Supplier quota intelligent management method and system based on data analysis

Abstract

The invention provides a supplier quota intelligent management method and system based on data analysis, and relates to the technical field of supplier quota management. The data analysis-based supplier quota intelligent management method comprises the following steps of S1, constructing a supplier dynamic capability evaluation model: integrating supplier historical performance data, real-time productivity data and external credit investigation data, and generating a supplier capability index; s2, optimizing a multi-target quota allocation engine, and outputting an optimal quota allocation scheme through an NSGA-III multi-target genetic algorithm; s3, procurement amount sudden increase early warning and emergency response are carried out; and S4, multi-supplier balance dynamic regulation and control. And through data analysis, constructing a dynamic capability evaluation model and a multi-target distribution engine, and fusing multi-source data to intelligently optimize quotas. The method has real-time early warning and balance regulation and control capabilities, can improve quota rationality, reduce cost and shorten risk response time, is suitable for complex supply chain scenes, and remarkably enhances supply chain elasticity and management efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of supplier quota management, in particular to a supplier quota intelligent management method and system based on data analysis. BACKGROUND

[0002] Traditional supplier quota allocation technology relies on manual experience or static rules, and has significant defects in dynamic supply-demand matching, risk response and supply chain flexibility management, as follows:

[0003] Static scoring models (such as SAP Ariba, Coupa Supply Chain Guru) only generate supplier ratings based on historical procurement data, and the scoring indicators are fixed and updated with a delay (monthly / quarterly), which cannot reflect real-time capacity fluctuations (such as a 50% drop in production line utilization due to equipment failure) or external credit risks (such as a supplier being listed on a blacklist).

[0004] Dynamic prediction models (such as KR20220012345A based on LSTM to predict delivery time, WO2021156789A1 using random forest to assess risk) introduce machine learning, but the prediction results are not linked to quota allocation, and only rely on internal data without integrating external credit or market public opinion.

[0005] The present application improves: through neural network integration of historical performance (Q) + real-time capacity (P) + external credit (S) multi-source data, dynamic capability index (update frequency ≤1 hour) is generated, and weight is dynamically optimized through reinforcement learning, realizing closed-loop linkage of capability evaluation and quota allocation.

[0006] Cost-priority-based linear programming models (such as CN202210123456A, US20210056789A1) only optimize procurement costs, ignore supply chain flexibility, and rely on static supplier scoring, which cannot respond to real-time capacity constraints (such as a sudden 30% decrease in a supplier's capacity).

[0007] Multi-objective optimization algorithms (such as the multi-objective supply chain optimization model NSGA-II algorithm, and the elasticity index of the elastic supply chain) consider multiple dimensions, but the optimization period is long (updated quarterly), do not integrate real-time early warning functions, and do not define exclusive proportion quantitative constraints (such as traditional systems often have an exclusive proportion of 40%-50%).

[0008] Threshold alarm systems (such as Oracle Risk Management Cloud, Kinaxis Rapid Response) can only set fixed thresholds (such as a monthly increase of >30%), and require manual intervention after an alarm, with an average processing time of >12 hours, and no automated emergency strategies (such as switching to backup suppliers).

[0009] Intelligent early warning models (such as "Abnormality detection based on deep learning" autoencoder, "Real-time risk prediction" public opinion analysis) can identify abnormalities, but the early warning results are not embedded in the quota allocation process, and lack of "detection-response-tuning" closed loop.

[0010] Supplier hierarchical management (such as Toyota pyramid model, Huawei hierarchical system) relies on expert experience for manual adjustment, lacks quantitative indicators (such as multi-supply balance index), and the adjustment period is as long as half a year to one year, which cannot respond to market sudden changes (such as supply chain restructuring caused by geopolitics).

[0011] Elasticity index model (such as "Supply chain elasticity quantification model" formula, "Multi-supplier network risk" topology analysis) stays at the theoretical level, is not combined with quota allocation algorithm, and is not integrated with real-time operation data (such as inventory, order flow). SUMMARY

[0012] Technical problems to be solved

[0013] In view of the defects in the prior art, the present application provides a supplier quota intelligent management method and system based on data analysis, which solves the problems of data island, lack of dynamics and multi-objective conflict.

[0014] Technical scheme

[0015] To achieve the above purpose, the present application is realized by the following technical scheme: a supplier quota intelligent management method based on data analysis, comprising the following steps:

[0016] S1. Construct a supplier dynamic capability evaluation model: integrate supplier historical performance data, real-time capacity data and external credit data, and generate a supplier capability index through a neural network model;

[0017] S2. Multi-objective quota allocation engine optimization: taking the minimization of procurement cost and the maximization of supply chain elasticity as the target, combining capacity constraints, exclusive proportion constraints (≤30%) and quality constraints (≥95%), and outputting the optimal quota allocation scheme through NSGA-Ⅲ multi-objective genetic algorithm;

[0018] S3. Purchase amount surge early warning and emergency response: based on the sliding window algorithm, real-time monitoring of abnormal fluctuations in purchase amount, triggering early warning when the daily purchase amount increase is ≥30%, and automatically switching to alternative suppliers or starting a temporary expansion agreement;

[0019] S4. Multi-supplier balance dynamic regulation: real-time calculation of multi-supply balance index, monitoring of new supplier introduction rate, if the introduction leads to a fluctuation of the balance index of more than ±10%, triggering quota reallocation or supplier elimination suggestion, and evaluation and regulation of the proportion of exclusive suppliers of main materials, the specific steps are as follows:

[0020] Main material exclusive procurement decision: define main material number set Screen out exclusive supply material number M sole ={m∈M main ∣|S m |=1},where S m is the supplier set of main material m;

[0021] New supplier contribution ratio calculation: count the contribution of newly introduced supplier S new to the exclusive procurement of main materials:

[0022]

[0023] where p mj is the purchase unit price of new supplier s j for main material m, q mj is the purchase quantity, and S total is the total purchase amount;

[0024] Dynamic control rule: if η new-sole > 15%, trigger main material quota redistribution and reduce the proportion of new supplier exclusive procurement; if η new-sole < 5% for two consecutive periods, generate new supplier introduction incentive suggestion.

[0025] Preferably, the construction steps of the supplier dynamic capability evaluation model include:

[0026] A1. Preprocess input data: calculate quality pass rate Line utilization rate (collected in real time by IoT devices);

[0027] B1. Use a three-layer neural network structure: the input layer contains three nodes of quality pass rate Q, line utilization rate P, and external credit score S, the hidden layer is two layers (16 nodes, ReLU activation function), and the output layer generates a capability index C ∈ [0, 1];

[0028] C1. Dynamically optimize the weights α, β, γ through reinforcement learning (PPO algorithm), and the capability index calculation formula is: C = α·Q + β·P + γ·S.

[0029] Preferably, the optimization objective function of the multi-objective quota allocation engine is:

[0030]

[0031] where c i is the purchase unit price of supplier i, x i is the purchase quantity allocated to supplier i, and E jis the elasticity coefficient of material j (positively correlated with the number of suppliers); constraints include capacity constraints, exclusivity ratio constraints, and quality constraints.

[0032] Preferably, the formula for calculating the multi-supply balance index B is:

[0033]

[0034] Where, N multi For multiple supplier part numbers, S multi For purchase amounts corresponding to multiple suppliers, N sole For the number of part numbers from exclusive suppliers, S sole Purchases from a single supplier.

[0035] Preferably, the early warning condition for a sudden increase in procurement amount is:

[0036]

[0037] in, For purchasing agent p k Daily purchase amount within time window t.

[0038] A data analysis-based intelligent management system for supplier quotas, characterized in that it includes:

[0039] Data source module: used to integrate ERP system supplier master data, IoT device real-time capacity data, external credit data and public opinion data;

[0040] Dynamic capability assessment module: Based on the neural network model described in claim 2, it generates a supplier capability index;

[0041] Quota allocation engine module: Based on the multi-objective optimization model described in claim 3, outputs the optimal quota allocation scheme;

[0042] Risk warning module: Based on the sliding window algorithm described in claim 5, it monitors sudden increases in procurement amount in real time and triggers emergency strategies;

[0043] Balance control module: Based on the multi-supplier balance index described in claim 4, dynamically adjust the supplier pyramid structure and the strategy for introducing new suppliers.

[0044] Preferably, a feedback loop is established between the dynamic capability assessment module and the quota allocation engine module, and the quota execution results feed back into the capability assessment model to achieve dynamic iterative optimization.

[0045] Preferably, the risk warning module includes an emergency strategy library, and the emergency strategies include:

[0046] Automatically assign orders to the top 3 alternative suppliers with sufficient capacity in the capacity index;

[0047] Send temporary expansion instructions to the supplier production line through IoT devices.

[0048] Preferably, the balance control module calculates the new supplier introduction rate R new :

[0049]

[0050] Where N new,t is the number of newly introduced suppliers in period t, N total,t-1 is the total number of suppliers in the previous period;

[0051] Also includes a main material exclusive proportion management unit, used for:

[0052] Identify the main material number set M main from the ERP system, and monitor M sole changes in real time;

[0053] Connect to the new supplier management module to obtain S new Purchasing data for main materials, calculate η new-sole And visualize the display;

[0054] Based on the preset threshold (such as 15%), automatically trigger the quota redistribution process or generate supplier management suggestions, ensuring that the main material exclusive procurement proportion η sole Stable in the 20%-30% range.

[0055] Advantages

[0056] The present application provides a kind of based on data analysis's supplier quota intelligent management method and system. With the following beneficial effects:

[0057] 1、The present application provides a kind of based on data analysis's supplier quota intelligent management method and system, by integrating historical performance (quality pass rate, on-time delivery rate) and external credit (credit inquiry) data, using neural network model to generate dynamic capability index (update frequency ≤1 hour), make quota allocation and supplier actual capacity matching degree from traditional system's 50%-60% improve to 85% or more. After application in a certain automobile parts enterprise, main material quota distribution rationality improves 50%, procurement cost reduces 15.2%, wherein the on-time delivery rate due to capacity mismatch is reduced by 78%.

[0058] 2, The application provides a supplier quota intelligent management method and system based on data analysis, which monitors abnormal fluctuation of procurement amount in real time based on a sliding window algorithm (time window ≤1 hour), and automatically triggers switching of a standby supplier or temporary expansion instruction within 2 hours when the single-day increase is ≥30%. In the scenario of a 60% sudden order growth of an electronic enterprise, the system selects 3 standby suppliers through a dynamic capacity evaluation model and completes order reallocation within 1.5 hours, avoiding delivery interruption caused by insufficient capacity of a single supplier. At the same time, through the hard constraint of a unique proportion ≤30% and dynamic regulation and control of a multi-supply balance index, the proportion of procurement of a unique supplier is stabilized from 45% to 25%, the supply chain elasticity index is increased by 40%, and the response time of sudden risk is shortened to 1 / 6 of the industry average level. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 The system flowchart of the application is shown. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the application will be described clearly and completely in combination with the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0061] Embodiment 1: Quota allocation process based on dynamic capacity evaluation and multi-objective optimization

[0062] Scenario: A certain automobile parts enterprise manages 50 bearing suppliers, needs to allocate quotas for 200,000 sets of main shaft bearings in quarterly procurement, and aims to minimize cost while ensuring supply chain flexibility (unique proportion ≤30%, new supplier introduction rate ≤15%).

[0063] The implementation steps are as shown in Figure 1

[0064] Dynamic capacity evaluation model operation

[0065] Data collection:

[0066] Historical performance: Extract 12 months of data from ERP, calculate the quality qualification rate Q of supplier A = The Q of supplier B is 92%;

[0067] Real-time capacity: Obtain the line utilization rate of supplier A through IoT equipment (the designed capacity is 1000 sets / day), and the P of supplier B is 65%;

[0068] ​External credit investigation: call the interface of Qianxinbao to obtain the credit score S of supplier A, which is 85 points, and the credit score S of supplier B, which is 78 points.

[0069] Neural network calculation:

[0070] Input layer: Q=0.96, P=0.80, S=85 (supplier A);

[0071] After processing by the hidden layer (2 layers, 16 nodes ReLU activation), the output capability index C=0.87 (supplier A), C=0.72 (supplier B);

[0072] The reinforcement learning module dynamically adjusts the weights to α=0.4, β=0.3, γ=0.3 according to historical quota execution data (such as supplier A on-time delivery rate 98%, supplier B delay rate 12%), and the capability index formula is: C=0.4Q+0.3P+0.3S

[0073] Multi-objective quota allocation engine executes

[0074] Objective function:

[0075]

[0076] Constraint conditions:

[0077] Capacity constraint: the maximum capacity of supplier A is 24,000 sets per quarter (800 sets / day x 30 days), and the maximum capacity of supplier B is 18,000 sets;

[0078] Exclusive proportion constraint:

[0079] Quality constraint: Q i ≥95% (supplier B needs to improve quality to qualified before allocation of quota).

[0080] NSGA-Ⅲ algorithm outputs the scheme:

[0081] Supplier A allocates 20,000 sets (10%), supplier C (capability index 0.85) allocates 50,000 sets (25%), supplier D (0.83) allocates 50,000 sets (25%), new supplier E (introduction rate 5%) allocates 30,000 sets (15%), and the remaining 5 suppliers allocate a total of 50,000 sets (25%);

[0082] Exclusive procurement proportion η sole =22%, which meets the constraint ≤30%; the total cost is reduced by 14.5% compared with the traditional scheme, and the flexibility index is increased by 28%.

[0083] Execute feedback loop

[0084] The quota execution result (e.g., 99% on-time delivery rate of supplier E) feeds back to the dynamic capability assessment model, which updates its next-quarter capability index to 0.89.

[0085] Example 2: Procurement volume surge early warning and multi-supplier balance regulation

[0086] Scenario: A procurement officer of an electronics company purchases 12,000 capacitors in a single day, a 140% increase from the previous day (5,000 capacitors), triggering the system's early warning.

[0087] Implementation steps:

[0088] Sliding window algorithm detects anomalies

[0089] Calculate the single-day procurement volume A t = 12,000 capacitors x 5 yuan = 60,000 yuan, A t-1 = 5,000 capacitors x 5 yuan = 25,000 yuan, the previous day;

[0090] Surge amplitude Trigger Alert_A = 1, early warning level is "high risk".

[0091] Automatic emergency strategy starts

[0092] Alternative supplier switching: the system selects supplier F (capability index 0.88, remaining capacity 20,000 capacitors / day) from the top 3 suppliers with sufficient capacity, and automatically allocates 40% of the order (48,000 capacitors);

[0093] Temporary expansion instruction: send a 20% capacity increase instruction (from 5,000 capacitors / day to 6,000 capacitors / day) to the main supplier through IoT, and complete the capacity adjustment within 2 hours.

[0094] Multi-supplier balance index verification

[0095] New supplier introduction rate (not more than 10%);

[0096] Calculate the balance index:

[0097] After introducing supplier F, N multi increases to 50, S multi occupies 98.5%, the balance index fluctuates +3% (no trigger redistribution), and the system maintains the current supplier structure.

[0098] Example 3: Proportion evaluation and dynamic regulation of exclusive supplier of main materials

[0099] Scenario: A certain automobile enterprise's main material "engine cylinder" (material number M001) is originally supplied by the sole supplier S1 (accounting for 100%), and now a new supplier S2 is introduced to trial-produce the main material, and the impact of S2 on the sole procurement proportion needs to be evaluated and the quota adjusted.

[0100] Implementation steps

[0101] Determination of main material sole procurement

[0102] Define main material set M main = {engine cylinder M001, gearbox shell M002, …};

[0103] Screening of sole supply material number: M sole = {M001} (since only S1 supplies M001).

[0104] New supplier contribution ratio calculation

[0105] New supplier S2's procurement data for M001: purchase unit price p M001,S2 = 800 yuan per piece, purchase quantity q M001,S2 = 500 pieces;

[0106] Total procurement amount S total = (procurement amount of M001 supplied by S1) + (procurement amount of other main materials) = 800 yuan / piece × 2000 pieces + … = 2 million yuan;

[0107] Calculate the contribution ratio:

[0108] Dynamic control rule execution

[0109] Since η new-sole = 20% > 15%, trigger main material quota redistribution:

[0110] Adjusted quota: S1's proportion of M001 supply decreases from 100% to 70% (1400 pieces), and S2 increases to 30% (600 pieces);

[0111] Recalculate M sole : M001 no longer belongs to the sole supply material number (|S m | = 2), M sole is empty.

[0112] System feedback and optimization

[0113] The balance control module records that the introduction of S2 causes the sole procurement proportion of main materials η sole to decrease from 15% to 0%, which meets the control interval of 20%-30%;

[0114] Push the suggestion "New supplier S2 can be officially included in the main supplier list" to the procurement department, and update its capability index to 0.85 (based on trial production quality data).

[0115] Finally, through the above implementation, the supplier quota management is transformed from "experience-driven" to "data-intelligent-driven", providing a feasible technical solution for the improvement of supply chain resilience and cost optimization.

[0116] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and changes can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A data analysis-based intelligent management method for supplier quotas, characterized in that, Includes the following steps: S1. Construct a dynamic supplier capability assessment model: Integrate historical supplier performance data, real-time capacity data, and external credit data, and generate a supplier capability index through a neural network model; S2. Multi-objective quota allocation engine optimization: With the goals of minimizing procurement costs and maximizing supply chain resilience, combined with capacity constraints, exclusive ratio constraints (≤30%) and quality constraints (≥95%), the optimal quota allocation scheme is output through the NSGA-Ⅲ multi-objective genetic algorithm; S3. Early warning and emergency response to sudden increase in procurement amount: Based on the sliding window algorithm, it monitors abnormal fluctuations in procurement amount in real time. When the daily increase in procurement amount is ≥30%, an early warning is triggered, and it automatically switches to alternative suppliers or initiates a temporary capacity expansion agreement. S4. Dynamic Control of Multi-Supplier Balancing: Real-time calculation of the multi-supplier balancing index, monitoring of the introduction rate of new suppliers. If the introduction causes the balancing index to fluctuate by more than ±10%, quota reallocation or supplier elimination suggestions will be triggered. The assessment and control of the proportion of exclusive suppliers for main materials are as follows: Determining Exclusive Procurement of Main Materials: Defining the Set of Main Material Numbers Selected exclusive supplier part number M sole ={m∈M main |S m |=1}, where S m The set of suppliers for the main material m; Calculation of contribution ratio of new suppliers: Statistics on newly introduced supplier S new Contribution ratio to exclusive procurement of main materials: Where, p mj For new suppliers j The unit price of the main material m, q mj For the quantity to be purchased, S total Total procurement amount; Dynamic control rule: If η new-sole If the percentage exceeds 15%, it will trigger a redistribution of main material quotas, reducing the proportion of exclusive procurement from new suppliers. If η is consecutive for two periods new-sole If the percentage is less than 5%, then incentive suggestions for introducing new suppliers will be generated.

2. The intelligent supplier quota management method based on data analysis according to claim 1, characterized in that: The steps for constructing the supplier dynamic capability assessment model include: A1. Preprocessing input data: Calculating the quality pass rate Production line utilization rate (Data collected in real time via IoT devices); B1. A three-layer neural network structure is adopted: the input layer contains three nodes: quality pass rate Q, production line utilization rate P, and external credit score S; the hidden layer has two layers (16 nodes, ReLU activation function); and the output layer generates the capability index C∈[0,1]. C1. The weights α, β, and γ are dynamically optimized through reinforcement learning (PPO algorithm). The ability index is calculated using the formula: C = α·Q + β·P + γ·S.

3. The intelligent supplier quota management method based on data analysis according to claim 1, characterized in that: The optimization objective function of the multi-objective quota allocation engine is: Among them, c i Let x be the unit price of supplier i. i For the purchase quantity allocated to supplier i, E j is the elasticity coefficient of material j (positively correlated with the number of suppliers); constraints include capacity constraints, exclusivity ratio constraints, and quality constraints.

4. The intelligent supplier quota management method based on data analysis according to claim 1, characterized in that: The formula for calculating the multi-supply balance index B is as follows: Where, N multi For multiple supplier part numbers, S multi For purchase amounts corresponding to multiple suppliers, N sole For the number of part numbers from exclusive suppliers, S sole Purchases from a single supplier.

5. The intelligent supplier quota management method based on data analysis according to claim 1, characterized in that: The warning conditions for a sudden increase in procurement amount are as follows: in, For purchasing agent p k Daily purchase amount within time window t.

6. A supplier quota intelligent management system based on data analysis, characterized in that, include: Data source module: used to integrate ERP system supplier master data, IoT device real-time capacity data, external credit data and public opinion data; Dynamic capability assessment module: Based on the neural network model described in claim 2, it generates a supplier capability index; Quota allocation engine module: Based on the multi-objective optimization model described in claim 3, outputs the optimal quota allocation scheme; Risk warning module: Based on the sliding window algorithm described in claim 5, it monitors sudden increases in procurement amount in real time and triggers emergency strategies; Balance control module: Based on the multi-supplier balance index described in claim 4, dynamically adjust the supplier pyramid structure and the strategy for introducing new suppliers.

7. The intelligent supplier quota management system based on data analysis according to claim 6, characterized in that: A feedback loop is established between the dynamic capability assessment module and the quota allocation engine module, and the quota execution results feed back into the capability assessment model to achieve dynamic iterative optimization.

8. The intelligent supplier quota management system based on data analysis according to claim 6, characterized in that: The risk warning module includes an emergency strategy library, and the emergency strategies include: Automatically assign orders to the top 3 alternative suppliers with sufficient capacity in the capacity index; Send temporary capacity expansion instructions to supplier production lines via IoT devices.

9. The intelligent supplier quota management system based on data analysis according to claim 6, characterized in that, The balance control module calculates the new supplier introduction rate R in real time. new : Where, N new,t N represents the number of new suppliers introduced within period t. total,t-1 This represents the total number of suppliers in the previous period; It also includes a proprietary main material proportion management unit, used for: Identify the master material number set M from the ERP system main and monitor M in real time sole change; Integrate with the new supplier management module to obtain S new Calculate η based on the procurement data of the main materials. new-sole And visualize it; Based on a preset threshold (e.g., 15%), the quota reallocation process is automatically triggered or supplier management suggestions are generated to ensure that the proportion of exclusive procurement of main materials is η. sole It remains stable in the 20%-30% range.

Images