A method and system for real-time analysis of enterprise operation status based on multi-source data fusion

Abstract

This invention discloses a method and system for real-time analysis of enterprise operational status based on multi-source data fusion, belonging to the field of enterprise operation management technology. The method includes: collecting enterprise production, market, and supply chain data for multi-source data fusion; constructing a business resource association graph to identify resource competition hotspots; calculating resource gaps and idle resources and establishing a risk function to predict the probability of resource mismatch; and finally optimizing resource scheduling based on the principle of time-series reuse. This invention achieves real-time monitoring of enterprise operational status and precise resource allocation, improves resource utilization efficiency, reduces operational risks, and has significant practical value.

Description

Technical Field

[0001] This invention relates to the field of enterprise operation management technology, specifically to a method and system for real-time analysis of enterprise operation status based on multi-source data fusion. Background Technology

[0002] As businesses expand and their processes become increasingly complex, the data generated during operations becomes multi-sourced, heterogeneous, and highly real-time. Traditional methods for analyzing business operations often focus on only a single dimension of data, failing to comprehensively reflect the overall operational status and leading to inefficient resource allocation.

[0003] Enterprise operational analysis primarily relies on offline analysis of historical data, lacking the ability to integrate and analyze enterprise production, market, and supply chain data in real time. Due to the complex resource competition relationships among various business segments, traditional analysis methods struggle to accurately identify resource competition hotspots and the degree of congestion in transmission paths, failing to promptly detect potential operational risks.

[0004] Most existing resource scheduling methods employ static allocation, failing to adequately consider the dynamic changes in business operations and hindering flexible resource allocation. Furthermore, the lack of quantitative assessment of resource mismatch risks during resource allocation easily leads to resource waste or insufficient supply. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for real-time analysis of enterprise operation status based on multi-source data fusion. By real-time fusion analysis of multi-source heterogeneous data of an enterprise, resource competition hotspots can be accurately identified, dynamic optimization of enterprise resources can be achieved, and enterprise operation efficiency can be improved.

[0006] This invention provides a method for real-time analysis of enterprise operational status based on multi-source data fusion, comprising the following steps:

[0007] Collect enterprise production data, market data, and supply chain data; extract data features based on principal component analysis; and fuse multi-source data to generate comprehensive operational data.

[0008] A business resource association diagram is constructed based on comprehensive operational data. A resource competition conflict matrix is ​​constructed based on the business resource association diagram to identify resource competition hotspots and transmission path congestion levels, and to generate node resource occupancy and business operation status data.

[0009] Based on business operation status data and node resource occupancy, calculate the resource gap and idle resources of each node, establish a resource allocation risk function, predict the probability of resource mismatch based on the gradient of the risk function, optimize the scheduling strategy, and generate a resource allocation scheme.

[0010] Based on the resource allocation scheme, the resource scheduling time interval is divided, the business tasks are decomposed into parallel sub-tasks, and the idle resources are scheduled to supplement the resource gaps using a time-sharing multiplexing method. Resource scheduling instructions are generated and executed to complete the resource allocation.

[0011] Furthermore, enterprise production data, market data, and supply chain data are collected. Principal component analysis is used to extract data features, and multi-source data is fused to generate comprehensive operational data, including:

[0012] Collect enterprise production data, market data, and supply chain data, and standardize the collected data to obtain a standardized data matrix;

[0013] Calculate the feature weight of each indicator data in the standardized data matrix, and perform bidirectional weighted fusion on the standardized data matrix to obtain a bidirectional weighted feature matrix;

[0014] The bidirectional weighted feature matrix is ​​decomposed into features and the principal feature vectors are selected. The standardized data matrix is ​​then mapped to the feature space constructed by the principal feature vectors to obtain the principal component data.

[0015] An adaptive weight matrix is ​​constructed based on the correlation between the principal component data and the standardized data matrix in each dimension. The principal component data is then weighted and fused using the adaptive weight matrix to obtain comprehensive operational data.

[0016] Furthermore, a business resource association graph is constructed based on comprehensive operational data. A resource competition conflict matrix is ​​then built based on this graph to identify resource competition hotspots and transmission path congestion levels, generating node resource occupancy and business operation status data, including:

[0017] Based on comprehensive operational data analysis, the resource usage status of resource nodes and the strength of business associations between nodes are analyzed. Based on the resource usage status, the resource occupancy of nodes is calculated, and based on the strength of business associations, the dependency between nodes is calculated.

[0018] A business resource association graph is constructed by using resource nodes as vertices and inter-node dependencies as edge weights. Based on the edge weights and node resource occupancy in the business resource association graph, the resource competition metric between nodes is calculated, and a resource competition conflict matrix is ​​constructed based on the inter-node dependencies.

[0019] Analyze the time-series cumulative amount of competition metrics in the resource competition conflict matrix, and identify resource competition hotspots based on the changing trend of the time-series cumulative amount;

[0020] Based on the location of resource competition hotspots in the business resource association graph, calculate the degree of congestion of the transmission path along the direction of inter-node dependency.

[0021] The competitive metrics of resource competition hotspots are weighted and combined with the degree of congestion of transmission paths to generate business operation status data and node resource occupancy.

[0022] Furthermore, based on business operation status data and node resource occupancy, the resource gap and idle resources of each node are calculated, including:

[0023] The node competition intensity is calculated based on the competition metric value in the business operation status data, and the upstream competition impact value is calculated based on the congestion level of the transmission path. The node competition intensity and the upstream competition impact value are combined to obtain the resource demand baseline value.

[0024] The node resource occupancy is adjusted based on the resource demand baseline value, and the node resource gap and idle amount are calculated. When the node resource occupancy is less than the resource demand baseline value, it is determined as the resource gap. When the node resource occupancy is greater than the resource demand baseline value, it is determined as the idle amount.

[0025] Furthermore, a resource allocation risk function is established, and the resource mismatch probability is predicted based on the gradient of the risk function, and the scheduling strategy is optimized to generate a resource allocation scheme, including:

[0026] The risk degree function is obtained by multiplying the resource gap and the resource idle amount of each node. The risk degree function is sampled and accumulated according to the time interval to obtain the cumulative risk value. The risk degree gradient is obtained by performing a difference operation on the cumulative risk values ​​of adjacent time points.

[0027] Based on the changing trend of risk gradient, the probability of resource mismatch is predicted, and a priority sequence of resource allocation between nodes is generated.

[0028] The priority sequence is transformed into constraints for the scheduling policy, and optimization is performed using a group search method to generate a scheduling policy for node resources.

[0029] The scheduling strategy is divided into multiple scheduling time intervals. Within each time interval, resources are allocated sequentially according to node priority. The resource occupancy of each node is counted and the total amount of allocable resources is updated to generate a resource allocation scheme.

[0030] Furthermore, based on the resource allocation scheme, the resource scheduling time interval is divided, the business tasks are decomposed into parallelizable subtasks, and idle resources are scheduled to supplement resource gaps using a time-sharing multiplexing method. Resource scheduling instructions are generated and executed to complete resource configuration, including:

[0031] Obtain the resource occupancy difference between adjacent time points in the resource allocation scheme, and divide the resource scheduling time interval based on the fluctuation of the resource occupancy difference;

[0032] Within the resource scheduling time interval, analyze the resource usage correlation of business tasks, group the business tasks according to the resource usage correlation, and divide the business tasks into parallelizable subtasks based on the grouping results.

[0033] The resources of parallelizable subtasks are arranged in time sequence within each resource scheduling time interval. The idle amount in the resource idle interval is allocated to the adjacent resource gap interval according to the time reuse method, and resource scheduling instructions are generated.

[0034] Execute resource scheduling instructions to replenish idle resources to the interval where the resource shortage is located according to the scheduling path, and complete resource allocation.

[0035] Furthermore, the resources of parallelizable subtasks are time-sequentially allocated within each resource scheduling time interval. The idle resources in the idle resource intervals are distributed to adjacent resource gap intervals according to the time reuse method, and resource scheduling instructions are generated, including:

[0036] The resource usage of parallelizable subtasks within the resource scheduling time interval is statistically analyzed, the change in resource usage between adjacent time intervals is calculated, and a time-series arrangement index is generated based on the time-series distribution of the change in resource usage.

[0037] The time-series arrangement index is used as the scheduling weight to weight the idle amount in the resource idle interval and the gap amount in the resource gap interval to generate a scheduling benchmark value.

[0038] Based on the temporal distribution pattern of the scheduling benchmark value and the weighting coefficient of the temporal arrangement index, the time reuse coefficient is calculated for the resource idle interval, and the temporal arrangement is performed on the resource idle interval based on the time reuse coefficient.

[0039] Select the idle resource interval corresponding to the time reuse coefficient from the time sequence arrangement result, calculate the resource matching degree of the adjacent intervals of the idle resource interval, select the target gap interval based on the resource matching degree, allocate the idle amount of the idle resource interval to the target gap interval according to the time reuse method, and generate resource scheduling instructions.

[0040] This invention provides a real-time enterprise operation status analysis system based on multi-source data fusion, the system comprising:

[0041] The data acquisition module is used to collect enterprise production data, market data, and supply chain data. It extracts data features based on principal component analysis and performs multi-source data fusion to generate comprehensive operational data.

[0042] The resource analysis module is used to construct a business resource association diagram based on comprehensive operational data, construct a resource competition conflict matrix based on the business resource association diagram, identify resource competition hotspots and transmission path congestion levels, and generate node resource occupancy and business operation status data.

[0043] The scheme generation module is used to calculate the resource gap and idle resources of each node based on business operation status data and node resource occupancy, establish a resource allocation risk degree function, predict the resource mismatch probability based on the gradient of the risk degree function, optimize the scheduling strategy, and generate a resource allocation scheme.

[0044] The scheduling and execution module is used to divide the resource scheduling time interval according to the resource allocation scheme, decompose the business tasks into parallel sub-tasks, use time-sharing multiplexing to schedule idle resources to supplement the resource gap, generate resource scheduling instructions and execute them to complete resource configuration.

[0045] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0046] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the steps in any of the aforementioned methods.

[0047] This invention achieves real-time monitoring and evaluation of enterprise operational status through multi-source fusion analysis of enterprise production data, market data, and supply chain data, improving the accuracy and timeliness of enterprise operational analysis. Data features extracted based on principal component analysis effectively reduce data redundancy and improve data processing efficiency. By constructing a business resource association graph and a resource competition conflict matrix, it accurately identifies resource competition hotspots and the degree of congestion in transmission paths, providing a reliable decision-making basis for enterprise resource scheduling. The establishment of a resource allocation risk function enables quantitative assessment of resource mismatch risk, optimizes resource scheduling strategies, decomposes business tasks into parallel sub-tasks, and adopts a time-sharing reuse approach for resource scheduling, improving resource utilization and reducing resource waste. The generation and execution of dynamic resource allocation schemes ensure the flexibility and adaptability of enterprise resource allocation, effectively improving the overall operational efficiency of the enterprise. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A flowchart illustrating a real-time analysis method for enterprise operational status based on multi-source data fusion, provided as an embodiment of the present invention;

[0050] Figure 2 This is a schematic diagram comparing the trend of risk degree function over time in an embodiment of the present invention;

[0051] Figure 3 This is a schematic diagram of the structure of a real-time enterprise operation status analysis system based on multi-source data fusion, provided as an embodiment of the present invention. Detailed Implementation

[0052] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. In the following description relating to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements.

[0053] The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with one or more embodiments of this specification. It should be noted that the steps of the corresponding methods are not necessarily performed in the order shown and described in this specification in other embodiments. In some other embodiments, the methods may include more or fewer steps than described in this specification. Furthermore, a single step described in this specification may be broken down into multiple steps in other embodiments; and multiple steps described in this specification may be combined into a single step in other embodiments.

[0054] like Figure 1 As shown, Figure 1 A flowchart of a real-time analysis method for enterprise operation status based on multi-source data fusion provided in this embodiment of the invention, the method comprising the following steps:

[0055] Collect enterprise production data, market data, and supply chain data; extract data features based on principal component analysis; and fuse multi-source data to generate comprehensive operational data.

[0056] A business resource association diagram is constructed based on comprehensive operational data. A resource competition conflict matrix is ​​constructed based on the business resource association diagram to identify resource competition hotspots and transmission path congestion levels, and to generate node resource occupancy and business operation status data.

[0057] Based on business operation status data and node resource occupancy, calculate the resource gap and idle resources of each node, establish a resource allocation risk function, predict the probability of resource mismatch based on the gradient of the risk function, optimize the scheduling strategy, and generate a resource allocation scheme.

[0058] Based on the resource allocation scheme, the resource scheduling time interval is divided, the business tasks are decomposed into parallel sub-tasks, and the idle resources are scheduled to supplement the resource gaps using a time-sharing multiplexing method. Resource scheduling instructions are generated and executed to complete the resource allocation.

[0059] In one optional embodiment, enterprise production data, market data, and supply chain data are collected. Data features are extracted based on principal component analysis, and multi-source data is fused to generate comprehensive operational data, including:

[0060] Collect enterprise production data, market data, and supply chain data, and standardize the collected data to obtain a standardized data matrix;

[0061] Calculate the feature weight of each indicator data in the standardized data matrix, and perform bidirectional weighted fusion on the standardized data matrix to obtain a bidirectional weighted feature matrix;

[0062] The bidirectional weighted feature matrix is ​​decomposed into features and the principal feature vectors are selected. The standardized data matrix is ​​then mapped to the feature space constructed by the principal feature vectors to obtain the principal component data.

[0063] An adaptive weight matrix is ​​constructed based on the correlation between the principal component data and the standardized data matrix in each dimension. The principal component data is then weighted and fused using the adaptive weight matrix to obtain comprehensive operational data.

[0064] Data generated during enterprise operations can be categorized into three main types: production data, market data, and supply chain data. Production data includes production line efficiency, product qualification rate, equipment utilization rate, and energy consumption; market data includes sales revenue, market share, customer satisfaction, and product return rate; and supply chain data includes raw material inventory turnover rate, supplier on-time delivery rate, and logistics efficiency. After data collection, the collected data undergoes standardization to eliminate differences in dimensions and orders of magnitude between different indicators. The standardization process employs zero-mean standardization, subtracting the mean of each indicator from its data and then dividing by the standard deviation to obtain a standardized data matrix with a mean of 0 and a standard deviation of 1.

[0065] Calculate the feature weights of each indicator in the standardized data matrix. Feature weights can be determined based on the business importance of the indicator and data quality assessment. Business importance assessment considers the relevance of the indicator to the company's core business objectives; data quality assessment considers the completeness, consistency, and accuracy of the data. Combining these two factors, the feature weights of each indicator are obtained. A two-way weighted feature matrix is ​​obtained by bidirectionally weighting and fusing the standardized data matrix. The bidirectional weighting and fusing considers weights in both the indicator dimension and the time dimension. The indicator dimension weight is the feature weight calculated above; the time dimension weight is determined using a time decay function, giving higher weights to recent data to reflect data timeliness. The specific method for bidirectional weighted fusing is: multiply each element in the standardized data matrix by its corresponding indicator weight and time weight to form the bidirectional weighted feature matrix.

[0066] Eigenvalue decomposition is performed on the bidirectional weighted feature matrix to extract principal eigenvectors. Eigenvalue decomposition includes calculating the covariance matrix, solving for eigenvalues ​​and eigenvectors, and sorting them in descending order of eigenvalue magnitude. The criterion for selecting principal eigenvectors is that their cumulative contribution rate reaches a preset threshold; typically, a set of eigenvectors with a high cumulative contribution rate is selected. The standardized data matrix is ​​mapped to the feature space constructed from the principal eigenvectors to obtain principal component data. The mapping process involves multiplying the standardized data matrix with the principal eigenvector matrix, achieving data dimensionality reduction and feature extraction. The principal component data retains the main information of the original data while reducing data dimensionality, redundant information, and noise interference.

[0067] An adaptive weight matrix is ​​constructed based on the correlation between principal component data and each dimension of the standardized data matrix. The correlation coefficient method is used to measure the degree of correlation between the principal component data and the original indicators. A correlation coefficient matrix is ​​formed by calculating the correlation coefficient between each principal component in the principal component data and each indicator in the standardized data matrix. The adaptive weight matrix is ​​constructed based on the correlation coefficient matrix, assigning greater weight to indicators with high correlation. The adaptive weights reflect the actual importance of different indicators in the current enterprise operation and are dynamically adjusted according to environmental changes. The principal component data is then weighted and fused using the adaptive weight matrix to obtain comprehensive operational data. This weighted fusion combines the principal component data with the adaptive weight matrix to form a comprehensive indicator that fully reflects the enterprise's operational status.

[0068] Real-time analysis of enterprise operational status is achieved through multi-source data fusion, demonstrating significant technical benefits. By integrating heterogeneous data from multiple sources, including production, market, and supply chain, a comprehensive view of enterprise operational status is provided. Principal component analysis is employed to extract key features and reduce data dimensionality, minimizing redundant information interference. The introduction of bidirectional weighting and adaptive weight matrices makes the data fusion process more flexible, adapting to different enterprise characteristics and dynamically changing market environments. Real-time reflection of changes in enterprise operational status provides decision support for enterprise managers, helping enterprises adjust production strategies promptly, optimize resource allocation, improve market responsiveness and supply chain efficiency, and enhance overall enterprise competitiveness.

[0069] In one optional embodiment, a business resource association graph is constructed based on comprehensive operational data. A resource competition conflict matrix is ​​then constructed based on the business resource association graph to identify resource competition hotspots and transmission path congestion levels. Node resource occupancy and business operation status data are generated, including:

[0070] Based on comprehensive operational data analysis, the resource usage status of resource nodes and the strength of business associations between nodes are analyzed. Based on the resource usage status, the resource occupancy of nodes is calculated, and based on the strength of business associations, the dependency between nodes is calculated.

[0071] A business resource association graph is constructed by using resource nodes as vertices and inter-node dependencies as edge weights. Based on the edge weights and node resource occupancy in the business resource association graph, the resource competition metric between nodes is calculated, and a resource competition conflict matrix is ​​constructed based on the inter-node dependencies.

[0072] Analyze the time-series cumulative amount of competition metrics in the resource competition conflict matrix, and identify resource competition hotspots based on the changing trend of the time-series cumulative amount;

[0073] Based on the location of resource competition hotspots in the business resource association graph, calculate the degree of congestion of the transmission path along the direction of inter-node dependency.

[0074] The competitive metrics of resource competition hotspots are weighted and combined with the degree of congestion of transmission paths to generate business operation status data and node resource occupancy.

[0075] Enterprise operations involve multiple resource nodes, including computing resource nodes, storage resource nodes, network resource nodes, and business resource nodes. Computing resource nodes can be server clusters that process orders, storage resource nodes can be database systems that store enterprise data, network resource nodes can be internal network devices, and business resource nodes can be various business departments such as production, sales, and logistics. These resource nodes are interconnected through business processes, forming a complex enterprise resource network.

[0076] Based on the comprehensive operational data obtained in the previous step, analyze the resource usage status of each resource node and the strength of business connections between nodes. Resource usage status is obtained by monitoring the resource occupancy of each node, such as CPU and memory usage of compute resource nodes, disk usage and read / write speeds of storage resource nodes, and bandwidth utilization and packet loss rate of network resource nodes. Taking compute resource nodes as an example, when the CPU utilization of a server cluster is monitored to be 75% and the memory utilization to be 82%, its resource occupancy can be calculated as the weighted average of the two, i.e., 0.75 × 0.5 + 0.82 × 0.5 = 0.785, indicating that the resource occupancy level of this node is 78.5%. The strength of business connections is determined by analyzing the interaction frequency and data traffic between nodes in the business process.

[0077] A business resource association graph is constructed using resource nodes as vertices and inter-node dependencies as edge weights. In this graph, each resource node is a vertex, and the lines connecting vertices represent business relationships between nodes. The weight of the line represents the inter-node dependency. The higher the dependency, the stronger the business relationship between the two nodes. Based on the edge weights and node resource usage in the business resource association graph, a resource competition metric is calculated between nodes. When two highly dependent nodes both have high resource usage, their resource competition metric will be high. Assuming the order processing system has a resource usage of 0.85, the inventory management system has a resource usage of 0.78, and their dependency is 0.8, then their resource competition metric can be calculated as 0.85 × 0.78 × 0.8 = 0.5304. A resource competition conflict matrix is ​​constructed based on the inter-node dependencies. Each element in this matrix represents the resource competition metric between the corresponding two nodes.

[0078] The analysis focuses on the time-series cumulative value of competition metrics in the resource competition conflict matrix, which is the sum of competition metrics over a certain period. By comparing the time-series cumulative values ​​across different time periods, trends can be observed. For example, the resource competition metric between the order processing system and the inventory management system showed a continuous increase over the past hour, rising from 0.4 to 0.53, while the competition metric between other node pairs remained relatively stable. This suggests a potential resource competition hotspot between the order processing system and the inventory management system. Resource competition hotspots are identified based on the trends in the time-series cumulative values; node pairs with changes exceeding a preset threshold are marked as resource competition hotspots.

[0079] Based on the location of resource contention hotspots in the business resource association graph, the congestion level of the transmission path is calculated along the direction of inter-node dependencies. Starting from the resource contention hotspot, the path is transmitted upstream and downstream along the dependencies, and the combined value of the resource consumption of each node and the inter-node dependencies is calculated to obtain the congestion level of the transmission path. When the order processing system and the inventory management system are identified as resource contention hotspots, the path is transmitted downstream to the logistics management system along the dependencies. If the resource consumption of the logistics management system is 0.65, and the dependency between the inventory management system and the logistics management system is 0.7, then the congestion level of this transmission path can be calculated as 0.5304 × 0.65 × 0.7 = 0.24153.

[0080] The competition metrics for resource contention hotspots are weighted and combined with the congestion level of transmission paths to generate business operation status data and node resource utilization. The weighting can be determined based on the company's specific circumstances; for example, the competition metric could have a weight of 0.6, and the transmission path congestion level a weight of 0.4. The generated business operation status data reflects the operational status of each business segment, while the node resource utilization reflects the load level of each resource node. This data can be visualized for management personnel, helping them to promptly identify and resolve resource contention issues in the company's operations.

[0081] This invention achieves accurate identification and location of internal resource competition hotspots within an enterprise by constructing a business resource association graph and a resource competition conflict matrix. Compared with traditional methods, this method not only considers the resource usage status of each resource node but also the strength of business associations and dependencies between nodes, providing a more comprehensive reflection of the true situation of enterprise resource competition. By analyzing the transmission path of resource competition hotspots, it can predict the potential business impact of resource competition, providing strong support for enterprise resource scheduling and business optimization. It enables real-time monitoring of enterprise resource competition, timely identification of potential resource bottlenecks, and helps enterprises optimize resource allocation, improve resource utilization efficiency, reduce business operating costs, and enhance the enterprise's market competitiveness and risk resistance.

[0082] In one optional embodiment, calculating the resource gap and idle resources of each node based on business operation status data and node resource occupancy includes:

[0083] The node competition intensity is calculated based on the competition metric value in the business operation status data, and the upstream competition impact value is calculated based on the congestion level of the transmission path. The node competition intensity and the upstream competition impact value are combined to obtain the resource demand baseline value.

[0084] The node resource occupancy is adjusted based on the resource demand baseline value, and the node resource gap and idle amount are calculated. When the node resource occupancy is less than the resource demand baseline value, it is determined as the resource gap. When the node resource occupancy is greater than the resource demand baseline value, it is determined as the idle amount.

[0085] In an enterprise resource network, nodes compete with each other due to business operations. Accurately calculating node competition intensity is fundamental to the rational allocation of resources. Node competition intensity is calculated based on competition metrics from the aforementioned generated business operation status data. Node competition intensity reflects the intensity of resource competition between that node and other nodes. For each node, the competition metrics with all its neighboring nodes are statistically analyzed and normalized to obtain the node competition intensity. By collecting these competition metrics and weighting them according to business importance, the overall node competition intensity can be obtained.

[0086] The upstream competition impact value is calculated based on the congestion level of the transmission path. The upstream competition impact value represents the degree to which the resource competition status of upstream nodes affects the current node. During calculation, the node competition intensity of upstream nodes and their dependence on the current node need to be considered. The higher the dependence, the greater the impact of the upstream node's competition status on the current node. By considering the impact of all upstream nodes and weighting them according to their importance, the total upstream competition impact value of the current node can be obtained.

[0087] The baseline resource demand value is obtained by combining the node's competitive intensity with the upstream competitive influence value. This baseline value represents the amount of resources a node should theoretically be allocated in the current business environment. A weighted average approach can be used for this combination, with weights determined based on the company's specific circumstances. For core business nodes, the focus may be more on their own competitive intensity; while for auxiliary business nodes, they may be more influenced by upstream nodes. This flexible weighting configuration better adapts to the characteristics of different types of nodes.

[0088] The node resource utilization is adjusted based on the resource demand baseline to calculate the node resource gap and idle resources. Node resource utilization reflects the actual resource usage monitored. When a node's resource utilization is less than the resource demand baseline, it indicates insufficient resources for that node, and the calculated difference is the resource gap. When a node's resource utilization is greater than the resource demand baseline, it indicates redundant resources for that node, and the calculated difference is the idle resources. For an order processing system, if its actual resource utilization is lower than the resource demand baseline calculated based on competition, a resource gap exists. For an inventory management system, if its actual resource utilization is higher than the resource demand baseline, idle resources exist.

[0089] The calculation of resource gaps and idle resources uses a direct difference method, which involves subtracting the actual resource usage (or vice versa) from the baseline resource demand. For nodes with resource gaps, the difference is positive; for nodes with idle resources, the difference is negative. For ease of understanding and operation, the absolute value of the difference can be taken, and the resource status (gap or idle) can be marked. Enterprise managers can intuitively understand the resource status of each node, providing a basis for subsequent resource adjustments.

[0090] Based on resource gaps and idle resources, dynamic resource adjustment strategies can be further developed. For nodes with significant resource gaps, resources can be allocated from nodes with more idle resources, achieving optimal allocation of internal enterprise resources. During the adjustment process, the cost and time of resource migration need to be considered to avoid system instability caused by frequent adjustments.

[0091] Resource adjustments can be implemented in various ways, including but not limited to: dynamically adjusting server resource allocation, adjusting database connection pool size, adjusting network bandwidth allocation, and optimizing task scheduling strategies. Different adjustment methods can be used for different types of resource nodes. For example, for compute resource nodes, CPU and memory allocation can be dynamically adjusted using virtualization technology; for storage resource nodes, resource allocation can be optimized by adjusting I / O priorities and caching strategies.

[0092] After resource adjustments are completed, their effectiveness needs to be evaluated. This involves monitoring the resource usage and operational status of each node after the adjustments to verify whether the adjustments have achieved the expected results. If the results are unsatisfactory, it may be necessary to reassess the node competition situation, adjust the baseline resource requirements, and conduct a new round of resource optimization. This closed-loop feedback mechanism ensures continuous optimization of resource allocation.

[0093] This invention accurately identifies resource competition hotspots within an enterprise's resource network and, combined with the business relationships between nodes, precisely calculates baseline values ​​for resource demand, providing a scientific basis for optimal resource allocation. Compared to traditional resource allocation methods, this method considers the competitive relationships between nodes and upstream / downstream influences, making it more aligned with the actual operational characteristics of enterprises. By dynamically adjusting resource allocation, it effectively alleviates resource competition hotspots, improves resource utilization efficiency, and reduces operational costs. Its real-time and adaptive nature enables enterprises to quickly respond to changes in the business environment, promptly adjust resource allocation strategies, and enhance the flexibility and stability of enterprise operations.

[0094] In one optional embodiment, a resource allocation risk function is established, and the resource mismatch probability is predicted based on the gradient of the risk function, and the scheduling strategy is optimized to generate a resource allocation scheme, including:

[0095] The risk degree function is obtained by multiplying the resource gap and the resource idle amount of each node. The risk degree function is sampled and accumulated according to the time interval to obtain the cumulative risk value. The risk degree gradient is obtained by performing a difference operation on the cumulative risk values ​​of adjacent time points.

[0096] Based on the changing trend of risk gradient, the probability of resource mismatch is predicted, and a priority sequence of resource allocation between nodes is generated.

[0097] The priority sequence is transformed into constraints for the scheduling policy, and optimization is performed using a group search method to generate a scheduling policy for node resources.

[0098] The scheduling strategy is divided into multiple scheduling time intervals. Within each time interval, resources are allocated sequentially according to node priority. The resource occupancy of each node is counted and the total amount of allocable resources is updated to generate a resource allocation scheme.

[0099] Inappropriate allocation of enterprise resources can lead to operational risks. To quantify these risks, a risk degree function is derived by calculating the product of resource gaps and idle resources at each node. Resource gaps reflect the degree of resource insufficiency at each node, while idle resources reflect the degree of resource waste. Their product comprehensively reflects the severity of resource mismatch. By calculating the risk degree function for all node pairs in the entire enterprise resource network, the overall risk distribution of the network can be obtained.

[0100] The risk level function is sampled at time intervals and accumulated to obtain the cumulative risk value. The time interval can be set to 5 minutes, 10 minutes, or longer, depending on the rate of change in the business. At each sampling time point, the current risk level function value is calculated and added to the previous cumulative risk value to obtain the new cumulative risk value. If the initial cumulative risk value is 0, and the risk level function value at the first sampling time point is 0.018, then the cumulative risk value at the first sampling time point is 0 + 0.018 = 0.018; if the risk level function value at the second sampling time point is 0.025, then the cumulative risk value at the second sampling time point is 0.018 + 0.025 = 0.043. Through continuous sampling, a time series of cumulative risk values ​​can be obtained.

[0101] The risk gradient is obtained by subtracting the cumulative risk values ​​at adjacent time points. The risk gradient reflects the rate of change in risk and is an important indicator for predicting future risk changes. If the cumulative risk value at the second sampling time point is 0.043 and the cumulative risk value at the third sampling time point is 0.075, then the risk gradient at the third sampling time point is 0.075 - 0.043 = 0.032. By calculating the risk gradients at multiple consecutive time points, a time series of risk gradients can be obtained.

[0102] The probability of resource mismatch is predicted based on the changing trend of the risk gradient. If the risk gradient is consistently positive and increasing, it indicates that the resource mismatch is worsening and the probability of resource mismatch is high. If the risk gradient is positive but tends to stabilize, it indicates that the resource mismatch exists but has stabilized and the probability of resource mismatch is moderate. If the risk gradient is negative, it indicates that the resource mismatch is improving and the probability of resource mismatch is low. Based on the predicted resource mismatch probability, a priority sequence for resource allocation among nodes is generated. Node pairs with higher resource mismatch probabilities have higher adjustment priority.

[0103] The priority sequence is transformed into constraints for the scheduling strategy, and optimization is performed using a swarm search approach to generate a scheduling strategy for node resources. Swarm search is a heuristic optimization algorithm that finds the optimal solution by simulating swarm behavior. In the resource scheduling problem, each possible resource allocation scheme can be regarded as a point in the search space, and the optimal solution is found by evaluating the fitness of different points. Constraints include: limited total resources, guaranteed minimum resource requirements for nodes, and adherence to the priority sequence. The fitness function can be designed as the negative of the risk function, i.e., the lower the risk, the higher the fitness. Through the swarm search algorithm, the resource allocation scheme with the lowest risk under the current constraints, i.e., the optimal scheduling strategy, can be obtained.

[0104] The scheduling strategy is divided into multiple scheduling time intervals. Within each time interval, resources are allocated sequentially according to node priority. The division of time intervals can be determined based on the characteristics of the enterprise's business, such as dividing them according to peak and off-peak periods, or dividing them according to fixed time intervals. Within each time interval, resources are allocated to each node in order of priority. During allocation, the resource needs of high-priority nodes are met first, followed by lower-priority nodes.

[0105] The resource allocation plan involves statistically analyzing the resource usage of each node, updating the total available resources, and generating a resource allocation scheme. After resource allocation, the actual amount of resources received by each node needs to be recorded, and the remaining total available resources calculated to prepare for resource allocation in the next time interval. The resource allocation scheme should include the amount of resources each node should receive in each time interval, as well as specific operational guidelines for resource adjustments.

[0106] This invention achieves dynamic optimization of enterprise resources by quantifying resource mismatch risk and predicting resource demand trends. Compared with traditional resource allocation methods, it considers not only current resource usage but also historical data and future trends, demonstrating foresight and adaptability. By introducing a risk degree function and the concept of cumulative risk value, it can accurately identify the severity and changing trends of resource mismatch, providing a scientific basis for resource scheduling. Employing a swarm search optimization algorithm, it can find near-optimal resource allocation schemes under complex constraints, improving resource utilization efficiency. Through time interval division and priority ranking, it can flexibly respond to the time-varying characteristics of enterprise business load, ensuring that the resource needs of critical businesses are met first.

[0107] like Figure 2 The diagram illustrates a comparison of the risk function's change over time in this embodiment. It compares the risk function changes of the traditional method and the proposed solution over time. The horizontal axis represents the time interval (minutes), and the vertical axis represents the risk function value. The red curve represents the traditional method, whose risk value mostly fluctuates between 40% and 80% and shows an upward trend, indicating that the instability of system resource allocation increases over time. The blue curve represents the proposed solution, whose risk value remains stable between 15% and 32%, with an average reduction of approximately 60%, demonstrating that the solution effectively controls resource mismatch risk. Particularly noteworthy is the surge in risk from the traditional method to over 70% at time points 8-12, while the proposed solution maintains a risk value of around 30%, showcasing a significant advantage under high load conditions. This result verifies that the resource scheduling strategy based on the risk function has higher stability and reliability during long-term operation.

[0108] In one optional embodiment, the resource scheduling time interval is divided according to the resource allocation scheme, the business task is decomposed into parallelizable subtasks, idle resources are scheduled to supplement the resource gap using a time-sharing multiplexing method, resource scheduling instructions are generated and executed, and the resource configuration is completed by including:

[0109] Obtain the resource occupancy difference between adjacent time points in the resource allocation scheme, and divide the resource scheduling time interval based on the fluctuation of the resource occupancy difference;

[0110] Within the resource scheduling time interval, analyze the resource usage correlation of business tasks, group the business tasks according to the resource usage correlation, and divide the business tasks into parallelizable subtasks based on the grouping results.

[0111] The resources of parallelizable subtasks are arranged in time sequence within each resource scheduling time interval. The idle amount in the resource idle interval is allocated to the adjacent resource gap interval according to the time reuse method, and resource scheduling instructions are generated.

[0112] Execute resource scheduling instructions to replenish idle resources to the interval where the resource shortage is located according to the scheduling path, and complete resource allocation.

[0113] During the execution of an enterprise resource allocation plan, it is crucial to accurately grasp the timing and granularity of resource scheduling. This involves obtaining the resource usage difference between adjacent time points within the resource allocation plan to identify trends in resource usage changes. The resource usage difference refers to the difference in resource usage for the same node between two adjacent time points. For example, if the resource usage of an order processing system is 0.65 at 8:00 and 0.72 at 8:05, the resource usage difference between these two time points is 0.07. The resource usage difference is calculated for all nodes at all sampled time points to obtain a resource usage difference sequence.

[0114] Resource scheduling time intervals are defined based on fluctuations in resource utilization differences. These fluctuations can be measured by the variance or range of the difference sequence. When fluctuations exceed a preset threshold, it can be considered that resource demand has changed significantly, requiring resource scheduling. For example, business data from a company shows that 9:00-11:00 AM and 2:00-4:00 PM are peak order processing times with larger fluctuations in resource utilization differences; while fluctuations are smaller during other time periods. Based on this, a day can be divided into multiple resource scheduling time intervals: 0:00-9:00 AM is a low-load interval, 9:00-11:00 AM is the morning peak interval, 11:00-2:00 PM is the medium-load interval, 2:00-4:00 PM is the afternoon peak interval, and 4:00-12:00 AM is the low-load interval. Resource demands and allocation strategies differ across these intervals.

[0115] Analyze the resource usage correlation of business tasks within the resource scheduling time interval. Resource usage correlation reflects the similarity of resource usage among different business tasks. The correlation between tasks is calculated by analyzing the temporal distribution, intensity, and patterns of resource usage. Order processing and payment processing are both business tasks that require intensive use of computing and network resources and have a high degree of temporal overlap, thus their resource usage correlation is high. Conversely, order processing and data backup differ significantly in resource usage, thus their resource usage correlation is low.

[0116] Business tasks are grouped based on resource consumption correlation, and then divided into parallelizable subtasks based on the grouping results. Tasks with high correlation are not suitable for parallel execution because they may compete for the same resources; tasks with low correlation can be executed in parallel because they use different types of resources or their usage times are staggered. The enterprise's business tasks are divided into four groups: order processing and payment processing (Group 1), inventory management and logistics management (Group 2), user management and permission verification (Group 3), and data analysis and backup (Group 4). Tasks within the same group have high resource consumption correlation and are not suitable for parallel execution; tasks between different groups have low correlation and can be parallelized. Based on the grouping results, each business task can be further divided into multiple parallelizable subtasks. The order processing task can be divided into subtasks such as order receiving, order verification, and order distribution, which can be executed in parallel on different resource nodes.

[0117] The resources for parallelizable subtasks are scheduled in a time sequence across different resource scheduling intervals. This scheduling needs to consider task priority, dependencies, and resource requirements. High-priority tasks should be allocated resources first, tasks with dependencies should be executed in the order of dependency, and tasks with high resource requirements may need to wait for sufficient resource accumulation. For example, during the morning peak period (9:00-11:00), the order receiving subtask has the highest priority and needs sufficient resources; the order verification subtask depends on the order receiving subtask and needs to be executed after order receiving is completed; the data backup subtask has a lower priority and can be postponed to a lower load period.

[0118] The idle resources in the resource-idle intervals are allocated to adjacent resource-deficient intervals using a time-reuse method, generating resource scheduling instructions. By comparing resource demands across different time intervals, resource-idle intervals and resource-deficient intervals are identified. For example, the computing resource utilization rate is low in the low-load interval (0:00-9:00), resulting in idle resources; while the computing resource demand is high in the morning peak interval (9:00-11:00), resulting in a resource deficit. Idle resources in the low-load intervals can be allocated to the peak intervals using a time-reuse method. Time reuse refers to flexibly adjusting resource allocation strategies across different time periods, allowing the same resource to serve different tasks at different times. Resource scheduling instructions include information such as the time of resource scheduling, resource type, quantity, source node, and target node.

[0119] Execute resource scheduling instructions to replenish idle resources to the resource gap area according to the scheduling path, completing resource configuration. Resource scheduling can be achieved through various technical means, such as virtual machine migration, container orchestration, and load balancing. During low-load periods (0:00-9:00), some server computing resources can be allocated to data backup and analysis tasks; while during peak periods (9:00-11:00), these resources are reallocated to order processing and payment processing tasks. During resource scheduling, service continuity must be ensured to avoid business interruptions due to resource adjustments. After resource scheduling is completed, the resource usage and business operation status of each node need to be monitored to verify the effectiveness of resource scheduling.

[0120] This invention achieves optimized spatiotemporal allocation of enterprise resources by accurately identifying resource scheduling time intervals and analyzing the resource consumption characteristics of business tasks. Compared with traditional resource scheduling methods, it not only considers the spatial distribution of resources but also fully utilizes the temporal characteristics of resource usage, improving resource utilization efficiency through time reuse. By grouping business tasks and dividing them into subtasks, it can fully tap the parallel potential of business tasks, accelerate task execution, and improve enterprise response speed. The task scheduling strategy based on resource consumption correlation avoids resource competition between highly correlated tasks, reduces system resource conflicts, and improves system stability. By dynamically adjusting resource allocation, it can adapt to the temporal variation characteristics of enterprise business load, ensuring sufficient resources during peak business periods while avoiding resource waste during off-peak periods.

[0121] In one optional embodiment, the resources of parallelizable subtasks are time-sequentially arranged within each resource scheduling time interval, and the idle amount in the resource idle interval is allocated to adjacent resource gap intervals according to the time reuse method. The generated resource scheduling instructions include:

[0122] The resource usage of parallelizable subtasks within the resource scheduling time interval is statistically analyzed, the change in resource usage between adjacent time intervals is calculated, and a time-series arrangement index is generated based on the time-series distribution of the change in resource usage.

[0123] The time-series arrangement index is used as the scheduling weight to weight the idle amount in the resource idle interval and the gap amount in the resource gap interval to generate a scheduling benchmark value.

[0124] Based on the temporal distribution pattern of the scheduling benchmark value and the weighting coefficient of the temporal arrangement index, the time reuse coefficient is calculated for the resource idle interval, and the temporal arrangement is performed on the resource idle interval based on the time reuse coefficient.

[0125] Select the idle resource interval corresponding to the time reuse coefficient from the time sequence arrangement result, calculate the resource matching degree of the adjacent intervals of the idle resource interval, select the target gap interval based on the resource matching degree, allocate the idle amount of the idle resource interval to the target gap interval according to the time reuse method, and generate resource scheduling instructions.

[0126] Enterprise resource optimization requires a precise understanding of the temporal characteristics of resource usage. This involves statistically analyzing the resource usage of parallelizable subtasks within a resource scheduling time interval and recording the resource usage of each subtask in each time interval. As a preferred embodiment of this invention, in an e-commerce platform, the order processing task is divided into three parallelizable subtasks: order receiving, order verification, and order distribution. The computational, memory, and network resource usage of these subtasks in different time intervals is statistically analyzed to obtain complete temporal distribution data of resource usage. This data reflects the temporal distribution characteristics of the enterprise's business load and serves as a crucial basis for resource scheduling.

[0127] Calculate the resource utilization changes between adjacent time intervals, i.e., the resource utilization differences of each subtask within adjacent time intervals. For order processing-related subtasks, calculate the resource utilization differences between adjacent time periods, such as the morning interval and the noon interval, and the noon interval and the afternoon interval. These changes directly reflect the fluctuations in business load; the larger the change, the greater the difference in business load between adjacent time intervals, and the more drastic the changes in resource demand. Generate a time-series distribution index based on the temporal distribution of resource utilization changes. The time-series distribution index reflects the drasticness of resource demand changes; the more drastic the changes, the higher the time-series distribution index. The time-series distribution index can be calculated using the standard deviation or coefficient of variation of resource utilization changes.

[0128] Using the time-series distribution index as the scheduling weight, a weighted average is applied to the idle resources in resource-idle intervals and the resource shortage intervals to generate a scheduling benchmark value. This benchmark value comprehensively considers both the degree of resource idleness or shortage and the drastic changes in resource demand, providing a basis for resource scheduling. The scheduling benchmark value for resource-idle intervals is the idle amount multiplied by the time-series distribution index, while the benchmark value for resource-short intervals is the shortage amount multiplied by the time-series distribution index. A higher scheduling benchmark value indicates a higher priority for resource scheduling in that interval. This weighted approach prioritizes scheduling time intervals with drastic changes in resource demand and significant resource shortages.

[0129] Based on the temporal distribution pattern of the scheduling baseline value and the weighting coefficient of the temporal arrangement index, a time reuse coefficient is calculated for idle resource intervals. The time reuse coefficient reflects the potential of idle resources to support other time intervals. The calculation considers the magnitude and distribution pattern of the scheduling baseline value, as well as the weighting coefficient of the temporal arrangement index. The weighting coefficient of the temporal arrangement index can be determined based on the enterprise's business characteristics and management strategies, reflecting the importance of scheduling different types of resources. For e-commerce platforms, the temporal arrangement of computing resources may be more important, thus assigning a higher weighting coefficient to computing resources; while for content distribution platforms, the temporal arrangement of network resources may be more important, thus assigning a higher weighting coefficient to network resources.

[0130] Resource idle intervals are time-series arranged based on time reuse coefficients, ranked from highest to lowest, with priority given to intervals with high time reuse coefficients for resource scheduling. This ensures the efficiency and effectiveness of resource scheduling. Intervals with high time reuse coefficients typically have greater resource scheduling potential and can more effectively alleviate resource shortages in other intervals. A time reuse coefficient threshold can be set; only idle intervals exceeding the threshold will be included in the scheduling scope, avoiding system overhead caused by small-scale scheduling.

[0131] From the time-series scheduling results, select the idle resource intervals corresponding to the time reuse coefficients and calculate the resource matching degree between adjacent intervals of this idle interval. The resource matching degree reflects the degree of matching between the resource characteristics of the idle interval and the resource demand of the gap interval. Factors such as resource type, quantity, and time distribution are considered during the calculation. Idle computing resources in the night interval have a high matching degree with the gap computing resources in the morning interval; however, idle storage resources in the night interval have a low matching degree with the gap computing resources in the morning interval. The calculation of the resource matching degree needs to comprehensively consider the portability, substitutability, and compatibility of resources to ensure that the resources after scheduling can effectively meet business needs.

[0132] Target gap intervals are selected based on resource matching degree, with gap intervals having high matching degree being prioritized as scheduling targets. Idle resources in idle intervals are allocated to target gap intervals using a time-reuse method, generating resource scheduling instructions. Resource scheduling instructions include information such as resource type, quantity, source interval, target interval, and scheduling time. Time-reuse can be implemented through various technologies, such as task delay execution, pre-execution, and resource virtualization. The generation of resource scheduling instructions must consider scheduling costs and business impact to avoid system instability caused by frequent scheduling.

[0133] After resource scheduling instructions are generated, they need to be executed through the enterprise resource management system to allocate resources from idle intervals to gap intervals according to the specified scheduling paths. During execution, the resource scheduling effect needs to be monitored in real time, such as changes in resource utilization and business response time, to ensure that the scheduling achieves the expected results. If the scheduling effect is found to be unsatisfactory, the scheduling strategy needs to be adjusted in a timely manner to optimize the resource allocation scheme. After scheduling is completed, the scheduling results need to be recorded to provide a reference for the next round of resource optimization.

[0134] This invention achieves precise time-series scheduling of enterprise resources by introducing a time-series arrangement index and a time reuse coefficient. Compared with traditional resource scheduling methods, this method places greater emphasis on the time dimension of resource usage, fully utilizing resource reserves in different time intervals through a time reuse mechanism, significantly improving overall resource utilization efficiency. A target interval selection strategy based on resource matching ensures the accuracy and effectiveness of resource scheduling, avoiding ineffective scheduling and resource waste. The design of the time-series arrangement index as a scheduling weight makes resource scheduling more focused on intervals with significant business fluctuations, enhancing the system's adaptability to changes in business load. Through matching calculations between idle and gap intervals, the optimal resource scheduling path can be found, minimizing resource scheduling costs and impact on business operations.

[0135] like Figure 3 As shown, Figure 3 A schematic diagram of a real-time enterprise operation status analysis system based on multi-source data fusion provided in this embodiment of the invention, the system comprising:

[0136] The data acquisition module is used to collect enterprise production data, market data, and supply chain data. It extracts data features based on principal component analysis and performs multi-source data fusion to generate comprehensive operational data.

[0137] The resource analysis module is used to construct a business resource association diagram based on comprehensive operational data, construct a resource competition conflict matrix based on the business resource association diagram, identify resource competition hotspots and transmission path congestion levels, and generate node resource occupancy and business operation status data.

[0138] The scheme generation module is used to calculate the resource gap and idle resources of each node based on business operation status data and node resource occupancy, establish a resource allocation risk degree function, predict the resource mismatch probability based on the gradient of the risk degree function, optimize the scheduling strategy, and generate a resource allocation scheme.

[0139] The scheduling and execution module is used to divide the resource scheduling time interval according to the resource allocation scheme, decompose the business tasks into parallel sub-tasks, use time-sharing multiplexing to schedule idle resources to supplement the resource gap, generate resource scheduling instructions and execute them to complete resource configuration.

[0140] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0141] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing a computer program, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0142] The specific embodiments described above are preferred embodiments of the present invention and are not intended to limit the specific scope of the present invention. The scope of the present invention includes, but is not limited to, these specific embodiments. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention.

Claims

1. A method for real-time analysis of enterprise operational status based on multi-source data fusion, characterized in that, Includes the following steps: Collect enterprise production data, market data, and supply chain data; extract data features based on principal component analysis; and fuse multi-source data to generate comprehensive operational data. A business resource association diagram is constructed based on comprehensive operational data. A resource competition conflict matrix is ​​constructed based on the business resource association diagram to identify resource competition hotspots and transmission path congestion levels, and to generate node resource occupancy and business operation status data. Based on business operation status data and node resource occupancy, calculate the resource gap and idle resources of each node, establish a resource allocation risk function, predict the probability of resource mismatch based on the gradient of the risk function, optimize the scheduling strategy, and generate a resource allocation scheme. Obtain the resource occupancy difference between adjacent time points in the resource allocation scheme, and divide the resource scheduling time interval based on the fluctuation of the resource occupancy difference; Within the resource scheduling time interval, analyze the resource usage correlation of business tasks, group the business tasks according to the resource usage correlation, and divide the business tasks into parallelizable subtasks based on the grouping results. The resources of parallelizable subtasks are arranged in time sequence within each resource scheduling time interval. The idle amount in the resource idle interval is allocated to the adjacent resource gap interval according to the time reuse method, and resource scheduling instructions are generated. Execute resource scheduling instructions to replenish idle resources to the interval where the resource shortage is located according to the scheduling path, and complete resource allocation.

2. The method according to claim 1, characterized in that, Collect enterprise production data, market data, and supply chain data; extract data features based on principal component analysis; and perform multi-source data fusion to generate comprehensive operational data, including: Collect enterprise production data, market data, and supply chain data, and standardize the collected data to obtain a standardized data matrix; Calculate the feature weight of each indicator data in the standardized data matrix, and perform bidirectional weighted fusion on the standardized data matrix to obtain a bidirectional weighted feature matrix; The bidirectional weighted feature matrix is ​​decomposed into features and the principal feature vectors are selected. The standardized data matrix is ​​then mapped to the feature space constructed by the principal feature vectors to obtain the principal component data. An adaptive weight matrix is ​​constructed based on the correlation between the principal component data and the standardized data matrix in each dimension. The principal component data is then weighted and fused using the adaptive weight matrix to obtain comprehensive operational data.

3. The method according to claim 1, characterized in that, Based on comprehensive operational data, a business resource association diagram is constructed. A resource competition conflict matrix is ​​then built upon this diagram to identify resource competition hotspots and transmission path congestion levels. This generates node resource occupancy and business operation status data, including: Based on comprehensive operational data analysis, the resource usage status of resource nodes and the strength of business associations between nodes are analyzed. Based on the resource usage status, the resource occupancy of nodes is calculated, and based on the strength of business associations, the dependency between nodes is calculated. A business resource association graph is constructed by using resource nodes as vertices and inter-node dependencies as edge weights. Based on the edge weights and node resource occupancy in the business resource association graph, the resource competition metric between nodes is calculated, and a resource competition conflict matrix is ​​constructed based on the inter-node dependencies. Analyze the time-series cumulative amount of competition metrics in the resource competition conflict matrix, and identify resource competition hotspots based on the changing trend of the time-series cumulative amount; Based on the location of resource competition hotspots in the business resource association graph, calculate the degree of congestion of the transmission path along the direction of inter-node dependency. The competitive metrics of resource competition hotspots are weighted and combined with the degree of congestion of transmission paths to generate business operation status data and node resource occupancy.

4. The method according to claim 1, characterized in that, Based on business operation status data and node resource usage, the resource gap and idle resources of each node are calculated, including: The node competition intensity is calculated based on the competition metric value in the business operation status data, and the upstream competition impact value is calculated based on the congestion level of the transmission path. The node competition intensity and the upstream competition impact value are combined to obtain the resource demand baseline value. The node resource occupancy is adjusted based on the resource demand baseline value, and the node resource gap and idle amount are calculated. When the node resource occupancy is less than the resource demand baseline value, it is determined as the resource gap. When the node resource occupancy is greater than the resource demand baseline value, it is determined as the idle amount.

5. The method according to claim 1, characterized in that, Establish a resource allocation risk function, predict the resource mismatch probability based on the gradient of the risk function, optimize the scheduling strategy, and generate a resource allocation scheme, including: The risk degree function is obtained by multiplying the resource gap and the resource idle amount of each node. The risk degree function is sampled and accumulated according to the time interval to obtain the cumulative risk value. The risk degree gradient is obtained by performing a difference operation on the cumulative risk values ​​of adjacent time points. Based on the changing trend of risk gradient, the probability of resource mismatch is predicted, and a priority sequence of resource allocation between nodes is generated. The priority sequence is transformed into constraints for the scheduling policy, and optimization is performed using a group search method to generate a scheduling policy for node resources. The scheduling strategy is divided into multiple scheduling time intervals. Within each time interval, resources are allocated sequentially according to node priority. The resource occupancy of each node is counted and the total amount of allocable resources is updated to generate a resource allocation scheme.

6. The method according to claim 1, characterized in that, The system performs time-series allocation of resources for parallelizable subtasks within each resource scheduling time interval, distributes idle resources in idle intervals to adjacent resource gap intervals according to time reuse, and generates resource scheduling instructions including: The resource usage of parallelizable subtasks within the resource scheduling time interval is statistically analyzed, the change in resource usage between adjacent time intervals is calculated, and a time-series arrangement index is generated based on the time-series distribution of the change in resource usage. The time-series arrangement index is used as the scheduling weight to weight the idle amount in the resource idle interval and the gap amount in the resource gap interval to generate a scheduling benchmark value. Based on the temporal distribution pattern of the scheduling benchmark value and the weighting coefficient of the temporal arrangement index, the time reuse coefficient is calculated for the resource idle interval, and the temporal arrangement is performed on the resource idle interval based on the time reuse coefficient. Select the idle resource interval corresponding to the time reuse coefficient from the time sequence arrangement result, calculate the resource matching degree of the adjacent intervals of the idle resource interval, select the target gap interval based on the resource matching degree, allocate the idle amount of the idle resource interval to the target gap interval according to the time reuse method, and generate resource scheduling instructions.

7. A real-time enterprise operation status analysis system based on multi-source data fusion, used to implement the method described in any one of claims 1-6, characterized in that, The system includes: The data acquisition module is used to collect enterprise production data, market data, and supply chain data. It extracts data features based on principal component analysis and performs multi-source data fusion to generate comprehensive operational data. The resource analysis module is used to construct a business resource association diagram based on comprehensive operational data, construct a resource competition conflict matrix based on the business resource association diagram, identify resource competition hotspots and transmission path congestion levels, and generate node resource occupancy and business operation status data. The scheme generation module is used to calculate the resource gap and idle resources of each node based on business operation status data and node resource occupancy, establish a resource allocation risk degree function, predict the resource mismatch probability based on the gradient of the risk degree function, optimize the scheduling strategy, and generate a resource allocation scheme. The scheduling and execution module is used to divide the resource scheduling time interval according to the resource allocation scheme, decompose the business tasks into parallel sub-tasks, use time-sharing multiplexing to schedule idle resources to supplement the resource gap, generate resource scheduling instructions and execute them to complete resource configuration.

8. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 6.

Images