CN112633796B - Knowledge-driven-based supplier management inventory optimization method - Google Patents

Abstract

The invention discloses a knowledge-driven supplier management inventory optimization method, which comprises the steps of firstly, determining entities and relations related to problems based on the domain knowledge of a supplier management inventory VMI, storing numerical attributes of examples, and generating OWL ontology files; then reading the OWL body file, and performing numerical calculation by using JAVA to obtain a commodity distribution scheme of a supplier for each store; finally, based on JENA API definition rules, relationship rule reasoning is carried out, and sold out commodity prompts, commodity minimum inventory modification prompts and delivery delay prompts are obtained; according to the invention, a distribution recommendation scheme is provided for suppliers according to the commodity sales data of the stores, so that the cost is reduced and the efficiency is improved; providing related early warning prompts and suggestions for commodity inventory and minimum inventory, and facilitating management of suppliers; weather factors are considered, early warning is carried out aiming at the possibly delayed delivery, and the accuracy of the recommended scheme is improved; the original VMI provider inventory management method is optimized, and the consumption of manpower, material resources and financial resources is reduced.

Description

Knowledge-driven-based supplier management inventory optimization method

Technical Field

The invention relates to the technical field of inventory optimization, in particular to a knowledge-driven-based inventory optimization method for supplier management.

Background

Supplier inventory management (VMI) is a mode of inventory operation in a supply chain environment that essentially changes a multi-level supply chain problem to a single-level inventory management problem relative to the traditional practice of restocking orders by traditional users. VMI is a solution to market demand forecast and inventory replenishment with actual or forecast consumption demand and inventory, i.e., the consumer demand information is derived from sales data, and suppliers can plan more effectively and more quickly to deal with market changes and consumer demand. At present, most VMIs adopt mathematical modeling, the modeling threshold is high, the function is fixed, emergency conditions (such as weather and other factors) must be modeled again, the real-time updating cannot be carried out according to actual requirements, and the applicability is poor.

OWL is the description language of the ontology. In the fields of computer science and information science, in theory, an ontology refers to a "formalized," explicit and detailed description of a shared concept system. An ontology provides a shared vocabulary, i.e., those object types or concepts that exist in a particular domain and their attributes and interrelationships. Ontologies can be used to infer properties of the domain and can also be used to define the domain (i.e., model the domain). In addition, people may also refer to "ontologies" as "ontologies". As a form of knowledge representation about the real world or some component thereof, the application areas of the ontology include (but are not limited to): artificial intelligence, semantic web, software engineering, biomedical informatics, librarian and information architecture.

Jena is an open source tool for the HP laboratory semantic Web research project group, which is a JAVA-based semantic Web application framework. Jena contains a general rule inference engine, which can be used in RDFS and OWL inference engines, or can be used alone. The inference engine supports reasoning on RDF graphs, providing forward chains, backward chains, and hybrid execution modes of both. Including RETE engines and one tabled datalog engine. The parameters may be configured by genericrule releaser, using various inference engines. To use genericrule releaser, a rule set is required to define its behavior.

Disclosure of Invention

The invention aims to: aiming at the problems that the prior VMI modeling threshold mentioned in the background technology is very high and the function is fixed, the emergency (such as weather and other factors) needs to be modeled again, the real-time updating can not be realized according to the actual requirement, and the applicability is poor, provides a knowledge-driven supplier management inventory optimization method, which is used for solving the problem of the prior VMI supplier inventory management; the invention can be used for a manager to automatically add or change axiom according to the real-time sales data of the user so as to cope with sudden factors, and has stronger universality.

The technical scheme is as follows: in order to achieve the above purpose, the invention adopts the following technical scheme:

a knowledge-driven vendor-managed inventory optimization method comprising the steps of:

step S1, based on the domain knowledge of the supplier management stock VMI, determining the entity and relation related to the problem, and storing the numerical attribute of the instance to generate an OWL ontology file;

s2, reading an OWL body file, and performing numerical calculation by using JAVA to obtain a commodity distribution scheme of a supplier for each store;

and S3, based on JENA API definition rules, carrying out relation rule reasoning to obtain a sold-out commodity prompt, a commodity minimum inventory modification prompt and a delivery delay prompt.

Further, the OWL body file in step S1 specifically includes:

(1) Physical characteristics;

the method comprises a shop entity drug store, a commodity entity product and an inventory entity inventory;

(2) Object attributes;

including has_inventory and has_product; wherein the drug entity comprises a has_product attribute, and the Product entity has a has_inventory attribute;

(3) Packaging information packageType;

including data attributes weight, preperationcost, and unitquality; wherein the package of each commodity is established with a corresponding packageType instance.

Further, in the step S2, JAVA is used to perform numerical calculation, so as to obtain the commodity distribution scheme of the provider for each store, which specifically includes the following steps:

s2.1, storing sales of commodities in shops in a plurality of days, taking the average value of the sales of the commodities in a unit day as the forecast of the sales in a unit day in the future, and marking the forecast as a history;

step S2.2, judging the distribution condition of a store when the commodity meets the requirement of present-history_arrival < minstock or present-history dailytrunk_arrival < minstock on the same day; triggering delivery when delivery does not occur within one week of the store;

wherein minstock is the minimum inventory of the commodity in the store, maxstock is the maximum inventory of the commodity in the store, present is the number of days spent in delivering the commodity in the store in the current inventory express_arrival mode, and dailytrunk_arrival is the number of days spent in delivering the commodity in the dailyrun mode;

step S2.3, further judging a distribution mode according to the commodity inventory judgment result in the step S2.2; when the commodity meets the condition of presentation-history_arrival < minstock, an EXPRESS delivery mode is adopted for delivery whether the condition of presentation-history_arrival < minstock is met or not; when the commodity meets the requirement that the present-history_arrival is more than or equal to minstock and the present-history_arrival is less than minstock, calculating the distribution expense of two distribution modes, namely EXPRESS and DAILYTRUNC, and selecting the one with smaller expense for distribution; the distribution behavior is performed for all commodities in the store at the moment, and the distribution amount is as follows: minstock+ (maxstock-minstock) 0.8-present+history 7.

Further, in the step S3, rules are defined based on the JENA API, and the specific steps of relationship rule reasoning are as follows:

s3.1, establishing a relationship between the ontology and the ontology according to a JAVA numerical calculation judgment result, and then carrying out relationship reasoning by utilizing a Jena reasoning machine;

step S3.2, listSubjectsWithProperty traversal based on JENA API searches for the ontology object meeting the relation.

Further, the JNA inference engine rules include the following three cases:

(1) [ (; namely triggering delay delivery when the crossing range of the delivery date contains a certain bad weather;

(2) [ (; when the stock quantity is smaller than or equal to zero stock, triggering sold-out early warning;

(3) The [ (; when the inventory early warning prompting times reach the preset maximum prompting times, the minimum inventory of the commodity is modified.

The beneficial effects are that: the invention has the following advantages:

the invention solves the problem of stock management of VMI suppliers by adopting a knowledge-based driving method, and solves the problems that the past mathematical modeling threshold is high, the function is fixed, the emergency (such as weather and other factors) must be modeled again, the real-time updating can not be carried out according to the actual requirement, and the applicability is poor by constructing an OWL body file. Meanwhile, a rule file is constructed, and the Jena inference engine is utilized to conduct early warning information inference so as to cope with sudden weather conditions. The realization of the method proves the expansibility of similar functions in other similar fields to a certain extent.

Drawings

FIG. 1 is an OWL local file architecture diagram provided by the present invention;

FIG. 2 is a program implementation level architecture diagram in an embodiment of the invention;

FIG. 3 is a JENA rule reasoning flow chart provided by the present invention;

FIG. 4 is a flowchart of program execution in an embodiment of the present invention.

Detailed Description

The invention will be further described with reference to the accompanying drawings.

A knowledge-driven vendor-based inventory optimization method as shown in fig. 1-4, comprising the steps of:

and step S1, determining entities and relations related to problems based on domain knowledge of the supplier management inventory VMI, storing numerical attributes of the examples, and generating an OWL ontology file.

The OWL local file is shown in fig. 1, and includes:

(1) Physical characteristics;

the method comprises a shop entity drug store, a commodity entity product and an inventory entity inventory;

(2) Object attributes;

including has_inventory and has_product; wherein the drug entity comprises a has_product attribute, and the Product entity has a has_inventory attribute;

(3) Packaging information packageType;

including weight of data attribute unit packaged product, packaging cost preparation cost and quality unit quality of each packaged product; wherein the package of each commodity is established with a corresponding package information packageType instance.

Specifically, the local OWL body file is read by the method shown in the following table 1:

table 1 read ontology File API

The code for the specific operation is as follows:

reading information of each class and corresponding instance in the body file, wherein partial codes are as follows:

(1) Reading information of a drug:

OWLClass drugstoreCls＝factory.getOWLClass(":drugstore",pm)；

NodeSet<OWLNamedIndividual>individualsNodeSet＝reasoner.getInstances(drugstoreCls,true)；

(2) Reading the information of the product:

OWLClass productCls＝factory.getOWLClass(":product",pm)；

NodeSet<OWLNamedIndividual>productSet＝reasoner.getInstances(productCls,true)；

(3) Read the information of the packageType (taking a data attribute unitquality as an example):

OWLObjectProperty hasPackageType＝factory.getOWLObjectProperty(":has_PackageType",pm)；

NodeSet<OWLNamedIndividual>hasPackageTypeSet＝reasoner.getObjectPropertyValues(p,hasPackageType)；

OWLNamedIndividual invInd＝factory.getOWLNamedIndividual(name,pm)；

OWLDataProperty unit_Quantity＝factory.getOWLDataProperty(":unitQuantity",pm)；

Set<OWLLiteral>unit_QuantityValue＝reasoner.getDataPropertyValues(invInd,unit_Quantity)。

step S2, reading an OWL body file, and performing numerical calculation by using JAVA to obtain a commodity distribution scheme of a provider for each store, wherein the commodity distribution scheme is specifically shown in FIG. 2, and English tools appearing in FIG. 2 are explained as follows:

OWL API: a JAVA application programming interface for creating, manipulating and serializing OWL ontologies;

JENA API: an open source JAVA development kit for semantic web application development;

swing: java designed graphical user interface toolkits include text boxes, buttons, split panes, and tables.

S2.1, storing sales of commodities in shops in a plurality of days, taking the average value of the sales of the commodities in a unit day as the forecast of the sales in a unit day in the future, and marking the forecast as a history;

step S2.2, judging the distribution condition of a store when the commodity meets the requirement of present-history_arrival < minstock or present-history dailytrunk_arrival < minstock on the same day; triggering delivery when delivery does not occur within one week of the store;

wherein minstock is the minimum inventory of the commodity in the store, maxstock is the maximum inventory of the commodity in the store, present is the number of days spent in delivering the commodity in the store in the current inventory express_arrival mode, and dailytrunk_arrival is the number of days spent in delivering the commodity in the dailyrun mode;

step S2.3, further judging a distribution mode according to the commodity inventory judgment result in the step S2.2; when the commodity meets the condition of presentation-history_arrival < minstock, an EXPRESS delivery mode is adopted for delivery whether the condition of presentation-history_arrival < minstock is met or not; when the commodity meets the requirement that the present-history_arrival is more than or equal to minstock and the present-history_arrival is less than minstock, calculating the distribution expense of two distribution modes, namely EXPRESS and DAILYTRUNC, and selecting the one with smaller expense for distribution; the distribution behavior is performed for all commodities in the store at the moment, and the distribution amount is as follows: minstock+ (maxstock-minstock) 0.8-present+history 7.

In a specific embodiment, firstly, a hash table is adopted to map the read body file information into a corresponding hash table, and a vector is used for carrying out subsequent operations such as traversing, accessing and the like;

drug hashmap = drug store. Getdrug info (r. Factory, drug cls, pm); hashmapdragstorindidualvector of the acquired drug store = drug store. Getdrug store vector (individalnodeset); vector for/(and taking drug

Product hashmap = product. GetProductInfo (product, product cls, pm); hashmap corresponding to product is obtained

Product individicon = product. Getproduct vector (product set); vector corresponding to product is obtained/obtained

PackageHashMap=initialPackageHashMap (releaser, factor, manager, pm); information of the initial packagehashmap is obtained, and packaging information corresponding to each commodity such as < p1, < box, < certainpackageinfo > isstored

initialSomeMap (hasProduct 1, ontologigy, inventioalnodeset, releaser, factor, man-agent, pm); information stock quantity of stop map at initial time, minimum limit of stock quantity and maximum limit of stock quantity are obtained/obtained

Daily inventory variation and distribution is then inferred by numerical calculations:

traversing the drug industry device, traversing the product industry device in the process to acquire the commodity of a certain store,

add < product, sold > to hashmap: tmpsellinghhashmap. Put (pro, ss);

adding the new sales into the historic sales hashMap: aversalex=aversaleshashmap. Get (drug). Get (pro);

adds < merchandise, < product, sales > > to the table: selhashmap.put (drug, tmpsellinghhashmap);

processing whether an order needs to be generated on this day according to daily sales history sales and the above-obtained information: dealEveryDay (currentDay);

in the dealevalyday () method, inventory pre-warning is judged, and the quantity of each commodity delivered is determined, and different modes of delivery are selected:

vector quality_prepared=new Vector (productive nummax); the quantity of goods/articles stored for each article

quality_prepared= dealPreparedProductNum (drug); parameters of a store to be transmitted for a store

total_preparation cost=dealtotal_preparation cost (quality_prepared); cost of preparing goods is calculated

Calculating the transportation cost of two delivery modes:

dailytrunc_transcost＝getDailytruncCost(total_weight,drug)；

express_franscost=getexpress cost (total_weight, drug); if return-1 represents not as much cargo as it is sent

The delivery mode is selected through comparing the current stock-today sales quantity-history sales quantity with the minimum stock quantity, if both modes can be used, the transportation cost of the two modes is compared, the selection cost is low, the final delivery cost is cost_all=total_preparation cost+express_transfer, and the generated delivery instance de_tmp is added into a priority queue delivery.

And S3, based on JENA API definition rules, carrying out relation rule reasoning to obtain a sold-out commodity prompt, a commodity minimum inventory modification prompt and a delivery delay prompt.

Apache Jena (or Jena for short) is a free and open source Java framework for building semantic Web and associated data applications. The framework consists of different APIs for processing RDF data. Before rule reasoning can be performed, the method provided by the Jena API shown in table 2 below is needed:

TABLE 2 Jena rule reasoning related APIs (taking possibly delayed delivery Pre-alarm as an example)

In the embodiment of the invention, the relationship between the ontology and the ontology is established by the JAVA numerical calculation judgment result, and then the relationship reasoning is carried out by using a Jena reasoning machine. ListSubjectsWithProperty traversal based on JENA API finds ontology objects that satisfy the relationship, as shown in FIG. 3.

The JENA inference engine rules include the following three cases:

(1) [ (; namely triggering delay delivery when the crossing range of the delivery date contains a certain bad weather;

(2) [ (; when the stock quantity is smaller than or equal to zero stock, triggering sold-out early warning;

(3) The [ (; when the inventory early warning prompting times reach the preset maximum prompting times, the minimum inventory of the commodity is modified.

Taking the judgment of the distribution which is possibly delayed as an example, the process of calling rule reasoning and result query is as follows:

all inference rules used in the embodiments of the present invention are shown in table 3 below:

TABLE 3 inference rule summarization

/>

The foregoing is only a preferred embodiment of the invention, it being noted that: it will be apparent to those skilled in the art that various modifications and adaptations can be made without departing from the principles of the present invention, and such modifications and adaptations are intended to be comprehended within the scope of the invention.

Claims (2)

1. A knowledge-driven vendor-managed inventory optimization method, comprising the steps of:

step S1, based on the domain knowledge of the supplier management stock VMI, determining the entity and relation related to the problem, and storing the numerical attribute of the instance to generate an OWL ontology file;

s2, reading an OWL body file, and performing numerical calculation by using JAVA to obtain a commodity distribution scheme of a supplier for each store;

s3, based on JENAAPI definition rules, carrying out relation rule reasoning to obtain a sold-out commodity prompt, a commodity minimum inventory modification prompt and a delivery delay prompt;

in the step S2, JAVA is adopted to perform numerical calculation, so as to obtain the specific steps of the commodity distribution scheme of the suppliers to each store, wherein the specific steps are as follows:

s2.1, storing sales of commodities in shops in a plurality of days, taking the average value of the sales of the commodities in a unit day as the forecast of the sales in a unit day in the future, and marking the forecast as a history;

step S2.2, judging the distribution condition of a store when the commodity meets the requirement of present-history_arrival < minstock or present-history dailytrunk_arrival < minstock on the same day; triggering delivery when delivery does not occur within one week of the store;

wherein minstock is the minimum inventory of the commodity in the store, maxstock is the maximum inventory of the commodity in the store, present is the current inventory of the commodity in the store, express_arrival is the number of days spent in the delivery by EXPRESS delivery, and dailyTrunc_arrival is the number of days spent in the delivery by DAILYTRUNC delivery;

step S2.3, further judging a distribution mode according to the commodity inventory judgment result in the step S2.2; when the commodity meets the condition of presentation-history_arrival < minstock, an EXPRESS delivery mode is adopted for delivery whether the condition of presentation-history_arrival < minstock is met or not; when the commodity meets the requirement that the present-history_arrival is more than or equal to minstock and the present-history_arrival is less than minstock, calculating the distribution expense of two distribution modes, namely EXPRESS and DAILYTRUNC, and selecting the one with smaller expense for distribution; the distribution behavior is performed for all the commodities in the store at the moment, and the distribution amount is as follows: minstock+ (maxstock-minstock) 0.8-present+history 7;

in the step S3, rules are defined based on JENAAPI, and the specific steps of relationship rule reasoning are as follows:

s3.1, establishing a relationship between the ontology and the ontology according to a JAVA numerical calculation judgment result, and then carrying out relationship reasoning by utilizing a JENA reasoning machine;

step S3.2, listSubjectsWithProperty traversal based on JENAAPI searches for the ontology objects meeting the relation;

the JENA inference engine rules include the following three cases:

(1) [ (; namely triggering delay delivery when the crossing range of the delivery date contains a certain bad weather;

(2) [ (; when the stock quantity is smaller than or equal to zero stock, triggering sold-out early warning;

(3) The [ (; when the inventory early warning prompting times reach the preset maximum prompting times, the minimum inventory of the commodity is modified.

2. The knowledge-driven vendor-based inventory optimization method according to claim 1, wherein the OWL ontology file in step S1 specifically includes:

(1) Physical characteristics;

the method comprises a shop entity drug store, a commodity entity product and an inventory entity inventory;

(2) Object attributes;

including has_inventory and has_product; wherein the drug entity comprises a has_product attribute, and the Product entity has a has_inventory attribute;

(3) Packaging information packageType;

including data attributes weight, preparationCost and unitquality; wherein the package of each commodity is established with a corresponding packageType instance.