieval efficiency and the like.
Therefore, an effective method for packaging and discovering the service of manufacturing resources is needed, which weakens the direct connection between users and complex manufacturing resources, abstracts and extracts the manufacturing capability and related attributes of different manufacturing resources, packages the manufacturing capability and related attributes into cloud services available in networked manufacturing environment, and realizes unified management of shared resources so as to enable a service demander to acquire a cloud manufacturing service set capable of meeting the self requirements.
Disclosure of Invention
The invention aims to provide a cloud manufacturing-oriented manufacturing service discovery method, which aims to realize manufacturing service packaging and manufacturing service discovery framework under the cloud manufacturing background, further improve the efficiency of integrating and sharing manufacturing resources of a cloud manufacturing service platform, improve the efficiency of acquiring a cloud manufacturing service set meeting the self requirements of a service demander, weaken the direct connection between a user and complex manufacturing resources and realize the integration and sharing of the manufacturing resources of the cloud manufacturing platform.
The invention is realized at least by one of the following technical schemes.
A cloud manufacturing-oriented manufacturing service discovery method meets the matching requirements of user requirements and manufacturing services in a cloud manufacturing platform, and comprises the following steps:
(1) Classifying the manufacturing resources according to the existence form and the use mode of the manufacturing resources;
(2) Extracting relevant attributes of the manufacturing equipment resources according to characteristics of the manufacturing equipment resources in the cloud manufacturing background, performing formal description on the manufacturing equipment resources, and establishing a manufacturing equipment resource formal description model;
(3) According to the manufacturing equipment resource formalized description model established in the step (2), extending OWL-S service description language, and constructing a cloud manufacturing service packaging model to obtain a cloud manufacturing service document set;
(4) The service document constructed in the step (3) is analyzed, topic information of the service document is mined through a topic model and a sampling algorithm, and a probability distribution set of the service document-topic is established;
(5) And (3) designing a similarity calculation method of the user demands and each manufacturing service in the cloud manufacturing platform according to the service document-topic probability distribution obtained in the step (4), setting a corresponding threshold value, wherein the threshold value is used for judging whether the demands are met, if the service similarity of a certain user demand and a certain manufacturing service in the cloud manufacturing platform is greater than the threshold value, the user demands and the cloud manufacturing platform are considered to be matched, otherwise, the user demands and the cloud manufacturing platform are considered to be not matched.
Preferably, the manufacturing resources are divided into hard manufacturing resources, soft manufacturing resources and other resources according to the existence form and the use mode of the manufacturing resources; the hard manufacturing resources comprise various manufacturing equipment resources, computing equipment resources, monitoring resources and material resources which are used in the actual production process; soft manufacturing resources include software resources, technology resources, knowledge resources, and human resources; other resources refer to resources other than the hard and soft fabrication resource classifications described above.
Preferably, the relevant attributes of the manufacturing equipment resource include a base attribute, a status attribute, a functional attribute, and a quality of service attribute.
Preferably, the manufacturing equipment resource formalized description model formalizes through a multi-tuple form, abstracts and extracts common features of manufacturing equipment resources that are represented in a production manufacturing campaign, and masks specific details.
Preferably, the formalized description format is:
EquipmentInfo＝<Basic,Function,Quality,State>
wherein Basic represents Basic attributes for describing Basic information of manufacturing equipment resources; functions represent functional attributes that describe the processing capabilities of the manufacturing equipment resources; quality represents a Quality attribute describing the ability of a manufacturing facility resource to provide a service; state represents a State attribute that describes the condition of a manufacturing facility resource throughout a production manufacturing and service cycle.
Preferably, the expansion OWL-S service description language expands the original OWL-S service description language according to the inherent characteristics of the manufacturing equipment resource, and the expansion part comprises manufacturing equipment resource body information and manufacturing service quality attributes; the manufacturing equipment resource ontology information comprises basic attributes, state attributes and functional attributes of the manufacturing equipment resource; the manufacturing quality of service attribute includes service cost, service time, service energy consumption and service Reliability of the manufacturing equipment resource, wherein Reliability refers to the capability of the manufacturing equipment to complete the production task correctly in the monitoring time interval, and is derived from the historical statistics, and if the total number of times the manufacturing equipment performs the production task in the monitoring time period is N total and the number of times the manufacturing equipment successfully completes the task is N success, the historical statistics indicate that the manufacturing equipment performs the production task in the monitoring time period
Preferably, the document needs to be preprocessed before mining the subject of the service document, wherein the preprocessing comprises the operations of compound word segmentation, word stem reduction and deactivation.
Preferably, the topic model and sampling algorithm is a BTM model and a Gibbs sampling algorithm;
The BTM model is used for matching basic attribute information of manufacturing services, filtering manufacturing services which do not meet the conditions, constructing a similarity function ts based on JS divergence when performing theme matching, and calculating the theme similarity among the manufacturing services.
Preferably, the BTM model and Gibbs sampling algorithm comprises the steps of:
1) Modeling the preprocessed service description document dataset as an input to the BTM;
2) Sampling using Gibbs: randomly selecting an initial state of a Markov chain, randomly assigning word pairs, words under each topic, and then calculating a conditional probability P (z|z -b, B, alpha, beta) of each word pair b= (w i,wj), wherein
Where w i、wj denotes any two words in the pre-processed document, B denotes a word pair, i.e. b= (w i,wj),z-b denotes the topic assignment of all word pairs except word pair B, B denotes the global word pair set, n z denotes the number of times a word pair is assigned to topic z, n w|z denotes the number of times a word w is assigned to topic z,Representing the number of times a word w i、wj is assigned to the topic z, M representing the number of different words in the corpus, α, β representing the parameters of a given a priori dirichlet distribution;
notably, here, symmetrical dirichlet priors α and β are employed, i.e. one word pair b is assigned to the topic z, then two of the words w i and w j are also assigned to the topic at the same time; using word pair topic assignment counter and word pair co-occurrence to conveniently estimate topic-word distribution And the global topic distribution θ is:
Wherein, Representing the probability of word w in topic z, θ z representing the probability of topic z, |b| is the total number of word pairs, z represents an unknown underlying topic, and is also the topic to which each word pair corresponds, θ and/>The Dirichlet prior distribution with parameters alpha and beta is respectively arranged, and K represents the number of topics;
3) Representing each manufacturing service description text as an implicit topic distribution vector:
d＝[p(z1|d),p(z2|d),…,p(zt|d)]
Where p (z t |d) represents the probability that topic z t is assigned to document d;
to infer document topics, the BTM assumes that the document-topic distribution is equal to the product of the word pair distribution p (b|d) and the word pair-topic distribution p (z|b) in the document, calculated as:
Wherein, taking b= (w i,wj) as an intermediate quantity, the conditional probability distribution p (b|d) of word pairs in the document d is calculated:
Wherein n d (b) represents the number of times b occurs in document d;
based on the parameters of BTM estimation, the word pair-topic distribution p (z|b) is calculated by Bayes formula:
where p (z) represents the probability of subject z in θ, i.e., θ z;p(wi|z)、p(wj |z represents the probabilities of words w i and w j of b= (w i,wj) in z, respectively, i.e.
4) Calculating a service document-topic distribution p (z|d):
5) Taking JS divergence as a basis for measuring the similarity of the manufactured service document theme, calculating the probability distribution similarity of the theme, and calculating a formula of the similarity of the service document theme:
Wherein p and q respectively represent two different document-topic probability distributions, x i represents a topic number, p (x i) and q (x i) respectively represent probabilities corresponding to topic x i in the two different document-topic probability distributions, D KL (p||q) represents relative entropy of the document-topic probability distributions p and q, ts (R, S) represents topic similarity of the two service documents R, S, R p represents document-topic distribution of the service request document R, S q represents document-topic distribution of the service document S, JS (R p||Sq) represents JS divergence values of the two probability distributions, α is an adjustment factor, and the smaller JS (R p||Sq) is, the higher the similarity is;
preferably, the similarity calculation method is as follows:
Setting two different document-topic probability distributions p and q, and measuring the calculation formulas of the two distribution differences as shown in the formula (1) and the formula (2):
Wherein p and q represent two different document-topic probability distributions, x i represents a topic number, p (x i) and q (x i) represent probabilities corresponding to topic x i in the two different document-topic probability distributions, D KL (p||q) represents relative entropy of the document-topic probability distributions p and q, and JS (p||q) represents JS divergence values of the document-topic probability distributions p and q, respectively;
When the distribution of p is closer to q, i.e., the p distribution is more fit to q, the divergence value is smaller, i.e., the difference between the two is smaller, while formula (2) satisfies symmetry, which can be used to measure the similarity between probability distributions. And the manufacturing service which does not meet the threshold requirement can be directly filtered, so that the problem scale of matching is reduced, and the matching speed is improved.
Compared with the prior art, the invention has the following advantages:
1. the invention firstly classifies manufacturing resources, and provides a manufacturing service packaging method based on an extended OWL-S service description language, which shields the complexity and the isomerism of cloud manufacturing resources and uniformly describes the cloud manufacturing resources.
2. The invention provides a semantic-based manufacturing service discovery method, which can fully utilize semantic information of a manufacturing service document, has higher service discovery efficiency compared with the traditional grammar-based keyword discovery method, and can more meet the requirement of a large number of heterogeneous manufacturing service discovery under the cloud manufacturing background.
Drawings
FIG. 1 is a schematic diagram of the cloud manufacturing resource classification type of the present invention;
FIG. 2 is a schematic diagram of the main attributes of the manufacturing equipment resources according to the present invention;
FIG. 3 is a schematic diagram of an extended service profile module structure according to the present invention;
FIG. 4 is a schematic diagram of an extended QoS module structure according to the present invention;
FIG. 5 is a schematic diagram of the mapping relationship between the manufacturing equipment and the extended OWL-S;
FIG. 6 is a schematic diagram of a correspondence between a manufacturing device and an abstract device according to the present invention;
FIG. 7 is a schematic diagram of a semantic-based manufacturing service discovery method framework of the present invention.
Detailed Description
The invention will be described in further detail with reference to the following examples, but the embodiments of the invention are not limited to the examples, and the invention comprises the following steps:
1. Manufacturing resource classification:
The cloud manufacturing resource classification essentially merges cloud manufacturing resources with common attributes and characteristics, is a premise and a foundation for resource description and modeling, and provides support for finally realizing virtualization and service of different types of resources. The types of manufacturing resources shared in the cloud manufacturing service system mainly include manufacturing resources and manufacturing capabilities.
As shown in fig. 1. The manufacturing resource refers to a physical resource which exists physically and has a resource form of static transmission media, such as processing equipment, simulation software and the like. The manufacturing capability is an intangible and dynamic resource form, which refers to subjective conditions required by a manufacturing enterprise to complete a certain production goal, and is a capability which is displayed by combining actual manufacturing resources in a production and manufacturing activity, such as design development capability, experimental simulation capability, production and manufacturing capability, management decision capability and the like, wherein the manufacturing capability is tightly combined with the manufacturing resources, and is separated from actual manufacturing resource elements, so that the manufacturing capability is never reflected.
Manufacturing resources can be classified into hard manufacturing resources, soft manufacturing resources, and other resources according to the form in which they exist and the manner in which they are used.
The hard manufacturing resources mainly comprise various manufacturing equipment resources, computing equipment resources, monitoring resources and material resources used in the actual production process. The manufacturing equipment resources refer to various physical equipment such as production, processing, experiment, simulation, transportation and the like used in the whole production and manufacturing process of the product, such as a cutting machine tool, forging equipment, a mechanical arm, an automatic guiding transportation vehicle and the like. Computing device resources refer to computing hardware infrastructure such as various servers, memories, etc. that support operations of manufacturing enterprises and cloud manufacturing service systems. The monitoring resources are mainly used for monitoring other manufacturing resources, such as sensors, cameras, radio frequency identification Readers (RFID), adapters and the like. The material resources are the raw materials, semi-finished products, additives and the like required in the actual production and manufacturing activities.
Soft manufacturing resources include software resources, technical resources, knowledge resources, and human resources. The software resource refers to various system software and application software involved in the processes of product design, simulation, analysis, process planning, production and manufacturing, and the like, such as AutoCAD, solidWorks, office, visual Studio, and the like; the technical resources refer to resources such as design standards, process specifications, experience models, marketing schemes, product case libraries and the like accumulated in the whole life cycle process of the product; knowledge resources refer to a series of related knowledge such as market information knowledge, manufacturing process knowledge, copyright, invention patent and other resources throughout the production and manufacturing process of the product. Human resources are teams or professional technicians, such as workshop operators, design specialists, simulation engineers, project authorities and the like, who are involved in activities such as certain operations, research and development design, market research, marketing planning, technical application, project management, after-sales service and the like in the whole production and manufacturing process of products.
Other resources refer to resources other than the above-mentioned hard manufacturing resources and soft manufacturing resources, such as user basic information resources for recording resource providers and resource users, service resources for providing various information consultation, technical training, logistics, after-sales services for cloud service users, and business process service management resources for searching and matching to optimal services to execute manufacturing tasks according to requests submitted by users.
2. Formalized description is carried out on the manufacturing equipment resources, and a manufacturing equipment resource formalized description model is established;
under the cloud manufacturing background, manufacturing equipment resources have the characteristics of isomerism, diversity, dispersibility, dynamic property and the like, and if the cloud manufacturing equipment resources are directly subjected to unified modeling, the workload is large and complex. Formalized description methods have powerful description and analysis capabilities and are readily understood by abstracting and extracting common features exhibited by manufacturing equipment resources in a manufacturing campaign and masking the complex specifics thereof.
The main attributes of the manufacturing equipment resources are shown in fig. 2, and can be described in a formal manner by using four tuples, specifically expressed as follows:
EquipmentInfo＝<Basic,Function,Quality,State>
Wherein a Basic attribute (Basic) is used to describe Basic information of manufacturing equipment resources. The functional attribute (Function) is used to describe the processing capabilities of the manufacturing equipment resource. Quality attributes (Quality) are used to describe the ability of a manufacturing facility resource to provide a service. The State attribute (State) is used to describe the condition of the manufacturing equipment resource throughout the manufacturing and service cycle.
3. According to the established manufacturing equipment resource formalized description model, the OWL-S service description language is expanded;
OWL-S service description language generally consists of three components: a Service Profile module (Service Profile), a Service base module (Service Grounding), and a Service Model module (Service Model). The invention provides a cloud manufacturing service description language based on extended OWL-S, wherein the extended content is based on two parts, namely a service configuration file module and a service quality module, and a cloud manufacturing service packaging model is constructed on the basis, so that a cloud manufacturing service document set is finally obtained.
The structure of the extended service profile module is shown in fig. 3, and it can be seen that the extended service profile module mainly comprises four parts, namely a profile, device resource information, a device resource provider and service behavior. The configuration file describes the main structure of the service configuration file module and is used for describing the relation among the bodies in the service configuration file module; the service behavior is used for describing the behavior of the service profile module, and comprises input, output, preconditions, service parameters and operation results of the service; the device resource provider is used for describing relevant information of the service provider, including names, addresses, telephone numbers, mailboxes and the like; the device resource information is used to describe information about the manufacturing device, including basic properties, function information, quality properties, and status properties of the device.
The structure of the extended qos module is shown in fig. 4. The service quality module is used for describing service quality attributes of cloud manufacturing service, and the service quality body constructed by the method comprises four attributes of service cost, service time, service energy consumption and service reliability of manufacturing equipment resources, wherein the owned service quality relationship points to the service quality body from the service quality module, and the owned service quality relationship is shown between the service quality module and the service quality body;
4. the method comprises the steps of carrying out service packaging, constructing a cloud manufacturing service packaging model on the basis of expansion of OWL-S service description language, and obtaining a cloud manufacturing service document set;
In the production and manufacturing process, the manufacturing equipment is usually small-sized and single-function equipment, the packaging process of the equipment is simple, only the information of the manufacturing equipment is required to be described in a semantic manufacturing service description language, and then one manufacturing equipment is packaged into one service, for the manufacturing equipment, a one-to-one packaging model is adopted, the basic attribute, the functional attribute and the state attribute in the manufacturing equipment body are mapped into the equipment resource information body of the service configuration file, the quality attribute of the manufacturing equipment resource is mapped into the quality of service body of the quality of service module, and the mapping relationship is shown in figure 5. For large, complex and functionally diverse manufacturing equipment, if such equipment directly adopts a one-to-one packaging model, the obtained semantic manufacturing service is not flexible enough, which reduces the utilization rate of the manufacturing equipment. For such manufacturing equipment, using a one-to-many packaging model, the information of the manufacturing equipment resources is composed of two parts: a functional part and a nonfunctional part, the formal descriptions of which are shown in formula (1):
R＝(F,NF) (1)
Wherein R represents a manufacturing equipment resource, F represents a functional part, and a plurality of functions can be included, namely F= (F 1,f2,f3,…,fn), n (n is more than or equal to 2) represents the number of the functions of the manufacturing equipment, and F i represents one function in the functional part of the manufacturing equipment resource; NF then represents a non-functional part of the manufacturing equipment. The present invention will introduce the concept of abstract manufacturing facilities for describing single function manufacturing facilities that are separated from multi-function facilities. The manufacturing equipment information f i corresponds to an abstract equipment AR i, and the formal description of the abstract equipment AR i is shown in formula (2):
ARi＝(fi,NF) (2)
The mapping relation is shown in figure 6.
5. Manufacturing service discovery, namely mining service document theme information through a theme model and a sampling algorithm, and establishing a probability distribution set of a service document-theme;
And the manufacturing service discovery process performs semantic analysis on the manufacturing service request and the published manufacturing service in the cloud manufacturing service system to obtain semantic information of each service, and then performs matching in the cloud manufacturing service system according to the semantic information to return the manufacturing service meeting the threshold requirement. And combining semantic analysis, basic attribute matching and functional attribute parameter matching based on the topic model, and finally realizing the target of cloud manufacturing service discovery. The specific process is as follows:
(1) Extracting relevant information such as a service configuration file module from an extended OWL-S manufacturing service description document;
(2) And extracting relevant information of equipment resources in the service configuration file module, wherein the relevant information comprises basic attribute information and functional attribute parameter information, and constructing a theme model of the manufacturing service document based on the basic attribute information.
(3) The BTM model is adopted to match the basic attribute information of the manufacturing service, and the manufacturing service which does not meet the condition is filtered. When subject matching is performed, a similarity function ts is constructed based on JS divergence, and subject similarity between manufacturing services is calculated.
(4) And in addition, when the quality of service attribute parameters are matched, the service with the attribute value larger than the attribute value corresponding to the task requirement is directly filtered, so that the operation speed can be improved.
The integral frame is shown in fig. 7. And a data preprocessing stage for preprocessing data of the manufacturing service description document semantically described by using the extended OWL-S, and extracting key characteristic information in the manufacturing service to form a service text, such as basic attribute information, functional attribute information, cited ontology concepts and the like of the manufacturing service. And then, carrying out operation on the service text by means of compound word segmentation, word stem reduction, stop word removal and the like to obtain final service text description, storing the preprocessed data in a text form, wherein one service corresponds to one text description. The detailed process is as follows:
(1) In order to obtain key information of a service document as accurately and effectively as possible, remove irrelevant information at the same time, reduce processing load, the chapter extracts key semantic information such as basic attribute information, functional attribute information, service quality information and the like in an extended OWL-S manufacturing service description document, and particularly can complete process description information, which is a main component of the manufacturing service description document.
(2) And splitting the compound words in the extracted text.
(3) And removing the stop words and symbols by using the stop word list, so as to avoid the influence of the stop words on modeling.
(4) Part of speech reduction is carried out, words with the same stem have the same meaning, and for the convenience of word matching, the part of speech reduction is carried out by utilizing NLTK packages in a Python library.
(5) And removing the stop words by using the stop word list again, and preventing the words after part-of-speech reduction from being stop words.
In the topic model construction stage, a preprocessed service description document dataset is used as an input of a BTM to model, an implicit topic is learned, because some implicit variables and parameters exist in the BTM, an inference method can be divided into accurate inference and approximate inference, and the accurate inference is difficult to use, so that the parameters are generally inferred by using an approximate inference method, in the existing approximate inference method, gibbs sampling is widely applied, the basic idea is that parameters are alternately estimated, namely, the value of one variable is replaced by the value of the variable, the value of the variable is determined by the distribution of the values of the variable, and the inference method such as variational inference (variational inference) and maximum posterior estimation (maximum posterior estimation) of other potential variable models has two advantages compared with the reasoning method of the other potential variable models. First, it is more accurate because it progressively approaches the correct distribution. Second, it is more memory efficient because it only requires maintenance of counters and state variables, which is more appropriate for large-scale data sets. For Gibbs sampling, the initial state of the markov chain is chosen randomly, i.e. word pairs, words are randomly assigned under each topic, and then the conditional probability P (z|z -b, B, α, β) of each word pair b= (w i,wj) is calculated, where z -b represents the topic assignment of all word pairs except word pair B, and B represents the global word pair set. Wherein the method comprises the steps of
Where w i、wj denotes any two words in the pre-processed document, b denotes a word pair, i.e. b= (w i,wj),nz denotes the number of times a word pair is assigned to the topic z, n w|z denotes the number of times a word w is assigned to the topic z, Representing the number of times a word w i、wj is assigned to the topic z, M representing the number of different words in the corpus, α, β representing the parameters of a given a priori dirichlet distribution, respectively. Notably, here, symmetrical dirichlet priors α and β are employed, i.e. one word pair b is assigned to the topic z, then two of the words w i and w j are also assigned to the topic at the same time. Finally, the topic-word distribution/>, can be conveniently estimated by using the word pair topic distribution counter and the word pair co-occurrenceAnd the global topic distribution θ is:
Wherein, Representing the probability of word w in topic z, θ z representing the probability of topic z, |b| is the total number of word pairs, z represents an unknown underlying topic, and is also the topic to which each word pair corresponds, θ and/>There is a dirichlet prior distribution with parameters α and β, respectively, K representing the number of topics.
One major difference between BTMs and traditional topic models is that BTMs do not model the document generation process directly, but rather model word-to-word corpora in the document. Therefore, in the topic learning process, the service document-topic distribution p (z|d) cannot be directly acquired, and an inference calculation needs to be performed on the parameter, so that each manufacturing service description text is represented as an implicit topic distribution vector, as shown in a formula (6).
d＝[p(z1|d),p(z2|d),…,p(zt|d)] (6)
Where p (z t |d) represents the probability that document d contains the topic z t.
To infer document topics, the BTM assumes that the document-topic distribution is equal to the product of the word pair distribution p (b|d) and the word pair-topic distribution p (z|b) in the document, calculated as:
Wherein, taking b= (w i,wj) as an intermediate quantity, a conditional probability distribution p (b|d) of word pairs in the document d can be calculated, as shown in equation (8), where n d (b) represents the number of times b occurs in the document d.
Based on the parameters of the BTM estimation, the word pair-topic distribution p (z|b) can be calculated by Bayes formula:
where p (z) represents the probability of subject z in θ, i.e., θ z;p(wi|z)、p(wj |z represents the probabilities of words w i and w j of b= (w i,wj) in z, respectively, i.e.
On this basis, a service document-topic distribution p (z|d) can be further calculated, which can be expressed as:
For the calculation of the topic probability distribution similarity, JS divergence is used as a basis for measuring the topic similarity of the manufacturing service document, and the calculation formulas of the topic similarity of the service document are shown in formulas (11), (12) and (13).
Wherein p and q represent two different document-topic probability distributions, x i represents a topic number, p (x i) and q (x i) represent probabilities corresponding to topic x i in the two different document-topic probability distributions, D KL (p||q) represents relative entropy of the document-topic probability distributions p and q, JS (p|q) represents JS divergence values of the document-topic probability distributions p and q, ts (R, S) represents topic similarity of two service documents, R p represents document-topic distribution of the service request document R, S q represents document-topic distribution of the service document S, JS (R p||Sq) represents JS divergence values of the two probability distributions, α is an adjustment factor, and the smaller JS (R p||Sq) is, the higher the similarity is.
For the function parameter similarity calculation, let a ni denote one piece of function attribute parameter information in a certain manufacturing service request, and B nj denote one piece of function attribute parameter information in a certain published manufacturing service B. When the functional parameter information of a ni and B nj is represented by the numerical intervals a n and B n, the similarity calculation formula of a n and B n is shown in formula (14):
Where || denotes the length of the number of intervals, such as |0, 10|=10.
The preferred embodiments of the invention disclosed above are intended only to assist in the explanation of the invention. The preferred embodiments are not exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and the practical application, to thereby enable others skilled in the art to best understand and utilize the invention. The invention is limited only by the claims and the full scope and equivalents thereof.
Claims (7)
1. The cloud manufacturing-oriented manufacturing service discovery method is characterized by meeting the matching requirements of user requirements and manufacturing services in a cloud manufacturing platform, and comprises the following steps:
(1) Classifying the manufacturing resources according to the existence form and the use mode of the manufacturing resources;
(2) Extracting relevant attributes of the manufacturing equipment resources according to characteristics of the manufacturing equipment resources in the cloud manufacturing background, performing formal description on the manufacturing equipment resources, and establishing a manufacturing equipment resource formal description model;
(3) According to the manufacturing equipment resource formalized description model established in the step (2), extending OWL-S service description language, and constructing a cloud manufacturing service packaging model to obtain a cloud manufacturing service document set;
(4) The service document constructed in the step (3) is analyzed, topic information of the service document is mined through a topic model and a sampling algorithm, and a probability distribution set of the service document-topic is established;
The topic model and the sampling algorithm are a BTM model and a Gibbs sampling algorithm;
The BTM model is used for matching basic attribute information of manufacturing services, filtering the manufacturing services which do not meet the conditions, constructing a similarity function ts based on JS divergence when performing theme matching, and calculating the theme similarity among the manufacturing services;
the BTM model and Gibbs sampling algorithm comprises the steps of:
1) Modeling the preprocessed service description document dataset as an input to the BTM;
2) Sampling using Gibbs: randomly selecting an initial state of a Markov chain, randomly assigning word pairs, words under each topic, and then calculating a conditional probability P (z|z -b, B, alpha, beta) of each word pair b= (w i,wj), wherein
Where w i、wj denotes any two words in the pre-processed document, B denotes a word pair, i.e. b= (w i,wj),z-b denotes the topic assignment of all word pairs except word pair B, B denotes the global word pair set, n z denotes the number of times a word pair is assigned to topic z, n w|z denotes the number of times a word w is assigned to topic z,Representing the number of times a word w i、wj is assigned to the topic z, M representing the number of different words in the corpus, α, β representing the parameters of a given a priori dirichlet distribution;
parameters alpha and beta of symmetric prior dirichlet distribution are adopted, namely a word pair b is allocated to a theme z, and two words w i and w j are also allocated to the theme at the same time; the word-to-topic distribution counter and the word-to-topic co-occurrence are utilized to conveniently estimate the topic-word distribution And the global topic distribution θ is:
Wherein, Representing the probability of the word w in the topic z, theta z representing the probability of the topic z in theta, b| being the total number of word pairs, K representing the number of topics;
3) Representing each manufacturing service description text as an implicit topic distribution vector:
d＝[p(z1|d),p(z2|d),...,p(zt|d)]
Where p (z t |d) represents the probability that topic z t is assigned to document d;
To infer document topics, the BTM assumes that the document-topic distribution is equal to the product of the word pair-document distribution p (b|d) and the word pair-topic distribution p (z|b) in the document, calculated as:
Wherein, taking b= (w i,wj) as an intermediate quantity, calculating a conditional probability distribution p (b|d) of word pairs b in the document d:
Wherein n d (b) represents the number of times b occurs in document d;
based on the parameters of BTM estimation, the word pair-topic distribution p (z|b) is calculated by Bayes formula:
Where p (z) represents the probability of subject z in θ, i.e., θ z;p(wi|z)、p(wj |z represents the probabilities of words w i and w j of b= (w i,wj) in subject z, respectively, i.e.
4) Calculating a topic distribution p (z|d) of the service document:
5) Taking JS divergence as a basis for measuring the similarity of the manufactured service document theme, calculating the probability distribution similarity of the theme, and calculating a formula of the similarity of the service document theme:
Wherein p and q represent the topic probability distributions of two different documents, respectively, x i represents the topic number, p (x i) and q (x i) represent the probabilities corresponding to topic x i in the topic probability distributions of two different documents, respectively, The relative entropy of the document-topic probability distribution p and q is represented, JS (p||q) represents the JS divergence value of the document-topic probability distribution p and q, when the distribution of p is closer to q, namely the p distribution is more fit to q, the divergence value is smaller, namely the difference between the two is smaller, the symmetry is satisfied, the similarity between probability distributions can be measured, and the similarity is higher as JS (R p||Sq) is smaller; ts (R p,Sq) represents the topic similarity of the two service documents R p、Sq, R p represents the document-topic distribution of the service request document R, S g represents the document-topic distribution of the service document S, JS (R p||Sq) represents the JS divergence value of the two service documents R p、Sq distribution, ω is an adjustment factor;
(5) And (3) designing a similarity calculation method of the user demands and each manufacturing service in the cloud manufacturing platform according to the service document-topic probability distribution obtained in the step (4), setting a corresponding threshold value, wherein the threshold value is used for judging whether the demands are met, if the service similarity of a certain user demand and a certain manufacturing service in the cloud manufacturing platform is greater than the threshold value, the user demands and the cloud manufacturing platform are considered to be matched, otherwise, the user demands and the cloud manufacturing platform are considered to be not matched.
2. The cloud manufacturing-oriented manufacturing service discovery method according to claim 1, wherein: manufacturing resources are divided into hard manufacturing resources, soft manufacturing resources and other resources according to the existence form and the use mode of the manufacturing resources; the hard manufacturing resources comprise various manufacturing equipment resources, computing equipment resources, monitoring resources and material resources which are used in the actual production process; soft manufacturing resources include software resources, technology resources, knowledge resources, and human resources; other resources refer to resources other than the above-mentioned hard manufacturing resources and soft manufacturing resources, including user basic information resources for recording resource providers and resource users, service resources for providing various information consultations, technical training, logistics and after-sales services for cloud service users, and business process service management resources for searching and matching to optimal services to execute manufacturing tasks according to requests submitted by users.
3. The cloud manufacturing-oriented manufacturing service discovery method according to claim 1, wherein: the relevant attributes of the manufacturing facility resources include basic attributes, status attributes, functional attributes, and quality of service attributes.
4. The cloud manufacturing-oriented manufacturing service discovery method according to claim 1, wherein: the formal description model of the manufacturing equipment resources performs formal description through a multi-element group form, so as to abstract and extract common characteristics of the manufacturing equipment resources in production and manufacturing activities and shield specific details.
5. The cloud manufacturing-oriented manufacturing service discovery method according to claim 1, wherein: the formalized description format is:
EquipmentInfo＝＜Basic，Function，Quality，State＞
wherein Basic represents Basic attributes for describing Basic information of manufacturing equipment resources; functions represent functional attributes that describe the processing capabilities of the manufacturing equipment resources; quality represents a Quality attribute describing the ability of a manufacturing facility resource to provide a service; state represents a State attribute that describes the condition of a manufacturing facility resource throughout a production manufacturing and service cycle.
6. The cloud manufacturing-oriented manufacturing service discovery method according to claim 1, wherein: the expansion OWL-S service description language expands the original OWL-S service description language according to the inherent characteristics of the manufacturing equipment resource, and the expansion part comprises manufacturing equipment resource body information and manufacturing service quality attributes; the manufacturing equipment resource ontology information comprises basic attributes, state attributes and functional attributes of the manufacturing equipment resource; the manufacturing quality of service attribute includes service cost, service time, service energy consumption and service Reliability of the manufacturing equipment resource, wherein Reliability refers to the capability of the manufacturing equipment to complete the production task correctly in the monitoring time interval, and is derived from the historical statistics, and if the total number of times the manufacturing equipment performs the production task in the monitoring time period is N total and the number of times the manufacturing equipment successfully completes the task is N success, the historical statistics indicate that the manufacturing equipment performs the production task in the monitoring time period
7. The cloud manufacturing-oriented manufacturing service discovery method according to claim 1, wherein: before mining the service document theme, the document needs to be preprocessed, wherein the preprocessing comprises the operations of compound word segmentation, word stem reduction and word deactivation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111264538.XA CN114066669B (en) | 2021-10-28 | 2021-10-28 | Cloud manufacturing-oriented manufacturing service discovery method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111264538.XA CN114066669B (en) | 2021-10-28 | 2021-10-28 | Cloud manufacturing-oriented manufacturing service discovery method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114066669A CN114066669A (en) | 2022-02-18 |
| CN114066669B true CN114066669B (en) | 2024-05-03 |
Family
ID=80235884
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111264538.XA Active CN114066669B (en) | 2021-10-28 | 2021-10-28 | Cloud manufacturing-oriented manufacturing service discovery method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114066669B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570080A (en) * | 2016-10-18 | 2017-04-19 | 河海大学常州校区 | Multilevel semantic matching method for cloud manufacturing resource services |
| CN111582561A (en) * | 2020-04-23 | 2020-08-25 | 华南理工大学 | Small-batch multi-variety-oriented reconfigurable production line scheduling optimization method |
| CN112632215A (en) * | 2020-12-01 | 2021-04-09 | 重庆邮电大学 | Community discovery method and system based on word-pair semantic topic model |
| CN113139599A (en) * | 2021-04-22 | 2021-07-20 | 北方工业大学 | Service distributed clustering method fusing word vector expansion and topic model |
-
2021
- 2021-10-28 CN CN202111264538.XA patent/CN114066669B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106570080A (en) * | 2016-10-18 | 2017-04-19 | 河海大学常州校区 | Multilevel semantic matching method for cloud manufacturing resource services |
| CN111582561A (en) * | 2020-04-23 | 2020-08-25 | 华南理工大学 | Small-batch multi-variety-oriented reconfigurable production line scheduling optimization method |
| CN112632215A (en) * | 2020-12-01 | 2021-04-09 | 重庆邮电大学 | Community discovery method and system based on word-pair semantic topic model |
| CN113139599A (en) * | 2021-04-22 | 2021-07-20 | 北方工业大学 | Service distributed clustering method fusing word vector expansion and topic model |
Non-Patent Citations (1)
| Title |
|---|
| 基于BTM的物联网服务发现方法;王舒漫;李爱萍;段利国;付佳;陈永乐;;计算机应用;20200210(02);第459-461页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114066669A (en) | 2022-02-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sarker | Machine learning: Algorithms, real-world applications and research directions | |
| US11430065B2 (en) | Subscription-enabled news recommendation system | |
| US8719192B2 (en) | Transfer of learning for query classification | |
| US11232365B2 (en) | Digital assistant platform | |
| US10565528B2 (en) | Analytic system for feature engineering improvement to machine learning models | |
| US8898092B2 (en) | Leveraging user-to-tool interactions to automatically analyze defects in it services delivery | |
| Paramesh et al. | Automated IT service desk systems using machine learning techniques | |
| US20200090059A1 (en) | Utilizing machine learning models to automatically generate contextual insights and actions based on legal regulations | |
| CN110968701A (en) | Relationship map establishing method, device and equipment for graph neural network | |
| CN110909165A (en) | Data processing method, device, medium and electronic equipment | |
| CN110780965A (en) | Vision-based process automation method, device and readable storage medium | |
| CN113392920B (en) | Method, apparatus, device, medium, and program product for generating cheating prediction model | |
| US20170109640A1 (en) | Generation of Candidate Sequences Using Crowd-Based Seeds of Commonly-Performed Steps of a Business Process | |
| US11062330B2 (en) | Cognitively identifying a propensity for obtaining prospective entities | |
| CN112287111B (en) | Text processing method and related device | |
| CN115222433A (en) | Information recommendation method and device and storage medium | |
| CN113051324A (en) | User portrait construction method and device based on big data and storage medium | |
| Chen et al. | Automatic web services classification based on rough set theory | |
| CN114066669B (en) | Cloud manufacturing-oriented manufacturing service discovery method | |
| CN116823321A (en) | Method and system for analyzing economic management data of electric business | |
| CN113822390B (en) | User portrait construction method and device, electronic equipment and storage medium | |
| CN113342988B (en) | Method and system for constructing service knowledge graph to realize service combination optimization based on LDA cross-domain | |
| US20220156285A1 (en) | Data Tagging And Synchronisation System | |
| CN110232184A (en) | The machining service two-way semantic matching method of supply and demand under a kind of cloud manufacturing mode | |
| Li | Naive Bayes algorithm for Twitter sentiment analysis and its implementation in MapReduce |
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 |