Intelligent matching method for upstream and downstream enterprises in chemical industry chain based on big data

Abstract

This invention provides a big data-based intelligent matching method for upstream and downstream enterprises in the chemical industry chain, belonging to the technical field of matching upstream and downstream enterprises in the chemical industry chain. The method includes collecting multi-dimensional feature data of the chemical industry chain; analyzing basic enterprise data using an industry chain attribute matching algorithm; analyzing the dynamic changes in supply and demand relationships with market cycles; constructing a quantitative model of supply and demand impact, analyzing the influence weight of dynamic supply and demand data on initial enterprise matching data; and building an intelligent matching model of the industry chain to evaluate the matching process of upstream and downstream enterprises in real time. By integrating multi-dimensional feature data of enterprises and utilizing industry chain attribute matching algorithms, dynamic supply and demand analysis methods, quantitative supply and demand impact models, and intelligent matching models of the industry chain, this invention achieves accurate matching and dynamic optimization of upstream and downstream enterprises in the chemical industry chain, solving problems such as reliance on manual experience in matching, untimely response to dynamic changes in supply and demand, and insufficient risk management capabilities in existing technologies.

Description

Technical Field

[0001] This invention belongs to the field of matching technology for upstream and downstream enterprises in the chemical industry chain, specifically a smart matching method for upstream and downstream enterprises in the chemical industry chain based on big data. Background Technology

[0002] The chemical industry chain covers multiple links such as raw material supply, production and processing, and product sales. Efficient matching of upstream and downstream enterprises is the core to ensure the stable operation of the industry chain and enhance overall competitiveness.

[0003] Currently, the matching models in the chemical industry largely rely on traditional offline connections and industry exhibitions, which suffer from prominent problems such as information asymmetry, limited matching dimensions, and insufficient dynamic adaptability. On the one hand, corporate cooperation decisions often focus only on superficial factors such as product type and price, neglecting key compatibility indicators such as qualification compliance, production capacity flexibility, and environmental protection standards, resulting in poor cooperation stability and high performance risks. On the other hand, the market demand for chemical products fluctuates frequently, and regulations are becoming increasingly stringent. Traditional static matching methods are unable to respond to market cycle changes and policy adjustments, and cannot optimize cooperation combinations in a timely manner. In addition, the lack of a standardized matching system within the industry, coupled with inconsistent data formats and poor information sharing among enterprises, further exacerbates problems such as low matching efficiency and resource waste, hindering the coordinated development and transformation and upgrading of the chemical industry chain. Summary of the Invention

[0004] This invention provides a big data-based intelligent matching method for upstream and downstream enterprises in the chemical industry chain, in order to address the shortcomings of existing technologies.

[0005] This invention provides a method for intelligent matching of upstream and downstream enterprises in the chemical industry chain based on big data, including: Collect multidimensional feature data of upstream and downstream enterprises in the chemical industry chain. The multidimensional feature data includes basic static feature data of enterprises and core business feature data of enterprises.

[0006] Basic attribute items are extracted from the basic static feature data of enterprises to form basic enterprise data. The basic enterprise data is then analyzed using the industry chain attribute matching algorithm to obtain the core matching feature set of enterprises.

[0007] By employing dynamic supply and demand analysis methods, we can analyze the dynamic changes in the supply and demand relationship of enterprises in the industrial chain with the market cycle and generate dynamic supply and demand characteristic data.

[0008] By associating the enterprise's core business characteristic data with the enterprise's core matching feature set, the enterprise's initial matching data is obtained.

[0009] A quantitative model of the impact of supply and demand is constructed to analyze the influence weight of dynamic supply and demand data on the initial matching data of enterprises, and to obtain the supply and demand matching influence coefficient.

[0010] Based on the core matching feature set of enterprises, the initial matching data of enterprises, and the supply and demand matching influence coefficient, an intelligent matching model for the industrial chain is built to evaluate the matching process of upstream and downstream enterprises in real time and generate matching optimization and cooperation risk management instructions.

[0011] The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by this invention includes the following process for analyzing basic enterprise data using an industry chain attribute matching algorithm: The core attribute parameters are extracted from the enterprise's basic static feature data, including product type, production capacity, geographical location, and qualification level. These are combined with the enterprise's technological R&D capabilities, supply chain response speed, and environmental compliance level as auxiliary attribute parameters.

[0012] Based on the characteristics of the sub-sectors of the industrial chain and the market circulation demand parameters, scenario matching parameters are obtained. Combined with core attribute parameters and auxiliary attribute parameters, the enterprise matching priority and cooperation mode are determined.

[0013] By associating scenario matching parameters with cooperation stability requirements and auxiliary attribute parameters with enterprise cooperation potential, a set of core matching features for enterprises is obtained through multi-dimensional weighted analysis.

[0014] The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by this invention includes the following process for generating dynamic supply and demand characteristic data: Based on the cyclical characteristics of the chemical product market and the production response cycle of enterprises, three key monitoring cycles are divided into the market expansion period, the stable period and the contraction period, and dynamic monitoring data on the supply and demand status of enterprises are collected within the corresponding cycles.

[0015] The dynamic monitoring data is processed by outlier removal, trend smoothing and time series alignment to eliminate data noise and unify the time base, resulting in standardized dynamic data.

[0016] Based on standardized dynamic data, the product supply and demand balance coefficient, price fluctuation response rate, order delivery timeliness rate, and cooperation elasticity coefficient are calculated as core dynamic characteristic indicators.

[0017] For different sub-sectors of the chemical industry and enterprise size types, the core dynamic characteristic indicators are calibrated for industry-specific differences to obtain supply and demand dynamic characteristic data.

[0018] According to the intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by this invention, the process of calculating core dynamic characteristic indicators includes: The average product supply and market demand for each key cycle are selected from standardized dynamic data. The market matching trend is analyzed by the supply-demand difference, and the product supply-demand balance coefficient is obtained by combining the enterprise's capacity elasticity adjustment.

[0019] By filtering product price fluctuations and corporate response adjustment cycles from standardized dynamic data over different periods, an initial response coefficient is calculated, and the price fluctuation response rate is obtained by adjusting the coefficient based on corporate decision-making efficiency.

[0020] The actual delivery time and agreed delivery time of orders are selected from standardized dynamic data to calculate the basic value of delivery on time rate, and then the order delivery timeliness rate is obtained by combining the product transportation radius.

[0021] The number of cooperation adjustments and cooperation satisfaction scores during the duration of the cooperation relationship are screened from standardized dynamic data to calculate the initial elasticity value, and then adjusted according to industry cooperation practices to obtain the cooperation elasticity coefficient.

[0022] According to the intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by the present invention, the process of obtaining initial matching data for enterprises includes: The core business characteristic data of enterprises are normalized with the core matching feature set, and a mapping relationship matrix is ​​established from four dimensions: product matching, capacity matching, cost matching, and risk matching.

[0023] By combining the policy requirements of the chemical industry with the qualification standards for cooperation with enterprises, a rigid screening rule is set to retain a combination of enterprises that meet environmental protection standards, have sufficient production capacity and complete qualifications, forming an initial matching pool.

[0024] Construct a deep neural network model, collect historical data on cooperation between companies in the industry chain for training, input initial matching pool data, and output historical cooperation success rate scores.

[0025] The random forest algorithm is used, and the optimized model parameters are verified through Bayesian optimization. The matching degree of the enterprise combination in the initial matching pool is predicted to obtain the initial matching score.

[0026] The initial matching score is dynamically corrected based on the dynamic characteristics of supply and demand data to obtain the initial matching data for enterprises.

[0027] The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data, provided by this invention, includes the following process for obtaining the supply and demand matching influence coefficient: Collect cases of failed collaborations in the chemical industry chain due to dynamic imbalances in supply and demand, establish a mapping database of dynamic characteristics of supply and demand and types of collaboration risks, and clarify the transmission path of the impact.

[0028] A risk weight assessment system is constructed based on the frequency of risk occurrence and the degree of economic loss. The weight value of each risk type is determined by the analytic hierarchy process, and the correlation strength between core dynamic characteristic indicators and cooperation risks is calculated.

[0029] By combining the current deviation, risk weight, and correlation strength of supply and demand dynamic characteristic data, a weighted summation method is used to calculate the impact score of each dynamic characteristic indicator on the enterprise's initial matching data.

[0030] By integrating the impact scores of all dynamic characteristics, an initial impact coefficient is obtained. This coefficient is then calibrated using current industry chain market sentiment data to obtain a supply and demand matching impact coefficient.

[0031] According to the intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by this invention, the process of building an intelligent matching model for the industry chain includes: The core matching feature set of enterprises, the initial matching data of enterprises, and the supply and demand matching influence coefficient are standardized in format and unified in dimension. They are then correlated by time according to the batch ID of enterprise cooperation and aligned by business dimension according to the links in the industrial chain to obtain the fused data.

[0032] Three types of fusion features—static matching basis, dynamic risk warning, and weight adjustment factor—are extracted from the fused data to construct a multi-dimensional matching feature set.

[0033] A hybrid architecture combining time-series prediction and static classification models is adopted, and historical data on cooperation among chemical industry chain enterprises is collected as a training set for model training and parameter optimization.

[0034] Based on a multi-dimensional matching feature set, matching qualification thresholds, optimization thresholds, and risk thresholds are set for different sub-sectors of the industrial chain to obtain an intelligent matching model for the industrial chain.

[0035] The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by this invention includes the following process for obtaining a multi-dimensional matching feature set: The final matching score in the initial matching data of enterprises is converted into a standardized matching score, and a static matching basis is obtained by combining the enterprise's historical cooperation performance statistics.

[0036] The dynamic risk score is obtained by weighting each core dynamic characteristic indicator based on the supply and demand matching impact coefficient, and the dynamic risk warning feature is obtained by combining the time-series change slope of the dynamic characteristic indicator.

[0037] A weighted adjustment factor is constructed based on the supply and demand matching influence coefficient. The influence ratio of static matching basis and dynamic risk warning is determined by the entropy weight method. The three factors are weighted and calculated to obtain a multi-dimensional matching feature set.

[0038] According to the intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by this invention, the process of generating matching optimization and cooperation risk management instructions includes: Real-time data collection of supply and demand dynamics, transaction execution data, and market environment data from enterprises in the chemical industry chain is used. After alignment with a multi-dimensional matching feature set, the data is input into the intelligent matching model of the industry chain, and the real-time matching status level and anomaly contribution analysis results are output.

[0039] If the matching status is "high-quality match," continuously monitor and record data, and update the company's cooperation credit profile. If the status is "potentially optimized," trigger a matching optimization alert, extract the fluctuation threshold of supply and demand dynamic data, and generate suggestions for adjusting cooperation conditions based on the company's initial matching data.

[0040] If the situation is a risk-matched state, the risk type is identified based on the abnormal contribution analysis results, and targeted matching adjustment instructions and risk control measures are generated in combination with the supply and demand matching impact coefficient.

[0041] According to the intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data provided by this invention, the risk control measures include: To address the risk of mismatched production capacity, generate suggestions for capacity adjustment or instructions to supplement cooperation with partner companies. To address the risk of price fluctuations, generate suggestions for adjusting dynamic pricing mechanisms or guidelines for hedging operations. To address the risk of delivery delays, generate logistics route optimization solutions or suggestions for setting safety stock levels.

[0042] After all instructions are executed, the adjusted enterprise cooperation data is collected in real time and sent back to the intelligent matching model of the industrial chain for effect verification and parameter iteration optimization.

[0043] This invention provides a big data-based intelligent matching method for upstream and downstream enterprises in the chemical industry chain. By constructing a multi-dimensional feature system, it comprehensively covers core information such as static qualifications, dynamic operations, and market connections of enterprises, breaking the limitations of single-dimensional matching, improving the accuracy and suitability of matching results, and reducing the risk of breach of cooperation. By introducing a dynamic supply and demand analysis mechanism and adjusting the matching strategy in real time according to market cycle changes, the matching results can dynamically respond to market demand fluctuations and policy adjustments, enhancing the risk resistance and flexibility of the industry chain. Through standardized data processing and intelligent algorithm models, the matching process is automated and intelligent, significantly reducing manual intervention, improving matching efficiency, and lowering the cost of enterprise cooperation. By establishing a scientific cooperation priority and risk management system, it provides enterprises with clear basis for cooperation decisions, helping leading enterprises to integrate high-quality resources and creating fair cooperation opportunities for SMEs, thus promoting the coordinated upgrading and sustainable development of the chemical industry chain. Attached Figure Description

[0044] The invention will now be further described with reference to the accompanying drawings.

[0045] Figure 1 This is a flowchart illustrating the intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data in this invention. Figure 2 This is a schematic diagram of the process for generating dynamic supply and demand characteristic data in this invention; Figure 3 This is a flowchart illustrating the calculation of core dynamic feature indicators in this invention. Detailed Implementation

[0046] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0047] like Figures 1 to 3 As shown in the embodiment of the present invention, the intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data includes: This data collection comprises multi-dimensional characteristic data from upstream and downstream enterprises in the chemical industry chain. The multi-dimensional characteristic data includes four main categories: basic static characteristic data, core business characteristic data, dynamic operational characteristic data, and market and external relationship characteristic data. Basic static characteristic data includes information on enterprise qualifications, geographical location, and production capacity. Specifically, enterprise qualifications include registered capital, years of establishment, chemical industry-specific qualification levels, and safety production standardization levels. Geographical location includes the latitude and longitude of production bases and the coverage area of ​​transportation radius. Production capacity includes the designed capacity of core products, actual capacity utilization rate, and production line specifications. Core business characteristic data includes product-related, transaction and cooperation, and compliance and risk control information. Product-related data includes core product models / specifications, product implementation standards, and key technical parameters. Transaction and cooperation data includes settlement methods, minimum order quantities, and delivery cycles. Compliance and risk control data includes records of violations and penalties in the past three years and the issuance of compliance audit reports. Dynamic operational characteristic data includes supply chain response, production operation, and delivery and fulfillment information. Supply chain response includes order response time and raw material procurement cycles. Production and operation data includes production equipment failure rate and production plan completion rate. Delivery and fulfillment data includes historical order on-time delivery rate and delivered product qualification rate. Market and external correlation characteristic data includes market matching, policy matching, and cooperation potential information. Market matching data includes the annual demand of the target market and the demand fluctuation coefficient. Policy matching data includes whether it complies with the latest environmental protection policies and whether it enjoys industry support policies. Cooperation potential data includes capacity expansion plans and the proportion of technology research and development investment.

[0048] Basic attribute items are extracted from the enterprise's basic static feature data to form the enterprise's basic data. The specific selection rules for basic attribute items strictly follow the following standards: For entities with specific qualifications, priority should be given to those with registered capital, years of establishment, chemical industry-specific qualification level, and safety production standardization level.

[0049] For geographical location, select the latitude and longitude of the production base (to facilitate calculation of the distance to the target cooperation area) and the transportation radius coverage (which must cover the target cooperation area).

[0050] For capacity scale categories, select the core product's designed capacity (which must meet the minimum capacity requirements of the target cooperation order), actual capacity utilization rate (not less than 60%), and production line specifications (which must match the production requirements of the cooperation product).

[0051] For product-related categories, select the core product model / specification (consistent with the product required for cooperation), product execution standards (meeting the standards required by the partner), and key technical parameters (all indicators must be within the range required by the partner).

[0052] For transaction cooperation, select the settlement method (matching the settlement method acceptable to the partner), minimum order quantity (not exceeding the partner's expected order quantity), and delivery cycle (not exceeding the longest delivery cycle required by the partner).

[0053] The basic data of enterprises is analyzed using an industry chain attribute matching algorithm to obtain the core matching feature set of enterprises. The specific process is as follows: The core attribute parameters are extracted from the enterprise's basic static feature data, including product type, production capacity, geographical location, and qualification level. The specific definitions and quantification methods of each parameter are as follows: According to the classification standards of the chemical industry, it is divided into subcategories such as basic chemical raw material manufacturing, chemical fertilizer manufacturing, pesticide manufacturing, coating manufacturing, and synthetic material manufacturing, with each subcategory corresponding to a unique classification code.

[0054] The actual annual production capacity of core products is used as a quantitative indicator, with the unit being tons / year. The data comes from the company's annual production report and actual production capacity statistics.

[0055] The straight-line distance between the production base and the target cooperation area is used as a quantitative indicator, with the unit being kilometers, and is calculated using latitude and longitude coordinates.

[0056] The qualification level is assigned points based on specific qualifications: Level A is worth 4 points, Level B is worth 3 points, Level C is worth 2 points, and Level D is worth 1 point. The safety production standardization level is worth 3 points for Level 1, 2 points for Level 2, and 1 point for Level 3. The scores from both levels are added together to obtain the final quantitative score for the qualification level.

[0057] Combining a company's technological R&D capabilities, supply chain responsiveness, and environmental compliance level as auxiliary attribute parameters, the quantification methods for each parameter are as follows: Technological R&D capability is quantified using the average R&D investment ratio over the past three years as a quantitative indicator. The calculation formula is as follows: .

[0058] Supply chain response speed is quantified by the average order response speed, which refers to the time from receiving an order to confirming it, measured in hours. The average response speed of the last 100 orders is calculated.

[0059] Environmental compliance is quantified using the emission compliance rate over the past 12 months. The emission compliance rate refers to the proportion of months with compliant emissions out of the total number of monitored months. The calculation formula is as follows: .

[0060] Based on the characteristics of the sub-sectors of the industry chain and market circulation demand parameters, scenario matching parameters are obtained and specifically set as follows: The characteristics of the industrial chain sub-sectors are assigned differentiated weights according to sub-sectors such as petrochemicals, fine chemicals, coal chemicals, and biochemicals. The weight of petrochemicals is 0.3, fine chemicals is 0.25, coal chemicals is 0.25, and biochemicals is 0.2. The weighting is based on the market size and cooperation demand of each sub-sector.

[0061] Market circulation demand parameters include the annual demand of the target market and the demand volatility coefficient (calculated using the following formula: ), and transportation timeliness requirements.

[0062] The matching priority and cooperation mode for enterprises are determined by combining core attribute parameters and auxiliary attribute parameters. The matching priority is calculated using a weighted scoring method, and the specific weight allocation and scoring criteria are shown in the table below: ; Matching priority score = score of each parameter × sum of their corresponding weights, with a maximum score of 10. Based on the scores, cooperation modes are categorized into the following four types: A score of ≥9 indicates a strategic partnership, enabling long-term, in-depth cooperation, including joint R&D and capacity sharing.

[0063] Suppliers rated 7-8.9 are given priority in cooperation under the same conditions, and will be given certain price discounts and order preferences.

[0064] Suppliers rated 5-6.9 are considered regular partners, operating according to normal cooperation procedures without any special preferential policies.

[0065] A score of 5 or less indicates a backup partner, considered only as an emergency alternative resource if other partners are unable to meet the needs.

[0066] The scenario matching parameters are linked to the cooperation stability requirements, which are quantified by the average cooperation duration over the past 3 years and the cooperation default rate. The scenario matching parameter score = the characteristic weight score of the subdivided field of the industrial chain + the market circulation demand parameter matching score (each accounting for 50%). The cooperation stability requirement score = (3 - cooperation default rate × 100) × (average cooperation duration in the past 3 years ÷ 3) (full score is 3 points). The product of the scenario matching parameter score and the cooperation stability requirement score is the cooperation stability matching value.

[0067] The auxiliary attribute parameters are linked to the enterprise's cooperation potential, which is quantified by capacity expansion potential and the proportion of investment in technology research and development. Capacity expansion potential is categorized based on the proportion of new capacity added in the next 3 years to current capacity: ≥50% earns 3 points, 30%-50% earns 2 points, 10%-30% earns 1 point, and <10% earns 0 points.

[0068] 3 points for R&D investment ≥ 5%, 2 points for 3%-5%, 1 point for 1%-3%, and 0 points for < 1%.

[0069] Enterprise cooperation potential score = capacity expansion potential score + technology R&D investment ratio score (full score 6 points), auxiliary attribute parameter score = technology R&D capability score × 0.4 + supply chain response speed score × 0.3 + environmental compliance level score × 0.3 (full score 10 points, converted to a 6-point system for use), the product of auxiliary attribute parameter score and enterprise cooperation potential score is the cooperation potential matching value.

[0070] The core matching feature set of enterprises is obtained through multi-dimensional weighted analysis. The weighting formula is: F = a × cooperation stability fit value + b × cooperation potential fit value + c × matching priority score, where a, b, and c are weight coefficients that satisfy a + b + c = 1 and are dynamically adjusted according to the market cycle.

[0071] The final core matching feature set is presented in the form of a three-dimensional vector (F, cooperation mode, sub-domain matching label), where the sub-domain matching label is the classification code of the corresponding sub-domain of the industrial chain.

[0072] Using dynamic supply and demand analysis methods, we analyze the dynamic changes in the supply and demand relationship of enterprises in the industrial chain with market cycles, and generate dynamic supply and demand characteristic data. The specific process is as follows: Based on the cyclical characteristics of the chemical product market and the production response cycle of enterprises, three key monitoring periods are defined: market expansion, stability, and contraction. The criteria for defining each period and the data collection requirements are as follows: The criteria for determining a market expansion period are a month-on-month increase in market demand of ≥5% for three consecutive months. The criteria for determining a stable period are a month-on-month increase in market demand between -2% and 2%. The criteria for determining a contraction period are a month-on-month decrease in market demand of ≥3% for three consecutive months.

[0073] Collect dynamic monitoring data on the supply and demand status of enterprises within the corresponding period, specifically including data such as enterprise product supply, market demand, product price, order delivery time, number of cooperation adjustments, and cooperation satisfaction rating. Data collection sources include enterprise internal ERP systems, industry databases, third-party monitoring platforms, and feedback data from partners.

[0074] Outlier removal, trend smoothing, and time series alignment are performed on dynamic monitoring data to eliminate data noise and unify the time base, resulting in standardized dynamic data.

[0075] Based on standardized dynamic data, the product supply and demand balance coefficient, price fluctuation response rate, order delivery timeliness rate, and cooperation elasticity coefficient are calculated as core dynamic characteristic indicators. The specific calculation process for each indicator is as follows: The average product supply and market demand for each key cycle are selected from standardized dynamic data. Market matching trends are analyzed through the supply-demand difference. Combined with the enterprise's capacity elasticity adjustment, the product supply-demand balance coefficient is obtained. The calculation formula is as follows:

[0076]

[0077]

[0078] In the formula, This is the product supply and demand balance coefficient. This represents the average supply volume of enterprises. This represents the fluctuation value of supply. Let be the supply quantity for the i-th period. This represents the average market demand. This represents the fluctuation in demand. Let be the duration requirement for the i-th cycle. For the number of periods, This is the enterprise's capacity elasticity adjustment coefficient (range 1.0-1.5).

[0079] By filtering product price fluctuations and corporate response adjustment periods from standardized dynamic data over different periods, an initial response coefficient is calculated. This coefficient is then adjusted based on corporate decision-making efficiency to obtain the price fluctuation response rate. The calculation formula is as follows:

[0080] In the formula, Price volatility response rate The difference in price changes, As the baseline, In response to the adjustment cycle, This is a correction factor for enterprise decision-making efficiency (range: 0.8-1.2).

[0081] By filtering actual delivery times and agreed delivery times from standardized dynamic data, a baseline value for on-time delivery rate is calculated. This value is then adjusted for product transportation radius to obtain the on-time delivery rate. The calculation formula is as follows:

[0082] In the formula, To improve the on-time delivery rate of orders, To ensure timely delivery of orders, Where L is the total number of orders, and L is the actual transportation radius. For standard transportation radius, To achieve the maximum reasonable transportation radius.

[0083] The number of cooperation adjustments and cooperation satisfaction scores during the duration of the partnership are selected from standardized dynamic data to calculate the initial elasticity value. The cooperation elasticity coefficient is then obtained by adjusting the data according to industry cooperation practices. The calculation formula is as follows:

[0084] In the formula, This is the cooperation flexibility coefficient, ranging from 0.7 to 1.3. A higher value indicates greater flexibility and satisfaction in cooperation adjustments. The term "number of adjustments to cooperation" refers to the total number of times that both parties negotiate and adjust the terms of cooperation (such as order quantity, delivery time, price, etc.) during the cooperation period. The duration of the partnership is represented by S, which is the partnership satisfaction score, calculated on a 10-point scale based on the average value of feedback from the partner survey. To give a perfect score, This is an adjustment factor based on industry practice, ranging from 0.9 to 1.1. It is set according to the cooperation practices of different sub-sectors of the chemical industry; for example, adjustments are more frequent in the fine chemical sector. With values ​​ranging from 1.0 to 1.1, cooperation in the field of basic chemical raw material manufacturing remains relatively stable. The value ranges from 0.9 to 1.0.

[0085] For different sub-sectors and enterprise sizes in the chemical industry, core dynamic characteristic indicators are calibrated industry-specifically to obtain supply and demand dynamic characteristic data. The calibration process is as follows: Based on the sub-sectors of the chemical industry (petrochemicals, fine chemicals, coal chemicals, biochemicals, etc.) and the size of the enterprise (large enterprises: annual operating revenue ≥ 1 billion yuan; medium-sized enterprises: 100-1 billion yuan; small enterprises: < 100 million yuan), the categories are divided into 4 × 3 = 12 categories.

[0086] Collect core dynamic characteristic indicator data of benchmark enterprises in the industry under each combination category, and calculate the industry mean and standard deviation of each indicator.

[0087] The Z-score standardization method is used to calibrate the core dynamic characteristic indicators of the target enterprise to form supply and demand dynamic characteristic data.

[0088] The initial matching data of an enterprise is obtained by associating its core business characteristic data with its core matching feature set. The specific process is as follows: The core business characteristic data of the enterprise and the core matching feature set are normalized, and the Min-Max standardization method is used to transform all indicator data into the 0-1 range.

[0089] A mapping matrix is ​​established based on four dimensions: product matching, capacity matching, cost matching, and risk matching. The matrix is ​​in the following form:

[0090] Where n is the number of companies participating in the matching process. (i=1,2,...,n,j=1,2,3,4) represent the matching degree values ​​(between 0 and 1) of the i-th enterprise in the four dimensions of product matching, capacity matching, cost matching, and risk matching, respectively. The matching degree value is obtained by calculating the cosine similarity between the enterprise's core business feature data and the core matching feature set. The calculation formula is as follows:

[0091] In the formula, Let be the normalized value of the k-th characteristic indicator of the i-th company. Let p be the standard value of the k-th feature index in the j-th dimension of the core matching feature set, and p be the number of feature indices in that dimension.

[0092] Based on the requirements of chemical industry policies and the qualification standards for enterprise cooperation, strict screening rules have been established. The specific rules are as follows: The company has achieved an emission compliance rate of ≥90% over the past 12 months and has no record of major environmental violations or penalties. The actual capacity utilization rate of its core products is ≥60%, and it can meet the minimum capacity requirements of target orders. It possesses a chemical industry-specific qualification (no lower than Class C) and a safety production standardization level (no lower than Level III), and all relevant qualifications are valid. It has no record of major violations or penalties in the past 3 years, and its compliance audit reports are complete.

[0093] The participating companies were screened strictly according to the above rules, and all eligible company combinations were retained to form the initial matching pool.

[0094] A deep neural network model is constructed, consisting of an input layer, three hidden layers (each with 64, 32, and 16 neurons respectively), and an output layer. The ReLU function is used as the activation function, and the mean squared error function is used as the loss function.

[0095] We collected historical data on cooperation between companies in the industry chain over the past 5 years as a training set. The data included the core business characteristics of the cooperating companies, the core matching feature set, and the cooperation results (success / failure). We then trained the model by setting the number of training iterations and the learning rate.

[0096] The initial matching pool data is input into the trained deep neural network model, which outputs a historical cooperation success rate score (0-10 points). The higher the score, the greater the probability of historical cooperation success for the enterprise combination.

[0097] The random forest algorithm was adopted, with 100 decision trees, a maximum tree depth of 10, and a minimum number of sample splits of 5.

[0098] The optimal model parameters were validated using Bayesian optimization. The optimization objective was to improve the model's prediction accuracy. The optimization range included the number of decision trees (50-200), the maximum tree depth (5-15), and the minimum number of sample splits (2-10). The number of iterations was set to 50 to obtain the optimal model parameters.

[0099] The enterprise portfolio data in the initial matching pool is input into the optimized random forest algorithm model to predict the matching degree of the enterprise portfolio and obtain the initial matching score (0-10 points).

[0100] The initial matching score is dynamically adjusted based on supply and demand dynamic characteristics data. The adjustment formula is as follows:

[0101] In the formula: This is the initial matching data for enterprises (final initial matching score), with a value range of 0-10. Initial match score. This is the product supply and demand balance coefficient. This refers to the price fluctuation response rate. To ensure on-time order delivery rate. This represents the cooperation elasticity coefficient.

[0102] This revised formula incorporates dynamic supply and demand factors into the initial matching score, making the initial matching data more reflective of a company's matching potential in a dynamic market environment.

[0103] A quantitative model of the impact of supply and demand is constructed to analyze the influence weight of dynamic supply and demand data on the initial matching data of enterprises, and the supply and demand matching influence coefficient is obtained. The specific process is as follows: We collected nearly 10 years of cases of failed collaborations in the chemical industry chain due to dynamic imbalances in supply and demand, categorized and organized these cases, and clarified the dynamic characteristics of supply and demand for each case (such as severe supply and demand imbalance, excessive price fluctuations, delivery delays, etc.) and the corresponding types of collaboration risks (such as capacity mismatch risk, price fluctuation risk, delivery delay risk, insufficient collaboration flexibility risk, etc.).

[0104] Establish a mapping database between dynamic characteristics of supply and demand and types of cooperation risks, clarifying the transmission paths of influence. For example: Supply-demand balance coefficient < 0.8 → capacity mismatch risk → cooperation failure; Price fluctuation response rate < 0.7 → price fluctuation risk → cooperation failure; Order delivery timeliness rate < 0.8 → delivery delay risk → cooperation failure; Cooperation elasticity coefficient < 0.8 → insufficient cooperation elasticity risk → cooperation failure.

[0105] The mapping database will be continuously updated to include new cases of failed collaborations, thereby improving the accuracy and comprehensiveness of the mapping relationships.

[0106] A risk weighting assessment system is constructed based on the frequency of risk occurrence and the extent of economic loss. The specific steps are as follows: The frequency of each type of cooperation risk in historical failure cases is calculated using the following formula: .

[0107] Calculate the average economic loss for each risk type, expressed as the average of the direct economic losses of the enterprise in failed cases (such as penalties for breach of contract, inventory backlog losses, and the cost of finding a new partner).

[0108] The Analytic Hierarchy Process (AHP) was used to determine the weight values ​​of each risk type, constructing a hierarchical model. The target layer represents the risk weights, the criterion layer represents the frequency of occurrence and the degree of economic loss, and the alternative layer represents each risk type. A judgment matrix was constructed through pairwise comparisons, the weight vectors were calculated, and a consistency test (consistency ratio CR < 0.1) was performed to finally obtain the weight values ​​for each risk type. (The sum is 1).

[0109] Calculate the correlation strength between core dynamic characteristic indicators and cooperation risks. The correlation coefficient is calculated using the Pearson correlation coefficient, which ranges from -1 to 1. The absolute value is taken as the correlation strength value, and the closer it is to 1, the higher the correlation.

[0110] Combining the current deviation, risk weight, and correlation strength of supply and demand dynamic characteristic data, a weighted summation method is used to calculate the impact score of each dynamic characteristic indicator on the enterprise's initial matching data. The formula is as follows:

[0111] In the formula, The score represents the influence of the i-th dynamic feature. A higher score indicates that the dynamic feature has a greater influence on the initial matching data. This represents the current deviation of the feature. This corresponds to the risk weight value. This represents the strength of the association between this feature and risk.

[0112] By integrating the influence scores of all dynamic features, an initial influence coefficient is obtained. (i=1,2,3,4 correspond to the four core dynamic feature indicators respectively).

[0113] The calibration is performed using current industry chain market sentiment data. The market sentiment data adopts the chemical industry sentiment index (published monthly by the industry association, with a value range of 0-200, and 100 being the critical point for prosperity). The calibration formula is as follows:

[0114] In the formula, The supply and demand matching impact coefficient ranges from 0 to 1. I represents the current market prosperity index of the industrial chain. When I ≥ 150, 0.8 + 0.002 × I is calculated as 1.3. When I ≤ 50, it is calculated as 0.9, ensuring the calibrated coefficient is within a reasonable range.

[0115] Based on the core matching feature set of enterprises, initial matching data of enterprises, and supply and demand matching influence coefficients, an intelligent matching model for the industrial chain is built. This model performs real-time evaluation of the matching process between upstream and downstream enterprises and generates matching optimization and cooperation risk management instructions. The specific process is as follows: The core matching feature set of enterprises, the initial matching data of enterprises, and the supply and demand matching influence coefficient are standardized in format and unified in dimension.

[0116] Time-linked data is established based on enterprise cooperation batch IDs to ensure consistency across all data within the same cooperation batch. Business-level alignment is performed according to supply chain links (such as raw material supply, production and processing, and product sales) to clarify the specific role of each enterprise within the supply chain, resulting in integrated and processed data.

[0117] Three types of fusion features—static matching basis, dynamic risk warning, and weight adjustment factor—are extracted from the fused data to construct a multi-dimensional matching feature set. The specific process is as follows: The final match score (0-10) in the initial matching data of enterprises is converted into a standardized match score in the range of 0-1. The conversion formula is as follows: .

[0118] Based on the statistical analysis of historical cooperation fulfillment rates of enterprises, a static matching basis is obtained, and the calculation formula is as follows:

[0119] In the formula, This is the basis for static matching, and its value range is 0-1. The historical contract fulfillment rate of the enterprise is calculated using the following formula: It is used after being converted to the 0-1 range.

[0120] The dynamic risk score is obtained by weighting each core dynamic characteristic indicator based on the supply and demand matching impact coefficient. The calculation formula is as follows:

[0121] In the formula, The score is a dynamic risk score, ranging from 0 to 1. A higher score indicates a greater dynamic risk. , , , All values ​​are calibrated core dynamic characteristic index values ​​(0-1 range).

[0122] Dynamic risk warning characteristics are obtained by combining the time-series change slope of dynamic feature indicators. The time-series change slope is calculated using a linear regression method, and the formula is as follows: (where t is a time series, (where t is the core dynamic characteristic index value at time t, and n is the length of the time series).

[0123] Dynamic risk warning features The value range is 0-1.5.

[0124] A weighted adjustment factor is constructed based on the supply and demand matching influence coefficient, and the calculation formula is as follows: .

[0125] The influence ratio of static matching basis and dynamic risk warning is determined using the entropy weight method. The specific steps are as follows: Calculate the information entropy of each feature (where m is the number of enterprises, , (This refers to the value of the j-th feature of the i-th company).

[0126] Calculate weights (j=1,2 correspond to static matching basis and dynamic risk warning features, respectively).

[0127] The impact percentage of static matching basis. The impact of dynamic risk warning features .

[0128] By weighting the static matching basis, dynamic risk warning features, and weight adjustment factors, a multi-dimensional matching feature set is obtained.

[0129] A hybrid architecture combining temporal prediction and static classification models is used to build an intelligent matching model for the industrial chain. The temporal prediction model employs a Long Short-Term Memory (LSTM) network, taking the time-series data of multi-dimensional matching features as input and outputting the predicted matching feature values ​​for the next period. The model structure includes an input layer, an LSTM layer, a fully connected layer, and an output layer. The static classification model uses a Support Vector Machine (SVM), taking the current values ​​of the multi-dimensional matching features as input and outputting the matching status level. A radial basis function (RBF) is used as the kernel function, with a penalty coefficient C set to 1.0 and a gamma value set to 0.1.

[0130] We collected nearly five years of historical data on cooperation among chemical industry chain enterprises as a training set. The data includes multi-dimensional matching features, matching status levels, and cooperation results, and was used to train the hybrid architecture model. First, a time series prediction model is trained using an adaptive moment estimator optimizer with a learning rate of 0.001, 500 iterations, and mean squared error as the loss function.

[0131] Then, the output of the time-series prediction model is combined with static feature data to train a static classification model. Cross-validation is used to optimize the model parameters to ensure that the model classification accuracy is ≥85%.

[0132] Based on multi-dimensional matching features, matching qualification thresholds, optimization thresholds, and risk thresholds are set for different sub-sectors of the industry chain. The specific thresholds are shown in the table below: ; When the multi-dimensional matching features are greater than or equal to the matching qualification threshold, it is judged as a high-quality match. When the optimization threshold is less than or equal to the multi-dimensional matching features but less than the matching qualification threshold, it is judged as a potential optimization state. When the multi-dimensional matching features are less than the risk threshold, it is judged as a risky match, and the final intelligent matching model of the industrial chain is obtained.

[0133] Real-time data collection of supply and demand dynamics, transaction execution data, and market environment data from enterprises in the chemical industry chain is used. After being aligned with multi-dimensional matching features, the data is input into the intelligent matching model of the industry chain, and the real-time matching status level and abnormal contribution analysis results are output (identifying the main abnormal feature indicators that lead to the current matching status).

[0134] Based on different matching status levels, corresponding processing measures shall be taken: If the matching status is high-quality, we will continuously monitor and record data, including enterprise supply and demand dynamic data, transaction execution data, cooperation satisfaction scores, etc. We will update the enterprise cooperation credit profile every quarter. The credit profile includes basic enterprise information, matching characteristic data, cooperation history, and credit score (calculated based on matching status, performance, etc., from 0 to 100 points). Enterprises with a credit score of ≥90 points can be upgraded to strategic partners and enjoy more preferential cooperation policies.

[0135] If the situation is in a potential optimization state, trigger a matching optimization alert and notify the relevant personnel. Extract the fluctuation threshold of the supply and demand dynamic data (standard deviation of each core dynamic characteristic indicator × 1.5), and generate adjustment suggestions for cooperation conditions based on the company's initial matching data, specifically including: If product matching is low, it is recommended to adjust product specifications or models to better align with the needs of partners. If capacity utilization is insufficient, it is recommended to adjust capacity appropriately to improve capacity matching. Based on price fluctuation response rate, it is recommended to set a more flexible dynamic pricing mechanism to cope with market price fluctuations. Combined with order delivery timeliness, it is recommended to optimize production plans or logistics solutions to shorten delivery cycles.

[0136] If the situation is one of risk matching, the risk type (capacity mismatch risk, price fluctuation risk, delivery delay risk, etc.) is identified based on the abnormal contribution analysis results. Then, targeted matching adjustment instructions and risk control measures are generated by combining the supply and demand matching impact coefficient. If the risk arises due to insufficient enterprise capacity, generate capacity adjustment suggestions (such as adding production lines, optimizing production processes, and clarifying capacity improvement targets and timelines). If the risk arises due to overcapacity, generate matching instructions for supplementary partners (selecting suitable companies from candidate partners, expanding the scope of cooperation, and absorbing excess capacity).

[0137] If price fluctuations lead to risks, suggestions for adjusting the dynamic pricing mechanism will be generated (such as setting a price fluctuation range and linking it to the market price index) or hedging operation guidelines (suggesting the use of financial instruments such as futures and options to hedge against price fluctuation risks, and specifying the types of instruments to be traded and position recommendations).

[0138] If delivery delays lead to risks, generate a logistics route optimization plan (plan the optimal logistics route based on transportation radius and actual transportation conditions to shorten transportation time) or a safety stock setting suggestion (calculate a reasonable safety stock level based on historical demand data and delivery cycle to avoid delivery delays due to stockouts; the safety stock calculation formula is: safety stock = average daily demand × maximum delivery cycle × safety factor, with the safety factor set at 1.5).

[0139] If risks arise due to insufficient cooperation flexibility, it is recommended to negotiate with partners to establish a more flexible cooperation mechanism, such as increasing the upper limit on the number of cooperation adjustments and optimizing the negotiation process for cooperation terms, so as to improve the flexibility of cooperation.

[0140] After all instructions are executed, the adjusted enterprise cooperation data is collected in real time and sent back to the intelligent matching model of the industrial chain for effect verification and parameter iteration optimization.

[0141] In summary, this embodiment provides a big data-based intelligent matching method for upstream and downstream enterprises in the chemical industry chain. By constructing a multi-dimensional feature system, it comprehensively covers core information such as static qualifications, dynamic operations, and market connections of enterprises, breaking the limitations of single-dimensional matching, improving the accuracy and suitability of matching results, and reducing the risk of breach of cooperation. By introducing a dynamic supply and demand analysis mechanism and adjusting the matching strategy in real time in conjunction with market cycle changes, the matching results can dynamically respond to market demand fluctuations and policy adjustments, enhancing the risk resistance and flexibility of the industry chain. Through standardized data processing and intelligent algorithm models, the matching process is automated and intelligent, significantly reducing manual intervention, improving matching efficiency, and lowering the cost of enterprise cooperation. By establishing a scientific cooperation priority and risk management system, it provides enterprises with clear basis for cooperation decisions, which helps leading enterprises integrate high-quality resources and creates fair cooperation opportunities for SMEs, promoting the coordinated upgrading and sustainable development of the chemical industry chain.

[0142] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0143] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent matching of upstream and downstream enterprises in a chemical industry chain based on big data, characterized in that, include: Collect multidimensional feature data of upstream and downstream enterprises in the chemical industry chain. The multidimensional feature data includes basic static feature data of enterprises and core business feature data of enterprises. Basic attribute items are extracted from the enterprise's basic static feature data to form enterprise basic data. The enterprise basic data is then analyzed using an industry chain attribute matching algorithm to obtain the enterprise's core matching feature set. By employing dynamic supply and demand analysis methods, we can analyze the dynamic changes in the supply and demand relationship of enterprises in the industrial chain with the market cycle and generate dynamic supply and demand characteristic data. The enterprise's core business characteristic data is associated with the enterprise's core matching feature set to obtain the enterprise's initial matching data; A quantitative model of the impact of supply and demand is constructed to analyze the influence weight of the dynamic characteristics of supply and demand on the initial matching data of the enterprises, and to obtain the supply and demand matching influence coefficient. Based on the core matching feature set of enterprises, the initial matching data of enterprises, and the supply and demand matching influence coefficient, an intelligent matching model for the industrial chain is built to evaluate the matching process of upstream and downstream enterprises in real time and generate matching optimization and cooperation risk management instructions. 2.The big data-based intelligent matching method for upstream and downstream enterprises in a chemical industry chain according to claim 1, characterized in that, The process of analyzing enterprise basic data using industry chain attribute matching algorithms includes: The product type, production capacity, geographical location and qualification level of the enterprise are extracted from the basic static feature data of the enterprise as core attribute parameters, and combined with the enterprise's technological research and development capabilities, supply chain response speed and environmental compliance level as auxiliary attribute parameters. Based on the characteristics of the sub-sectors of the industrial chain and the market circulation demand parameters, scenario matching parameters are obtained. Combined with core attribute parameters and auxiliary attribute parameters, enterprise matching priority and cooperation mode are determined. By associating scenario matching parameters with cooperation stability requirements and auxiliary attribute parameters with enterprise cooperation potential, a set of core matching features for enterprises is obtained through multi-dimensional weighted analysis. 3.The big data-based intelligent matching method for upstream and downstream enterprises in a chemical industry chain according to claim 1, characterized in that, The process of generating dynamic supply and demand characteristic data includes: Based on the market cycle characteristics of chemical products and the production response cycle of enterprises, three key monitoring cycles are divided into the market expansion period, the stable period and the contraction period, and dynamic monitoring data of the supply and demand status of enterprises are collected in the corresponding cycles. Outlier removal, trend smoothing, and time series alignment are performed on dynamic monitoring data to eliminate data noise and unify the time base, resulting in standardized dynamic data. Based on the standardized dynamic data, the product supply and demand balance coefficient, price fluctuation response rate, order delivery timeliness rate, and cooperation elasticity coefficient are calculated as core dynamic characteristic indicators. For different sub-sectors of the chemical industry and enterprise size types, the core dynamic characteristic indicators are calibrated for industry-specific differences to obtain supply and demand dynamic characteristic data.

4. The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data as described in claim 3, characterized in that, The process of calculating the core dynamic feature index includes: The average product supply and market demand for each key period are selected from the standardized dynamic data. The market matching trend is analyzed by the supply and demand difference. The product supply and demand balance coefficient is obtained by combining the enterprise's capacity elasticity adjustment. The product price fluctuation range and enterprise response adjustment cycle for different periods are selected from the standardized dynamic data to calculate the initial response coefficient, and the price fluctuation response rate is obtained by correcting it according to the enterprise decision-making efficiency. The actual delivery time and the agreed delivery time of the order are filtered from the standardized dynamic data to calculate the basic value of the on-time delivery rate, and the order delivery timeliness rate is obtained by combining the product transportation radius correction. The number of cooperation adjustments and cooperation satisfaction scores during the duration of the cooperation relationship are screened from the standardized dynamic data to calculate the initial elasticity value, and the cooperation elasticity coefficient is obtained by adjusting it according to industry cooperation practices.

5. The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data as described in claim 1, characterized in that, The process of obtaining the initial matching data for the enterprise includes: The core business characteristic data of the enterprise and the core matching feature set are normalized, and a mapping relationship matrix is ​​established from four dimensions: product matching, capacity matching, cost matching, and risk matching. By combining the policy requirements of the chemical industry with the qualification standards for cooperation with enterprises, a hard screening rule is set to retain the combination of enterprises that meet environmental protection standards, meet production capacity standards and have complete qualifications, forming an initial matching pool; Construct a deep neural network model, collect historical data on cooperation between companies in the industrial chain for training, input initial matching pool data, and output historical cooperation success rate scores; The random forest algorithm is used, and the optimized model parameters are verified through Bayesian optimization. The matching degree of the enterprise combination in the initial matching pool is predicted to obtain the initial matching score. The initial matching score is dynamically corrected based on the supply and demand dynamic characteristic data to obtain the initial matching data for enterprises.

6. The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data as described in claim 1, characterized in that, The process of obtaining the supply and demand matching influence coefficient includes: Collect cases of failed collaborations in the chemical industry chain due to dynamic imbalances in supply and demand, establish a mapping database of dynamic characteristics of supply and demand and types of collaboration risks, and clarify the transmission path of the impact; A risk weight assessment system is constructed based on the frequency of risk occurrence and the degree of economic loss. The analytic hierarchy process is used to determine the weight value of each risk type and to calculate the correlation strength between core dynamic characteristic indicators and cooperation risks. By combining the current deviation, risk weight value, and correlation strength of supply and demand dynamic characteristic data, a weighted summation method is used to calculate the impact score of each dynamic characteristic indicator on the enterprise's initial matching data. By integrating the impact scores of all dynamic characteristics, an initial impact coefficient is obtained. This coefficient is then calibrated using current industry chain market sentiment data to obtain a supply and demand matching impact coefficient.

7. The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data as described in claim 1, characterized in that, The process of building the intelligent matching model for the industrial chain includes: The core matching feature set of enterprises, the initial matching data of enterprises and the supply and demand matching influence coefficient are standardized in format and unified in dimension. They are then correlated by time according to the enterprise cooperation batch ID and aligned by business dimension according to the industrial chain links to obtain the fused data. Three types of fusion features—static matching basis, dynamic risk warning, and weight adjustment factor—are extracted from the fused data to construct a multi-dimensional matching feature set. A hybrid architecture combining time-series prediction and static classification models is adopted, and historical data on cooperation among enterprises in the chemical industry chain is collected as a training set for model training and parameter optimization. Based on a multi-dimensional matching feature set, matching qualification thresholds, optimization thresholds, and risk thresholds are set for different sub-sectors of the industrial chain to obtain an intelligent matching model for the industrial chain.

8. The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data as described in claim 7, characterized in that, The process of obtaining the multi-dimensional matching feature set includes: The final matching score in the initial matching data of enterprises is converted into a standardized matching score, and a static matching basis is obtained by combining the enterprise's historical cooperation fulfillment rate statistics. The dynamic risk score is obtained by weighting each core dynamic characteristic indicator based on the supply and demand matching impact coefficient, and the dynamic risk warning feature is obtained by combining the time series change slope of the dynamic characteristic indicator. A weighted adjustment factor is constructed based on the supply and demand matching influence coefficient. The influence ratio of static matching basis and dynamic risk warning is determined by the entropy weight method. The three factors are weighted and calculated to obtain a multi-dimensional matching feature set.

9. The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data as described in claim 1, characterized in that, The process of generating matching optimization and cooperation risk management instructions includes: Real-time data collection of supply and demand dynamics, transaction execution data and market environment data of enterprises in the chemical industry chain; alignment with multi-dimensional matching feature sets; input into the intelligent matching model of the industry chain; output real-time matching status level and abnormal contribution analysis results. If the matching status is a high-quality match, continuously monitor and record the data, and update the enterprise's cooperation credit profile; if it is a potential optimization status, trigger a matching optimization warning, extract the fluctuation threshold of supply and demand dynamic data, and generate suggestions for adjusting cooperation conditions based on the enterprise's initial matching data. If the situation is a risk-matched state, the risk type is identified based on the abnormal contribution analysis results, and targeted matching adjustment instructions and risk control measures are generated in combination with the supply and demand matching impact coefficient.

10. The intelligent matching method for upstream and downstream enterprises in the chemical industry chain based on big data as described in claim 1, characterized in that, The risk management measures include: To address the risk of mismatched production capacity, generate suggestions for capacity adjustment or matching instructions from partner companies; to address the risk of price fluctuations, generate suggestions for adjusting dynamic pricing mechanisms or hedging operation guidelines; to address the risk of delivery delays, generate logistics route optimization solutions or safety stock setting suggestions. After all instructions are executed, the adjusted enterprise cooperation data is collected in real time and sent back to the intelligent matching model of the industrial chain for effect verification and parameter iteration optimization.

Images