CN117875724B

CN117875724B - Purchasing risk management and control method and system based on cloud computing

Info

Publication number: CN117875724B
Application number: CN202410275515.6A
Authority: CN
Inventors: 韩青周
Original assignee: Shenzhen Shengsheng Technology Co ltd
Current assignee: Shenzhen Shengsheng Technology Co ltd
Priority date: 2024-03-12
Filing date: 2024-03-12
Publication date: 2024-06-14
Anticipated expiration: 2044-03-12
Also published as: CN117875724A

Abstract

The invention relates to the technical field of purchasing risk management and control, in particular to a purchasing risk management and control method and system based on cloud computing. The method comprises the following steps: acquiring historical purchasing logs, logistics supplementary node data and warehouse data; heterogeneous edge migration calculation is carried out on the logistics supplementary node data, and supplementary node distributed edge data are generated; performing neural network optimization and transformation on the complementary node distributed edge data to construct a node distributed neural network; performing purchasing behavior time sequence analysis on the historical purchasing log to generate historical purchasing behavior time sequence data; performing risk perception association analysis on the historical purchasing behavior time sequence data to generate purchasing risk influence association data; and carrying out transient risk evolution analysis on the purchasing risk influence associated data so as to obtain risk evolution response data. The invention realizes timely and accurate purchasing risk management and control.

Description

Purchasing risk management and control method and system based on cloud computing

Technical Field

The invention relates to the technical field of purchasing risk management and control, in particular to a purchasing risk management and control method and system based on cloud computing.

Background

In modern business environment, purchasing management is a key link in enterprise operation, and with the development of globalization and complex supply chain networks, purchasing risk management becomes more and more challenging, traditional purchasing risk management methods generally rely on manual processing and manual judgment, and have the problems of untimely information and inaccurate decision, so that an intelligent purchasing risk management and control method is needed to meet the increasing purchasing risk management demands.

Disclosure of Invention

The invention provides a purchasing risk management and control method and system based on cloud computing for solving at least one technical problem.

In order to achieve the above purpose, the invention provides a purchasing risk management and control method based on cloud computing, which comprises the following steps:

step S1: acquiring historical purchasing logs, logistics supplementary node data and warehouse data; heterogeneous edge migration calculation is carried out on the logistics supplementary node data, and supplementary node distributed edge data are generated; performing neural network optimization and transformation on the complementary node distributed edge data to construct a node distributed neural network;

Step S2: performing purchasing behavior time sequence analysis on the historical purchasing log to generate historical purchasing behavior time sequence data; performing risk perception association analysis on the historical purchasing behavior time sequence data to generate purchasing risk influence association data;

Step S3: performing transient risk evolution analysis on the purchasing risk influence associated data so as to obtain risk evolution response data; performing dynamic risk time-varying network optimization on the node distributed neural network based on the risk evolution response data, and constructing a risk time-varying multidimensional network map;

Step S4: performing virtual space risk simulation processing on the risk time-varying multidimensional network map to generate risk scenario simulation data; carrying out multi-stage risk situation evolution prediction on the risk scenario simulation data to generate multi-stage risk situation evolution prediction data;

step S5: performing circulation delay distribution rule analysis on the historical purchasing behavior time sequence data to generate circulation delay distribution data; carrying out situation evolution path analysis on the multi-stage risk situation evolution prediction data to generate situation evolution path prediction data; carrying out flow track deviation recognition on situation evolution path prediction data based on the flow delay distribution data so as to generate a flow risk area;

Step S6: performing risk adaptability monitoring treatment on the flow direction risk area to obtain a dynamic real-time risk value; and carrying out data-driven mining modeling on the warehouse data by utilizing the dynamic real-time risk value, and constructing a dynamic purchasing risk decision model so as to execute purchasing risk management and control operation.

According to the invention, key data in the purchasing process are obtained through obtaining historical purchasing logs, logistics supplementary node data and storage data, and a data basis is provided for subsequent risk analysis, heterogeneous edge migration calculation effectively performs distributed processing on the logistics supplementary node data, data processing efficiency and expandability are improved, the supplementary node distributed edge data are optimized and processed through neural network optimization transformation, a node distributed neural network is constructed, a more accurate and reliable data basis is provided for subsequent risk analysis, time-sequential processing can be performed on the historical purchasing logs through purchasing behavior time-sequential analysis, time-sequential characteristics of purchasing behaviors are extracted, a time-sequential data basis is provided for subsequent risk analysis, a correlation relation between purchasing behaviors is mined from the historical purchasing behavior time-sequential data through risk perception correlation analysis, potential risk factors are identified, purchasing risk influence correlation data are generated, basis is provided for subsequent risk analysis, dynamic analysis of the correlation data can be performed through transient risk evolution analysis, evolution and response characteristics are revealed, thus, the dynamic risk response data can be obtained, the dynamic risk analysis is based on the dynamic risk analysis, the real-time-sequential risk analysis is simulated, the real-time risk situation is simulated, the real-time situation is not predicted, the real-time risk is simulated, the real-time situation is provided for the risk is simulated, the multi-stage risk situation evolution prediction is carried out based on the risk situation simulation data, the multi-stage risk situation evolution prediction is carried out, the identified risk development trend and evolution path are provided with prediction basis for purchasing risk management and control decision, circulation delay distribution rule analysis is carried out, statistical characteristics and distribution rules of circulation delay are extracted from historical purchasing behavior time sequence data, circulation delay distribution data are generated, delay information is provided for subsequent risk analysis, situation evolution path analysis is carried out based on the multi-stage risk situation evolution prediction data, the risk situation evolution path between different stages is analyzed, situation evolution path prediction data are generated, path selection basis is provided for purchasing risk management and control decision, flow track deviation identification is carried out on the situation evolution path prediction data based on the circulation delay distribution data, a flow direction risk area, namely a purchasing flow direction area with risk is identified, real-time monitoring is carried out on the flow direction risk area according to real-time data and prediction model, dynamic real-time risk value is obtained, data-driven mining modeling is carried out on the storage data, dynamic risk decision support and purchasing risk management and control strategy is carried out, and purchasing risk management and control strategy is provided.

Preferably, step S1 comprises the steps of:

Step S11: acquiring historical purchasing logs, logistics supplementary node data and warehouse data;

step S12: analyzing the cross-modal deep structure data of the logistics supplementary node data to generate node multi-modal data;

step S13: heterogeneous edge migration calculation is carried out on the node multi-mode data, and complementary node distributed edge data are generated;

Step S14: and carrying out neural network optimization and reconstruction on the complementary node distributed edge data to construct a node distributed neural network.

According to the invention, key data in the purchasing process, including purchasing orders, supplier information, logistics node information, warehouse information and the like, are obtained through obtaining historical purchasing logs, logistics supplementary node data and warehouse data, multi-modal feature extraction and analysis are carried out on the logistics supplementary node data through cross-modal deep structure data analysis, potential information in the data is fully mined, different modalities of the node data such as text description, images and sounds are analyzed, the system obtains more comprehensive and rich node features, thereby providing a more accurate data basis for subsequent risk analysis, heterogeneous edge migration calculation effectively carries out distributed processing and calculation on the node multi-modal data, distributes the data to edge nodes for processing, reduces the cost of data transmission and calculation, improves the efficiency and expandability of data processing, can better adapt to a distributed computing environment, provides faster and flexible data support for subsequent risk analysis, and the neural network optimization and reconstruction utilizes the distributed edge data to carry out optimization and construction of the neural network.

Preferably, the specific steps of step S2 are:

Step S21: performing purchasing behavior time sequence analysis on the historical purchasing log to generate historical purchasing behavior time sequence data;

step S22: unstructured risk factor extraction processing is carried out on the time sequence data of the historical purchasing behavior so as to generate purchasing risk factor data;

step S23: analyzing risk potential driving factors of the purchasing risk factor data to obtain purchasing risk factors;

Step S24: and performing risk perception association analysis on the purchasing risk factors to generate purchasing risk influence association data.

According to the invention, time sequence modeling and analysis are carried out on a historical purchasing log through purchasing behavior time sequence analysis, time characteristics and rules of purchasing behavior are revealed, seasonal, periodical and trend changes of purchasing behavior and purchasing risk conditions occurring in different time periods are helped to be known, unstructured risk factor extraction processing extracts potential risk factors from historical purchasing behavior time sequence data, such as supplier credit conditions, market price fluctuation, raw material supply stability and the like, a purchasing risk factor data set is established through extracting the factors, a data basis is provided for subsequent risk analysis and evaluation, risk potential driving factor analysis carries out statistics and modeling on purchasing risk factor data, factors which have important influence on purchasing risk are identified and analyzed, including key factors related to purchasing risk and fundamental driving factors which have important influence on purchasing risk are determined through statistical analysis, machine learning and the like, guidance and decision support are provided for risk management and control through the analysis, association relation between purchasing risk factors is modeled and analyzed through risk perception association analysis, so that the degree of interaction and association between different factors is explored, the association relation between important risk factors is more accurately provided for risk management and risk control, and risk management and risk control are provided for comprehensively controlling and risk management and risk control.

Preferably, the specific steps of step S3 are:

Step S31: performing transient risk evolution analysis on the purchasing risk influence associated data to generate transient risk evolution data;

Step S32: performing risk evolution response simulation on the transient risk evolution data to obtain risk evolution response data;

step S33: and carrying out dynamic risk time-varying network optimization on the node distributed neural network based on the risk evolution response data, and constructing a risk time-varying multidimensional network map.

According to the invention, through analysis of transient risk evolution analysis on associated data of influence of purchasing risks, the transient variation situation of risks is known, rapid variation and sudden events of purchasing risks are captured, more timely and accurate risk assessment is provided, through generation of transient risk evolution data, dynamic characteristics of risks are better understood, real-time decision support is provided for purchasing risk management and control, risk evolution response simulation simulates and predicts transient risk evolution data, so that the evolution trend and response behavior of purchasing risks are known, through simulation of different scenes and paths of risk evolution, the development trend of purchasing risks is predicted, and identified risk burst points and key nodes are provided, through generation of risk evolution response data, decision basis with more predictability and initiative is provided for purchasing risk management and control, dynamic risk time-varying network optimization utilizes the risk evolution response data to optimize and update node distributed neural networks so as to adapt to time-varying characteristics of purchasing risks, through adjustment of network structure and parameters according to the risk evolution response data, adaptability and response capacity of the network are improved, the time-varying mode and the time-varying trend are more accurately captured, multidimensional network is constructed, multi-dimensional information of risks is integrated into a corresponding risk management and control of purchasing risks is provided, and corresponding risk management and control measures are provided, real-time-varying risk management and analysis is facilitated.

Preferably, the specific steps of step S31 are:

step S311: performing transient serialization processing on the purchasing risk influence associated data to generate a purchasing risk influence transient sequence;

Step S312: performing time-frequency analysis on the transient sequence of the purchasing risk influence, and extracting transient risk frequency component characteristic data;

Step S313: carrying out change trend track analysis on the transient risk frequency component characteristic data to generate transient risk trend track data;

Step S314: and carrying out transient risk evolution analysis on the transient risk trend track data to generate transient risk evolution data.

According to the invention, the transient serialization processing is used for converting the purchasing risk influence associated data into the transient sequence, the time relation of the risk influence is converted into the form of the sequence, so that more detailed and accurate analysis is facilitated to be performed on the transient change of the purchasing risk, the characteristics and the trend of the rapid change of the risk are captured, the frequency component characteristics in the transient sequence are extracted through the time-frequency analysis, namely, the risk change characteristics on different frequencies are extracted through the analysis of the purchasing risk influence transient sequence, the frequency characteristics of the purchasing risk, such as periodical fluctuation, trend change, sudden events and the like, the frequency characteristic data of the purchasing risk are well understood through the extraction of the transient risk frequency component characteristic data, the basis is provided for subsequent risk analysis and prediction, the change trend of the transient risk frequency component characteristic data is analyzed and modeled to understand the trend of the purchasing risk, the future development direction and the change speed of the transient risk component are predicted through the analysis, the dynamic change mode of the purchasing risk is facilitated to be revealed, the reference and decision support are provided for purchasing risk management, the transient risk track data is better understood, the rapid change trend analysis is better understood, the purchasing risk is easier to be accurately analyzed through the analysis of the transient risk evolution trend, the purchasing risk is better understood, and the real-time risk is better analyzed, and the purchasing risk is better analyzed through the real-time change, and has better risk is better obtained.

Preferably, the specific steps of step S4 are:

Step S41: carrying out situational risk prediction on the risk time-varying multidimensional network map to generate risk situational prediction data;

step S42: performing virtual space risk simulation processing on the risk time-varying multidimensional network map through the risk scenario prediction data to generate risk scenario simulation data;

step S43: performing risk simulation parameter variation analysis on the risk scenario simulation data to generate dynamic risk simulation data;

step S44: and carrying out multi-stage risk situation evolution prediction on the dynamic risk simulation data to generate multi-stage risk situation evolution prediction data.

The invention predicts future risk situations by simulating different situations and conditions based on a risk time-varying multidimensional network map, predicts the occurrence probability and influence degree of different risk events by analyzing the relation and the state change of nodes in the network map, generates risk situation prediction data to help identify potential risk situations, provides early warning and decision support for purchasing risk management and control, performs simulation on the risk time-varying multidimensional network map by using the risk situation prediction data by virtual space risk simulation processing to simulate the risk evolution process under different situations, obtains the evolution path and the result of the risk event by simulating the state change and the mutual influence of the risk nodes in different situations, generates risk situation simulation data to help deeply understand the evolution mechanism and the potential influence of the purchasing risk, provides reference for formulating a risk management strategy, changes the parameters in the risk situation simulation data by the analysis of the change of the different parameters to understand the influence of the risk process, and the change trend and the sensitivity of the analysis parameters to identify important influence factors and key nodes to generate dynamic risk simulation data to help comprehensively consider the change of the risk situation prediction data, and the risk situation prediction data to be more comprehensively considered and the influence the current risk situation, and the current risk situation is more accurately predicted by the evolution trend and the future risk situation prediction data, and the risk situation prediction is provided by the multiple-stage of the analysis of the current risk situation prediction and the future risk situation prediction, strategic decision support is provided for purchasing risk management and control.

Preferably, the specific steps of step S5 are:

step S51: carrying out order flow track analysis on the historical purchasing behavior time sequence data to generate historical flow track data;

Step S52: performing circulation delay distribution rule analysis on the historical flow track data to generate circulation delay distribution data;

step S53: carrying out situation evolution path analysis on the multi-stage risk situation evolution prediction data to generate situation evolution path prediction data;

step S54: carrying out flow direction track deviation recognition on situation evolution path prediction data based on the flow delay distribution data so as to generate flow direction track abnormal data;

Step S55: and carrying out regional risk marking on the node distributed neural network by using the flow direction track abnormal data so as to generate a flow direction risk area.

The invention identifies the flow direction and path of an order through analysis of time sequence data of historical purchasing behavior by order flow direction track analysis, understands the flow and path of the purchasing behavior by tracing the source, the destination and the middle circulation node of the order, generates the historical flow direction track data to help grasp the overall situation and trend of the purchasing behavior, provides basic data for subsequent risk analysis and management, the circulation delay distribution rule analysis counts and analyzes the circulation delay in the historical flow direction track data, knows the transfer delay of the purchasing behavior among different nodes, identifies the typical delay mode and abnormal situation by analyzing the distribution rule of the circulation delay, generates circulation delay distribution data to help evaluate the delivery efficiency and timeliness of the purchasing behavior, provides basis for subsequent risk analysis and optimization, the situation evolution path analysis is based on multi-stage risk situation evolution prediction data, analyzes the evolution path and the development direction of the purchasing risk, and predicts the data by analyzing the historical evolution and prediction data of risks, predicts the propagation path and influence range of the risk, generates trend prediction data to help pre-judge the purchase trend and propagation mode, provides support for management and control of the flow delay, identifies the flow direction deviation and potential deviation of the purchasing behavior by analyzing the historical flow path by analyzing the situation delay distribution data and the abnormal flow direction track, and the abnormal flow direction has no relation, and the situation has the basis for the situation prediction data to be compared with the historical flow path prediction data, the regional risk marking marks the node distributed neural network by utilizing the flow track abnormal data, identifies a risk aggregation region, namely a flow risk region, determines high risk sites and nodes in purchasing behavior by identifying the risk region, and generates the flow risk region so as to be beneficial to focusing on risk management resources and measures and improve the efficiency and accuracy of purchasing risk management and control.

Preferably, the specific steps of step S54 are:

Step S541: carrying out normal circulation mode space learning on the circulation time delay distribution data to generate normal circulation data;

Step S542: constructing a prediction reference model based on normal circulation data;

Step S543: carrying out flow direction track deviation recognition on situation evolution path prediction data by using a prediction reference model, and marking the flow direction deviation track data;

step S544: and carrying out deviation quantification on the flow direction deviation track data based on a preset predicted flow direction track deviation value, and generating flow direction track abnormal data when the preset predicted flow direction track deviation value is smaller than or equal to the flow direction deviation track data.

According to the method, a normal circulation mode in purchasing behavior is learned and identified through analysis of circulation delay distribution data through normal circulation mode space learning, normal circulation delay distribution is acquired through statistics and modeling, a typical circulation mode and behavior rules are captured, a standard mode in which normal circulation data is beneficial to establishing purchasing behavior is generated, reference is provided for subsequent abnormal identification and risk management, a prediction standard model is built based on the normal circulation data, a standard prediction model is established by utilizing characteristics and rules of the normal circulation data, the model is used for predicting future circulation tracks and delay of purchasing behavior, reference prediction under normal conditions is used, a standard line is provided through establishment of the prediction standard model, so that subsequent comparison analysis on abnormal conditions and risk assessment is facilitated, flow direction track deviation identification utilizes the prediction standard model to analyze situation evolution path prediction data, identify abnormal tracks with obvious differences from the normal prediction tracks, whether flow direction track deviation exists or not is judged through comparison of prediction results of the prediction paths and the standard model, marking flow direction deviation track data is beneficial to distinguishing the abnormal tracks from the normal conditions, basic data is provided for subsequent identification and management, deviation quantification is based on preset prediction flow direction deviation difference values, when the preset flow direction deviation is smaller than the preset deviation and risk deviation data are equal to the preset, and the abnormal flow direction deviation is accurately marked to be equal to the potential deviation positioning and risk deviation, and the abnormal flow deviation is accurately determined, and the abnormal flow direction deviation is equal to the abnormal situation has a standard deviation.

Preferably, the specific steps of step S6 are:

step S61: performing risk adaptability monitoring treatment on the flow direction risk area to generate flow direction risk area monitoring data;

Step S62: calculating a dynamic risk value of the flow-direction risk area monitoring data to obtain a dynamic real-time risk value;

Step S63: utilizing the dynamic real-time risk value to make predictive purchasing decision on the warehouse data so as to generate predictive purchasing decision data;

Step S64: and carrying out data-driven mining modeling on the predictive purchasing decision data by using the deep neural network, and constructing a dynamic purchasing risk decision model so as to execute purchasing risk management and control operation.

The invention carries out real-time monitoring and analysis on the flow risk area through risk adaptability monitoring to capture the change and evolution of the risk area, generates flow risk area monitoring data through monitoring processing on the flow risk area, including information such as position, range, risk degree and the like of the flow risk area, generates flow risk area monitoring data which is favorable for timely knowing the dynamic situation of purchasing risks, provides data support for subsequent risk assessment and decision, carries out quantitative calculation on the risks of the risk area based on the flow risk area monitoring data, obtains dynamic real-time risk values through considering factors such as position, range, historical data and the like of the flow risk area, reflects the degree and change trend of the current purchasing risk, generates a state of the dynamic real-time risk values which is favorable for real-time monitoring purchasing risk, provides reference basis for purchasing decision and risk management, predicts and optimizes purchasing decision based on the dynamic real-time risk values, carries out correlation and analysis on the storage risk and potential decision data through the dynamic real-time risk values, generates a favorable purchasing decision and has proper guarantee and purchasing decision-making and is favorable for purchasing decision-making and is favorable for driving a decision-making and is favorable for the performance of a network-based on the purchasing risk management and performance-control, the accuracy-oriented control and the performance-oriented control is improved by driving of the data, the purchasing risk management and the control is favorable for the performance of the performance and the control-oriented control model is based on the data and the purchasing management and the management-oriented control and the data and the control, and scientific basis and decision support are provided for purchasing decisions.

In this specification, a purchasing risk management and control system based on cloud computing is provided, which is configured to execute the purchasing risk management and control method based on cloud computing as described above, and includes:

the distributed node module is used for acquiring historical purchasing logs, logistics supplementary node data and warehouse data; heterogeneous edge migration calculation is carried out on the logistics supplementary node data, and supplementary node distributed edge data are generated; performing neural network optimization and transformation on the complementary node distributed edge data to construct a node distributed neural network;

The risk perception module is used for carrying out purchasing behavior time sequence analysis on the historical purchasing log to generate historical purchasing behavior time sequence data; performing risk perception association analysis on the historical purchasing behavior time sequence data to generate purchasing risk influence association data;

The risk evolution module is used for carrying out transient risk evolution analysis on the purchasing risk influence associated data so as to obtain risk evolution response data; performing dynamic risk time-varying network optimization on the node distributed neural network based on the risk evolution response data, and constructing a risk time-varying multidimensional network map;

The situation evolution prediction module is used for carrying out virtual space risk simulation processing on the risk time-varying multidimensional network map so as to generate risk scenario simulation data; carrying out multi-stage risk situation evolution prediction on the risk scenario simulation data to generate multi-stage risk situation evolution prediction data;

The flow track deviation module is used for carrying out flow delay distribution rule analysis on the historical purchasing behavior time sequence data so as to generate flow delay distribution data; carrying out situation evolution path analysis on the multi-stage risk situation evolution prediction data to generate situation evolution path prediction data; carrying out flow track deviation recognition on situation evolution path prediction data based on the flow delay distribution data so as to generate a flow risk area;

The risk decision module is used for carrying out risk adaptability monitoring treatment on the flow to the risk area so as to obtain a dynamic real-time risk value; and carrying out data-driven mining modeling on the warehouse data by utilizing the dynamic real-time risk value, and constructing a dynamic purchasing risk decision model so as to execute purchasing risk management and control operation.

The invention obtains the historical purchasing log, the logistics supplementary node data and the storage data through the distributed node module, the distributed node module obtains the multi-source data to provide a data basis for the subsequent purchasing risk management and control, the risk perception module generates the historical purchasing behavior time sequence data through carrying out purchasing behavior time sequence analysis on the historical purchasing behavior time sequence data and generates purchasing risk influence associated data through carrying out risk perception associated analysis on the historical purchasing behavior time sequence data, thereby being beneficial to identifying and understanding potential risk factors existing in purchasing activities, the risk evolution module carries out transient risk evolution analysis on the purchasing risk influence associated data to obtain risk evolution response data, carries out dynamic risk time-varying network optimization on the node distributed neural network based on the risk evolution response data and constructs a risk time-varying multi-dimensional network map, understanding and predicting the evolution trend of the purchasing risk, providing decision basis for risk management and control, carrying out virtual space risk simulation processing on the risk time-varying multidimensional network map by a situation evolution prediction module, generating risk situation simulation data, carrying out multi-stage risk situation evolution prediction on the risk situation simulation data, generating multi-stage risk situation evolution prediction data which is helpful for predicting the development trend of the future purchasing risk, providing reference for purchasing decision and risk management, carrying out circulation delay distribution rule analysis on historical purchasing behavior time sequence data by a flow track deviation module, generating circulation delay distribution data, carrying out situation evolution path analysis on the multi-stage risk situation evolution prediction data, generating situation evolution path prediction data, identifying a flow risk area, namely an area with deviation in purchasing activity based on the circulation delay distribution data, the risk decision module carries out risk adaptability monitoring processing on the flow risk area to obtain a dynamic real-time risk value, carries out data-driven mining modeling on storage data by utilizing the dynamic real-time risk value, builds a dynamic purchasing risk decision model, carries out risk assessment, decision recommendation and execution of risk management and control operation based on real-time data and historical experience, and is beneficial to improving accuracy and efficiency of purchasing risk management and control and ensuring that purchasing activities accord with risk strategies and targets.

Drawings

FIG. 1 is a schematic flow chart of the steps of a purchasing risk management and control method based on cloud computing;

FIG. 2 is a detailed implementation step flow diagram of step S1;

FIG. 3 is a detailed implementation step flow diagram of step S2;

fig. 4 is a detailed implementation step flow diagram of step S3.

Detailed Description

It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.

The embodiment of the application provides a purchasing risk management and control method and system based on cloud computing. The execution main body of the purchase risk management and control method and system based on cloud computing comprises, but is not limited to, the execution main body of the purchase risk management and control method and system based on cloud computing comprises the following steps: mechanical devices, data processing platforms, cloud server nodes, network uploading devices, etc. may be considered general purpose computing nodes of the present application, including but not limited to: at least one of an audio image management system, an information management system and a cloud data management system.

Referring to fig. 1 to 4, the invention provides a purchasing risk management and control method based on cloud computing, which comprises the following steps:

In the embodiment of the present invention, referring to fig. 1, a schematic flow chart of steps of a purchasing risk management and control method based on cloud computing according to the present invention is shown, where in this example, the steps of the purchasing risk management and control method based on cloud computing include:

In this embodiment, historical purchase log data including purchase orders, supplier information, purchase quantity, price, and the like are collected, the data including position information, arrival time, departure time, and the like of the logistics supplemental nodes are obtained from a purchase system or a related database, the data including storage capacity, storage quantity, access records, and the like of a warehouse are obtained from a warehouse system or a related database, heterogeneous edge migration methods and techniques are determined, the logistics supplemental node data is migrated from an original environment to a distributed edge environment, the logistics supplemental node data is formatted and prepared so as to perform heterogeneous edge migration calculation, the format and structure of the distributed edge data of the supplemental nodes are determined so as to adapt to the distributed edge environment, distributed edge data of the supplemental nodes are generated according to the result of heterogeneous edge migration calculation, the distributed edge data are distributed on different edge nodes to form a data structure for distributed storage and processing, the distributed edge data of the supplemental nodes are prepared into a form suitable for neural network optimization, the data format conversion, feature extraction, and the like, the distributed edge data of the supplemental nodes are optimized by using a proper neural network optimization algorithm and technique, the neural network optimization model is adjusted, the distributed edge network is optimized, the data structure of the distributed edge network is optimized, and the data structure of the distributed edge network is optimized is selected, and the edge network is optimized, and the edge network structure of the data is optimized is determined, and the edge network is optimized, and the edge network is distributed and has the characteristics of the edge network structure and has good distribution characteristics.

in this embodiment, data cleaning and preprocessing are performed on historical purchase log data, including removing duplicate data, processing missing values, and the like, and according to timestamps in the purchase log, the purchase behaviors are ordered and organized according to time sequence to form time sequence data of the purchase behaviors, the time sequence data includes generation time of a purchase order, response time of a provider, arrival time of a logistics node, and the like, and other relevant time sequence information, indexes such as delivery time rates, product quality, and the like of different providers are analyzed, association analysis is performed on the time sequence data of the purchase behaviors, influences of risks of the provider on the purchase behaviors are identified, indexes such as punctuality and damage rate of goods of the logistics node are analyzed, association analysis is performed on the time sequence data of the purchase behaviors, influences of risks of the logistics node on the purchase behaviors are identified, indexes such as market price fluctuation and supply and demand condition are analyzed, association analysis is performed on the time sequence data of the purchase behaviors, and influences of market risks on the purchase behaviors are identified.

in this embodiment, according to the relevance and the influence degree in the related data of the purchasing risk influence, a risk propagation model is established, a propagation path and an influence range of risks are described, the established risk propagation model is used to simulate and evolve the purchasing risk, the change and the propagation process of the risks in time are observed, according to the simulated risk evolution result, risk evolution response data including the information of the risk degree, the risk propagation path and the like at different moments are calculated, according to specific requirements and the risk evolution response data, a target of network optimization, such as minimizing the risk propagation path, maximizing the network robustness and the like, a dynamic optimization algorithm adapting to the time-varying risk is designed based on the determined target, the change of the risk evolution response data and the dynamic adjustment of the network are considered, the node distributed neural network is optimized by using the designed dynamic optimization algorithm, and the network structure, the connection weight and the like are adjusted to adapt to the time-varying property of the risk.

In this embodiment, different risk scenarios are defined according to nodes and connection information in a multidimensional network map, such as risk occurrence of a single node, joint risk occurrence of multiple nodes, and the like, based on the defined risk scenarios, simulation of risk occurrence is performed on the multidimensional network map, conditions of state change, connection interruption, and the like of the nodes are considered, risk scenario simulation data including information of time, position, degree, and the like of risk occurrence are generated according to simulated risk occurrence results, the risk scenario simulation data are analyzed, key features and indexes are extracted for risk situation evolution prediction, a suitable prediction model such as time sequence prediction, machine learning model, and the like is selected according to the analyzed features and indexes, a multi-stage risk situation evolution prediction model is established, the risk situations of multiple stages in the future are predicted by using the constructed prediction model, trend sum of risk development is inferred, and multi-stage risk situation evolution prediction data including information of risk states, risk degree changes, and the like of different stages are generated according to the prediction results.

In this embodiment, according to timestamp information of purchasing behavior, circulation delay between different purchasing nodes is calculated, relevant delay features such as average delay and delay distribution are extracted, statistical analysis and visualization are performed based on the extracted delay features, rules and distribution conditions of circulation delay are explored, circulation delay distribution data including delay probability density functions, cumulative distribution functions and the like are generated according to the analyzed delay rules, key situation features such as risk degree, risk type and inter-node connection relation are extracted from multi-stage risk situation evolution prediction data, analysis of situation evolution paths is performed based on the extracted features, propagation paths and evolution trends of risks are explored, situation path prediction data including information such as node evolution sequences and risk propagation paths are generated according to the analyzed situation evolution paths, indicators for measuring deviation of flow paths are defined according to the circulation delay distribution data, deviation conditions of flow paths are identified, namely paths inconsistent with the circulation delay distribution are analyzed based on the defined indicators, and a risk area where the flow paths exist are identified according to the identified flow direction deviation of the identified flow direction dangerous paths.

In this embodiment, according to the characteristics and requirements of the risk area, an appropriate risk index, such as a risk degree, a risk probability, and the like, is selected, according to the selected risk index and data collected in real time, a dynamic real-time risk value is calculated and obtained, the current risk state is reflected, key characteristics, such as inventory, supplier reliability, and the like, are extracted from storage data to construct a purchasing risk decision model, an appropriate data mining algorithm and a machine learning method are selected, the dynamic purchasing risk decision model is constructed by utilizing the data extracted by the characteristics, model training is performed by using historical data, the model is optimized by an optimization algorithm and a verification method, the accuracy and generalization capability of the model are improved, the constructed dynamic purchasing risk decision model is applied to an actual purchasing process, and purchasing risk management and control operations, including risk early warning, supplier selection, inventory management, and other decisions, are performed according to the real-time dynamic real-time risk value and storage data.

In this embodiment, referring to fig. 2, a detailed implementation step flow chart of the step S1 is shown, and in this embodiment, the detailed implementation step of the step S1 includes:

In this embodiment, historical purchase log data is extracted from a purchase system or a related database, including purchase orders, supplier information, purchase quantity, purchase time and the like, logistics supplementary node data is obtained, the logistics supplementary node data is a key link in logistics, such as a transportation node, a distribution center and the like, the data comprises information of transportation time, node position, cargo state and the like, storage data is obtained from a storage management system or a related database, including stock quantity, stock position, cargo type and the like, the logistics supplementary node data is preprocessed, such as data cleaning, denoising, normalization and the like, so as to ensure the quality and consistency of the data, a deep learning model is utilized to extract key features in different modal data, such as object recognition in images, emotion analysis in texts and the like, the extracted features are fused, node multi-modal data is generated, so as to synthesize representation of different modal information, nodes for edge calculation, such as a logistics node, a cloud server and the like are determined according to system requirements and resource conditions, the node multi-modal data is divided into proper data blocks, the data blocks are migrated to an edge calculation node for processing, the edge calculation capability is utilized, the distributed neural network calculation model is utilized to realize the realization of the distributed neural network calculation, the distributed neural network is better, the neural network is used for the data is better in the performance of the training, the neural network is better in the performance can be better assessed, the network is better in the performance is better by the training and the network is better by the training and better by the training method.

In this embodiment, referring to fig. 3, a detailed implementation step flow chart of the step S2 is shown, and in this embodiment, the detailed implementation step of the step S2 includes:

In this embodiment, the purchasing time fields are ordered to ensure that purchasing behavior is arranged in time sequence, the purchasing behavior data is organized into time sequence data according to the purchasing time sequence, which is a time sequence, wherein each time point corresponds to the information of one purchasing behavior, or is a sequence divided according to time periods, such as daily, weekly or monthly purchasing behavior data, text mining is performed on text fields in the purchasing log, information such as keywords, entities and the like, such as vendor reputation, product quality and the like is extracted, the purchasing behavior data is subjected to clustering analysis, a purchasing pattern or behavior group with similar characteristics is identified, an abnormal pattern in the purchasing behavior such as an abnormal order, vendor or product abnormality and the like is identified by using an abnormality detection algorithm, and the abnormal pattern is converted into structured purchasing risk factor data according to the extracted unstructured risk factor, the method comprises the steps of carrying out exploratory data analysis on purchasing risk factor data, revealing the characteristics of distribution, correlation and the like of the data through visualization and statistical indexes, selecting the most relevant risk factors as potential driving factors of purchasing risks based on data analysis and domain knowledge, constructing a model to verify the prediction capability and influence degree of the potential driving factors by using modeling technologies such as regression analysis, machine learning and other methods, calculating the correlation among different purchasing risk factors, analyzing by using statistical indexes such as correlation coefficients, covariance and the like, discovering frequent correlation rules among different purchasing risk factors by using a correlation rule mining algorithm such as Apriori algorithm, generating purchasing risk influence correlation data according to the result of the correlation analysis, which is a correlation matrix or network, and displaying the association degree and influence relation among different purchasing risk factors.

In this embodiment, as described with reference to fig. 4, a detailed implementation step flow diagram of the step S3 is shown, and in this embodiment, the detailed implementation step of the step S3 includes:

In this embodiment, the procurement risk influence association data is converted into a time sequence, where each time point represents a time period (e.g., daily, weekly), an appropriate evolution model is selected according to a specific situation, e.g., an ARIMA model, an exponential smoothing model, etc., and is used to describe and predict a transient evolution trend of a risk factor, the selected evolution model is utilized to perform model fitting on the time sequence data, and predict the transient risk evolution data, so as to generate transient risk evolution data, an appropriate simulation method, e.g., a monte carlo simulation, a system dynamics simulation, etc., is selected, and is used to simulate an evolution response process of the risk factor, according to requirements of the simulation method, set related parameters, e.g., simulation times, time steps, etc., execute the risk evolution response simulation, generate risk evolution response data, wherein the structure of the risk evolution path sum includes the simulated risk evolution, the node and edge is designed, and the topology structure of the network is defined, and an objective function of network optimization is defined according to a specific optimization objective, e.g., minimizing the length of the risk propagation path, maximizing the robustness of the network, and the network is utilized to optimize algorithm, e.g., gradient descent, genetic algorithm, etc., optimizing the parameters of the network are optimized, so as to meet the optimization objective function, and perform optimization algorithm, and perform optimization of the optimization algorithm is performed to the optimal decision-making the optimal relation, or the optimal structure is performed until the optimal structure is reached, and the time-varying the iteration relation is optimized, based on the iteration relation is reached, and the iteration structure is further optimized.

In this embodiment, the specific steps of step S31 are as follows:

In this embodiment, the purchasing risk influence associated data is sliced according to time, for example, daily, weekly or monthly, so as to obtain data at different time points, the data at each time point is converted in a serialization manner, different methods are used, for example, the data is converted into a time sequence, a vector sequence and the like, so as to obtain a transient sequence of risk factors, a proper time-frequency analysis method is selected according to data characteristics and analysis requirements, parameters of the time-frequency analysis method, for example, a wavelet basis function, a time window length and the like are set, the purchasing risk influence transient sequence is subjected to time-frequency analysis, so as to obtain frequency component characteristic data, a proper trend analysis method, for example, linear regression, movement average and the like, parameters of the trend analysis method, for example, a time window size, fitting order and the like are set, trend analysis is performed on the transient risk frequency component characteristic data, so as to obtain track data of a changing trend, transient risk evolution analysis is performed on the transient risk trend track data, so as to know the evolution process and the evolution trend of the risk factors, a proper time-frequency analysis method, for example, a time-sequence analysis method is selected, a time-sequence model and the like, parameters of the time-frequency analysis method are set, for the transient analysis method are such as to obtain the number of clustering trend, the number, the time-sequence trend, the time-sequence average trend is used for the transient risk analysis, and the evolution data is used for the transient risk analysis, and the transient risk analysis is obtained, and the transient risk evolution is predicted, and the risk is obtained.

In this embodiment, the specific steps of step S4 are as follows:

In this embodiment, the data required by the risk time-varying multidimensional network map is managed and prepared, including historical risk data, related influence factor data, environmental data and the like, to ensure the accuracy and the integrity of the data, select a suitable situational risk prediction model according to the characteristics of the risk time-varying multidimensional network map, such as time series analysis, machine learning and the like, use the historical risk data for model training, adjust model parameters to obtain better prediction effects, predict future risks by using the trained model, predict the risk according to the current environmental data and related influence factors, input the model for prediction, generate risk scenario prediction data, consider generating a plurality of prediction results under different situations to provide more choices and references for a decision maker, establish a corresponding virtual space model according to the structure and the attributes of the risk time-varying multidimensional network map, consider the relationship and interaction between the risk elements, represent the model as a virtual space, set different risk scenario parameters according to the risk scenario prediction data, such as the degree of risk, occurrence probability and the like, combine these parameters with the virtual space model, construct a simulation setting, process the risk scenario simulation, process the risk scenario parameters according to the virtual space simulation, process the risk scenario parameters, change the risk scenario parameters, and simulate the risk scenario parameters according to the risk scenario parameters, and the result is changed according to the risk scenario parameters, the simulation parameters and the risk scenario parameters are changed and the risk scenario parameters is changed and the corresponding to be changed, generating dynamic risk simulation data at different time points, selecting proper multi-stage prediction models according to the characteristics of the dynamic risk simulation data, wherein the models are used for predicting the evolution trend of a risk situation in a plurality of stages in the future by taking time sequence analysis, system dynamics and the like into consideration, analyzing and predicting the dynamic risk simulation data by using the selected multi-stage prediction models, predicting the evolution of the risk situation in the plurality of stages in the future according to the current risk situation and parameter variation trend, and generating multi-stage risk situation evolution prediction data.

In this embodiment, the specific steps of step S5 are as follows:

In this embodiment, a node relation network is analyzed to extract a circulation path of an order to form flow track data of the order, circulation delay of the order between different nodes is calculated according to the flow track data of the order, statistical analysis is performed on delay data, such as average delay, variance, quantile and the like, so as to understand distribution rules of circulation delay, circulation delay distribution data is generated according to analysis results of the circulation delay distribution rules, the circulation delay distribution situation of the order between different nodes is described by the data, multi-stage risk situation evolution prediction data is preprocessed, such as redundancy information removal, missing data processing and the like, the prediction data is analyzed to extract the evolution path information of the risk situation, the information comprises a node sequence and a time sequence, the predicted evolution path is matched with a historical flow track, the node sequence and the time sequence are compared, the identified deviation situation is compared, the matched track is subjected to circulation delay analysis with actual circulation delay, the node and flow track abnormality data are taken as input, a distributed neural network model is constructed, the flow track abnormality data is used for model training, the node is marked as different risk levels, such as high risk level, low risk level, risk level and risk level is different risk level is described by the node, and risk area is generated, and risk area is described according to the risk level.

In this embodiment, the specific steps of step S54 are as follows:

In this embodiment, meaningful features, such as average delay, variance, quantile, etc., are extracted from the circulation delay distribution data, a machine learning algorithm (such as a clustering algorithm, a dimension reduction algorithm, etc.) is used to spatially learn the features, similar circulation patterns are clustered together, a suitable feature is selected as an input variable of a model, such as a node, time, etc., a prediction reference model is trained by using normal circulation data, so that the normal circulation pattern can be predicted, a situation evolution path is predicted by using the prediction reference model to carry out flow track prediction, a predicted track is compared with an actual historical flow track, deviation conditions therein are identified, flow direction deviation track data is marked, a difference between the flow direction deviation track data and a preset predicted flow direction track deviation value is calculated, for example, by calculating a euclidean distance of track deviation or other suitable measurement mode, the deviation value is compared with the preset predicted flow direction track deviation value, the flow direction deviation track data is quantized to determine a deviation degree, and when the preset predicted flow direction deviation value is smaller than or equal to a deviation quantization result of the flow direction deviation track data, the flow direction deviation track data is marked as abnormal flow direction track data, so that abnormal flow direction track data is generated.

In this embodiment, the specific steps of step S6 are as follows:

In this embodiment, relevant data of a flow risk area is obtained, including risk marks of nodes, abnormal data of flow paths, etc., so as to ensure accuracy and integrity of the data, the flow risk area data is subjected to risk adaptability monitoring processing, including updating of the risk marks of the nodes, detection and processing of abnormal flow paths, etc., according to actual conditions, processing is performed by adopting technical means such as a rule engine, machine learning, etc., according to the results of the risk adaptability monitoring processing, flow risk area monitoring data are generated, these data reflect the current state and risk level of the flow risk area, calculation indexes of dynamic risk values are determined, including the risk level of the nodes, the number of abnormal data, the area of the risk area, etc., appropriate indexes are selected according to specific conditions, calculation is performed on the flow risk area monitoring data according to the calculation indexes, so as to obtain dynamic real-time risk values, calculating by adopting methods such as weighted summation, rule definition and the like, wherein a dynamic real-time risk value reflects the overall risk level of a current time point flowing to a risk area, making a predictive purchasing decision according to the dynamic real-time risk value and storage data, determining decision contents such as purchasing plans, supplier selection, inventory allocation and the like according to the height of the risk value and the requirement of storage conditions, generating predictive purchasing decision data according to the making result of the predictive purchasing decision, recording the content and related information of the purchasing decision for subsequent risk decision model construction, constructing a dynamic purchasing risk decision model by adopting a deep neural network and other data-driven mining modeling method, learning and mining the relevance between the predictive purchasing decision and the risk result by using the predictive purchasing decision data, and model optimization is carried out to improve accuracy and performance, common deep learning algorithms and technologies, such as forward and backward propagation algorithms of a neural network, gradient descent and the like are used, a dynamic purchasing risk decision model is constructed according to the trained and optimized model, risks brought by purchasing decisions are predicted according to real-time predictive purchasing decision data, corresponding risk management and control suggestions are provided, purchasing risk management and control operation is executed by using the constructed dynamic purchasing risk decision model, risks of purchasing decisions are evaluated by the model according to the real-time predictive purchasing decision data, and corresponding decision suggestions and risk control measures are provided to help purchasing decision makers to make decisions and manage and control.

In this embodiment, a purchasing risk management and control system based on cloud computing is provided, which is configured to execute the purchasing risk management and control method based on cloud computing as described above, and includes:

The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

The foregoing is merely a specific embodiment of the invention to enable a person skilled in the art to understand or practice the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. The purchasing risk management and control method based on cloud computing is characterized by comprising the following steps of:

step S3: performing transient risk evolution analysis on the purchasing risk influence associated data so as to obtain risk evolution response data; performing dynamic risk time-varying network optimization on the node distributed neural network based on the risk evolution response data, and constructing a risk time-varying multidimensional network map; the dynamic risk time-varying network optimization specifically comprises the following steps: designing a structure of a risk time-varying network, comprising setting nodes and edges and a topological structure of the network, defining a network optimization objective function according to a specific optimization target, optimizing and adjusting parameters of the network by utilizing an optimization algorithm to meet the optimization objective function to the maximum extent, executing the network optimization algorithm, performing iterative optimization on the network until a convergence condition or preset optimization iteration times are reached, and constructing a risk time-varying multidimensional network map according to an optimization result;

Step S5: performing circulation delay distribution rule analysis on the historical purchasing behavior time sequence data to generate circulation delay distribution data; carrying out situation evolution path analysis on the multi-stage risk situation evolution prediction data to generate situation evolution path prediction data; carrying out flow track deviation recognition on situation evolution path prediction data based on the flow delay distribution data so as to generate a flow risk area; the specific steps of the step S5 are as follows:

Step S52: performing circulation delay distribution rule analysis on the historical flow track data to generate circulation delay distribution data; the analysis of the distribution rule of the circulation delay is specifically as follows: calculating the circulation time delay of the order among different nodes according to the flow track data of the order, and carrying out statistical analysis on the time delay data so as to know the distribution rule of the circulation time delay;

Step S54: carrying out flow direction track deviation recognition on situation evolution path prediction data based on the flow delay distribution data so as to generate flow direction track abnormal data; the specific steps of step S54 are:

Step S544: carrying out deviation quantification on the flow direction deviation track data based on a preset predicted flow direction track deviation value, and generating flow direction track abnormal data when the preset predicted flow direction track deviation value is smaller than or equal to the flow direction deviation track data;

step S55: carrying out regional risk marking on the node distributed neural network by utilizing the flow direction track abnormal data so as to generate a flow direction risk area;

2. The cloud computing-based purchasing risk management and control method according to claim 1, wherein the specific steps of step S1 are as follows:

3. The cloud computing-based purchasing risk management and control method according to claim 1, wherein the specific steps of step S2 are as follows:

4. The cloud computing-based purchasing risk management and control method according to claim 1, wherein the specific steps of step S3 are as follows:

5. The cloud computing-based purchasing risk management method according to claim 4, wherein the specific steps of step S31 are as follows:

6. The cloud computing-based purchasing risk management and control method according to claim 1, wherein the specific steps of step S4 are as follows:

7. The cloud computing-based purchasing risk management and control method according to claim 1, wherein the specific steps of step S6 are as follows:

8. A cloud computing-based procurement risk management and control system for implementing the cloud computing-based procurement risk management and control method of claim 1, comprising:

The risk evolution module is used for carrying out transient risk evolution analysis on the purchasing risk influence associated data so as to obtain risk evolution response data; performing dynamic risk time-varying network optimization on the node distributed neural network based on the risk evolution response data, and constructing a risk time-varying multidimensional network map; the dynamic risk time-varying network optimization specifically comprises the following steps: designing a structure of a risk time-varying network, comprising setting nodes and edges and a topological structure of the network, defining a network optimization objective function according to a specific optimization target, optimizing and adjusting parameters of the network by utilizing an optimization algorithm to meet the optimization objective function to the maximum extent, executing the network optimization algorithm, performing iterative optimization on the network until a convergence condition or preset optimization iteration times are reached, and constructing a risk time-varying multidimensional network map according to an optimization result;

The flow track deviation module is used for carrying out flow delay distribution rule analysis on the historical purchasing behavior time sequence data so as to generate flow delay distribution data; carrying out situation evolution path analysis on the multi-stage risk situation evolution prediction data to generate situation evolution path prediction data; carrying out flow track deviation recognition on situation evolution path prediction data based on the flow delay distribution data so as to generate a flow risk area; the analysis of the distribution rule of the circulation delay is specifically as follows: calculating the circulation time delay of the order among different nodes according to the flow track data of the order, and carrying out statistical analysis on the time delay data so as to know the distribution rule of the circulation time delay;