Enterprise product pricing system and method based on hybrid multiple models

Abstract

The invention provides an enterprise product pricing system and method based on hybrid multiple models, the pricing system comprises a support layer, a data layer, a model layer, a decision layer and an application layer, the support layer deploys a server and a storage device to provide underlying resource support for the system; the data layer completes data acquisition and preprocessing, deploys a distributed database, divides the distributed database into an original database module, a process database module, an achievement database module and a rule database module, stores data in a classified manner, deploys a block chain system and stores the data in a chaining manner; the model layer designs 11 product pricing models, calls data calculation of the data layer to output results respectively, and outputs final prices of products after weighted fusion by using a machine learning algorithm; the decision-making layer generates a product price report and formulates a final product price strategy; and the application layer displays the final pricing result for the user. According to the invention, flexible pricing decision is realized through enriching pricing models, the process is transparent and efficient, and price maximum rationalization of enterprise products under various market environment requirements is realized.

Description

Technical Field

[0001] The present invention relates to artificial intelligence, machine learning, distributed databases, blockchain, smart contracts, natural language processing, and big data analysis technologies, and is applied to the field of enterprise product pricing. It is an enterprise product pricing system and method based on a hybrid multi-model. Background Art

[0002] Existing enterprise product pricing solutions usually adopt static and single processing flows, and cannot effectively integrate data from multiple channels and systems, resulting in a slow response to market dynamics during product pricing, and obvious lags and inconsistencies in overall data processing. At the same time, the centralized data storage method adopted by enterprises is inefficient in the face of large-scale and diverse data, and cannot achieve the high efficiency and scalability required for data processing, thereby leading to data islands and security risks. In addition, due to the lack of automated data verification and full-process traceability mechanisms, the authenticity and compliance of data cannot be fully guaranteed, bringing risks to the enterprise decision-making process. Moreover, traditional pricing methods fail to achieve in-depth mining and real-time feedback of complex market behaviors, making pricing decisions always stay at the superficial statistical and simple regression stages, without intelligent adaptive capabilities, and it is difficult to meet the rapidly changing market demands. Therefore, overall, it cannot provide accurate, timely, and credible pricing support. To sum up, due to the lack of real-time integration, automated verification, and intelligent adaptive mechanisms for multi-source heterogeneous data, traditional solutions lead to lagging information updates, serious data islands, and frequent security risks, resulting in inaccurate pricing decisions and difficulty in coping with rapidly changing market demands. Summary of the Invention

[0003] The objective of the present invention is to propose an enterprise product pricing system and method based on a hybrid multi-model to solve the problems raised in the above background art.

[0004] On the one hand, the present invention proposes an enterprise product pricing system based on a hybrid multi-model. The system structure is divided into five layers hierarchically, namely the support layer, the data layer, the model layer, the decision layer, and the application layer.

[0005] Support layer: Deploy servers and storage devices;

[0006] Data layer: Data collection, preprocessing, and storage;

[0007] Model layer: Output after weighted fusion of the results of multiple price calculation models;

[0008] Decision layer: Generate product price reports and charts, and formulate the final product price strategy;

[0009] Application layer: Provide a user interface for interaction with users and display the results.

[0010] Furthermore, the support layer deploys servers and storage devices to provide computing resources, storage resources, and network resources for the system, and uses load balancing and fault-tolerant backup technical means to maintain the stable availability of the system.

[0011] Furthermore, for the data requirements of each enterprise product pricing model in the system, data collection and entry are carried out in the data layer by means of automated data collection, manual data entry, and expert research.

[0012] Furthermore, the data preprocessing operations in the data layer include: data deduplication, outlier detection and handling, missing value handling, data consistency checking, format conversion and standardization, redundant field deletion, text data cleaning, data verification and compliance checking, and data integration.

[0013] Furthermore, the data in the data layer is stored in a distributed database, which is divided into four modules: the original database, the process database, the achievement database, and the rule database.

[0014] Furthermore, 11 independent enterprise product price calculation models are deployed in the model layer, and a machine learning model is trained. The output results of each pricing model are fused by means of weighted averaging to formulate a final product pricing result for different demand scenarios.

[0015] Furthermore, product price reports and charts are generated in the decision-making layer, and the final price strategy is formulated.

[0016] Furthermore, the application layer is developed using Web, AJAX, BI, and front-end framework technologies to provide a user operation interface and display the pricing results of products in multiple scenarios to users.

[0017] Furthermore, the above system includes the following functions and contents:

[0018] The support layer, which is the underlying cornerstone of the entire system, is mainly responsible for providing computing resources, storage resources, and network resources. This layer uses high-performance servers and storage devices to ensure the efficient operation of the system and data security; in order to cope with possible sudden traffic and failure situations, load balancing and fault-tolerant backup technical means are used to improve the stability and availability of the system, providing a solid and reliable underlying support for the entire system.

[0019] The data layer. Furthermore, for the data required by each model in the system, data collection is carried out by means of automated data collection, manual data entry, and expert research, and the data update frequency is set.

[0020] Further, for the data preprocessing operation, after collecting data, data deduplication, outlier detection and handling, missing value handling, data consistency checking, format conversion and standardization, redundant field deletion, text data cleaning, data verification and compliance checking, and data integration operations are carried out.

[0021] Further, for the data storage operation, this system adopts a distributed database and stores data in four database modules respectively: the raw database, the process database, the achievement database, and the rule database. Each module has its independent and clear function, and the entire system realizes data flow and collaborative work among modules through the API gateway and the message bus.

[0022] Further, the raw database is the foundation of the entire system and bears all the raw data collected from the outside. This database is designed through the distributed storage technology Hadoop to handle massive heterogeneous data.

[0023] Preferably, before the data flows into the raw database, it first goes through the preprocessing stage.

[0024] Preferably, the raw database design includes different types of data tables, such as market data tables: recording raw material prices, competitor prices, market demand fluctuations, etc.; industry data tables: industry gross profit margins, net profit margins, typical processing costs, etc.; historical data tables: historical costs, past sales data, production data, etc.; real-time data tables: real-time price fluctuations, market news, supply chain events, etc. The data storage format adopts the Parquet columnar storage format, which supports efficient data query and compression.

[0025] Preferably, the data access interface provides a standard RESTAPI, allowing other modules and external systems to obtain raw data; for real-time updates, the message queue RabbitMQ mechanism is adopted to transmit external data flows to the raw database in real time, and the data is instantaneously processed and stored through stream processing.

[0026] Preferably, the original database module deploys a blockchain system, which uses blockchain technology to record the hash values, sources, and generation times of important price node data items for each product, ensuring data immutability and transparency. It deploys smart contracts to automatically verify the legality, integrity, and consistency of data sources, and only the verified data can be recorded on the blockchain, thus ensuring the reliability of the original data. Utilizing the distributed characteristics of the blockchain, it provides an efficient query engine for the original database. By recording hot data or frequently queried data in the blockchain and using the distributed query architecture IPFS, it accelerates data access and query efficiency. An access control module for each user or system is established in the blockchain, and all data access requests are automatically verified through smart contracts, ensuring that only authorized users can access sensitive data. All query operations are recorded in the blockchain, ensuring the traceability and transparency of operations, thereby enabling different system users to share data through the blockchain and ensuring the security and consistency during data transmission.

[0027] Furthermore, for the process database design, the process database stores the data obtained by processing and transforming the original data through the pricing large model algorithm. According to the nature and use of the data, the data is classified and archived, and the milestone data generated stage by stage forms the process database, facilitating subsequent data retrieval and analysis.

[0028] Preferably, the process database stores data by stage. During the data generation process, every time a milestone task is completed, its data is stored as an independent stage dataset: a process stage table is constructed to record the key data and results of each process node, such as the output of the price calculation model, the data processed by the AI algorithm, the stage cost analysis, etc.; a data classification table is constructed to classify and archive the data according to its nature, such as "market data", "cost data", "product pricing data", etc.

[0029] Preferably, the processed data in the process database is stored in both a distributed database and the blockchain. Enterprises can query all the data in the process database through smart contracts or API interfaces, and at the same time can access each data change record on the blockchain, ensuring complete transparency when querying data.

[0030] Furthermore, for the result database design, the result database integrates all the data results processed by the system, such as cost analysis, price prediction, optimization suggestions, etc. These data are stored in the database in a structured form, facilitating subsequent query and analysis. Through the result database, the results of system analysis can be intuitively viewed, such as the cost situation after price strategy adjustment, etc. The data and analysis results provided by the result database can support enterprise decision-making. For example, based on the results of price prediction, enterprises can formulate reasonable price strategies to maximize profits.

[0031] Preferably, an analysis report form is constructed in the achievement database module to store various pricing analysis and market prediction reports, including cost changes, expected profits, market impacts, etc. after pricing strategy adjustments; an optimization suggestion form is constructed, and the system generates pricing optimization suggestions based on the analysis of the large model to help enterprise decision-makers understand the best pricing plan.

[0032] Preferably, through a BI tool or a customized data analysis platform, visual reports are generated in this storage layer, and enterprises can comprehensively view the data in the achievement database, facilitating enterprise data analysis.

[0033] Preferably, the core verification information of each achievement data (such as data ID, hash value, version, timestamp, etc.) is stored on the blockchain to ensure non-tampering and traceability; the complete achievement data is stored in an off-chain database, which contains a large amount of detailed information and is stored after encryption to ensure data security and privacy protection. Each update, adjustment, and decision-making process of the achievement data will generate a new block to record the relevant modification information, and all data operations will be transparently recorded on the blockchain, enabling managers, auditors, and relevant decision-makers to view the processing process of the achievement data in real time and increasing data transparency.

[0034] Furthermore, in the design of the rule database, the rule database is a key component to ensure that the entire system can make reasonable and compliant decisions when conducting price and cost control.

[0035] Preferably, various verification rules, compliance discrimination rules, rationality evaluation rules, sufficiency confirmation rules, etc. are stored in the rule base data. Each rule will include the type of the rule, applicable scenarios, input and output, etc.; a version number is assigned to each rule to ensure version control of the rules, facilitating system updates and backtracking.

[0036] Preferably, a rule association table is designed to record the relationships between rules, data sets, and analysis modules, ensuring a clear application scope of the rule base and facilitating calls when processing data.

[0037] Preferably, a blockchain system is deployed, and all rules in the rule base, such as "authenticity verification rules", "compliance discrimination rules", "rationality evaluation rules", etc., are embedded in the blockchain in the form of smart contracts to ensure rule transparency and automated execution; all change histories of the rules are recorded through the blockchain to generate audit logs, ensuring the transparency and compliance of rule changes. When new pricing rules are introduced, all relevant nodes and systems can be updated and executed automatically in real time.

[0038] In the model layer, 11 independent price calculation models are set, and the weighted fusion of the calculation results of multiple models is output to obtain the final product pricing.

[0039] Furthermore, the model layer includes a social price model for product cost element prediction, factor analysis, and benchmarking management.

[0040] Furthermore, the model layer includes an industry price model for calculating the target cost of products and process cost benchmarking analysis.

[0041] Furthermore, the model layer includes a competitor price model for calculating the target cost of the target product, process cost benchmarking analysis, and price analysis and decision-making.

[0042] Furthermore, the model layer includes a historical cost model for monitoring the product cost process, target cost management, standard cost maintenance, price analysis, and evaluation.

[0043] Furthermore, the model layer includes a base period economic indicator model for product cost accounting, target cost management, price analysis, and performance evaluation.

[0044] Furthermore, the model layer includes an empirical statistical model for estimating the target price of products, profit forecasting, comprehensive project evaluation, and price analysis and decision-making.

[0045] Furthermore, the model layer includes a technical and economic model for estimating the target price of enterprise products, making decisions on research and development (including improvement, upgrade, and customized development) project establishment, monitoring process costs, and evaluating the cost-effectiveness of project construction and the contribution rate of the system.

[0046] Furthermore, the model layer includes a life cycle cost model for estimating the life cycle cost of enterprise products, calculating and analyzing after-sales service costs, and making project establishment decisions.

[0047] Furthermore, the model layer includes a labor cost model for calculating the cost of enterprise products, evaluating personnel performance, and conducting business analysis and decision-making.

[0048] Furthermore, the model layer includes a standard cost model for making pricing decisions for enterprise products, target cost management, and business analysis and decision-making.

[0049] Furthermore, the model layer includes a rule model for data processing, data auditing, data analysis, and evaluation.

[0050] Furthermore, train a machine learning model, input the output results of each pricing model as features, as well as the actual market performance and sales data, so that the model can learn which mixed output of the pricing models performs better in specific environments with different demands, and fuse the output results of each pricing model through weighted average to formulate a final pricing result for different demand scenarios.

[0051] Preferably, the system dynamically adjusts the product pricing strategy according to different market environments and application scenarios.

[0052] Furthermore, the data required for each independent price calculation model is stored in the original database for calling; the calculation results generated by each model are stored in the process database for calling; the mixed weighted output results of each model for different scenarios are stored in the achievement database for use; and the verification data such as relevant industry rules is stored in the rule database for use.

[0053] The decision-making layer analyzes the product price, generates detailed reports and charts, and formulates the final price strategy.

[0054] The application layer is developed using technologies such as Web, AJAX, BI, and front-end frameworks, provides an intuitive and easy-to-use operation interface for users, interacts with users and displays the results. Users can obtain the pricing information of the enterprise product full solution in multiple scenarios in real time at this layer and make the final pricing decision for the enterprise product in different scenarios.

[0055] In another aspect of the present invention, the present invention provides the following technical solution: a method for pricing enterprise products based on a hybrid multi-model, the method comprising the following steps:

[0056] S1. Deploy servers and storage devices to provide the required computing resources, storage resources, and network resources for the system.

[0057] S2. Collect the data required by the system model, perform data preprocessing, and store it in the database.

[0058] S3. Set multiple price calculation models, and perform weighted fusion on the calculation results of multiple models and then output.

[0059] S4. Analyze the product price, generate reports and charts, and formulate the final price strategy.

[0060] S5. Provide a user interface to display the pricing results for users.

[0061] According to the above content, the present invention has the following advantages and beneficial effects:

[0062] This system can intelligently integrate the output results of 11 enterprise product pricing models, and dynamically adjust the output weights of each pricing model according to different market changes, competition situations, and production environments to ensure the high efficiency and accuracy of the final product pricing. Through weighted fusion by machine learning algorithms, the system flexibly adjusts the influence of each price model according to different market demands and different stages of the product life cycle, ensuring that the final output product pricing is well-founded and reasonable, and maximizing the enterprise's profit. At the same time, the system introduces blockchain technology to ensure the transparency and immutability of all data processing processes, enabling each pricing decision to be traced in real time, greatly enhancing the credibility and compliance of the data, and being particularly applicable to industries that require high auditing and supervision. The system adopts a distributed database design, which can efficiently process a large amount of pricing data, ensuring the efficiency and reliability of data storage, query, and update. This design not only avoids the single-point failure problem but also can automatically expand according to the load situation, improving the fault tolerance and performance of the system. Generally speaking, through flexible pricing decisions, transparent operation processes, and efficient and reliable database support, this system realizes the maximum rationalization of the prices of enterprise products under various market environment demands, helps enterprises maintain competitiveness in long-term operations, and at the same time ensures that enterprises achieve the maximum profit target. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is the system architecture diagram of the present invention.

[0064] Figure 2 It is the flowchart of the method steps of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0065] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0066] In one or more embodiments, the present invention provides an enterprise product pricing system based on a hybrid multi-model as shown in Figure 1 The system structure is divided into five layers hierarchically, namely the support layer, the data layer, the model layer, the decision layer, and the application layer.

[0067] In this embodiment, the support layer is the underlying cornerstone of the entire system and is mainly responsible for providing computing resources, storage resources, and network resources. This layer uses high-performance servers and storage devices to ensure the efficient operation of the system and data security. To cope with possible sudden traffic and failure situations, load balancing and fault tolerance backup technical means are adopted to improve the stability and availability of the system, providing a solid and reliable underlying support for the entire system.

[0068] In this embodiment, for the data required by each model in the system, the data layer collects data by means of automated data collection, manual data entry, and expert research:

[0069] Automated data collection: Connect through API interfaces to directly obtain public data from the National Bureau of Statistics, local statistical bureaus, industry associations, major procurement platforms, etc. through standard APIs; for web crawlers, for some data sources without API support, such as website announcements and public reports, develop customized crawler programs to regularly crawl and parse data on the premise of compliance with website laws and regulations; support data formats including JSON, XML, CSV, and Excel formats to ensure the standardization of data structures.

[0070] Manual data entry and expert research: Manually enter or batch import data from the enterprise's internal financial system, R & D system, production management system, etc.; design an online questionnaire and research interview record system to collect data such as expert evaluations, experience data, and competitor price information and enter it into the system.

[0071] Set the data update frequency: Set real-time data such as raw material prices and competitor prices to be updated daily; set periodic data such as enterprise financial indicators and industry gross profit margins to be updated quarterly; set long-term stable data such as regulations and policies and internal rules to be updated when policies are adjusted.

[0072] In this embodiment, the data preprocessing operation preferably includes the following steps:

[0073] Data deduplication: The system checks for identical entries in different data sources, identifies and removes those duplicate records. For some fields with duplicates, use unique identifiers (such as product IDs, transaction numbers, etc.) for deduplication to ensure that each piece of data appears only once;

[0074] Outlier detection and handling: Use a rule engine to set a reasonable threshold range based on industry experience and rules. Data beyond the threshold is marked as an outlier, and the system selects processing methods such as removing, correcting (filling with adjacent values), replacing with the median or mean for outliers, and the specific processing method depends on the nature of the data;

[0075] Missing value handling: Predict missing values through the regression model of existing data. For more complex data relationships, use multiple imputation or filling methods based on machine learning;

[0076] Data Consistency Check: Ensure that the data between the same field or different fields is consistent in the same dimension. Check the correlation between different fields through rules or logical relationships. For example, ensure that the purchase amount is equal to the product of the purchase quantity and the unit price. If inconsistent, mark it as incorrect data for subsequent processing.

[0077] Format Conversion and Standardization: For data in different ranges, such as price, weight, area, etc., perform normalization processing to make them comparable and avoid the improper influence of certain large dimensional differences on the analysis results.

[0078] Redundant Field Deletion: Based on business requirements and analysis goals, delete redundant fields that are not helpful for data analysis or irrelevant to the goals. For example, when certain ID fields appear repeatedly in multiple data tables but have no impact on the analysis, merge or delete them.

[0079] Text Data Cleaning: Clean and format data from different text fields (such as product descriptions, user comments, etc.).

[0080] Data Verification and Compliance Check: According to preset rules, check whether the data meets relevant national or industry standards, such as whether the price conforms to the specified price range, whether the product meets the quality standards, etc.; verify the legality of the data to avoid the system generating illegal or non-compliant data and ensure its legal use in subsequent analysis.

[0081] Data Integration: Integrate data from different sources based on the primary key (such as product ID, order ID, etc.). Eliminate duplicates through a deduplication algorithm and retain only unique valid records to ensure the accuracy and consistency of the data.

[0082] In this embodiment, for the data storage operation, the system adopts a distributed database and stores the data in four database modules respectively: the original database, the process database, the result database, and the rule database. Each module has its independent and clear function, and the entire system will achieve data flow and collaborative work between modules through the API gateway and the message bus.

[0083] It should be further noted that in this embodiment, the original database is the foundation of the entire system and bears all the original data collected from the outside. This database is designed through the distributed storage technology Hadoop to handle massive heterogeneous data.

[0084] Before the data flows into the original database, it first goes through a preprocessing stage, and the steps include: data deduplication, outlier detection and processing, missing value processing, data consistency check, format conversion and standardization, redundant field deletion, text data cleaning, data verification and compliance check, data integration and deduplication.

[0085] The original database design includes different types of data tables, such as market data tables: recording raw material prices, competitor prices, market demand fluctuations, etc.; industry data tables: industry gross profit margins, net profit margins, typical processing costs, etc.; historical data tables: historical costs, past sales data, production data, etc.; real-time data tables: real-time price fluctuations, market news, supply chain events, etc. The data storage format uses the Parquet columnar storage format, which supports efficient data query and compression.

[0086] The data access interface provides a standard REST API, allowing other modules and external systems to obtain the original data; for real-time updates, the message queue RabbitMQ mechanism is adopted to transmit the external data flow to the original database in real time, and the data is immediately processed and stored through stream processing.

[0087] The original database module deploys a blockchain system. Blockchain technology is used to record the hash values, sources, and generation times of important price node data items for each product, ensuring data immutability and transparency. Smart contracts are deployed to automatically verify the legality, integrity, and consistency of the data sources. Only the data that passes the verification can be recorded on the blockchain, thus ensuring the reliability of the original data. Utilizing the distributed characteristics of the blockchain, an efficient query engine is provided for the original database. By recording the hot data or the data that needs to be frequently queried in the blockchain and using the distributed query architecture IPFS to accelerate the data access and query efficiency. An access control module for each user or system is established in the blockchain. All data access requests are automatically verified through smart contracts, ensuring that only authorized users can access sensitive data. All query operations will be recorded in the blockchain, ensuring the traceability and transparency of the operations. Thus, different system users can share data through the blockchain, ensuring the security and consistency during the data transmission process.

[0088] Regarding the process database design, the process database stores the data after processing and transforming the original data through the pricing large model algorithm. According to the nature and use of the data, the data is classified and archived, and the milestone data generated in stages forms the process database, which is convenient for subsequent data retrieval and analysis.

[0089] The process database stores data in stages. During the process of generating data, every time a milestone task is completed, its data is stored as an independent stage dataset: a process stage table is constructed to record the key data and results of each process node, such as the output of the price calculation model, the data processed by the AI algorithm, the stage cost analysis, etc.; a data classification table is constructed to classify and archive the data according to its nature, such as "market data", "cost data", "product pricing data", etc.

[0090] The processing data in the process database will be stored in both a traditional distributed database and the blockchain simultaneously. Enterprises can query all the data in the process database through smart contracts or API interfaces, and at the same time access each data change record on the blockchain, ensuring complete transparency when querying data.

[0091] Regarding the design of the achievement database, the achievement database integrates all the data results processed by the system, such as cost analysis, price prediction, optimization suggestions, etc. These data are stored in the database in a structured form for easy subsequent query and analysis. Through the achievement database, the results of system analysis can be intuitively viewed, such as the cost situation after price strategy adjustment, etc. The data and analysis results provided by the achievement database can support enterprise decision-making. For example, based on the results of price prediction, enterprises can formulate reasonable price strategies to maximize profits.

[0092] In the achievement database module, an analysis report form is constructed to store various pricing analysis and market prediction reports, including cost changes, expected profits, market impacts, etc. after pricing strategy adjustment; an optimization suggestion form is constructed, and the system generates pricing optimization suggestions based on the analysis of the large model to help enterprise decision-makers understand the best pricing plan.

[0093] Through BI tools or customized data analysis platforms, visual reports are generated in this storage layer, and enterprises can comprehensively view the data in the achievement database, facilitating enterprise data analysis.

[0094] The core verification information of each achievement data (such as data ID, hash value, version, timestamp, etc.) is stored on the blockchain to ensure immutability and traceability; the complete achievement data is stored in an off-chain database, which contains a large amount of detailed information and is stored after encryption to ensure data security and privacy protection. Each update, adjustment, and decision-making process of achievement data will generate a new block to record the relevant modification information, and all data operations will be transparently recorded on the blockchain, enabling management personnel, auditors, and relevant decision-makers to view the processing process of achievement data in real time, increasing data transparency.

[0095] Regarding the design of the rule database, the rule database is a key component to ensure that the entire system can make reasonable and compliant decisions when conducting price and cost control.

[0096] Various verification rules, compliance discrimination rules, rationality evaluation rules, sufficiency confirmation rules, etc. are stored in the rule database data. Each rule will include the type of rule, applicable scenarios, input and output, etc.; a version number is assigned to each rule to ensure version control of the rules, facilitating system updates and backtracking.

[0097] A design rule association table records the relationships between rules, data sets, and analysis modules, ensuring a clear scope of application for the rule library and facilitating its invocation during data processing.

[0098] Deploy a blockchain system. All rules in the rule library, such as "authenticity verification rules", "compliance discrimination rules", "rationality evaluation rules", etc., will be embedded in the blockchain in the form of smart contracts to ensure the transparency of rules and the automation of execution. Record all change histories of rules through the blockchain to generate audit logs, ensuring the transparency and compliance of rule changes. When new pricing rules are introduced, all relevant nodes and systems can be updated in real time and executed automatically.

[0099] In this embodiment, in the model layer, 11 independent price calculation models are deployed, and the weighted fusion of the calculation results of multiple models is performed to output the final product pricing.

[0100] It should be further noted in this embodiment that the model layer includes a social price model, an industry price model, a competitor price model, a historical cost model, a base-period economic indicator model, an empirical statistics model, a technical and economic model, a life-cycle cost model, a labor cost model, a standard cost model, and a rule model.

[0101] The social price model is used for product cost element prediction, factor analysis, and benchmarking management:

[0102] Data requirements: Collect the main raw material price indices, such as crude oil, polyethylene, non-ferrous metals, and ferrous metals, on the official websites of national and regional statistical bureaus, relevant industry association websites, and publicly released platforms; collect the manufacturers of common standard parts and their main parameters; collect the evaluation indicators of various scale enterprises affiliated to the State-owned Assets Supervision and Administration Commission of the State Council, the prices and grading standard values of electronic components, typical module components, and general supporting parts.

[0103] Data processing methods: Time series forecasting: Use ARIMA, seasonal decomposition, and LSTM methods to predict the price trends of raw materials and related items; Exponentially Weighted Moving Average (EWMA): Smooth historical data to identify trends and fluctuations.

[0104] Output: The price ranges of each raw material and a report on the overall market price trend.

[0105] Parameter tuning: Use historical data for model verification, automatically select ARIMA parameters and LSTM network structures; adjust for seasonal fluctuations.

[0106] The industry price model is used for product target cost calculation and process cost benchmarking analysis:

[0107] Data requirements: Collect the gross profit margin and net profit margin of typical enterprises in related industries through investigation and research; the processing prices of typical machining, optical processing, electrical assembly, and PCB board manufacturing; the prices of surface treatment, component screening, and environmental testing; the hourly expense rate and the proportion of period expenses of typical enterprises in related industries; the gross profit margin and net profit margin of large-scale enterprises in related industries, etc.

[0108] Data processing methods: Multiple regression analysis: Establish a statistical model of the typical cost structure and profit margin in the industry; Factor analysis and principal component analysis (PCA): Extract key cost factors and construct benchmarking indicators; Benchmarking analysis: Use the data of benchmark enterprises for horizontal comparison to form the industry average and the upper and lower floating ranges.

[0109] Output: Industry target cost benchmark, benchmarking coefficient and reference interval; Cost composition charts for different processes and expense items.

[0110] Parameter optimization: Regularly update the parameters of the regression model and recalibrate the factor weights in combination with the latest industry data.

[0111] Competitive product price model, used for target product cost calculation, process cost benchmarking analysis, price analysis and decision-making:

[0112] Data requirements: Collect the approved prices for single-source procurement and the winning bids for competitive procurement of typical products and components through investigation and research; data such as the hourly expense rate and the proportion of period expenses of the main participating enterprises in the same industry; data such as the historical bid prices of typical products of the main participating enterprises in the same industry.

[0113] Data processing methods: Regression and clustering analysis: Use multiple regression to model the price trend of competitive products and adopt K-means clustering to stratify competitive products; Dynamic comparison: Combine the time series data of competitive product prices and use moving window analysis to predict the future price range of competitive products.

[0114] Output: Suggestions on the price range of competitive products, competitive product comparison analysis report; Suggestions on the pricing strategies of competitive products for different products or components.

[0115] Parameter optimization: Adjust the number of clusters and the parameters of the regression model according to the actual winning bid data to form a more accurate price prediction model.

[0116] Historical cost model, used for product cost process monitoring, target cost management, standard cost maintenance, price analysis and evaluation:

[0117] Data requirements: Collect information through internal financial accounting and other systems, including:

[0118] Research stage: Research budgeted cost, material cost, subcontracted cost, special cost, administrative cost, fuel and power cost, fixed asset depreciation cost, employee compensation and labor cost, management cost collected during the implementation of the research project, research project acceptance final accounts cost; Target price in the project establishment and plan stage, target cost in the engineering development stage, actual cost in the finalization (appraisal) and small batch production stage;

[0119] Mass production stage: Direct material cost of each product component (purchase contract, invoice price, payment information, etc.), processing and manufacturing cost (working hours, rate), quality cost (test consumption, scrap loss, appraisal cost, etc.), tooling consumption, energy consumption, low-value consumables, warehousing and logistics, after-sales service cost, management cost (period expense distribution rate);

[0120] The correlation between the unit price of material procurement and factors such as procurement channels, procurement volume, supply cycle, payment cycle, etc.; The correlation between the unit price of main outsourced parts and relevant material price index, producer price index of industrial producers, purchase price index of industrial producers; The correlation between the unit price of main subcontracted parts and per capita salary in relevant industries, gross profit margin of enterprises above a certain economic scale, etc.

[0121] Data processing methods: Cost regression analysis: Using historical data to construct the relationship between each cost item and the final price of the product; Classification modeling: Distinguishing between research and mass production stages, direct costs and indirect costs, and modeling separately; Dynamic adjustment: Introducing price indices and exchange rate changes to correct costs.

[0122] Output: Historical cost estimation of each stage of the product, detailed cost composition statement and trend prediction chart.

[0123] Parameter optimization: Regularly correcting model parameters according to the latest financial data; Introducing correction factors considering regional and time differences.

[0124] Base period economic indicator model, used for product cost accounting, target cost management, price analysis and performance evaluation:

[0125] Data requirements: Collecting the wage consumption and work-hour attainment rate of various production personnel through various enterprise management systems; Depreciation and utilization rate of various factory buildings, production and testing equipment, automated production lines and other resources; Purchase cost, inventory days, etc. of various materials; Preparation, production, inspection and test cycles of typical product components and the input quantity of resources such as manpower and equipment; Expenses and output of auxiliary production, independent accounting departments, manufacturing departments, and functional management departments within a certain period.

[0126] Data processing methods: Ratio and rate analysis: Calculating the ratios of each key economic indicator (such as per capita output value, labor productivity); Normalization processing: Standardizing economic indicators of different departments and different products to form a unified indicator system.

[0127] Output: Base-period economic indicator statements and economic parameters for subsequent cost accounting and target cost management.

[0128] Parameter tuning: Adjust the weights of each indicator according to annual and quarterly data to maintain sensitivity to the enterprise's operating conditions.

[0129] Empirical statistical model for product target price estimation, profit prediction, project comprehensive evaluation, and price analysis and decision-making:

[0130] Data requirements: Through industry research, literature study, expert consultation, and regular statistical analysis of internal data, collect data such as the cost structure of the enterprise's typical products and components, the correlation between small-batch and large-batch costs, and the correlation between prototype trial production costs and mass production costs; data such as the cost structure of outsourced finished parts and components;

[0131] Data processing methods: Statistical regression and trend analysis: Use parametric methods, trend methods, and engineering methods to model empirical data; empirical ratio method: Construct an empirical ratio library for different production scales and different processes; data mining: Use historical data to find hidden cost correlations.

[0132] Output: Reference value for target price estimation; profit prediction and project comprehensive evaluation reports.

[0133] Parameter tuning: Continuously update and enrich the empirical library using historical project data and automatically correct the statistical model.

[0134] Technical and economic model for enterprise product target price estimation, research and development (including improvement, upgrade, and customized development) project approval decision-making, process cost monitoring, and project construction cost-effectiveness and system contribution rate evaluation:

[0135] Data requirements: Through user release, expert consultation, and special research and demonstration, collect the correlation between the enterprise's product technical indicators and costs, the correlation between different technical approaches and implementation costs, and the correlation between different process routes (methods) and implementation costs; product function requirements for typical application scenarios, product customized development and verification costs for typical requirements, technology iteration and update costs, and the correlation between the prices of basic products and customized products.

[0136] Data processing methods: Parametric estimation method: Calculate the costs of each subsystem using the parametric estimation method according to the engineering and functional breakdown structure; sensitivity analysis: Evaluate the impact of different technical solutions and process routes on costs; correlation analysis: Establish a relationship model between product technical indicators and costs.

[0137] Output: Cost comparison of each technical solution, pricing suggestions; project approval decision-making support report.

[0138] Parameter Tuning: Regularly update technical parameters and cost coefficients in combination with the feedback from actual research and development projects.

[0139] Total Life Cycle Cost Model, used for estimating the total life cycle cost of enterprise products, calculating and analyzing after-sales service costs, and making project establishment decisions.

[0140] Data Requirements: Collect the reliability, maintainability, and supportability indicators of enterprise products, quality costs during the production process, project after-sales service costs, and product usage, support, and retirement disposal costs through user releases, expert consultations, and regular statistical analyses of internal enterprise data.

[0141] Data Processing Methods: Monte Carlo Simulation: Consider the impact of uncertain factors on various costs during the usage period; Cost Allocation Model: Allocate initial investment, operating costs, maintenance costs, etc. over the product life cycle.

[0142] Output: Estimation of the total life cycle cost (TCO) of the product, after-sales service cost, and risk assessment report.

[0143] Parameter Tuning: Adjust parameters such as the repair rate and reliability according to actual usage data (such as repair records and user feedback).

[0144] Manpower Cost Model, used for calculating enterprise product costs, evaluating personnel performance, and conducting business analysis and decision-making.

[0145] Data Requirements: Collect the correlation between the total wage approved by the superior department, the actual paid amount of employee compensation payable, and the total output value and industrial added value, the correlation between the average compensation of on-the-job employees (including the average compensation of labor service personnel) and per capita output value and labor productivity, the correlation between the average compensation of different categories and per capita task working hours, and the correlation between the wage growth rate and output value, profit, labor productivity, etc. through the official websites of national and regional statistical bureaus and regular statistical analyses of internal data; the correlation between the average compensation and labor productivity per capita in the report form and the average compensation and labor productivity per capita of enterprises of the same scale in the same industry during the same period.

[0146] Data Processing Methods: Benchmarking Analysis: Construct a comparison model of compensation and efficiency between the enterprise internal and industry benchmarks; Regression and Correlation Analysis: Analyze the relationship between wage growth, labor productivity, and output value and profit; Dynamic Forecasting: Forecast the trend of manpower costs in the future based on historical data.

[0147] Output: Estimated manpower cost and the proportion of labor cost in a single product, labor productivity, and a comparison report of compensation levels.

[0148] Parameter Tuning: Adjust the regression coefficient in combination with the latest statistical data to timely reflect market changes.

[0149] Standard cost model, used for enterprise product pricing decisions, target cost management, and business analysis decisions.

[0150] Data requirements: Through special topic demonstration and review, various enterprise management systems of our unit collect data such as product BOM, raw material consumption quotas, main material process consumption coefficients, auxiliary material consumption quotas, raw material prices, component prices, component screening qualification rates, purchased part prices, outsourced part prices, labor hour quotas, hourly expense rates, transportation and miscellaneous expense rates, scrap loss rates, fuel and power distribution rates, per capita fixed asset depreciation, various inspection and test prices, after-sales service fee ratios, and period expense ratios.

[0151] Data processing methods: Activity-based costing (ABC): Cost allocation is carried out according to the product production process and resource consumption; Standard cost accounting: Calculate the standard cost based on the product BOM and various consumption quotas; Actual cost deviation analysis: Regularly compare with actual procurement and production costs and feedback the deviations.

[0152] Output: Product standard cost and detailed cost composition, cost deviation analysis report and improvement suggestions.

[0153] Parameter tuning: Regularly revise the consumption quotas and expense rates according to actual production data to ensure the practical guidance of the standard cost.

[0154] Rule model, used for data processing, data auditing, data analysis, and evaluation.

[0155] Data requirements: Through superior authorities, industry competent departments, user authorities, general units, and internal management departments, collect various rules included in system documents such as national price management regulations and policies issued for relevant products, unified accounting systems issued by the state, funds and cost management systems formulated by superior authorities, price management requirements and cost audit standards clarified by user authorities, and cost accounting systems and cost control standards formulated within the unit for relevant data auditing and calculation logics, including authenticity verification rules, compliance discrimination rules, rationality evaluation rules, and sufficiency confirmation rules.

[0156] Data processing methods: Data auditing and verification: Conduct authenticity, compliance, rationality, and sufficiency verification on each model data and calculation results; Automatic warning: When detecting non-compliance with regulations, internal standards, or abnormal deviations, automatically trigger an alarm and record detailed logs.

[0157] Output: Data audit report, compliance inspection results; Give correction suggestions for unreasonable or abnormal parts in pricing calculations.

[0158] Parameter tuning: Regularly update the rule library to ensure consistency with the latest policies and industry standards; Experts participate in rule verification to form a closed-loop feedback.

[0159] Train a machine learning model by taking the prediction results of each pricing model as features, along with the actual market performance and sales data, so that the model can learn which mixed outputs of the pricing models perform better in specific environments with different demands. Then, fuse the output results of each pricing model through weighted averaging to formulate a final pricing result for different demand scenarios.

[0160] In a specific embodiment, the system dynamically adjusts the product pricing strategy according to different market environments and application scenarios, including but not limited to the following situations:

[0161] In a highly competitive market, the system increases its reliance on competitor price models and industry price models to ensure that the final price has an advantage in the market competition.

[0162] In the initial stage of a new product or market, due to the lack of a large amount of market data and competitive information, the system focuses more on using the full-life cost model and empirical statistical model to determine the price, ensuring that the basic production cost is covered and sufficient market acceptance is obtained.

[0163] In a mature market or the long-term operation stage, the system relies more on long-term stability models such as the base-period economic indicator model and cost control model to ensure that the price strategy has sustainable and long-term competitiveness.

[0164] The data required for each independent price calculation model is stored in the original database for calling; the calculation results generated by each model are stored in the process database for calling; the mixed weighted output results of each model for different scenarios are stored in the result database for use; the verification data such as relevant industry rules is stored in the rule database for use.

[0165] In this embodiment, the decision-making layer analyzes the product price, generates detailed reports and charts, and formulates the final price strategy.

[0166] In this embodiment, the application layer: is developed using technologies such as Web, AJAX, BI, and front-end frameworks, provides an intuitive and easy-to-use operation interface for users, interacts with users and displays the results. Users can obtain the pricing information of the enterprise product full solution in multiple scenarios in real time at this layer and make final pricing decisions for the enterprise product in different scenarios.

[0167] In this embodiment, Figure 2 A method for pricing enterprise products based on a hybrid multi-model includes the following steps:

[0168] S1. Deploy servers and storage devices to provide the required computing resources, storage resources, and network resources for the system;

[0169] S2. Collect the data required by the system model, perform data preprocessing, and store it in the database;

[0170] S3. Set multiple price calculation models, and output after weighted fusion of the calculation results of multiple models;

[0171] S4. Analyze the product price, generate reports and charts, and formulate the final price strategy;

[0172] S5. Provide a user interface to display the pricing result to the user.

[0173] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An enterprise product pricing system based on hybrid multi-model, characterized in that: The system structure is divided into five layers, namely support layer, data layer, model layer, decision layer, and application layer: Support layer: deploy servers and storage devices; Data layer: data collection, preprocessing, and storage; Model layer: output after weighted fusion of multiple price calculation model results; Decision-making level: Generate product price reports and charts, and formulate final product price strategy; Application layer: provides a user interface to interact with users and display results.

2. The system according to claim 1, characterized in that The support layer deploys servers and storage devices to provide computing resources, storage resources and network resources for the system, uses load balancing and fault-tolerant backup technologies to maintain system stability and availability, and provides underlying support for the entire system.

3. The system according to claim 1, characterized in that The data layer and the data collection operation are carried out by using automated data collection, manual data entry and expert research to collect data required by the product pricing model of each enterprise in the system: Automated data collection: directly obtain public data from the National Bureau of Statistics, regional statistical bureaus, industry associations, and major procurement platforms through standard APIs; for some data sources without API support, design web crawlers to crawl required data regularly in compliance with website laws and regulations; after obtaining data, convert the format into JSON, XML, CSV, and Excel standard formats; Manual data entry and expert research: Manually enter or batch import data into the company's internal financial system, R&D system, and production management system; design online questionnaires, research and interview record systems, and collect expert evaluations, experience data, and competitive product price information data entry systems; Set data update frequency: set the real-time data of raw material prices and competitor prices to daily updates; set the corporate financial indicators and industry gross profit margin cycle data to quarterly updates; set the long-term stable data of laws, regulations, policies, and internal rules to update when policies are adjusted.

4. The system according to claim 1, characterized in that The data layer and the data preprocessing operations include the following data processing operations in sequence before the collected data is used for the product pricing model: data deduplication, outlier detection and processing, missing value processing, data consistency check, format conversion and standardization, redundant field deletion, text data cleaning, data validation and compliance check, and data integration.

5. The system according to claim 1, wherein: The data layer and the data storage operation are performed by using a distributed database, and the data is stored in four database modules: the original database, the process database, the results database, and the rule database. The data communication between the modules is realized through the API gateway and the message bus in the system: The original database is designed using the distributed storage technology Hadoop to store pre-processed data. The original database contains market data tables, industry data tables, historical data tables, and real-time data tables. The access interface provides a standard REST API, allowing other modules and external systems to obtain original data. The blockchain system is deployed in the original database module, and the blockchain technology is used to record the hash value, source, and generation time of the enterprise product price node data items. The distributed query architecture IPFS is used for data access and query. A user access rights control module is established in the blockchain, and all data access requests are automatically verified through smart contracts. The process database stores the calculation results output by each independent price calculation model and constructs a process stage table; the calculation results of each price calculation model are stored in both the traditional distributed database and the blockchain; The results database stores all the data results of each price calculation model after weighted fusion output by the system's machine learning algorithm, builds an analysis report table to store various pricing analysis and market forecast reports; builds an optimization suggestion table to store pricing optimization suggestions generated by the system based on large model analysis; The deployed blockchain system stores the data ID, hash value, version, and timestamp of each achievement data. The complete achievement data is stored in the off-chain database. Each update, adjustment, and decision-making process of the achievement data will generate a new block to record the modification information and data operations. The rule database stores data verification rules. Each rule includes the rule type, applicable scenario, input and output content, and version number. The rule association table records the relationship between the rule and the data set and analysis module. Deploy the blockchain system and embed all the rules in the rule base into the blockchain in the form of smart contracts. The rule verification is automatically executed and synchronized to all nodes when updated.

6. The system according to claim 1, wherein: The model layer sets 11 independent price calculation models, including: Social price model: Use ARIMA, seasonal decomposition, and LSTM methods to predict raw materials and related price trends; smooth historical data to identify trends and fluctuations; Industry price model: Establish a statistical model of the typical cost structure and profit margin of the industry; extract key cost factors and build benchmark indicators; use benchmark enterprise data for horizontal comparison to form the industry average and upper and lower floating ranges; Competitive product price model: Use multivariate regression to model competitive product price trends, and use K-means clustering to stratify competitive products; Combine competitive product price time series data and use moving window analysis to predict future competitive product price ranges; Historical cost model: Use historical data to build the relationship between each cost item and the final price of the product; distinguish between the scientific research and batch production stages, direct costs and indirect costs, and model them separately; introduce price index and exchange rate changes to correct costs; Base period economic indicator model: calculate the ratio of various economic indicators, standardize the economic indicators of different departments and products, and form a unified indicator system; Empirical statistical model: Use parameter method, trend method and engineering method to model empirical data, build an empirical ratio library between different production scales and different processes, and use historical data to find implicit cost correlations; Technical and economic model: Calculate the cost of each subsystem using parameter estimation method based on engineering and functional decomposition structure; evaluate the impact of different technical solutions and process routes on cost; establish a relationship model between product technical indicators and cost; Life cycle cost model: Monte Carlo simulation is used to consider the impact of uncertainties on various costs during the service life; a cost allocation model is established to allocate initial investment, operating costs, maintenance costs, etc. to the product life cycle; Human cost model: construct a salary and efficiency comparison model between the enterprise and industry benchmarks; analyze the relationship between wage growth, labor productivity, output value and profit; predict the trend of human cost in the future based on historical data; Standard cost model: Use activity-based costing (ABC) to allocate costs according to product production processes and resource consumption; calculate standard costs based on product BOM and various consumption quotas; regularly compare with actual procurement and production costs and report deviations; Rule model: Verify the authenticity and compliance of each model data and calculation results; when non-compliance with regulations, internal standards or abnormal deviations are detected, alarms are automatically triggered and logs are recorded.

7. The model layer according to claim 6, characterized in that: Train a machine learning model, input the prediction results of each pricing model as features, as well as actual market performance and sales data, let the model distinguish the mixed output results of pricing models under different demand environments, merge the output results of each pricing model by weighted average, and formulate a final pricing result for different demand scenarios.

8. The system of claim 1, wherein: The decision-making layer analyzes product prices and generates reports and charts to formulate the final pricing strategy.

9. The system according to claim 1, characterized in that The application layer is developed using Web, AJAX, BI, and front-end framework technologies to provide an operation interface for users, interact with users, and display results. Users obtain pricing information for enterprise products in multiple scenarios and across all solutions at this layer.

10. A method for enterprise product pricing based on hybrid multi-model, the method being used to implement the system described in any one of claims 1 to 9, characterized in that: The method comprises the following steps: S1. Deploy servers and storage devices to provide the required computing resources, storage resources, and network resources for the system; S2, collect the data required by the system model, perform data preprocessing and store it in the database; S3. Set up multiple price calculation models, perform weighted fusion on the calculation results of multiple models and output them; S4. Analyze product prices and generate reports and charts to formulate final pricing strategies; S5. Provide a user interface to display product pricing results to users.

Images