Supply chain evaluation method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a supply chain evaluation method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps of obtaining supply chain evaluation data related to a target interaction scene, wherein the supply chain evaluation data at least comprises supplier data of a first business type, inventory data of a second business type, sales data of a third business type and delivery data of a fourth business type; performing data preprocessing on the supply chain evaluation data to obtain target evaluation data; and calling an evaluation model constructed based on a neural network, and based on the target evaluation data, performing risk evaluation on the supply chain of the target interaction scene through the evaluation model to obtain a risk evaluation result. Therefore, the supply chain evaluation efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a supply chain evaluation method, apparatus, computer equipment, computer-readable storage medium, and computer program product. Background Art

[0002] With the deepening of the digital transformation of the retail industry, supply chain management has gradually developed from the traditional linear management model to the networking and intelligent direction. Today, the supply chain system is becoming increasingly complex, involving multiple links, and each link may have risks.

[0003] In the related art, evaluation is performed through collected supply chain data. However, the supply chain data collected in the prior art is relatively single and fixed, making it difficult to accurately evaluate the supply chain, that is, there is a problem of low accuracy in supply chain evaluation. Summary of the invention

[0004] Based on this, it is necessary to provide a supply chain evaluation method, apparatus, computer equipment, computer-readable storage medium and computer program product that can improve the accuracy of supply chain evaluation in response to the above technical problems.

[0005] In a first aspect, the present application provides a supply chain evaluation method, comprising:

[0006] Acquire supply chain evaluation data related to the target interaction scenario, wherein the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type;

[0007] Performing data preprocessing on the supply chain evaluation data to obtain target evaluation data;

[0008] An assessment model constructed based on a neural network is called, and based on the target assessment data, a risk assessment is performed on the supply chain of the target interaction scenario through the assessment model to obtain a risk assessment result.

[0009] In a second aspect, the present application also provides a supply chain evaluation device, comprising:

[0010] A data acquisition module, used to acquire supply chain evaluation data related to the target interaction scenario, wherein the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type;

[0011] A data preprocessing module, used to perform data preprocessing on the supply chain evaluation data to obtain target evaluation data;

[0012] The evaluation module is used to call the evaluation model built based on the neural network, and based on the target evaluation data, perform risk evaluation on the supply chain of the target interaction scenario through the evaluation model to obtain a risk evaluation result.

[0013] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0014] Acquire supply chain evaluation data related to the target interaction scenario, wherein the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type;

[0015] Performing data preprocessing on the supply chain evaluation data to obtain target evaluation data;

[0016] An assessment model constructed based on a neural network is called, and based on the target assessment data, a risk assessment is performed on the supply chain of the target interaction scenario through the assessment model to obtain a risk assessment result.

[0017] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0018] Acquire supply chain evaluation data related to the target interaction scenario, wherein the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type;

[0019] Performing data preprocessing on the supply chain evaluation data to obtain target evaluation data;

[0020] An assessment model constructed based on a neural network is called, and based on the target assessment data, a risk assessment is performed on the supply chain of the target interaction scenario through the assessment model to obtain a risk assessment result.

[0021] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0022] Acquire supply chain evaluation data related to the target interaction scenario, wherein the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type;

[0023] Performing data preprocessing on the supply chain evaluation data to obtain target evaluation data;

[0024] An assessment model constructed based on a neural network is called, and based on the target assessment data, a risk assessment is performed on the supply chain of the target interaction scenario through the assessment model to obtain a risk assessment result.

[0025] The above-mentioned supply chain evaluation method, device, computer equipment, computer-readable storage medium and computer program product obtain supply chain evaluation data related to the target interaction scenario, and the supply chain evaluation data at least includes supplier data of the first business type, inventory data of the second business type, sales data of the third business type, and delivery data of the fourth business type. That is, multiple dimensional data related to the target interaction scenario are obtained to avoid evaluation errors caused by evaluating a single data. The supply chain evaluation data is preprocessed to obtain the target evaluation data, thereby ensuring the validity and accuracy of the data for evaluation; the evaluation model constructed based on the neural network is called, and based on the target evaluation data, the supply chain of the target interaction scenario is risk evaluated through the evaluation model to obtain a more accurate and precise risk evaluation result. In this way, the accuracy of the supply chain evaluation is greatly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 A diagram of an application environment of a supply chain evaluation method in one embodiment;

[0028] Figure 2 A schematic diagram of a supply chain evaluation method in one embodiment;

[0029] Figure 3 It is a schematic diagram of evaluation in one embodiment;

[0030] Figure 4 A schematic diagram of risk management in one embodiment;

[0031] Figure 5 A schematic diagram of risk management in another embodiment;

[0032] Figure 6 A schematic diagram of a processing flow after an alarm is generated in one embodiment;

[0033] Figure 7 A schematic diagram of the priority order of nodes in one embodiment;

[0034] Figure 8 is a structural block diagram of a supply chain evaluation device in one embodiment;

[0035] Fig. 9 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0037] The supply chain evaluation method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be integrated on the server 104, or it can be placed on the cloud or other network servers. The supply chain evaluation method provided in the embodiment of the present application can be executed by the terminal 102 or the server 104 alone, or it can be executed by the terminal 102 and the server 104 in collaboration, without specific limitation.

[0038] In some embodiments, terminal 102 obtains supply chain evaluation data related to the target interaction scenario, and the supply chain evaluation data includes at least supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type; terminal 102 performs data preprocessing on the supply chain evaluation data to obtain target evaluation data; terminal 102 calls an evaluation model constructed based on a neural network, and based on the target evaluation data, performs a risk assessment on the supply chain of the target interaction scenario through the evaluation model to obtain a risk assessment result.

[0039] The terminal 102 may be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, etc. The server 104 may be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0040] In an exemplary embodiment, Figure 2 As shown, a supply chain evaluation method is provided, which is applied to computer equipment (which can be Figure 1 The terminal 102 in Figure 1 The server 104 in FIG. 1 is taken as an example to illustrate the method, which includes the following steps 202 to 206. Among them:

[0041] Step 202, obtaining supply chain evaluation data related to the target interaction scenario, the supply chain evaluation data at least including supplier data of the first business type, inventory data of the second business type, sales data of the third business type, and delivery data of the fourth business type.

[0042] Among them, the target interaction scenario is an interaction scenario for supply chain evaluation. The target interaction scenario can be a mall; it can also be a manufacturing factory, restaurant, hospital, etc. Supply chain evaluation data includes data of different business types (i.e. business data). For example, supplier data belonging to the first business type, inventory data belonging to the second business type, sales data belonging to the third business type, and delivery data belonging to the fourth business type. Among them, the supplier data can be supplier basic information, qualification certificates, credit ratings and other data. The inventory data can be item inventory (such as commodity inventory in the mall interaction scenario), inventory turnover rate and other data. Sales data (also understood as sales data, or transaction data) can be sales, gross profit margin and other data. Delivery data can be data such as delivery timeliness rate and order fulfillment rate.

[0043] Exemplarily, supplier data, inventory data, sales data, and delivery data may be any combination of data acquired by a supplier management system, a warehouse management system, a sales system, and a logistics system. For example, when the supplier data is supplier credit rating data, the delivery data may be order fulfillment rate data.

[0044] Exemplarily, business data of other business types can also be added according to the target effect, or at least one of supplier data, inventory data, sales data, and delivery data can be reduced according to actual needs. For example, when more refined risk management is needed, virtual resource data (such as financial data) can be added to capture the risk characteristics of virtual resources of suppliers, thereby further improving the accuracy and comprehensiveness of early warning. In addition, considering the impact of the diversity of business types on data quality, an adaptive data verification algorithm can also be introduced to dynamically adjust the supply chain evaluation data to ensure that high-quality supply chain data can be obtained in various business scenarios, and the supply chain evaluation data in the following steps is replaced by the adjusted data for the following corresponding processing. This can meet the needs of actual work. For example, if the delivery dimension does not need to be considered, the delivery data can be omitted, thereby reducing the workload of data collection and processing. However, in order to ensure the comprehensiveness of the evaluation in the embodiment of the present application, the data is still retained.

[0045] It should be noted that the supply chain assessment data can fully cover all aspects of the supply chain, including but not limited to supplier management, inventory management, sales management, logistics distribution, etc., so as to ensure the accuracy and comprehensiveness of the assessment. At the same time, by introducing multiple business systems and data sources, such as supplier management system, warehouse management system, sales system and logistics system, we can make full use of the advantages of various types of data, complement each other, and improve the efficiency and accuracy of early warning. This multi-dimensional data fusion strategy not only improves the flexibility and adaptability of the assessment system, but also provides a rich data source for subsequent intelligent analysis.

[0046] Step 204, preprocessing the supply chain evaluation data to obtain target evaluation data.

[0047] Among them, data preprocessing includes but is not limited to legality verification and data supplementation processing. Among them, legality verification can be regarded as a detection operation, that is, detecting the legality of supply chain evaluation data.

[0048] In some embodiments, data preprocessing is performed on the supply chain evaluation data to obtain target evaluation data, including: performing a legality check on the supply chain evaluation data, and if the legality check passes, performing data supplementation on the supply chain evaluation data to obtain target evaluation data.

[0049] Exemplarily, after the computer device performs a first legality check and a second legality check on the supply chain evaluation data, if both the first legality check and the second legality check pass, the computer device performs data supplementation processing on the supply chain evaluation data.

[0050] Exemplarily, the first legality check and the second legality check may be processed by a tester, or may be implemented by a computer device using a corresponding verification algorithm.

[0051] The first legitimacy check is to check the supplier's qualifications and credit. The second legitimacy check is to make a preliminary estimate of the authenticity of data of all business types.

[0052] Exemplarily, the first legitimacy check can be an inspection operation on the supplier's qualifications, for example, it can be a check on the supplier's qualifications, business conditions, and credit records to ensure that the supplier information is accurate. The second legitimacy check is a preliminary estimate of the authenticity of the supply chain evaluation data. Based on the two legitimacy checks, the accuracy of subsequent supply chain evaluations can be ensured.

[0053] Exemplarily, data supplementation is a data improvement operation, the purpose of which is to complete business data. The embodiment of the present application is implemented by verifying and supplementing the supply chain assessment data that has passed the legality check. Specifically, data integrity check, logical relationship verification, missing value processing, etc. are applied.

[0054] Exemplarily, data supplementation processing can also be based on data mining processing, using data mining processing technology and historical data to reasonably estimate missing data to meet the special needs of supply chain evaluation. This method can not only save a lot of data collection time and human resources, but also improve the integrity of data by using the correlation between data. In addition, in order to further improve the effect of data supplementation, data cleaning and data verification can be performed, or data prediction can be performed based on the historical data of the target interaction scenario to obtain predicted data, and the predicted data and the actually obtained supply chain evaluation data can be used as target evaluation data to achieve better preprocessing effect.

[0055] In this embodiment, through the legality verification and data supplementation processing, the legality of the target evaluation data used for subsequent evaluation can be ensured, and the amount of effective detail information in the target evaluation data is increased, thereby improving the accuracy of the supply chain evaluation process.

[0056] In some embodiments, before performing a legality verification process on the supply chain evaluation data, the method also includes: determining multiple business types involved in the supply chain evaluation data; for each business type, obtaining data of the business type from the supply chain evaluation data, and performing corresponding data processing on the data based on a data processing strategy corresponding to the business type to obtain processed data of the business type.

[0057] Among them, different business types have different data processing strategies.

[0058] Exemplarily, for the supplier data collected by the supplier management system, the computer equipment performs qualification verification, credit assessment and other processing, which requires complete data to be provided. Therefore, the corresponding data processing strategy is to verify whether the supplier data is complete. If not, re-acquire the complete data to obtain business data. For the inventory data collected by the warehouse management system, the corresponding data processing strategy is to perform inventory counting and turnover rate calculation and other processing, verify whether the inventory quantity and inventory turnover rate of the goods in the inventory data are consistent with the processed data (the actual inventory quantity and actual inventory turnover rate are obtained by processing), and update the inventory data based on the corresponding verification results to obtain the corresponding business data; for example, retain the consistent data, and if there is inconsistency, replace the inconsistent data in the inventory data with the processed data to obtain the corresponding processed data. For the sales data obtained by the sales system, the corresponding data processing strategy is to perform sales analysis and profit accounting and other processing, verify whether the sales volume and gross profit margin in the sales data are consistent with the processed data, and update the sales data based on the corresponding verification results to obtain the corresponding business data. For the delivery data collected by the logistics system, the corresponding data processing strategy is to conduct order tracking and timeliness analysis, verify whether the delivery timeliness rate and order fulfillment rate in the delivery data are consistent with the processed data, and obtain the corresponding verification results. Based on the corresponding verification results, the sales data is updated to obtain the corresponding business data.

[0059] In this embodiment, the correctness of the data itself can be verified through corresponding data processing to ensure the evaluation quality and availability of the subsequent supply chain.

[0060] In some embodiments, after obtaining the business data corresponding to each business type, the supply chain evaluation data is subjected to a legality verification process, including: performing a legality verification process on the processing data corresponding to each business type to obtain a legality verification result.

[0061] Exemplarily, when the legality check passes, data supplementation is performed on the processing data corresponding to each business type to obtain target evaluation data.

[0062] In this embodiment, a legality verification process is performed based on the processed data to improve the credibility of the legality verification and ensure the evaluation quality and availability of the subsequent supply chain.

[0063] Step 206, calling the evaluation model built based on the neural network, based on the target evaluation data, through the evaluation model, the risk evaluation of the supply chain of the target interaction scenario is performed to obtain the risk evaluation result.

[0064] The evaluation model is a neural network-based model, which is a trained model. The evaluation model is used to evaluate the risk of the supply chain in the target interaction scenario.

[0065] Exemplarily, the evaluation model can be based on an LSTM (Long Short-Term Memory) neural network as a basic model, and can also be constructed using other deep learning frameworks or algorithms, such as GRU (Gated Recurrent Unit), Transformer, and the like.

[0066] In some embodiments, based on the target assessment data, a risk assessment is performed on the supply chain of the target interaction scenario through an assessment model to obtain a risk assessment result, including: inputting the target assessment data into the assessment model to obtain the risk probability of the target interaction scenario, and obtaining a risk assessment result based on the risk probability.

[0067] Exemplarily, the computer device evaluates the risk level according to the risk probability, and determines the risk evaluation result based on the risk probability and the evaluated risk level. For example, the risk probability is compared to see whether it exceeds the threshold probability, and the risk level is determined based on the comparison result. Correspondingly, the risk evaluation result includes the comparison result and the risk level.

[0068] In this embodiment, after the target assessment model is directly input into the assessment model, the risk assessment of the target interaction scenario can be completed directly, quickly and accurately to obtain the risk probability, thereby generating a risk assessment result with high accuracy.

[0069] In some embodiments, the method further includes: when the risk assessment result indicates that the risk probability exceeds a threshold probability, issuing an alarm message matching the risk probability.

[0070] It should be noted that after exceeding the threshold probability, it is indicated that the risk level is high risk, and an alarm is required at this time, thereby generating alarm information matching the risk probability. Exemplarily, when the risk assessment result indicates that the risk probability exceeds the threshold probability, or when the risk level in the risk assessment result represents a high risk, matching alarm information is directly generated.

[0071] In addition, when the risk probability exceeds the threshold probability, the computer equipment will send the supply chain assessment data and alarm information to the cloud database in real time for subsequent analysis and recording. In addition, in order to further improve the accuracy and efficiency of early warning, the system will also dynamically adjust relevant parameters and strategies based on historical data and current alarm information, such as optimizing the setting algorithm of threshold probability and adjusting the weight coefficient of target loss. Through such continuous optimization, the system can better adapt to the supply chain assessment needs in different environments and scenarios, and improve the intelligence and automation level of alarms.

[0072] The computer equipment adopts a front-end and back-end separation design. The front-end uses the Vue.js framework (a framework for building interfaces) to develop interactive interfaces, and the back-end uses the Spring Boot framework (a lightweight framework) to provide interface services. The database uses MySQL (a relational database) to store structured data, and Redis (an open source memory data structure storage system) as a cache layer to improve system response speed. The system uses Nginx (a proxy service) to achieve load balancing to ensure stable operation in high-concurrency scenarios.

[0073] like Figure 3 As shown, it is an evaluation diagram in one embodiment. First, select the evaluation scope currently being evaluated. For example, divide the area where the target interaction scene exists into the evaluation scope, and collect data for the target interaction scene in the area to obtain the business data corresponding to the supplier management system, storage management system, sales system, and logistics system, and obtain the supply chain evaluation data. Next, preprocess the supply chain evaluation data, and after obtaining the target evaluation data, call the evaluation model, perform supply chain evaluation based on the target evaluation data, and generate an evaluation report. Among them, if the risk probability is less than the threshold probability, it is sent to the evaluation application logged in by the inspector to be displayed to the inspector for confirmation.

[0074] Of course, in other examples, different risk levels can also be handled accordingly, such as Figure 4 As shown, it is a schematic diagram of risk processing in an embodiment. For example, if the risk probability exceeds the probability threshold, it is determined that the risk level is high risk, and an alarm message is generated to warn. If the risk probability does not exceed the probability threshold and is greater than the first threshold, the risk level is determined to be medium, and real-time detection is performed. If the risk probability does not exceed the first threshold, the risk level is determined to be low risk, and the supply chain needs to be reviewed regularly. After determining the risk level, the supply chain data evaluated at the time is marked based on the risk level and stored in the risk database accordingly. Among them, the probability threshold is greater than the first threshold.

[0075] In other examples, such as Figure 5 FIG. 1 is a schematic diagram of risk processing in another embodiment. After determining that an alarm is required, or that a test is required, or that a review is required, a manual review is performed to re-verify the risk level. After the review is passed, the corresponding processing record is archived. If the review fails, new supply chain assessment data is obtained again for re-assessment.

[0076] In other examples, such as Figure 6 The figure is a schematic diagram of the processing flow after an alarm in an embodiment. Figure 6The warning information in the middle is the aforementioned alarm information. That is to say, after the high risk is determined, the corresponding application (APP) will receive the warning (alarm), and then the corresponding re-evaluation will be carried out. The evaluation process is shown in Figure 6 .

[0077] In this embodiment, once the risk probability is higher than the threshold probability, it is determined that the risk assessment result indicates that the risk probability exceeds the threshold probability, thereby automatically and quickly issuing an alarm message matching the risk probability.

[0078] Exemplarily, when the evaluation model is trained, the model training is performed based on the target loss. The target loss can be a composite loss, and the composite loss must at least include business-related losses. Other losses can be designed in other types according to the specific needs of different staff members. For example, time series losses can be added to more accurately capture the risk evolution trend, or classification losses can be added to more accurately judge the risk type. The business-related loss, time series loss, and classification loss are integrated to construct a composite loss for model training to obtain a trained model. In this way, the model can be more comprehensive and accurate in identifying supply chain risks. In addition, in order to improve the generalization ability and robustness of the model, after obtaining the sample evaluation data, the sample evaluation data is smoothed, normalized, and standardized to increase the training sample size of the model and improve its adaptability to different business scenarios. Finally, by continuously optimizing and adjusting the structure and parameters of the model, the recognition model can show higher accuracy and efficiency in practical applications, providing strong support for risk management of the supply chain. The sample evaluation data in the above training process is a sample. The sample evaluation data is determined based on sample supplier data of the first business type, sample inventory data of the second business type, sample sales data of the third business type, and sample delivery data of the fourth business type. Based on the sample and the model to be trained, the sample prediction result is obtained, and based on the sample prediction result and the label corresponding to the sample, the target loss is determined, and the model parameters are updated to obtain a trained recognition model.

[0079] For example: LSTM neural network is selected as the basic model, and the weight parameters are initialized, and the corresponding hyperparameters are set according to the actual application requirements. The setting of learning rate (Learning Rate, η) determines the pace at which the model updates the weights in each iteration. Usually, the learning rate needs to be debugged based on experience to balance the convergence speed and stability. Batch size (Batch Size, m) determines the number of samples used for each gradient update. A larger batch size will increase memory usage, but it can better estimate the gradient, and vice versa, it will help capture the local optimum. The number of iterations (Epochs, T) refers to the number of times the entire data set is fully learned by the model, usually set to 1000 to 10000 times, depending on the size and complexity of the data set. In terms of the choice of optimizer, the Adam optimizer is used, which combines the advantages of momentum and adaptive learning rate, which helps to converge faster.

[0080] Then, after the sample is input into the basic model, forward propagation and back propagation are performed, and the weight parameters are continuously adjusted to minimize the target loss (such as cross entropy loss).

[0081] For example, model training is carried out through a cycle of forward propagation and back propagation. During forward propagation, the input sample passes through each layer of the model and outputs the sample prediction result, such as the predicted risk probability p and the risk type classification c. The target loss L is a standard for evaluating the error between the sample prediction result predicted by the model (at least including the predicted business relevance, predicted risk probability, predicted risk type) and the actual label (label, at least including the actual business relevance, the actual risk probability, the actual risk type)). During training, after the sample is input into the model, the sample prediction result is obtained, and the business relevance loss is calculated based on the sample prediction result. (Using mean square error, MSE Loss), time series related loss (Using Dynamic Time Warping, DTW Loss) and classification loss Next, based on business-related losses (Using mean square error, MSE Loss), time series related loss (Using Dynamic Time Warping, DTW Loss) and classification loss (Using cross entropy, Cross-Entropy Loss), construct the target loss L:

[0082]

[0083] in,

[0084] in, Indicates the predicted business relevance, Indicates the actual business relevance.

[0085]

[0086] in, represents the true risk probability (i.e., a binary label indicating whether there is a risk), Shows the predicted risk probability (included in the sample prediction results). and Similarly, for the cross entropy loss used for multi-class classification, during backpropagation, the weight parameters of the model are adjusted through the gradient descent algorithm to minimize the target loss. , , Represents the weight coefficients of each loss function. These weight coefficients can be adjusted according to actual needs to balance the impact of different loss terms on model training. In practical applications, the selection of these weight coefficients often requires multiple experiments and tuning to achieve the best estimation effect.

[0087] The model performance is evaluated on the validation set, and the model accuracy and generalization ability are improved by adjusting hyperparameters and structural optimization. The validation set is used to evaluate the performance of the model during the model training process to avoid overfitting problems. In the embodiment of the present application, 20% of the sample evaluation data is reserved as the validation set when preparing the sample evaluation data. These sample evaluation data do not overlap with the training set and represent the performance of the model on unknown data. The performance of the model is mainly measured by indicators such as accuracy, precision, recall, and F1 score. Through these indicators, the embodiment of the present application can intuitively evaluate the estimated accuracy and stability of the model in different scenarios.

[0088] The trained model is deployed on a cloud server, and real-time reasoning is accelerated using GPU (graphics processing unit) to ensure efficient prediction and early warning. The final performance of the model is demonstrated through multiple indicators. For example, in the actual supply chain prediction process, the model can output risk probability, risk type, and confidence score in real time. By deploying the model on a cloud server and combining it with the computing power of GPU (graphics processing unit), it can quickly process the target evaluation data after receiving it and output accurate prediction results. The entire process can improve response speed and further improve real-time performance.

[0089] It should be noted that a deep learning time series model suitable for supply chain evaluation is selected, and multiple batches of model training and optimization are performed. Indicators such as target loss and accuracy are monitored during the training process to ensure that the model gradually converges on the training set. The accuracy of different models and different batches is compared to ensure that the target accuracy must reach at least 95% to pass.

[0090] The steps to determine the above samples include: obtaining historical data of multiple interaction scenarios, totaling about 1,000 samples, to ensure that multiple business types (such as supplier management, inventory management, etc.) are covered. The collected data is labeled and classified to generate a training set containing 500 training samples and a validation set of 100 samples. Data supplementation is also performed, such as supplementing 300 missing data, verifying 200 abnormal data, and correcting 150 erroneous data. The training set finally generated has a total of 1,850 samples. Data supplementation can expand the original 1,000 data samples to 1,850 samples. This data improvement technology not only helps the model better learn the various characteristics of supply chain risks, but also improves the generalization ability of the model in complex business environments. During the training process, the model will be exposed to more complete business data, so that it can better identify supply chain risks hidden in different scenarios.

[0091] It should be noted that the model evaluation data collected through analysis is used for model performance evaluation and optimization, including accuracy and speed. The estimated results are compared with the results of manual inspection one by one. In the selected test area, the estimated results and manual inspection are performed simultaneously. If an early warning is required, data is collated during the early warning process, and early warnings with an accuracy rate below 95% are manually reviewed. For special application scenarios, a manual review mechanism will also be introduced. The model's estimated results are reconfirmed through manual review to ensure the accuracy and reliability of the results.

[0092] It should be noted that after the user uses the algorithm, a large amount of data will continue to be collected in the embodiments of the present application, and the data set will be divided into multiple subsets, one of which will be used as a training set, and the remaining subsets will be used as test sets. Through multiple cross-validations, the performance of the algorithm on different data sets is evaluated. At the same time, manual inspections will be carried out from time to time, test samples will be updated, sample prediction results will be randomly checked, potential improvements will be found, and active communication will be carried out with users to collect user feedback and opinions and understand their problems and needs during use.

[0093] In the above supply chain evaluation method, by obtaining supply chain evaluation data related to the target interaction scenario, the supply chain evaluation data at least includes supplier data of the first business type, inventory data of the second business type, sales data of the third business type, and delivery data of the fourth business type. That is, multiple dimensional data related to the target interaction scenario is obtained to avoid evaluation errors caused by single data evaluation. The supply chain evaluation data is preprocessed to obtain target evaluation data, thereby ensuring the validity and accuracy of the data for evaluation; the evaluation model constructed based on the neural network is called, and based on the target evaluation data, the supply chain of the target interaction scenario is risk evaluated through the evaluation model to obtain a more accurate and precise risk evaluation result. In this way, the accuracy of the supply chain evaluation is greatly improved.

[0094] In some embodiments, obtaining supply chain evaluation data related to the target interaction scenario includes: determining a site set in the target interaction scenario, the site set including multiple different types of sites; constructing a to-be-visited set based on the business relationship between each site in the site set and other sites; obtaining the business data of each site in the to-be-visited set in turn based on the planning order indicated by the interaction rule optimization method; and determining the supply chain evaluation data related to the target interaction scenario based on the business data of each site.

[0095] The site set is a set of sites in the target interaction scenario, and the site set includes at least supplier sites that can obtain supplier data, sites that can obtain inventory data, sites that can obtain sales data, and sites that can obtain delivery data. For example, the site of supplier A, storage point B, interaction point C, logistics point D, inspection point E, the site of supplier F, and storage point G.

[0096] The business relationship refers to the interaction relationship between sites, such as the frequency or time course of interaction. The to-be-accessed set includes sites with business relationships. The business data of a site refers to the business data of the business type corresponding to the site.

[0097] The interaction rule optimization method is a method of optimizing node selection based on interaction information, wherein the planning order may refer to the site priority order or the importance priority order, which is pre-set. For example, the planning order is to select nodes according to the business relevance.

[0098] Exemplarily, the nodes that need to be accessed may be determined according to demand, and a to-be-accessed set may be formed based on each node that needs to be accessed and other nodes that have a business relationship with each node that needs to be accessed.

[0099] Exemplarily, the set to be visited can be constructed based on the interaction rule optimization method. For example, after determining the business relationship between each site in the site set and other sites, each site in the target interaction scenario is regarded as a node of the graph, and the business relationship between the nodes is regarded as an edge, and the weight of the edge represents the frequency or time process of the interaction; any node of the graph is set as the first point, and other nodes are selected according to the planning order indicated by the interaction rule optimization method, and all nodes are added to the set to be visited. In other words, each node is added to the set to be visited in turn according to the planning order.

[0100] Exemplarily, after the set to be visited is determined, the selection operation is repeated, and the process of each selection operation is: determine the starting point of the current selection operation (the current node described below) from the set to be visited, and select other nodes with business relationships from the starting point according to the planning order of the interaction rule optimization method; when other nodes in the set to be visited are visited, that is, after the business data corresponding to other nodes are obtained, the other nodes to be visited are deleted from the set to be visited, and the current selection operation is completed. This cycle is repeated until the set to be visited is an empty set, and the data interaction is completed. The starting point of the first selection operation can be randomly selected.

[0101] Take an example to illustrate the process of each selection operation: when the planning order is to select nodes according to business relevance, after selecting the current node, a node with the highest relevance is selected as the next node (access node) of the current node based on the candidate nodes that have a business relationship with the current node, and the business data of the business type to which the next node belongs is accessed, and the next node is removed from the set to be accessed, and the current operation is completed. Among them, the next node in the current selection operation is the current node of the next selection operation, and the selection operation is repeated until there are no nodes in the set to be accessed, and the accessed business data is used as the supply chain evaluation data.

[0102] In this way, business data acquisition based on the planning sequence not only ensures the collection of effective business data through the optimal interaction path, but also effectively reduces the time and system resource consumption of data acquisition, thereby improving the overall interaction efficiency. At the same time, by continuously collecting and analyzing business data from each node, the system can also more accurately identify risk points in the supply chain, providing strong support for subsequent processing work.

[0103] In other examples, the planning order of the interaction rule optimization method can also be to select nodes according to weights, and the weights can be node weights designed by technicians or users themselves, that is, the weights of sites. For example, if certain suppliers or checkpoints have higher priorities or importance, they can be given higher weights. When planning interaction paths, these high-weight nodes will be given priority, and these areas will be visited first for data interaction. This weight design method can flexibly respond to different business needs. For example, some areas may require more frequent or limited data interactions due to large business volumes, high risk levels or other factors. In this way, the interaction paths can be planned more intelligently to ensure that data acquisition tasks in important areas are given priority, further improving the pertinence and efficiency of the interaction. Figure 7 FIG. 1 is a schematic diagram of the priority order of nodes in an embodiment. Figure 7 The inspection point E in is the checkpoint E. The priority can be determined based on the weight. The higher the weight, the higher the priority.

[0104] In the above selection operation, after accessing the service data of the service type to which the access node belongs, it can be pre-determined whether the accessed service data meets the requirements. If so, the service data is obtained and the access node is removed from the to-be-accessed set. If not, the service data of the service type to which the access node belongs is obtained again.

[0105] Exemplarily, after the set to be accessed is an empty set, the data interaction is completed, that is, it is determined that the supply chain evaluation data is obtained.

[0106] The above process is essentially to generate the optimal interaction route according to the planning order, and to access the business data of each node in order according to the optimal interaction route. For example, the optimal interaction route from the starting point to the end point can be generated according to the optimal path tree. It can be expressed as:

[0107] initialization:

[0108] Update distance:

[0109]

[0110] in, For edge The weight of represents the currently known optimal distance from the starting point to the node v, and u represents the current node (the node with the highest association in the set to be visited). It refers to the distance from the starting point to itself.

[0111] In this embodiment, the interaction rule optimization method not only ensures that the interaction path is optimal, but also can efficiently cover all sites to avoid omissions or repeated visits. In addition, by judging whether it meets the requirements, it is possible to perform strict quality checks on the business data of each node, which can ensure that the collected business data meets the requirements of subsequent analysis and processing, and improve the accuracy and reliability of the overall work. In addition, the optimal interaction path can also provide clear guidance for the subsequent evaluation process, further improving the efficiency and effectiveness of the evaluation.

[0112] In a specific embodiment, the steps are specifically included:

[0113] The computer device determines a target interaction scenario and a site set in the target interaction scenario, wherein the site set includes multiple sites of different types; a to-be-visited set is constructed based on the business relationship between each site in the site set and other sites; business data of each site in the to-be-visited set is sequentially acquired based on the planning sequence indicated by the interaction rule optimization method; supply chain evaluation data related to the target interaction scenario is determined based on the business data of each site. The supply chain evaluation data includes at least supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type.

[0114] The computer device determines the multiple business types involved in the supply chain evaluation data; for each business type, the data of the business type is obtained from the supply chain evaluation data, and based on the data processing strategy corresponding to the business type, the data is processed accordingly to obtain the processed data of the business type.

[0115] The computer device performs a legality check on the processed data corresponding to each business type to obtain a legality check result. If the legality check passes, the computer device performs a data supplement process on the processed data corresponding to each business type to obtain target evaluation data.

[0116] If the quality of the target evaluation data does not pass, the supply chain data is collected again to obtain the target evaluation data that passes the quality. If the quality of the target evaluation data passes, the computer device calls the evaluation model built based on the neural network, inputs the target evaluation data into the evaluation model, obtains the risk probability of the target interaction scenario, and obtains the risk evaluation result based on the risk probability. If the risk evaluation result indicates that the risk probability exceeds the threshold probability, an alarm message matching the risk probability is issued. Specifically, after determining the risk probability, refer to the previous article Figure 4-6 processing process.

[0117] It should be noted that the above process is a process of conducting a supply chain evaluation on the supply chain of the target interaction scenario. In other embodiments, multiple supply chain evaluations can be conducted within a preset period to determine the corresponding risk probabilities (as predicted probabilities). And each time the supply chain is evaluated, the results of the risk evaluation manually conducted by the inspectors are also recorded. Therefore, within the preset period, the probability of manual evaluation during each supply chain evaluation is used as the correct probability. Therefore, based on the predicted probabilities estimated by each supply chain evaluation within the preset period and the correct probabilities corresponding to each supply chain evaluation, the accuracy is calculated. If the accuracy is higher than the target threshold, a report corresponding to the preset period is generated. If it is less than the target threshold, the estimation model needs to be manually reviewed to correct the evaluation model.

[0118] In this embodiment, by obtaining supply chain evaluation data related to the target interaction scenario, the supply chain evaluation data includes at least supplier data of the first business type, inventory data of the second business type, sales data of the third business type, and delivery data of the fourth business type. That is, multiple dimensional data related to the target interaction scenario is obtained to avoid evaluation errors caused by single data evaluation. The supply chain evaluation data is preprocessed to obtain target evaluation data, thereby ensuring the validity and accuracy of the data for evaluation; the evaluation model constructed based on the neural network is called, and based on the target evaluation data, the supply chain of the target interaction scenario is risk evaluated through the evaluation model to obtain a more accurate and precise risk assessment result. In this way, the accuracy of the supply chain evaluation is greatly improved. In addition, the data interaction path is adaptively adjusted by optimizing the interaction rules. This method significantly optimizes the data acquisition efficiency. Compared with the traditional fixed path acquisition method, the data acquisition time is reduced by 40%, and the acquisition strategy can be dynamically adjusted according to business changes. In addition, the business association loss term is introduced in the training of the above-mentioned evaluation model. Through the design of the composite loss function, the model can simultaneously consider business relevance, timing characteristics and classification accuracy. This multi-objective optimization design enables the model to achieve an early warning accuracy of more than 95% in practical applications, significantly exceeding the 72% accuracy of the traditional single evaluation model. In addition, standardized interfaces can be designed to flexibly access new data sources and alarm rules. When the business scenario changes, only the corresponding parameters need to be adjusted without changing the overall architecture, which greatly reduces maintenance costs and improves the adaptability of the system. Using the front-end and back-end separation architecture and GPU acceleration technology, combined with the optimized interactive path algorithm, the system can complete the entire process of data collection, processing and alarm analysis within 8 minutes. Compared with the traditional manual inspection method, it saves more than 90% of the time cost and significantly improves the efficiency of supply chain evaluation.

[0119] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0120] Based on the same inventive concept, the embodiment of the present application also provides a supply chain evaluation device for implementing the supply chain evaluation method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more supply chain evaluation device embodiments provided below can refer to the limitations of the supply chain evaluation method above, and will not be repeated here.

[0121] In an exemplary embodiment, Figure 8 As shown, a supply chain evaluation device 800 is provided, comprising: a data acquisition module 802, a data preprocessing module 804 and an evaluation module 806, wherein:

[0122] A data acquisition module 802 is used to acquire supply chain evaluation data related to the target interaction scenario, where the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type;

[0123] The data preprocessing module 804 is used to preprocess the supply chain evaluation data to obtain target evaluation data;

[0124] The evaluation module 806 is used to call the evaluation model built based on the neural network, and based on the target evaluation data, perform risk evaluation on the supply chain of the target interaction scenario through the evaluation model to obtain a risk evaluation result.

[0125] In some embodiments, the data acquisition module 802 is used to determine a site set in a target interaction scenario, the site set including multiple different types of sites; construct a to-be-visited set based on the business relationship between each site in the site set and other sites; obtain the business data of each site in the to-be-visited set in turn based on the planning order indicated by the interaction rule optimization method; and determine the supply chain evaluation data related to the target interaction scenario based on the business data of each site.

[0126] In some embodiments, the data preprocessing module 804 is used to perform a legality check on the supply chain evaluation data, and if the legality check passes, perform data supplementation on the supply chain evaluation data to obtain target evaluation data.

[0127] In some embodiments, the data preprocessing module 804 is also used to determine the multiple business types involved in the supply chain evaluation data; for each business type, the business type data is obtained from the supply chain evaluation data, and based on the data processing strategy corresponding to the business type, the data is processed accordingly to obtain the processed data of the business type; the data preprocessing module 804 is also used to perform legality verification processing on the processed data corresponding to each business type to obtain a legality verification result.

[0128] In some embodiments, the evaluation module 806 is used to input the target evaluation data into the evaluation model to obtain the risk probability of the target interaction scenario, and obtain the risk evaluation result based on the risk probability.

[0129] In some embodiments, the device further includes an information issuing module, which is configured to issue warning information matching the risk probability when the risk assessment result indicates that the risk probability exceeds a threshold probability.

[0130] Each module in the above supply chain evaluation device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0131] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Fig. 9 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a supply chain evaluation method is implemented.

[0132] Those skilled in the art will understand that Fig. 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0133] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0134] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0135] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0136] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0137] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0138] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0139] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A supply chain evaluation method, characterized in that: The method comprises: Acquire supply chain evaluation data related to the target interaction scenario, wherein the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type; Performing data preprocessing on the supply chain evaluation data to obtain target evaluation data; An assessment model constructed based on a neural network is called, and based on the target assessment data, a risk assessment is performed on the supply chain of the target interaction scenario through the assessment model to obtain a risk assessment result.

2. The method according to claim 1, characterized in that The obtaining of supply chain evaluation data related to the target interaction scenario includes: Determine a site set in a target interaction scenario, wherein the site set includes a plurality of sites of different types; Building a to-be-visited set based on the business relationship between each site in the site set and other sites; Based on the planning order indicated by the interaction rule optimization method, sequentially obtaining the service data of each site in the to-be-visited set; Based on the business data of each site, determine the supply chain evaluation data related to the target interaction scenario.

3. The method according to claim 1, characterized in that The data preprocessing of the supply chain evaluation data to obtain target evaluation data includes: The supply chain evaluation data is subjected to a legality check, and if the legality check passes, the supply chain evaluation data is subjected to a data supplementation process to obtain target evaluation data.

4. The method according to claim 3, characterized in that Before performing the legality verification process on the supply chain evaluation data, the method further includes: Determining multiple business types involved in the supply chain assessment data; For each business type, obtaining data of the business type from the supply chain evaluation data, and performing corresponding data processing on the data based on a data processing strategy corresponding to the business type to obtain processed data of the business type; The legitimacy verification process of the supply chain evaluation data includes: The processing data corresponding to each business type is respectively subjected to legality verification processing to obtain legality verification results.

5. The method according to claim 1, characterized in that Based on the target assessment data, the supply chain of the target interaction scenario is risk assessed by the assessment model to obtain a risk assessment result, including: The target assessment data is input into the assessment model to obtain the risk probability of the target interaction scenario, and based on the risk probability, a risk assessment result is obtained.

6. The method according to claim 5, characterized in that The method further comprises: When the risk assessment result indicates that the risk probability exceeds a threshold probability, an alarm message matching the risk probability is issued.

7. A supply chain evaluation device, characterized in that: The device comprises: A data acquisition module, used to acquire supply chain evaluation data related to the target interaction scenario, wherein the supply chain evaluation data at least includes supplier data of a first business type, inventory data of a second business type, sales data of a third business type, and delivery data of a fourth business type; A data preprocessing module, used to perform data preprocessing on the supply chain evaluation data to obtain target evaluation data; The evaluation module is used to call the evaluation model built based on the neural network, and based on the target evaluation data, perform risk evaluation on the supply chain of the target interaction scenario through the evaluation model to obtain a risk evaluation result.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images