Business collaborative management system and method based on data analysis

Abstract

The invention discloses a business collaborative management system and method based on data analysis, and relates to the technical field of business collaborative management. According to the business collaborative management system and method, a change point analysis method and a local anomaly factor LOF algorithm are adopted, multi-dimensional density distribution of operation data is subjected to quantitative evaluation, discrete anomaly points and accumulation anomaly points are marked, and a business collaborative management result is obtained; the source of the abnormal point is judged through correlation analysis, so that the abnormal data processing capability is remarkably improved; the feature analysis module is combined with association rule mining and a multi-dimensional regression analysis method, analyzes formation conditions of abnormal points, extracts key influence factors, quantifies the influence degree of the key influence factors on supply chain operation, and establishes a factor association model; and the decision-making unit performs high-dimensional simulation on the supply scheme of the supply chain through a reinforcement learning algorithm, and generates a dynamic adjustment scheme including transportation resource optimization, inventory strategy adjustment and high-risk supplier substitution through multiple times of strategy iteration.

Description

Technical Field

[0001] The present invention relates to the technical field of business collaboration management, and in particular to a business collaboration management system and method based on data analysis. Background Art

[0002] In traditional business collaboration management, enterprises rely on centralized management systems to coordinate core links such as production, procurement and sales. Such systems monitor the operating status of supply chain nodes by periodically collecting and analyzing data. However, with the increasing volatility of market demand, the increasing complexity of the supply chain, and the rapid increase in data volume, the limitations of traditional methods have gradually become apparent.

[0003] An obvious problem is the lag in data processing. Traditional systems usually obtain and update information at fixed time intervals, making it difficult to promptly capture real-time abnormal situations in the supply chain, such as delayed delivery, inventory shortages, or surges in demand. Information lag may not only lead to incorrect response decisions, but also easily lead to a crisis of trust between upstream and downstream supply chains. In addition, traditional solutions have limited abnormal data processing capabilities, and the data sources are diverse and of varying quality, which directly affects the accuracy of supply chain collaboration. Traditional systems rely on fixed rules or manual intervention when identifying anomalies, resulting in low efficiency and accuracy.

[0004] In this case, the company's risk response capabilities are also restricted. The lack of a rapid response mechanism to risks can easily lead to increased supply chain disruptions, further affecting the company's overall operational efficiency and market competitiveness. Although current traditional methods attempt to address these problems by increasing manual monitoring and optimizing static rules, they are unable to fundamentally resolve the contradiction between data real-time, processing accuracy, and collaborative efficiency. Therefore, a business collaborative management solution based on data analysis is urgently needed to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a business collaboration management system and method based on data analysis to solve the problems that traditional business collaboration management solutions are poor in identifying and processing abnormal data in complex and dynamic supply chain environments, and are not optimized in a timely manner.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a business collaboration management system based on data analysis, which includes:

[0009] Data collection network, which is used to obtain the operation data of each node in the supply chain, including inventory level, transportation status and order fluctuation, and divide the operation data into dimensions based on time, space and node attributes;

[0010] The anomaly detection module is used to identify sudden changes in the trend of operating data. It combines the local anomaly factor (LOF) algorithm to quantitatively evaluate the density distribution of multidimensional data points, mark discrete anomalies and accumulated anomalies, and determine the source of anomalies by correlating data dimensions.

[0011] The feature analysis module is used to analyze the marked abnormal data, use the association rule mining method to mine the conditions for the formation of abnormal points, and extract the influencing factors through multidimensional regression analysis technology;

[0012] The decision-making unit uses a reinforcement learning algorithm to perform high-dimensional simulation of the supply chain adjustment plan based on the key influencing factors output by the feature analysis module, and generates an adjustment plan after multiple strategy iterations;

[0013] Supply chain nodes are used to execute adjustment plans issued by the decision-making unit, including reallocating transportation resources, adjusting inventory strategies, or replacing suppliers;

[0014] The data feedback module is used to monitor the feedback effects of the adjustment plans executed by the supply chain nodes and evaluate the adjustment plans.

[0015] In a second aspect, the present invention provides a business collaborative management method based on data analysis, comprising:

[0016] Step S1, obtaining the operation data of each node in the supply chain and dividing it into dimensions;

[0017] Step S2, using the change point analysis method to identify the trend mutation of the operation data, detect possible abnormal events in the supply chain, and combine the local abnormal factor LOF algorithm to quantitatively evaluate the multidimensional density distribution of data points in the operation data, mark discrete abnormal points and accumulated abnormal points, analyze the abnormal points, and determine the source of the multidimensional data points;

[0018] Step S3, import the marked abnormal data into the feature analysis module, combine the association rule mining method, analyze the formation conditions of the abnormal points, use multidimensional regression analysis technology to extract the key influencing factors in the abnormal data, and quantify the specific impact of the abnormal data on the supply chain operation. Based on the extracted factors, establish a factor association model;

[0019] The abnormal data includes abnormal events, discrete abnormal points and accumulated abnormal points;

[0020] Step S4: Based on the influencing factors, a reinforcement learning algorithm is used to perform high-dimensional simulation of the supply chain's supply plan. After multiple strategy iterations, an adjustment plan is generated and sent to relevant nodes of the supply chain through a decision engine.

[0021] Step S5: The supply chain node receives and executes the adjustment plan, monitors the feedback effect during the execution process in real time, analyzes the feedback effect data, and evaluates the effectiveness of the adjustment plan.

[0022] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, the operation data is divided into dimensions based on time, space and node attributes, including:

[0023] Inventory level dimensions, including current inventory value, replenishment frequency, and historical trends,

[0024] Transportation status dimensions, including transportation delay duration, number of orders currently in transit, and transportation route attributes

[0025] Order fluctuation dimensions include changes in order volume, demand forecast errors, and customer order frequency.

[0026] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, the step of dimensional division is:

[0027] Define the running data set as D, expressed as:

[0028] D={d 1 ,d 2 ,…,d n},d i ={t i ,l i ,p i ,a i ,b i ,c i},

[0029] Where D represents the total set of running data, d i represents the i-th data point, t i Represents the time attribute of the data point, l i represents the spatial attributes of the data points, p i Represents the node attributes of the data point, a i represents the inventory level of a data point, b i Indicates the transportation status of the data point, c i represents the order fluctuation of the data points;

[0030] The data is divided based on the time dimension, and the division formula is:

[0031] Dt ={d i |t i ∈T},

[0032] Among them, D t Represents a data set divided by time dimension, d i represents the i-th data point, t i represents the time attribute of the data point, T represents the time interval set,

[0033] The data is divided based on the spatial dimension, and the division formula is:

[0034] D s ={d i |t i ∈S},

[0035] Among them, D s Represents a data set divided by spatial dimension, d i represents the i-th data point, l i represents the spatial attribute of the data point, S represents the spatial range set,

[0036] The data is divided based on node attributes, and the division formula is:

[0037] D p ={d i |t i ∈P},

[0038] Among them, D p Represents a data set divided by node attributes, d i represents the i-th data point, p i represents the node attribute of the data point, and P represents the set of supply chain node attributes.

[0039] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, the discrete abnormal points include node imbalance caused by transportation delays,

[0040] The said accumulation anomalies include insufficient inventory within a specific time period;

[0041] The method for analyzing the abnormal points is as follows:

[0042] Combine inventory with transportation data to analyze whether inventory anomalies are caused by transportation delays.

[0043] Order fluctuations are combined with inventory levels to determine whether there is replenishment pressure caused by a surge in demand.

[0044] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, wherein: the change point analysis method is used to identify the trend mutation of the operation data, detect the abnormal events that may exist in the supply chain, and the local abnormal factor LOF algorithm is combined to quantitatively evaluate the multidimensional density distribution of the data points in the operation data, mark the discrete abnormal points and the accumulated abnormal points and analyze the abnormal points, and the step of determining the source of the multidimensional data points is as follows:

[0045] Define the time series data as X, the formula is:

[0046] X={x 1 ,x 2 ,…,x n},

[0047] Among them, X represents the time series data set, x i represents the observation value at the i-th time point, i∈1,2,...,n;

[0048] The change point detection algorithm is used to calculate the change in log likelihood, and the calculation formula is:

[0049]

[0050] Among them, C(t) represents the mutation point at time t,

[0051] Indicates the selection of the maximum value among all time points t, ΔL(t) indicates the change in log likelihood before and after time t, represents the probability of data from time 1 to t, Represents the probability of data from time t+1 to n;

[0052] The local outlier factor LOF is used to quantify the density distribution of multidimensional data. The quantification formula is:

[0053]

[0054] Among them, LOF(p i ) represents the data point p i The local anomaly factor, N k (p i ) indicates p i The k-nearest neighbor set, lrd(q) represents the local density inverse of data point q, lrd(p i ) represents the data point p i The local density is reversed, |N k (p i )| represents the set N k (p i ) size;

[0055] Define the local density inversion, the formula is:

[0056]

[0057] Among them, dist(p i ,q) indicates p i The distance to q, k-dist(q) represents the k-nearest neighbor distance of data point q;

[0058] Mark outlier point A:

[0059] A={p i |LOF(p i )>λ},

[0060] Among them, A represents the set of outliers, p i represents the data point, λ is the abnormality determination threshold,

[0061] Discrete outlier analysis is performed, and the analysis formula is:

[0062] R 离散 ={(a,b)|a∈A 离散 ,b∈D, the association condition is satisfied},

[0063] in,

[0064] R 离散 Represents the association relationship of discrete outliers, a represents the discrete outlier, b represents other data points associated with the discrete outlier, A 离散 represents a set of discrete outliers, D represents a data set,

[0065] Perform accumulation outlier analysis, the analysis formula is:

[0066] R 堆积 ={(c,d)|c∈A 堆积 ,d∈D, the association condition is satisfied},

[0067] in,

[0068] R 堆积 represents the correlation between the accumulated abnormal points, c represents the accumulated abnormal points, d represents other data points associated with the accumulated abnormal points, A 堆积 represents a set of accumulated outlier points, and D represents a data set.

[0069] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, the step of establishing a factor association model based on the extracted factors is as follows:

[0070] Extract and mark abnormal data D a , the extraction formula is:

[0071] Da ={x i |x i ∈D,x i is marked as abnormal},

[0072] Among them, D a represents the data set marked as abnormal, x i represents the i-th data point in the data set D, where D represents the original running data set;

[0073] The formation conditions are extracted using the association rule mining method. The extraction formula is:

[0074]

[0075] Among them, R represents the generated association rule set, X represents the condition item of the rule, which is the combination of the outlier features, Y represents the result item of the rule, which is the combination of the outlier results, conf represents the confidence of the rule, and sup represents the support of the rule;

[0076] Multidimensional regression analysis was used to extract the influencing factors, and the extraction formula was:

[0077]

[0078] Among them, y represents the result indicator of abnormal formation, β 0 represents the constant term of the regression model, x i represents the i-th potential influencing factor, β i represents the regression coefficient of the i-th influencing factor, n represents the number of influencing factors, and ∈ represents the error term;

[0079] The quantitative formula for the contribution of the influencing factors to the abnormal results is:

[0080]

[0081] Among them, Impact(x i ) represents the impact factor x i Relative contribution to abnormal results, β i Indicates the impact factor x i The regression coefficient, β j Indicates the impact factor x j The regression coefficient of , n represents the number of influencing factors;

[0082] Establish a factor association model, the model formula is:

[0083] M={(x i ,Impact(x i ))|x i ∈ impact factor},

[0084] Among them, M represents the model of impact factor and its contribution, x i Represents a single impact factor, Impact(x i ) represents the impact factor x i relative contribution.

[0085] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, the objectives of the high-dimensional simulation include:

[0086] Reallocate transportation resources to reduce the frequency of inventory shortages,

[0087] Adjust inventory strategies to reduce supply risks.

[0088] Replace high-risk suppliers and improve supply chain stability.

[0089] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, the steps of: based on the influencing factors, using the reinforcement learning algorithm to perform high-dimensional simulation of the supply chain supply plan, and generating the adjustment plan through multiple strategy iterations are as follows:

[0090] Define the core components of reinforcement learning, including:

[0091] State space S: S = {s 1 ,s 2 ,…,s m},

[0092] Where S represents the state space, si represents the i-th state of the supply chain, and m represents the number of all possible states.

[0093] Action space A: A={a 1 ,a 2 ,…,a k},

[0094] Among them, A represents the action space, a j represents the jth adjustment plan, k represents the number of all possible actions,

[0095] Reward function R(s,a):

[0096] R(s,a)=f(cost,stability,risk),

[0097] Among them, R(s,a) represents the reward value of executing action a in state s, f(cost, stability, risk) is the target comprehensive function; the reward function f is expressed as:

[0098]

[0099] Among them, w 1 and w2 is the weight parameter;

[0100] Use the Qlearning algorithm to update the Q value. The update formula is:

[0101]

[0102] Among them, Q(s,a) represents the Q value of the current state s and action a, α represents the learning rate, R(s,a) represents the value of the reward function, and γ represents the discount factor.

[0103] Indicates the maximum Q value of the next state, s ′ represents the next state, a ′ Indicates the next action;

[0104] Set high-dimensional simulation goals, including:

[0105] Adjust the allocation of transportation resources. The adjustment formula is:

[0106]

[0107] Among them, C i represents the transportation cost of the i-th resource, x i represents the allocation amount of the i-th resource, m represents the total number of resource types,

[0108] To reduce the risk of supply chain disruption, the adjustment formula is:

[0109] maxP(s t ),

[0110] Among them, P(s t ) means in state s t The probability that the supply chain remains stable under t Indicates the state at a certain moment.

[0111] The strategy generation formula is:

[0112]

[0113] Among them, π * represents the optimal strategy,

[0114] represents the action a that maximizes Q(s,a), and Q(s,a) represents the value of action a in the current state s.

[0115] As a preferred solution of the business collaborative management method based on data analysis described in the present invention, the effectiveness evaluation method includes:

[0116] Whether reallocated transportation resources reduce the frequency of inventory shortages;

[0117] Whether inventory strategies are effective in reducing the risk of supply chain disruptions.

[0118] The beneficial effects of the present invention are as follows: the present invention adopts a change point analysis method and a local outlier factor LOF algorithm to quantitatively evaluate the multidimensional density distribution of operating data, mark discrete outliers and accumulated outliers, and determine the source of outliers through association analysis, thereby significantly improving the ability to process outlier data; the feature analysis module combines association rule mining and multidimensional regression analysis methods to analyze the formation conditions of outliers, extract key influencing factors and quantify their influence on supply chain operations, and establish a factor association model; the decision-making unit performs high-dimensional simulation of the supply plan of the supply chain through a reinforcement learning algorithm, and after multiple strategy iterations, generates a dynamic adjustment plan including transportation resource optimization, inventory strategy adjustment and high-risk supplier replacement; the data feedback module monitors the execution effect of the adjustment plan in real time, re-inputs the feedback data into the optimization model, verifies the effectiveness of the plan and makes continuous improvements. BRIEF DESCRIPTION OF THE DRAWINGS

[0119] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0120] Figure 1 It is a schematic diagram of the framework of the business collaboration management system based on data analysis of the present invention.

[0121] Figure 2 It is a flowchart of the business collaborative management method based on data analysis of the present invention. DETAILED DESCRIPTION

[0122] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0123] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0124] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0125] Example 1, reference Figure 1 and Figure 2 , this embodiment provides a business collaboration management system based on data analysis, including:

[0126] Data collection network, which is used to obtain the operation data of each node in the supply chain, including inventory level, transportation status and order fluctuation, and divide the operation data into dimensions based on time, space and node attributes;

[0127] The anomaly detection module is used to identify sudden changes in the trend of operating data. It combines the local anomaly factor (LOF) algorithm to quantitatively evaluate the density distribution of multidimensional data points, mark discrete anomalies and accumulated anomalies, and determine the source of anomalies by correlating data dimensions.

[0128] The feature analysis module is used to analyze the marked abnormal data, use the association rule mining method to mine the conditions for the formation of abnormal points, and extract the influencing factors through multidimensional regression analysis technology;

[0129] The decision-making unit uses a reinforcement learning algorithm to perform high-dimensional simulation of the supply chain adjustment plan based on the key influencing factors output by the feature analysis module, and generates an adjustment plan after multiple strategy iterations;

[0130] Supply chain nodes are used to execute adjustment plans issued by the decision-making unit, including reallocating transportation resources, adjusting inventory strategies, or replacing suppliers;

[0131] The data feedback module is used to monitor the feedback effects of the adjustment plans executed by the supply chain nodes and evaluate the adjustment plans.

[0132] This embodiment also provides a business collaborative management method based on data analysis, including:

[0133] Step S1, obtaining the operation data of each node in the supply chain and dividing it into dimensions;

[0134] The operation data is divided into dimensions based on time, space and node attributes, including:

[0135] Inventory level dimensions, including current inventory value, replenishment frequency, and historical trends,

[0136] Transportation status dimensions, including transportation delay duration, number of orders currently in transit, and transportation route attributes

[0137] Order fluctuation dimensions, including changes in order volume, demand forecast errors, and customer order frequency;

[0138] The steps for dimension partitioning are:

[0139] Define the running data set as D, expressed as:

[0140] D={d 1 ,d 2 ,…,d n},d i ={t i ,l i ,p i ,a i ,b i ,c i},

[0141] Where D represents the total set of running data, d i represents the i-th data point, t i Represents the time attribute of the data point, l i represents the spatial attributes of the data points, p i Represents the node attributes of the data point, a i represents the inventory level of a data point, b i Indicates the transportation status of the data point, c i represents the order fluctuation of the data points;

[0142] The data is divided based on the time dimension, and the division formula is:

[0143] D t ={d i |t i ∈T},

[0144] Among them, D t Represents a data set divided by time dimension, d i represents the i-th data point, t i represents the time attribute of the data point, T represents the time interval set,

[0145] The data is divided based on the spatial dimension, and the division formula is:

[0146] D s ={d i |t i ∈S},

[0147] Among them, D s Represents a data set divided by spatial dimension, d i represents the i-th data point, l i represents the spatial attribute of the data point, S represents the spatial range set,

[0148] The data is divided based on node attributes, and the division formula is:

[0149] D p ={d i |t i ∈P},

[0150] Among them, D p Represents a data set divided by node attributes, d i represents the i-th data point, p i represents the node attribute of the data point, and P represents the set of supply chain node attributes;

[0151] Specifically, the operation data is divided into several subsets through three dimensions: time, space and node attributes. The time dimension captures dynamic changes, the space dimension reveals regional distribution characteristics, and the node attributes reflect the individual differences of each node in the supply chain.

[0152] Step S2, using the change point analysis method to identify the trend mutation of the operation data, detect possible abnormal events in the supply chain, and combine the local abnormal factor LOF algorithm to quantitatively evaluate the multidimensional density distribution of data points in the operation data, mark discrete abnormal points and accumulated abnormal points, analyze the abnormal points, and determine the source of the multidimensional data points;

[0153] Discrete anomalies, including node imbalances caused by transportation delays,

[0154] Accumulation of outliers, including insufficient inventory during a specific time period;

[0155] The method for analyzing outliers is as follows:

[0156] Combine inventory with transportation data to analyze whether inventory anomalies are caused by transportation delays.

[0157] Combine order fluctuations with inventory levels to determine whether there is replenishment pressure caused by a surge in demand;

[0158] The change point analysis method is used to identify the trend mutation of the operation data and detect possible abnormal events in the supply chain. The local anomaly factor LOF algorithm is combined to quantitatively evaluate the multidimensional density distribution of data points in the operation data, mark discrete anomalies and accumulated anomalies, and analyze the anomalies. The steps to determine the source of multidimensional data points are as follows:

[0159] Define the time series data as X, the formula is:

[0160] X={x 1 ,x 2 ,…,x n},

[0161] Among them, X represents the time series data set, x irepresents the observation value at the i-th time point, i∈1,2,...,n;

[0162] The change point detection algorithm is used to calculate the change in log likelihood, and the calculation formula is:

[0163]

[0164] Among them, C(t) represents the mutation point at time t,

[0165] Indicates the selection of the maximum value among all time points t, ΔL(t) indicates the change in log likelihood before and after time t, represents the probability of data from time 1 to t, Represents the probability of data from time t+1 to n;

[0166] The local outlier factor LOF is used to quantify the density distribution of multidimensional data. The quantification formula is:

[0167]

[0168] Among them, LOF(p i ) represents the data point p i The local anomaly factor, N k (p i ) indicates p i The k-nearest neighbor set, lrd(q) represents the local density inverse of data point q, lrd(p i ) represents the data point p i The local density is reversed, |N k (pi)| represents the set N k (p i ) size;

[0169] Define the local density inversion, the formula is:

[0170]

[0171] Among them, dist(p i ,q) indicates p i The distance to q, k-dist(q) represents the k-nearest neighbor distance of data point q;

[0172] Mark outlier point A:

[0173] A={p i |LOF(p i )>λ},

[0174] Among them, A represents the set of outliers, p i represents the data point, λ is the abnormality determination threshold,

[0175] Discrete outlier analysis is performed, and the analysis formula is:

[0176] R 离散 ={(a,b)|a∈A 离散 ,b∈D, the association condition is satisfied},

[0177] in,

[0178] R 离散 Represents the association relationship of discrete outliers, a represents the discrete outlier, b represents other data points associated with the discrete outlier, A 离散 represents a set of discrete outliers, D represents a data set,

[0179] Perform accumulation outlier analysis, the analysis formula is:

[0180] R 堆积 ={(c,d)|c∈A 堆积 ,d∈D, the association condition is satisfied},

[0181] in,

[0182] R 堆积 represents the correlation between the accumulated abnormal points, c represents the accumulated abnormal points, d represents other data points associated with the accumulated abnormal points, A 堆积 represents a set of accumulated outlier points, and D represents a data set;

[0183] Specifically, change point detection identifies trend mutation points, providing a basis for the discovery of anomalies. Combined with local anomaly factors, it can effectively distinguish between discrete and accumulated anomalies, and clarify the associated sources and potential impacts of anomalies through correlation analysis.

[0184] Step S3, import the marked abnormal data into the feature analysis module, combine the association rule mining method, analyze the formation conditions of the abnormal points, use multidimensional regression analysis technology to extract the key influencing factors in the abnormal data, and quantify the specific impact of the abnormal data on the supply chain operation. Based on the extracted factors, establish a factor association model;

[0185] Abnormal data include abnormal events, discrete abnormal points and accumulated abnormal points;

[0186] Based on the extracted factors, the steps to establish the factor association model are:

[0187] Extract and mark abnormal data D a , the extraction formula is:

[0188] D a ={x i |x i ∈D,x i is marked as abnormal},

[0189] Among them, Da represents the data set marked as abnormal, x i represents the i-th data point in the data set D, where D represents the original running data set;

[0190] The formation conditions are extracted using the association rule mining method. The extraction formula is:

[0191]

[0192] Among them, R represents the generated association rule set, X represents the condition item of the rule, which is the combination of the outlier features, Y represents the result item of the rule, which is the combination of the outlier results, conf represents the confidence of the rule, and sup represents the support of the rule;

[0193] Multidimensional regression analysis was used to extract the influencing factors, and the extraction formula was:

[0194]

[0195] Among them, y represents the result indicator of abnormal formation, β 0 represents the constant term of the regression model, x i represents the i-th potential influencing factor, β i represents the regression coefficient of the i-th influencing factor, n represents the number of influencing factors, and ∈ represents the error term;

[0196] The quantitative formula for the contribution of the influencing factors to the abnormal results is:

[0197]

[0198] Among them, Impact(x i ) represents the impact factor x i Relative contribution to abnormal results, β i Indicates the impact factor x i The regression coefficient, β j Indicates the impact factor x j The regression coefficient of , n represents the number of influencing factors;

[0199] Establish a factor association model, the model formula is:

[0200] M={(x i ,Impact(x i ))|x i ∈ impact factor},

[0201] Among them, M represents the model of impact factor and its contribution, x i Represents a single impact factor, Impact(x i ) represents the impact factor x i relative contribution of

[0202] Specifically, the potential conditions for the formation of abnormal data are analyzed through association rule mining methods to reveal the occurrence mechanism of anomalies. Combined with multidimensional regression analysis technology, the contribution of each influencing factor to the anomaly is quantified, and the factor association model is finally generated.

[0203] Step S4: Based on the influencing factors, a reinforcement learning algorithm is used to perform high-dimensional simulation of the supply chain's supply plan. After multiple strategy iterations, an adjustment plan is generated and sent to relevant nodes of the supply chain through a decision engine.

[0204] The goals of high-dimensional simulation include,

[0205] Reallocate transportation resources to reduce the frequency of inventory shortages,

[0206] Adjust inventory strategies to reduce supply risks.

[0207] Replace high-risk suppliers and improve supply chain stability;

[0208] Based on the influencing factors, the reinforcement learning algorithm is used to perform high-dimensional simulation of the supply chain supply plan. Through multiple strategy iterations, the steps to generate the adjustment plan are as follows:

[0209] Define the core components of reinforcement learning, including:

[0210] State space S: S = {s 1 ,s 2 ,…,s m},

[0211] Where S represents the state space, s i represents the i-th state of the supply chain, m represents the number of all possible states,

[0212] Action space A: A={a 1 ,a 2 ,…,a k},

[0213] Among them, A represents the action space, a j represents the jth adjustment plan, k represents the number of all possible actions,

[0214] Reward function R(s,a):

[0215] R(s,a)=f(cost,stability,risk),

[0216] Among them, R(s,a) represents the reward value of executing action a in state s, f(cost, stability, risk) is the target comprehensive function; the reward function f is expressed as:

[0217]

[0218] Among them, w 1 and w 2 is the weight parameter;

[0219] Use the Qlearning algorithm to update the Q value. The update formula is:

[0220]

[0221] Among them, Q(s,a) represents the Q value of the current state s and action a, α represents the learning rate, R(s,a) represents the value of the reward function, and γ represents the discount factor.

[0222] Indicates the maximum Q value of the next state, s ′ represents the next state, a ′ Indicates the next action;

[0223] Set high-dimensional simulation goals, including:

[0224] Adjust the allocation of transportation resources. The adjustment formula is:

[0225]

[0226] Among them, C i represents the transportation cost of the i-th resource, x i represents the allocation amount of the i-th resource, m represents the total number of resource types,

[0227] To reduce the risk of supply chain disruption, the adjustment formula is:

[0228] maxP(s t ),

[0229] Among them, P(s t ) means in state s t The probability that the supply chain remains stable under t Indicates the state at a certain moment.

[0230] The strategy generation formula is:

[0231]

[0232] Among them, π * represents the optimal strategy,

[0233] represents the action a that maximizes Q(s,a), Q(s,a) represents the value of action a under the current state s;

[0234] Specifically, the supply chain simulation based on reinforcement learning transforms the supply chain adjustment problem into a multidimensional decision-making problem by defining the state space, action space and reward function. The Qlearning algorithm generates the optimal strategy through iterative learning. The goals include optimizing transportation resources, reducing supply chain risks and improving stability, thereby providing adjustment plans for supply chain management.

[0235] Step S5, the supply chain node receives and executes the adjustment plan, monitors the feedback effect during the execution process in real time, analyzes the feedback effect data, and evaluates the effectiveness of the adjustment plan;

[0236] Effectiveness is assessed by:

[0237] Whether reallocated transportation resources reduce the frequency of inventory shortages;

[0238] Whether inventory strategies are effective in reducing the risk of supply chain disruptions.

[0239] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A business collaboration management system based on data analysis, characterized in that: include, Data collection network, which is used to obtain operation data of each node in the supply chain, including inventory level, transportation status and order fluctuation, and divide the operation data into dimensions based on time, space and node attributes; The anomaly detection module is used to identify sudden changes in the trend of operating data. It combines the local anomaly factor (LOF) algorithm to quantitatively evaluate the density distribution of multidimensional data points, mark discrete anomalies and accumulated anomalies, and determine the source of anomalies by correlating data dimensions. The feature analysis module is used to analyze the marked abnormal data, use the association rule mining method to mine the conditions for the formation of abnormal points, and extract the influencing factors through multidimensional regression analysis technology; The decision-making unit uses a reinforcement learning algorithm to perform high-dimensional simulation of the supply chain adjustment plan based on the key influencing factors output by the feature analysis module, and generates an adjustment plan after multiple strategy iterations; Supply chain nodes are used to execute adjustment plans issued by the decision-making unit, including reallocating transportation resources, adjusting inventory strategies, or replacing suppliers; The data feedback module is used to monitor the feedback effects of the adjustment plans executed by the supply chain nodes and evaluate the adjustment plans.

2. A business collaborative management method based on data analysis, based on the business collaborative management system based on data analysis according to claim 1, characterized in that: include: Step S1, obtaining the operation data of each node in the supply chain and dividing it into dimensions; Step S2, using the change point analysis method to identify the trend mutation of the operation data, detect possible abnormal events in the supply chain, and combine the local abnormal factor LOF algorithm to quantitatively evaluate the multidimensional density distribution of data points in the operation data, mark discrete abnormal points and accumulated abnormal points, analyze the abnormal points, and determine the source of the multidimensional data points; Step S3, import the marked abnormal data into the feature analysis module, combine the association rule mining method, analyze the formation conditions of the abnormal points, use multidimensional regression analysis technology to extract the key influencing factors in the abnormal data, and quantify the specific impact of the abnormal data on the supply chain operation. Based on the extracted factors, establish a factor association model; The abnormal data includes abnormal events, discrete abnormal points and accumulated abnormal points; Step S4: Based on the influencing factors, a reinforcement learning algorithm is used to perform high-dimensional simulation of the supply chain's supply plan. After multiple strategy iterations, an adjustment plan is generated and sent to relevant nodes of the supply chain through a decision engine. Step S5: The supply chain node receives and executes the adjustment plan, monitors the feedback effect during the execution process in real time, analyzes the feedback effect data, and evaluates the effectiveness of the adjustment plan.

3. A business collaborative management method based on data analysis as claimed in claim 2, characterized in that: The operation data is divided into dimensions based on time, space and node attributes, including: Inventory level dimensions, including current inventory value, replenishment frequency, and historical trends, Transportation status dimensions, including transportation delay duration, number of orders currently in transit, and transportation route attributes Order fluctuation dimensions include changes in order volume, demand forecast errors, and customer order frequency.

4. A business collaborative management method based on data analysis as claimed in claim 3, characterized in that: The steps of dimension division are: Define the running data set as D, expressed as: D={d1,d2,...,d n },d i ={t i ,l i ,p i ,a i ,b i ,c i }, Where D represents the total set of running data, d i represents the i-th data point, t i Represents the time attribute of the data point, l i represents the spatial attributes of the data points, p i Represents the node attributes of the data point, a i represents the inventory level of a data point, b i Indicates the transportation status of the data point, c i represents the order fluctuation of the data points; The data is divided based on the time dimension, and the division formula is: D t ={d i |t i ∈T}, Among them, D t Represents a data set divided by time dimension, d i represents the i-th data point, t i represents the time attribute of the data point, T represents the time interval set, The data is divided based on the spatial dimension, and the division formula is: D s ={d i |t i ∈S}, Among them, D s Represents a data set divided by spatial dimension, d i represents the i-th data point, l i represents the spatial attribute of the data point, S represents the spatial range set, The data is divided based on node attributes, and the division formula is: D p ={d i |t i ∈P}, Among them, D p Represents a data set divided by node attributes, d i represents the i-th data point, p i represents the node attribute of the data point, and P represents the set of supply chain node attributes.

5. A business collaborative management method based on data analysis as claimed in claim 4, characterized in that: The discrete anomalies include node imbalances caused by transportation delays, The said accumulation anomalies include insufficient inventory within a specific time period; The method for analyzing the abnormal points is as follows: Combine inventory with transportation data to analyze whether inventory anomalies are caused by transportation delays. Order fluctuations are combined with inventory levels to determine whether there is replenishment pressure caused by a surge in demand.

6. A business collaborative management method based on data analysis as claimed in claim 5, characterized in that: The change point analysis method is used to identify the trend mutation of the operation data, detect the abnormal events that may exist in the supply chain, and combine the local abnormal factor LOF algorithm to quantitatively evaluate the multidimensional density distribution of the data points in the operation data, mark the discrete abnormal points and the accumulated abnormal points and analyze the abnormal points. The steps of determining the source of the multidimensional data points are as follows: Define the time series data as X, the formula is: X={x1,x2,...,x n }, Among them, X represents the time series data set, x i represents the observation value at the i-th time point, i∈1,2,...,n; The change point detection algorithm is used to calculate the change in log likelihood, and the calculation formula is: Among them, C(t) represents the mutation point at time t, Indicates the selection of the maximum value among all time points t, ΔL(t) indicates the change in log likelihood before and after time t, represents the probability of data from time 1 to t, Represents the probability of data from time t+1 to n; The local outlier factor LOF is used to quantify the density distribution of multidimensional data. The quantification formula is: Among them, LOF(p i ) represents the data point p i The local anomaly factor, N k (p i ) indicates p i The k-nearest neighbor set, lrd(q) represents the local density inverse of data point q, lrd(p i ) represents the data point p i The local density is reversed, |N k (p i )| represents the set N k (p i ) size; Define the local density inversion, the formula is: Among them, dist(p i ,q) represents p i The distance to q, k-dist(q) represents the k-nearest neighbor distance of data point q; Mark outlier point A: A6{p i |LOF(p i )>λ}, Among them, A represents the set of outliers, p i represents the data point, λ is the abnormality determination threshold, Discrete outlier analysis is performed, and the analysis formula is: R 离散 ={(a, b)|a∈A 离散 , b∈D, the association condition is satisfied}, in, R 离散 Represents the association relationship of discrete outliers, a represents the discrete outlier, b represents other data points associated with the discrete outlier, A 离散 represents a set of discrete outliers, D represents a data set, Perform accumulation outlier analysis, the analysis formula is: R 堆积 ={(c, d)|c∈A 堆积 , d∈D, the association condition is satisfied}, in, R 堆积 represents the correlation between the accumulated abnormal points, c represents the accumulated abnormal points, d represents other data points associated with the accumulated abnormal points, A 堆积 represents a set of accumulated outlier points, and D represents a data set.

7. A business collaborative management method based on data analysis as claimed in claim 6, characterized in that: The step of establishing the factor association model based on the extracted factors is: Extract and mark abnormal data D a , the extraction formula is: D a ={x i |x i ∈D,x i is marked as abnormal}, Among them, D a represents the data set marked as abnormal, x i represents the i-th data point in the data set D, where D represents the original running data set; The formation conditions are extracted using the association rule mining method. The extraction formula is: Among them, R represents the generated association rule set, X represents the condition item of the rule, which is the combination of the outlier features, Y represents the result item of the rule, which is the combination of the outlier results, conf represents the confidence of the rule, and sup represents the support of the rule; Multidimensional regression analysis was used to extract the influencing factors, and the extraction formula was: Among them, y represents the result indicator of abnormal formation, β0 represents the constant term of the regression model, and x i represents the i-th potential influencing factor, β i represents the regression coefficient of the i-th influencing factor, n represents the number of influencing factors, and ∈ represents the error term; The quantitative formula for the contribution of the influencing factors to the abnormal results is: Among them, Impact(x i ) represents the impact factor x i Relative contribution to abnormal results, β i Indicates the impact factor x i The regression coefficient, β j Indicates the impact factor x j The regression coefficient of , n represents the number of influencing factors; Establish a factor association model, the model formula is: M={(x i ,Impact(x i ))|x i ∈ impact factor}, Among them, M represents the model of impact factor and its contribution, x i Represents a single impact factor, Impact(x i ) represents the impact factor x i relative contribution.

8. A business collaborative management method based on data analysis as claimed in claim 7, characterized in that: The goals of the high-dimensional simulation include: Reallocate transportation resources to reduce the frequency of inventory shortages, Adjust inventory strategies to reduce supply risks. Replace high-risk suppliers and improve supply chain stability.

9. A business collaborative management method based on data analysis as claimed in claim 8, characterized in that: Based on the influencing factors, the reinforcement learning algorithm is used to perform high-dimensional simulation of the supply chain supply plan. Through multiple strategy iterations, the steps of generating the adjustment plan are as follows: Define the core components of reinforcement learning, including: State space S: S = {s1, s2, ..., s m }, Where S represents the state space, s i represents the i-th state of the supply chain, m represents the number of all possible states, Action space A: A = {a1, a2, ..., a k }, Among them, A represents the action space, a j represents the jth adjustment plan, k represents the number of all possible actions, Reward function R(s,a): R(s, a) = f(cost, stability, risk), Among them, R(s, a) represents the reward value of executing action a in state s, f(cost, stability, risk) is the target comprehensive function; the reward function f is expressed as: Among them, w1 and w2 are weight parameters; Use the Qlearning algorithm to update the Q value. The update formula is: Among them, Q(s, a) represents the Q value of the current state s and action a, α represents the learning rate, R(s, a) represents the value of the reward function, and γ represents the discount factor. represents the maximum Q value of the next state, s′ represents the next state, and a′ represents the next action; Set high-dimensional simulation goals, including: Adjust the allocation of transportation resources. The adjustment formula is: Among them, C i represents the transportation cost of the i-th resource, x i represents the allocation amount of the i-th resource, m represents the total number of resource types, To reduce the risk of supply chain disruption, the adjustment formula is: max P(s t ), Among them, P(s t ) means in state s t The probability that the supply chain remains stable under t Indicates the state at a certain moment. The strategy generation formula is: Among them, π * represents the optimal strategy, represents the action a that maximizes Q(s,a), and Q(s,a) represents the value of action a in the current state s.

10. A business collaborative management method based on data analysis as claimed in claim 9, characterized in that: Effectiveness is assessed by: Whether reallocated transportation resources reduce the frequency of inventory shortages; Whether inventory strategies are effective in reducing the risk of supply chain disruptions.

Images