Intelligent fixed asset transfer method and system, and electronic device

Abstract

The application provides a fixed asset intelligent transfer method and system and electronic equipment, and relates to the technical field of enterprise asset management. The application comprises the following steps: constructing an asset knowledge graph by using a multi-source data integration method; constructing a transfer rule knowledge graph by considering the double dimensions of asset categories and accounting subjects; setting necessary conditions for triggering the transfer operation, obtaining candidate order information to be transferred when the necessary conditions are met; determining the asset categories and asset amounts in the candidate order based on the matching relationship between the candidate order information and the nodes in the asset knowledge graph; determining the main equipment to which the assets in the candidate order are affiliated based on the matching relationship between the asset categories of the candidate order and the nodes in the transfer rule knowledge graph; and allocating the asset amounts of the candidate order to the main equipment to which the corresponding assets are affiliated. The application sequentially matches the candidate order information to be transferred with two knowledge graphs, and then obtains the main equipment to which the assets are affiliated for asset amount allocation, so that the transfer process is accurate and efficient.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of enterprise asset management and artificial intelligence, and particularly relates to a fixed asset intelligent transfer method, a system and an electronic device. BACKGROUND

[0002] In the whole life cycle management of fixed assets, "transfer" refers to the whole process of formally confirming the construction project (or contract assets, engineering materials) that has passed the completion acceptance and reached the predetermined usable state as "fixed assets" and being included in the financial books.

[0003] In the traditional fixed asset management, the transfer is performed manually: the information such as asset name, specification, amount, supplier and the like is checked manually, the fixed asset subject and the depreciation rule are matched manually, then the acceptance sheet, the contract and the invoice are checked by experience, and finally the result is input into the ERP.

[0004] Once the asset amount is large and the types are many, the mode will fall into a dead cycle of "inefficiency - backlog - high error rate": manual matching is not only slow, but also easy to misclassify and misuse the depreciation code, resulting in distorted financial data; the regulations and related regulations are updated frequently, and manual tracking cannot be performed in real time, and the compliance risk is soaring; and the procurement, contract and financial systems are independent, the data islands are serious, and the problems are difficult to trace. SUMMARY

[0005] To overcome the above-mentioned deficiencies of the prior art, the application provides a fixed asset intelligent transfer method, a system and an electronic device, two knowledge graphs are constructed to carry two semantic dimensions of financial rules and business facts respectively, the candidate order information to be transferred is matched with the two knowledge graphs in turn, and then the main equipment to be hung is obtained and the asset amount is allocated, the transfer process is accurate and efficient, and the problems of explainability, traceability and auditability of the dynamic allocation logic from auxiliary materials to main equipment are solved.

[0006] To achieve the above-mentioned purpose, one or more embodiments of the application provide the following technical solutions:

[0007] The first aspect of the application provides a fixed asset intelligent transfer method.

[0008] The fixed asset intelligent transfer method comprises the following steps:

[0009] An asset knowledge graph is constructed by using a multi-source data integration method;

[0010] A transfer rule knowledge graph is constructed considering the two dimensions of asset category and accounting subject;

[0011] A necessary condition for triggering the transfer operation is set, and when the necessary condition for triggering the transfer operation is met, candidate order information to be transferred is obtained;

[0012] determine the asset category and the asset amount in the candidate order based on the matching relationship between the candidate order information and the nodes in the asset knowledge graph;

[0013] determine the main equipment to which the assets in the candidate order are attached based on the matching relationship between the asset category of the candidate order and the nodes in the transfer capital rule knowledge graph;

[0014] distribute the asset amount of the candidate order to the main equipment to which the corresponding assets are attached, and complete the transfer capital process.

[0015] The second aspect of the present application provides a fixed asset intelligent transfer capital system.

[0016] The fixed asset intelligent transfer capital system comprises:

[0017] The asset knowledge graph construction module is configured to construct the asset knowledge graph by using a multi-source data integration method;

[0018] The transfer capital rule knowledge graph construction module is configured to construct the transfer capital rule knowledge graph by considering the asset category and the accounting subject in two dimensions;

[0019] The triggering module is configured to set the necessary conditions for triggering the transfer capital operation, and obtain the candidate order information to be transferred when the necessary conditions for the transfer capital are met;

[0020] The first matching module is configured to determine the asset category and the asset amount in the candidate order based on the matching relationship between the candidate order information and the nodes in the asset knowledge graph;

[0021] The second matching module is configured to determine the main equipment to which the assets in the candidate order are attached based on the matching relationship between the asset category of the candidate order and the nodes in the transfer capital rule knowledge graph;

[0022] The distribution module is configured to distribute the asset amount of the candidate order to the main equipment to which the corresponding assets are attached, and complete the transfer capital process.

[0023] The third aspect of the present application provides an electronic device, comprising a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to realize the steps in the fixed asset intelligent transfer capital method according to the first aspect of the present application.

[0024] The above one or more technical solutions have the following beneficial effects:

[0025] The application provides a fixed asset intelligent transfer method and system and electronic equipment, solves the rule fragmentation, data fragmentation and compliance tracing problem in the fixed asset transfer scene, proposes a hierarchical progressive closed-loop method of "project internal inspection trigger-knowledge graph unified semantics-rule engine dynamic arrangement-blockchain storage audit", adopts a multi-source data integration method to construct an asset knowledge graph, considers the asset category and accounting subject double dimensions, constructs a transfer rule knowledge graph, sets the hot-pluggable transfer necessary condition of triggering the transfer operation, significantly improves the level of data integration, the breadth and depth of rule coverage, and further improves the landing value of the large model in the financial asset field.

[0026] In the specific implementation of the transfer operation, based on the matching relationship of the candidate order information and the nodes in the asset knowledge graph, the asset category and the asset amount in the candidate order are determined; based on the matching relationship of the asset category of the candidate order and the nodes in the transfer rule knowledge graph, the main equipment to which the assets in the candidate order are attached is determined; the asset amount of the candidate order is allocated to the main equipment to which the corresponding assets are attached, and the transfer process is completed. The candidate order is matched with two knowledge graphs respectively, the attached main equipment is screened out, and dynamic updating of the knowledge graph can be realized, and the matching process is efficient and accurate.

[0027] The construction of the asset knowledge graph of the application covers multiple source data categories, solves the data island problem, and updates the regulation and provision nodes in the asset knowledge graph according to the latest regulations and provision texts, so that more accurate matching is realized.

[0028] The advantages of the additional aspects of the application will be partially given in the following description, partially will become obvious from the following description, or will be known by the practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0029] The drawings accompanying the specification of the application form part of the application and serve to provide further understanding of the application, the illustrative embodiments of the application and their description serve to explain the application and do not constitute an improper limitation of the application.

[0030] Figure 1 The method flowchart of example one.

[0031] Figure 2 The transfer necessary condition execution flowchart of example one.

[0032] Figure 3 The asset knowledge graph schematic diagram of example one.

[0033] Figure 4 The method execution block diagram of example one. DETAILED DESCRIPTION

[0034] It should be noted that the following detailed description is exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0035] It should be noted that the terms used herein are only intended to describe specific embodiments and are not intended to limit exemplary embodiments according to the present application.

[0036] In the absence of conflicts, the embodiments in the present application and the features in the embodiments can be combined with each other.

[0037] Embodiment one

[0038] In the field of fixed asset management, ERP, contract, OA, image and other systems are isolated, and asset data is scattered in different business links in a fragmented form. The traditional transfer process relies on manual checking of asset name, amount, acceptance sheet, invoice and depreciation subject, which is not only time-consuming and prone to error, but also cannot synchronize the latest financial and tax regulations and provisions in real time, resulting in high compliance risk and difficulty in tracing.

[0039] The transfer process is expressed in accounting language as:

[0040] Borrow: fixed assets;

[0041] Lend: construction in progress / engineering materials / contract assets, etc.

[0042] In the relevant provisions on fixed assets, construction in progress that reaches the predetermined usable state should be recognized as fixed assets and start to depreciate. The annual final accounts, audit, tax inspection of multiple departments all list "timely transfer" as a key compliance indicator.

[0043] Once the fixed assets are accounted for, they can be depreciated and deducted before enterprise income tax, and delayed transfer equals delayed tax deduction.

[0044] If not timely transfer → miss depreciation → increase profits → audit reservations → tax + penalties. Only by completing the transfer, the assets can enter the ERP / EAM system, and the subsequent depreciation, inventory, insurance, mortgage, impairment testing, disposal and other businesses. Group companies assess project companies and infrastructure project teams through "timely transfer rate" projects, which directly affects performance bonuses.

[0045] Therefore, a new intelligent transfer method capable of "second-level batch processing and zero-code configuration" is urgently needed to significantly improve the construction efficiency, level, breadth and depth of fixed asset data sets, and fully release the application value of large models in the field of financial assets.

[0046] In recent years, some manufacturers have tried to achieve automation through scripts or single-point plug-ins, but lack a unified semantic model and a configurable rule engine, which cannot cover complex scenarios such as multi-asset categories, cross-system data fusion, and dynamic updates of regulations and provisions, and the efficiency is limited and the maintenance cost is high.

[0047] In addition, traditional manual transfer or single-point script only solves the problem of writing the amount into the fixed asset card, and cannot answer the three questions that audit is most concerned about: why the order amount is recorded to device A instead of device B, which accounting subject regulations and provisions, and how to quickly take effect and traceability after the change of subsequent regulations and provisions. Therefore, the transfer process is a black box and cannot be seen through the above technical means.

[0048] When the embodiment of the application implements the transfer process by using knowledge graph technology, if only one large and complete knowledge graph is built, the following fatal defects will occur: business facts and financial rules are coupled, auxiliary material to main device allocation logic is a black box, regulations and provisions cannot be responded in time, and the like.

[0049] Based on a single knowledge graph, it is impossible to simultaneously carry business facts and financial rules in two semantic dimensions, and it is also impossible to solve the problems of explainability, traceability, and auditability of the dynamic allocation logic from auxiliary materials to main devices. The embodiment sets two knowledge graphs (asset knowledge graph + transfer rule knowledge graph) for cascaded matching.

[0050] The two graphs are not simply divided into two tables, but constitute an explainable, traceable, and updatable pipeline from business facts to financial rules to allocation results, solving the three problems of traditional transfer black box, allocation audit difficulty, and slow matching of regulations and provisions, enabling fixed asset transfer to achieve second-level batch processing, low-code configuration, and compliance traceability.

[0051] As shown in Figure 1 The embodiment provides a fixed asset intelligent transfer method, including the following steps:

[0052] A multi-source data integration method is used to construct an asset knowledge graph;

[0053] Considering the two dimensions of asset categories and accounting subjects, a transfer rule knowledge graph is constructed;

[0054] A necessary condition for triggering the transfer operation is set, and when the necessary condition for triggering the transfer operation is met, candidate order information to be transferred is obtained;

[0055] Based on the matching relationship between the candidate order information and the nodes in the asset knowledge graph, the asset category and the asset amount in the candidate order are determined;

[0056] Based on the matching relationship between the asset category of the candidate order and the nodes in the capital transfer rule knowledge graph, the main equipment to which the assets of the candidate order are attached is determined;

[0057] The asset amount of the candidate order is allocated to the main equipment to which the corresponding assets are attached, and the capital transfer process is completed.

[0058] The embodiment of the application effectively solves the problems of rule fragmentation, data fragmentation and compliance tracing in the fixed asset capital transfer scenario, and proposes a hierarchical progressive closed-loop method of "project internal inspection triggering-knowledge graph unified semantics-rule engine dynamic arrangement-blockchain evidence storage audit". A multi-source data integration method is used to construct an asset knowledge graph, and a hot-pluggable capital transfer rule knowledge graph is constructed according to the asset category and the accounting subject, which significantly improves the level of data integration, the breadth and depth of rule coverage, and further improves the landing value of large models in the field of financial assets.

[0059] In this embodiment, the meanings of project and candidate order are as follows:

[0060] Project refers to:

[0061] A set of engineering construction tasks established for the implementation of a specific project, with independent design documents, independent construction conditions, independent use functions, and the ability to form independent production capacity or use efficiency.

[0062] In the context of capital transfer, the project has completed engineering construction, acceptance and reached the predetermined usable state, and the physical assets have actually formed, meeting the conditions for transferring from "under-construction project" to "fixed asset" or "intangible asset", which is the complete object of capital transfer.

[0063] Candidate order refers to:

[0064] During the construction of the project, the corresponding physical assets have been formed and delivered, but the formal capital transfer procedures have not been completed, and whether they can be included in the project capital transfer range is still to be verified and confirmed.

[0065] These orders usually meet the preliminary capital transfer conditions (such as passing the acceptance test and complete cost collection), and are submitted to the capital transfer process as "candidates" to decide whether to officially include the project fixed assets or intangible asset value after further verification.

[0066] Next, the technical solutions of the embodiment will be explained in detail.

[0067] (I) Asset knowledge graph construction.

[0068] A multi-source data integration method is used to construct an asset knowledge graph, which specifically includes:

[0069] Extracting asset-related data from multi-source data, and preprocessing;

[0070] Defining entity types of the asset knowledge graph, including asset number, contract, supplier, invoice, acceptance sheet, accounting subject, asset category, depreciation rule, regulation and provision;

[0071] Defining relationship types of the asset knowledge graph, including the signing relationship between asset number and contract, the corresponding relationship between asset number and invoice, the association relationship between asset number and acceptance sheet, the belonging relationship between asset number and accounting subject, the belonging relationship between asset number and asset category, the applicable relationship between accounting subject and depreciation rule, the constraint relationship between accounting subject and regulation and provision, and the association relationship between contract and supplier;

[0072] Mapping the extracted asset-related data to the entity types of the asset knowledge graph, and matching the obtained entities to establish the asset knowledge graph;

[0073] Verifying and dynamically updating the asset knowledge graph.

[0074] Extracting asset-related data from multi-source data, preprocessing, specifically including:

[0075] Obtaining asset-related multi-source fields from multi-source data including ERP, contract management system, OA system, image system and IoT: real-time grabbing asset card data from ERP, obtaining parsed bill information from contract management system, obtaining internal acceptance conclusion and signature data from OA system, obtaining invoice and arrival receipt image data from image system, and obtaining equipment information from IoT;

[0076] Cleaning and standardizing the multi-source fields;

[0077] Using a globally unique key for entity disambiguation, the globally unique key including contract number, asset number and invoice number;

[0078] Using timestamp + primary key for bidirectional verification of incremental data.

[0079] More specifically, including:

[0080] 1. Data source integration:

[0081] Extracting asset-related data from data sources such as ERP, contract management system, procurement system and financial system, and automatically cleaning and standardizing the data, and using "timestamp + primary key" bidirectional verification for incremental data to ensure that the data is not lost or duplicated.

[0082] The incremental data refers to all business data records related to assets that are newly generated or changed in each data source (ERP, contract management system, procurement system, and financial system) since the last data collection completion time.

[0083] The system automatically pulls together five types of "capital transfer key multi-source data":

[0084] First, the relevant data captured by ERP (including project, procurement, inventory, and fixed asset modules) in real time through RFC / BAPI or API;

[0085] Second, the invoice information parsed by the contract system through OCR-NER;

[0086] Third, the internal acceptance conclusion and signed PDF pushed by the OA process;

[0087] Fourth, the images of invoices and delivery receipt documents from the image system;

[0088] Fifth, the device information reported by IoT.

[0089] Among them, ERP represents enterprise management system, OA represents office system, and IoT represents Internet of Things. RFC / BAPI represents internal business interface, API represents cross-system business interface, and OCR-NER represents image recognition.

[0090] From the above multi-source data, asset-related data is obtained and preprocessed. The core idea of "timestamp + primary key" bidirectional verification is:

[0091] Using "primary key" to determine a unique business object and using "timestamp" to judge whether the object is added or updated; maintaining a "cursor" in the "puller" and "source system", respectively, through "forward comparison (incremental pulling) + reverse review (missing number scanning)" double channels to ensure that the data is neither missed nor duplicated.

[0092] The technical details are given below according to the implementation steps.

[0093] 1) Unify the primary key and timestamp fields

[0094] The primary key includes:

[0095] Purchase order line: {PO_NO (purchase order number), PO_LINE_NO (purchase order line number)};

[0096] Contract: {CONTRACT_ID (contract unique identifier)};

[0097] Invoice: {INVOICE_CODE (invoice code), INVOICE_NO (invoice code)};

[0098] Asset Card: {ASSET_ID (Unique Asset Identifier)}.

[0099] Timestamp:

[0100] Use the last update time (UPDATE_TIME) recorded in the source system. If this field is not available, use the creation time plus an auto-incrementing business sequence number to create a pseudo timestamp.

[0101] 2) Create a "synchronization cursor table" in both the source system and the data gateway.

[0102] The fields of the synchronized cursor table include:

[0103] cursor_name (cursor name), last_max_ts (last synchronization maximum timestamp), last_max_pk (last synchronization maximum primary key), scan_start_time (scan start time), scan_end_time (scan end time), row_count (row count), hash_sum (hash sum).

[0104] effect:

[0105] last_max_ts: The maximum timestamp since the last successful synchronization.

[0106] last_max_pk (optional): Use this field to further segment when the timestamp granularity is down to the second level and concurrent updates may generate multiple records in the same second.

[0107] hash_sum: The CRC32 (checksum) calculated by the source system after sorting the primary key and timestamp of the records in this batch, which is used for subsequent back-checking.

[0108] 3) Positive incremental pull

[0109] Step 1: The data gateway reads the cursor table and obtains last_max_ts=T0, where T0 refers to the maximum timestamp time0.

[0110] Step 2: Call the source system API: Query the data in the data source table whose time is greater than the maximum timestamp since the last successful synchronization, sort the data, and return the data to the calling system.

[0111] Step 3: After the gateway receives the response, it performs an idempotent write locally:

[0112] First, update and insert data based on the primary key (update if it exists, insert if it does not exist).

[0113] Record the maximum timestamp T1 and the maximum primary key P1 of the current result set.

[0114] Step 4: Update the cursor table with the maximum timestamp and primary key after this synchronization.

[0115] last_max_ts = T1, last_max_pk = P1.

[0116] 4) Backward review

[0117] Purpose: To prevent "missing lines" caused by source system re-entry, API timeout, and network jitter. Compare local data with source data through hash value to find missing lines and immediately re-enter.

[0118] Steps:

[0119] Step 1: The gateway uses scan_start_time = T0 and scan_end_time = T1 in the cursor table to query the source system.

[0120] Step 2: Sort the records that have landed in the same interval locally and calculate hash_sum_local (local hash value).

[0121] Step 3: Compare: hash_sum_local ≠ hash_sum → There are missing or incorrect lines.

[0122] Compare the primary key list line by line, and immediately pull the missing lines for re-entry.

[0123] Step 4: Re-enter the data first according to the primary key (update if it exists, insert if it does not exist) to ensure eventual consistency; after completion, update the cursor table hash_sum field to equal hash_sum_local.

[0124] 2. Entity and relationship definition:

[0125] Define the entity types of the asset knowledge graph:

[0126] Asset number, contract, supplier, invoice, acceptance sheet, accounting subject, asset category, depreciation rule, regulation and provision, etc.

[0127] Define the relationship types of the asset knowledge graph:

[0128] Asset number-contract ("sign"), asset number-invoice ("corresponding"), asset number-accounting subject ("belongs to"), accounting subject-depreciation rule ("apply"), regulation and provision-accounting subject ("constraint"), asset number-acceptance sheet ("associated"), asset number-asset category ("belongs to"), contract-supplier ("associated").

[0129] 3. Establishment of asset knowledge graph

[0130] Step 1: Entity Definition and Ontology Modeling

[0131] In this embodiment, when mapping the extracted asset-related data to entity types in the asset knowledge graph, the "Label" of the graph database Neo4j is used to correspond to entity types: asset number, contract, supplier, invoice, acceptance form, accounting subject, asset category, depreciation rules, regulations and rules.

[0132] Store business fields using "node attributes", such as:

[0133] CREATE (:Asset{asset_id: 'FA20240829001', name: 'CNC Lathe', spec: 'CK6136×1000', cost: 680000, enable_date: '2024-08-29'}).

[0134] The node attributes created above have the following specific meanings:

[0135] Create (Asset {Asset ID: 'FA20240829001', Name: 'CNC Lathe', Specifications: 'CK6136×1000', Cost: 680000, Activation Date: '2024-08-29'}).

[0136] The second step involves dividing the extracted asset-related data into structured data and semi-structured / unstructured data, and mapping them to entity types in the asset knowledge graph:

[0137] (1) Structured data:

[0138] ERP asset card → Mapped to asset node;

[0139] Contract Management System → Mapped to Contract Node;

[0140] Invoice OCR results → Mapped to invoice node.

[0141] (2) Semi-structured / unstructured data:

[0142] PDF contract and acceptance report → Extract contract number and acceptance conclusion using NLP and regular expressions;

[0143] Regulations and rules → Parse official documents from the Ministry of Finance (PDF) and extract document number and effective date.

[0144] (3) Primary key alignment:

[0145] Globally unique keys such as contract number, asset number, and invoice number are used for entity disambiguation to avoid duplicate nodes.

[0146] The third step is to perform relationship matching on the obtained entities and write the relationships:

[0147] Example:

[0148] MATCH (a:asset{asset_id:'FA20240829001'}),

[0149] (c:Contract{contract_no:'HT20240156'})

[0150] CREATE (a)-[:sign]->(c);

[0151] The above example means:

[0152] (a:Asset{Assetid:'FA20240829001'}) matches (c:Contract{Contract Number:'HT20240156'}), and the relationship between the two is "signing". Therefore, (a)-[:signing]->(c) is created.

[0153] This example supports batch import:

[0154] By inputting nodes and the triplet of "starting primary key - relationship type - ending primary key", and matching the relationships between nodes, an asset knowledge graph is created.

[0155] The fourth step is to validate the created asset knowledge graph:

[0156] (1) Validation can be achieved using Neo4j's Cypher (query language for graph databases) query, for example:

[0157] Verify that "all assets have corresponding invoices":

[0158] MATCH (a:Asset) WHERE NOT (a)-[:Corresponding]-(:Invoice) RETURN a.

[0159] The specific meaning is:

[0160] Match assets that do not have corresponding invoices and return those assets.

[0161] Verify whether "regulations and rules cover all accounting subjects":

[0162] MATCH (k: accounting subject) WHERE NOT (k)<-[:constraints]-(:regulations and rules) RETURN k.

[0163] The specific meaning is:

[0164] Match out the accounting subjects that do not conform to the regulations and provisions, and return the accounting subjects that do not conform to the regulations and provisions.

[0165] (2) Calculate PageRank and connectivity in combination with the Neo4j Graph Data Science library, find isolated assets or abnormal suppliers, abnormal orders, etc.

[0166] (3) When new regulations and provisions are added, only the regulation and provision nodes and the new constraint edges need to be inserted, and the rule engine can be real-time aware without changing the code.

[0167] 4. Knowledge graph storage:

[0168] As shown in Figure 3 , the defined entities and relationships are stored in triples (entity-relation-entity).

[0169] In this embodiment, a graph database (such as Neo4j) is used to store triples (entity-relation-entity), which supports efficient query and reasoning.

[0170] In addition, the embodiment can also realize dynamic updating of rules:

[0171] Rules are automatically extracted from regulation and provision texts (such as accounting standards and tax laws) through NLP technology, and the asset knowledge graph is updated.

[0172] (II) Construction of transfer capital rule knowledge graph.

[0173] Considering the two dimensions of asset category and accounting subject, a transfer capital rule knowledge graph is constructed, which specifically includes:

[0174] Define the entity types of the transfer capital rule knowledge graph, including projects, main equipment, auxiliary materials, assembly rules, and allocation instances;

[0175] Define the relationship types of the transfer capital rule knowledge graph, including the inclusion relationship between projects and main equipment, the inclusion relationship between projects and auxiliary materials, the matching relationship between auxiliary materials and assembly rules, the affiliation relationship between auxiliary materials and main equipment, and the source / target relationship between allocation instances and auxiliary materials / main equipment;

[0176] Obtain multiple project information, map the multiple project information to the defined entity types of the transfer capital rule knowledge graph, and match the relationships between entities to establish the transfer capital rule knowledge graph.

[0177] There are five kinds of entity types (node types) of the transfer capital rule knowledge graph:

[0178] Project, Device, Material, AllocationRule, Allocation (optional, used for accounting).

[0179] There are seven types of relationship in the transfer fund rule knowledge graph, which are represented as follows in this embodiment:

[0180] Project ─hasMasterDevice→ Device indicates which master devices the project has;

[0181] Project ─hasMaterial→ Material indicates which auxiliary materials are in the project;

[0182] Material ─matchesRule→ AllocationRule indicates which allocation rule the auxiliary material hits;

[0183] Material ─allocatesTo→ Device indicates which master device the auxiliary material is finally hung to;

[0184] Project ─createsFallbackAsset→ Device indicates a fallback asset;

[0185] Device ─belongsTo→ Project indicates which project the master device belongs to (for reverse query);

[0186] Allocation ─source / target→ Material / Device indicates the allocation details.

[0187] After defining the entity types and relationship types of the transfer fund rule knowledge graph, multiple project information is obtained, the multiple project information is mapped to the defined entity types of the transfer fund rule knowledge graph, and the relationship between entities is matched, so as to complete the establishment of the transfer fund rule knowledge graph.

[0188] (Three) Transfer fund condition setting.

[0189] As shown in Figure 2 , the necessary conditions for triggering the transfer fund operation are set, specifically including:

[0190] The necessary conditions for setting the transfer fund operation include that the project has been internally accepted, the order material amount is normal, and the asset category is within the fixed asset directory; when the three conditions of the project having been internally accepted, the order material amount being normal, and the asset category being within the fixed asset directory are met at the same time, the transfer fund operation is performed;

[0191] The transfer capital necessary conditions are regularized.

[0192] More specifically, in the present embodiment, the "internal inspection" is the only trigger node, and after the internal inspection, the system automatically generates transfer capital tasks for all orders under the project that meet the transfer capital qualifications.

[0193] The transfer capital necessary conditions are set as: the project has been internally inspected, the order material amount is normal, and the asset category is within the fixed asset directory, which meets the necessary conditions to trigger the transfer capital.

[0194] It can be understood that the above-mentioned transfer capital necessary conditions need to be met at the same time to trigger the transfer capital execution process.

[0195] As shown in Figure 2 , the condition check is performed on the project that has completed the internal inspection, and when the condition check result is that the project has been internally inspected, further judgment is made in sequence whether the amount is normal and whether the asset category is in the directory. If yes, the transfer capital tasks are batch generated, and if no, the process is terminated.

[0196] After batch generating the transfer capital tasks, the rule engine is used again for verification, and if all the batch-generated transfer capital tasks pass the verification, the transfer capital process is executed, the asset card is created, the depreciation is started, and the transfer capital is completed; otherwise, manual review is performed.

[0197] In the present embodiment, the self-defined rule syntax of "natural language-like + keyword" is adopted, which takes into account the business readability and machine parsability, realizes the formalization of the trigger conditions, and is composed of three segments of "trigger condition (IF) → decision action (THEN) → authoritative basis (FROM)", which can be directly mapped to SWRL (rule language) / Drools (rule engine) statements, for example:

[0198] IF (project inspection status = "inspected");

[0199] THEN project transfer capital qualification = "compliant".

[0200] All rules are stored in the knowledge graph in the form of JSON-LD, support version number and release timestamp, and ensure that after the regulatory file is updated, the hot plug can be effective within 24 hours.

[0201] (4) Intelligent transfer capital decision.

[0202] 1) Based on the matching relationship between the candidate order information and the nodes in the asset knowledge graph, the asset category and asset amount in the candidate order are determined, which specifically includes:

[0203] The candidate order information includes asset number, asset name, contract number, invoice number and project number;

[0204] Based on the cosine similarity, the matching relationship of the candidate order information and the nodes in the asset knowledge graph is calculated, and the candidate order information is matched and associated with the nodes in the asset knowledge graph;

[0205] According to the priority order of asset number, asset name, contract number and invoice number, the matching and association are performed in sequence, and when the high-priority matching fails, the low-priority matching is performed;

[0206] After the matching and association are completed, the matching and association results are obtained, and the asset category and asset amount in the candidate order are extracted from the matching and association results.

[0207] 2) Based on the matching relationship of the asset category of the candidate order and the nodes in the asset conversion rule knowledge graph, the main equipment corresponding to the asset category of the candidate order is determined, which specifically includes:

[0208] Based on the project number in the candidate order information, the project entity in the asset conversion rule knowledge graph is matched to obtain the project entity matched with the candidate order;

[0209] Based on the matched project entity, the main equipment corresponding to the matched project entity is obtained;

[0210] The asset category of the candidate order is matched with the assembly rule entity in the asset conversion rule knowledge graph to obtain the assembly rule matched with the asset category of the candidate order;

[0211] Based on the assembly rule matched with the asset category of the candidate order, the main equipment to which the asset corresponding to the asset number of the candidate order is hung is determined from the main equipment corresponding to the project entity matched with the candidate order.

[0212] It can be understood that after the above steps are completed, in order to solve the explainability problem of the dynamic allocation logic of auxiliary materials to main equipment, the depreciation rules and regulations and provisions associated with the candidate order information in the asset knowledge graph and the assembly rules in the asset conversion rule knowledge graph that match the candidate order information can also be presented to the user.

[0213] For example:

[0214] Project number: P20240501, main equipment (converted) 0101-main transformer 1

[0215] Auxiliary materials / service (not allocated): M001 transformer oil 120000 yuan, M002 installation fee 50000 yuan

[0216] Objective: Assemble 170,000 yuan of value to the main equipment and generate assembly records.

[0217] The above example means that the assets classified as M001 transformer oil 120000 yuan and M002 installation fee 50000 yuan are allocated, and the final result is that they are allocated to the main transformer with project number P20240501 and main equipment (converted capital) number 0101.

[0218] It can be understood that M001 and M002 above are asset numbers of transformer oil and installation fee; 120000 yuan is the asset amount of M001 transformer oil, 50000 yuan is the asset amount of M002 installation fee, and the total asset amount to be allocated is 120000 yuan + 50000 yuan = 170000 yuan.

[0219] The execution logic of the whole intelligent capital conversion is as follows:

[0220] Step 1: Obtain the candidate order and verify whether the candidate order meets the necessary conditions for capital conversion, the candidate order information including asset number, asset name, contract number, invoice number and project number;

[0221] Step 2: If the necessary conditions for capital conversion are met, the candidate order is queried and reasoned in the asset knowledge graph to query the associated entities (contracts, invoices, acceptance forms, etc.). In specific implementation, the matching and association can be performed in order according to the priority order of asset number, asset name, contract number and invoice number. After the matching and association of high priority fails, the matching and association of low priority are performed. After the matching and association ends, the asset category and asset amount of the asset corresponding to the asset number in the candidate order are obtained from the matching and association result.

[0222] The purpose of the above steps is twofold: first, the asset category in the candidate order is matched from the asset knowledge graph, so as to match the assembly rule in the capital conversion rule knowledge graph in the next step, and then determine which main equipment the asset in the candidate order specifically relies on. Second, the asset amount in the candidate order is matched from the asset knowledge graph, so as to allocate the asset amount to the main equipment after determining the main equipment in the next step in the capital conversion rule knowledge graph.

[0223] Step 3: In order to determine the allocated main equipment, first, the project entity in the capital conversion rule knowledge graph is matched based on the project number in the candidate order, so as to filter out the project entity related to the project number;

[0224] Then, based on the matched project entity, the main equipment in the capital conversion rule knowledge graph is matched again, and the main equipment associated with the matched project entity is filtered out;

[0225] Determine the matching degree of the asset category of the candidate order and the assembly rule entity in the financing rule knowledge graph based on the assembly rule entity in the financing rule knowledge graph, and determine the assembly rule hit by the asset category of the candidate order.

[0226] Finally, based on the assembly rule hit by the asset category of the candidate order, the main equipment of the candidate order is filtered out from the main equipment corresponding to the project entity.

[0227] The execution process of a virtual candidate order for financing is described below:

[0228] (1) The ERP completion settlement process is completed, and the project order is pushed through KAFKA.

[0229] (2) After receiving, the listener does not perform any rule calculation, and directly stores the database. The order table adds a record, and the state is NEW (indicating a new order).

[0230] (3) Through a timing task, the order table state equal to NEW data is scanned every 30 seconds, and the database row is converted into a Java fact that can be recognized by Drools, representing the project, main equipment, and auxiliary material cost to be allocated.

[0231] (4) Create KieSession (session) and insert facts and execute the rule engine;

[0232] (5) Collect rule results, write back to the database and update the order state.

[0233] At this point, the assembly logic of an order is executed, and the assembly result is obtained.

[0234] The assembly result is synchronized to the asset, the assembly result is summarized; the amount is formed into asset value-added according to the unique value of the equipment, and the depreciation is recalculated according to the category of the asset. If it is the first financing, the related properties are obtained according to the material, the depreciation is calculated, and the asset card is formed.

[0235] (5) Automatic financing execution.

[0236] After the asset amount of the candidate order is allocated to the main equipment, the asset of the main equipment is correspondingly valued, the original value of the asset of the main equipment is adjusted, and a new asset card is formed;

[0237] Record the time, operator and basis documents of the whole financing process to generate tamper-proof data evidence.

[0238] The present embodiment is connected with the ERP and financial system through API, and the above interface is called to automatically complete the following operations: creating a fixed asset card or adjusting the asset original value in the financial system.

[0239] This embodiment records the whole process of capital transfer (time, operator, and according to the documents), generates tamper-proof data evidence, and realizes audit tracking.

[0240] (Six) Feedback and optimization.

[0241] 1. Abnormal learning:

[0242] The abnormal cases reviewed by artificial review are fed back to the knowledge graph, and the knowledge graph is updated.

[0243] 2. Model iteration:

[0244] Periodically use historical capital transfer data to train GNN model to improve the accuracy of rule matching.

[0245] As shown in Figure 4 , the method of the embodiment surrounds the four-step closed loop of "project internal inspection trigger→ unified semantic of knowledge graph→ dynamic arrangement of rule engine→ automatic writing of assets", and realizes intelligent capital transfer of fixed assets:

[0246] First, once the project internal inspection is triggered, the system automatically pulls together five types of "capital transfer key source data", corresponding to the data collection layer in Figure 4 :

[0247] First, the ERP (project, procurement, inventory, and fixed assets four modules) related data is real-time grabbed through RFC / BAPI or API;

[0248] Second, the bill information parsed by the contract system through OCR-NER;

[0249] Third, the internal acceptance conclusion and signed PDF pushed by the OA process;

[0250] Fourth, the invoice and arrival receipt single image of the image system, which can be obtained by screen grabbing through RPA robot;

[0251] Fifth, the device information reported by IoT.

[0252] After the multi-source fields obtained are mapped to unified asset master data by semantic model, the differences in caliber are eliminated, the asset knowledge graph is constructed, and the knowledge graph layer in Figure 4 corresponds.

[0253] Second, configure the three-layer hot plug rules of "necessary conditions-classification rules-exception exemptions" in the Drools+ self-developed DSL rule engine, complete the compliance judgment and capital transfer qualification score of 10,000 assets within 30 seconds, and the rule engine layer in Figure 4 corresponds.

[0254] Third step, intelligent transfer matching link (see the aforementioned intelligent transfer decision step), after matching, enter the execution layer, execute the transfer task, realize the update of fixed asset card, depreciation start and perfect the blockchain audit chain.

[0255] Fourth step, the system adopts containerized deployment, supports elastic expansion, rule update zero code, and financial personnel can maintain through the drag-and-drop interface. The time consumption of single transfer is reduced from days to seconds, meeting the modern asset management demand of high concurrency, high compliance and high traceability.

[0256] The embodiment significantly improves the construction efficiency of the fixed asset transfer scene data set, the hierarchical structure of the fixed asset transfer data set, and the coverage breadth and detail depth of the fixed asset transfer data set. Compared with the prior art, the embodiment of the present application realizes:

[0257] (1) Efficiency improvement: the transfer processing time is shortened from "day" level to "minute" level, supporting batch processing;

[0258] (2) Accuracy improvement: through the knowledge graph and rule engine, the error rate is reduced by more than 90%; compliance guarantee:

[0259] (3) Real-time synchronization of regulations and provisions, ensuring 100% compliance;

[0260] (4) Traceability: full-process recording, supporting audit and accountability.

[0261] Embodiment two

[0262] The embodiment discloses a fixed asset intelligent transfer system.

[0263] The fixed asset intelligent transfer system comprises:

[0264] The asset knowledge graph construction module is configured to construct an asset knowledge graph by using a multi-source data integration method;

[0265] The transfer rule knowledge graph construction module is configured to construct a transfer rule knowledge graph considering the asset category and the accounting subject;

[0266] The trigger module is configured to set the necessary conditions for triggering the transfer operation, and when the necessary conditions for triggering the transfer operation are met, the candidate order information to be transferred is obtained;

[0267] The first matching module is configured to determine the asset category and the asset amount in the candidate order based on the matching relationship between the candidate order information and the nodes in the asset knowledge graph;

[0268] The second matching module is configured to determine the main device to which the assets of the candidate order are affiliated based on a matching relationship between the asset category of the candidate order and the nodes in the asset transfer rule knowledge graph.

[0269] The apportionment module is configured to apportion the asset amount of the candidate order to the main device to which the corresponding assets are affiliated, thereby completing the asset transfer process.

[0270] Embodiment Three

[0271] An object of the present embodiment is to provide an electronic device.

[0272] The electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor, and the processor implements the steps in the fixed asset intelligent asset transfer method according to the embodiment 1 of the present disclosure when executing the program.

[0273] The steps and methods involved in the above embodiments two and three correspond to the embodiment one, and the specific implementation can refer to the related description part of the embodiment one. The term "computer readable storage medium" should be understood as including a single medium or multiple media of one or more instruction sets; it should also be understood as including any medium capable of storing, numbering, or carrying an instruction set for execution by a processor and causing the processor to execute any method in the present application.

[0274] Those skilled in the art should understand that the above-mentioned modules or steps of the present application can be realized by a general computer device, alternatively, they can be realized by program codes executable by a computing device, so that they can be stored in a storage device and executed by a computing device, or they can be made into individual integrated circuit modules, or a plurality of modules or steps among them can be made into a single integrated circuit module. The present application is not limited to any specific combination of hardware and software.

[0275] The above describes the specific embodiments of the present application in combination with the accompanying drawings, but is not a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications or variations made by those skilled in the art on the basis of the technical solutions of the present application without creative labor are still within the protection scope of the present application.

Claims

1. A method for intelligent transfer of fixed assets, characterized in that, Includes the following steps: An asset knowledge graph is constructed using a multi-source data integration method. Considering both asset categories and accounting items, a knowledge graph of capital transfer rules is constructed. Set the necessary conditions for triggering the capital transfer operation. When the necessary conditions for capital transfer are met, obtain the candidate order information to be transferred. Based on the matching relationship between candidate order information and nodes in the asset knowledge graph, the asset category and asset amount in the candidate order are determined. The candidate order information includes asset number, asset name, contract number, invoice number, and project number. Based on cosine similarity, the matching relationship between the candidate order information and nodes in the asset knowledge graph is calculated, and the candidate order information is matched and associated with the nodes in the asset knowledge graph. Matching and association are performed sequentially according to the priority order of asset number, asset name, contract number, and invoice number. When a high-priority match fails, a low-priority match is then performed. After the matching and association are completed, the matching and association results are obtained, and the asset category and asset amount in the candidate order are extracted from the matching and association results. Based on the matching relationship between the asset category of the candidate order and the nodes in the knowledge graph of transfer rules, the main equipment to which the asset of the candidate order is attached is determined; based on the project number in the candidate order information, it is matched with the project entity in the knowledge graph of transfer rules to obtain the project entity that matches the candidate order; based on the matched project entity, the main equipment of the corresponding matched project entity is obtained; the asset category of the candidate order is matched with the assembly rule entity in the knowledge graph of transfer rules to obtain the assembly rule that matches the asset category of the candidate order; based on the assembly rule that matches the asset category of the candidate order, the main equipment to which the asset corresponding to the asset number of the candidate order is attached is determined from the main equipment of the corresponding project entity that matches the candidate order. The asset amount of the candidate order is allocated to the main equipment to which the corresponding asset is attached, thus completing the asset transfer process.

2. The fixed asset intelligent transfer method as described in claim 1, characterized in that, An asset knowledge graph is constructed using a multi-source data integration method, specifically including: Extract asset-related data from multi-source data and perform preprocessing; Define the entity types of the asset knowledge graph, including asset number, contract, supplier, invoice, acceptance form, accounting subject, asset category, depreciation rules and regulations; Define the relationship types of the asset knowledge graph, including the relationship between asset number and contract signing, the correspondence between asset number and invoice, the association between asset number and acceptance form, the relationship between asset number and accounting subject, the relationship between asset number and asset category, the relationship between accounting subject and depreciation rule application, the relationship between accounting subject and regulatory constraints, and the relationship between contract and supplier. The extracted asset-related data is mapped to the entity types of the asset knowledge graph, and the relationships between the obtained entities are matched to build the asset knowledge graph. Perform verification and dynamic updates of the asset knowledge graph.

3. The fixed asset intelligent transfer method as described in claim 2, characterized in that, Asset-related data is extracted from multi-source data and preprocessed, specifically including: Obtain multi-source fields related to assets from multi-source data including ERP, contract management system, OA system, imaging system, and IoT: Real-time capture asset card data from ERP, obtain parsed bill information from the contract management system, obtain internal acceptance conclusions and signature data from the OA system, obtain invoice and arrival receipt imaging data from the imaging system, and obtain device information from IoT; Clean and standardize the multi-source fields; Perform entity disambiguation using a globally unique key, which includes contract number, asset number, and invoice number; Perform two-way verification on incremental data by combining timestamp and primary key.

4. The fixed asset intelligent transfer method as described in claim 1, characterized in that, Considering the two dimensions of asset category and accounting subject, construct a capital transfer rule knowledge graph, specifically including: Define the entity types of the capital transfer rule knowledge graph, including project, main equipment, auxiliary materials, assembly rules, and allocation instances; Define the relationship types of the capital transfer rule knowledge graph, including the inclusion relationship between project and main equipment, the inclusion relationship between project and auxiliary materials, the matching relationship between auxiliary materials and assembly rules, the attachment relationship between auxiliary materials and main equipment, and the source / target relationship between allocation instances and auxiliary materials / main equipment; Obtain multiple project information, map the multiple project information to the entity types of the defined capital transfer rule knowledge graph, and match the relationships between entities to establish a capital transfer rule knowledge graph.

5. The fixed asset intelligent transfer method as described in claim 1, characterized in that, Set the necessary conditions for capital transfer operations to trigger, specifically including: Set the necessary conditions for capital transfer operations, including that the project has been internally accepted, the order material amount is normal, and the asset category is within the fixed asset catalog; when the three conditions of the project being internally accepted, the order material amount being normal, and the asset category being within the fixed asset catalog are all met, perform the capital transfer operation; Regularize the necessary conditions for capital transfer.

6. The fixed asset intelligent transfer method as described in claim 1, characterized in that, Also include: After the asset amount of the candidate order is allocated to the main equipment, perform corresponding appreciation on the assets of the main equipment, adjust the original value of the assets of the main equipment, and form a new asset card; Record the time, operator, and basis documents of the entire process of capital transfer, and generate an immutable data evidence.

7. A fixed asset intelligent transfer system, characterized in that, Include: An asset knowledge graph construction module, configured to: construct an asset knowledge graph using a multi-source data integration method; A capital transfer rule knowledge graph construction module, configured to: considering the two dimensions of asset category and accounting subject, construct a capital transfer rule knowledge graph; A trigger module, configured to: set the necessary conditions for capital transfer operations to trigger, and when the necessary conditions for capital transfer are met, obtain the candidate order information to be transferred for capital. The first matching module is configured to: determine the asset category and asset amount in the candidate order based on the matching relationship between the candidate order information and nodes in the asset knowledge graph; the candidate order information includes asset number, asset name, contract number, invoice number, and project number; calculate the matching relationship between the candidate order information and nodes in the asset knowledge graph based on cosine similarity, and match and associate the candidate order information with the nodes in the asset knowledge graph; perform matching and association sequentially according to the priority order of asset number, asset name, contract number, and invoice number, and perform low-priority matching only when high-priority matching fails; after the matching and association are completed, obtain the matching and association results, and extract the asset category and asset amount in the candidate order from the matching and association results; The second matching module is configured to: determine the main equipment to which the assets of the candidate order are attached based on the matching relationship between the asset category of the candidate order and the nodes in the knowledge graph of transfer rules; match the project number in the candidate order information with the project entity in the knowledge graph of transfer rules to obtain the project entity that matches the candidate order; obtain the main equipment of the corresponding matching project entity based on the matching project entity; match the asset category of the candidate order with the assembly rule entity in the knowledge graph of transfer rules to obtain the assembly rule that matches the asset category of the candidate order; and determine the main equipment to which the asset corresponding to the asset number of the candidate order is attached from the main equipment of the corresponding project entity that matches the candidate order based on the assembly rule that matches the asset category of the candidate order. The allocation module is configured to allocate the asset amount of candidate orders to the main equipment to which the corresponding assets are attached, thereby completing the asset transfer process.

8. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the fixed asset intelligent transfer method as described in any one of claims 1-6.

Images