Enterprise cost optimization and budget management method based on big data analysis

Abstract

The invention discloses an enterprise cost optimization and budget management method based on big data analysis. The method comprises the steps that S1, real-time data of all departments are collected and preprocessed; s2, constructing a graph structure model of enterprise cost and budget, distributing a preliminary feature vector for each node, and constructing a node feature matrix; s3, the node feature matrix is processed through multi-layer convolution operation, and graph embedding representation of the nodes is generated; s4, based on the graph embedding representation generated by the graph convolutional network, performing budget optimization by using an adaptive simulated annealing algorithm, and outputting a budget allocation scheme; s5, automatically adjusting budget and resource configuration of each department and project of the enterprise; s6, displaying various information of the enterprise in real time through a visual decision support system; and S7, continuously optimizing cost management and budget control of the enterprise. According to the method, an efficient and scientific optimization scheme can be provided in enterprise cost optimization and budget management, and remarkable technical values and economic benefits are brought to practical application.

Description

Technical Field

[0001] The present invention relates to the technical field of enterprise cost optimization and budget management, and particularly to a method for enterprise cost optimization and budget management based on big data analysis. Background Art

[0002] With the expansion of enterprise scale and the increase in management complexity, how to effectively manage costs and optimize budgets has become an important task in enterprise management. Most of the existing enterprise cost management and budget optimization methods rely on traditional manual budget preparation and management processes, and these methods often struggle to meet the high-efficiency requirements of modern enterprises in aspects such as information flow, capital flow, and resource allocation. Traditional budget management is usually based on experience and static models, and fails to fully consider the complex relationships between departments and projects as well as changes in the external environment. Therefore, problems such as budget execution deviation, resource waste, and uneven capital allocation are likely to occur.

[0003] With the development of big data technology and artificial intelligence, many enterprises have begun to attempt to optimize budget management and cost control through big data analysis and intelligent algorithms. Traditional cost management and budget optimization methods are difficult to process massive amounts of data and have insufficient understanding of the relationships between various departments and projects within the enterprise. For this reason, data-driven analysis methods have gradually become a new trend. Currently, many enterprises have adopted budget prediction models based on historical data, rule-based cost control methods, and simple optimization algorithms, but these methods generally have the defect of insufficient handling of the complex structure and multi-objective optimization problems of enterprises. Many existing algorithms rely too much on static data analysis and ignore the dynamic associations between departments, projects, and the market environment, resulting in difficulties in real-time adjustment and optimization during the budget execution process.

[0004] In the prior art, as an advanced deep learning method, the graph convolutional network can effectively process graph data and extract the complex relationships between nodes therein. Through the structure of a multi-layer convolutional network, it can capture the spatial and structural dependency relationships between nodes and is suitable for analyzing and optimizing data containing complex interrelationships, such as department collaboration, resource allocation, and capital flow within an enterprise. However, the existing budget management systems have not widely applied the graph convolutional network to model and analyze the complex relationships between departments and projects, resulting in difficulties in accurately optimizing budgets and allocating resources in actual operations.

[0005] In addition, although some existing enterprise budget management methods have begun to introduce optimization techniques such as the simulated annealing algorithm, these methods usually face the following problems: First, the traditional simulated annealing algorithm is prone to falling into local optimal solutions in high-dimensional problems and is difficult to globally search for the optimal solution; Second, these algorithms fail to fully combine the dynamic environment of enterprises and lack a real-time feedback mechanism, resulting in the inability to effectively respond to changes in the external environment and fluctuations in internal demands during the budget optimization process. Therefore, the budget optimization methods of the existing technologies cannot flexibly adapt to the dynamic changes in the actual operation of enterprises, resulting in the enterprise's inability to quickly adjust budget allocation to cope with unexpected situations in actual operation.

[0006] Currently, the technology of combining graph convolutional networks with the adaptive simulated annealing algorithm for enterprise budget optimization is still in the exploratory stage. Especially, the research on dynamic adjustment and real-time feedback is relatively weak. Existing systems usually only rely on static analysis and historical data, and lack a real-time monitoring and adjustment mechanism, making it difficult to efficiently handle the complex resource flows and cost changes within the enterprise. Therefore, the existing technologies are difficult to meet the high-efficiency requirements of enterprises for cost optimization and budget management in a rapidly changing market environment. Summary of the Invention

[0007] An object of the present invention is to propose a method for enterprise cost optimization and budget management based on big data analysis. The present invention can provide an efficient and scientific optimization scheme in enterprise cost optimization and budget management, bringing significant technical value and economic benefits to practical applications.

[0008] A method for enterprise cost optimization and budget management based on big data analysis according to an embodiment of the present invention includes the following steps:

[0009] S1. Automatically collect real-time data from each department through various management systems of the enterprise and perform preprocessing;

[0010] S2. Construct a graph structure model of enterprise costs and budgets, assign preliminary feature vectors to each node, and construct a node feature matrix based on the preliminary feature vectors of each node;

[0011] S3. Use a graph convolutional network to process the node feature matrix through multi-layer convolutional operations, capture the long-term dependence relationships between nodes and the resource allocation information between departments, further fuse the interactions between nodes, and generate a graph embedding representation of the nodes;

[0012] S4. Based on the graph embedding representation generated by the graph convolutional network, use the adaptive simulated annealing algorithm for budget optimization, define a multi-objective optimization function for budget optimization, and perform global search through the temperature adjustment mechanism in simulated annealing, dynamically adjust the temperature drop speed and step size, and output a budget allocation plan;

[0013] S5. Automatically adjust the budgets and resource allocations of each department and project in the enterprise based on the budget allocation plan, and monitor the budget execution process in real time. Use the feedback mechanism to dynamically adjust the budget.

[0014] S6. Through the integrated visual decision support system, display the enterprise's cost data, budget execution status, and optimization suggestion information in real time.

[0015] S7. During the budget execution process, the system continuously optimizes the enterprise's cost management and budget control by continuously tracking the actual execution results, external environment changes, and internal adjustments.

[0016] Optionally, S1 includes the following steps:

[0017] S11. Automatically collect the real-time data of each department from various management systems of the enterprise.

[0018] S12. Preprocess the collected data. The preprocessing steps include removing missing data, filling in missing values, and removing outliers.

[0019] S13. Use data cleaning technology to detect and correct outliers in the data and perform standardization processing.

[0020] Optionally, S2 includes the following steps:

[0021] S21. Construct a graph structure model of the enterprise's costs and budgets, with each department, resource, and project as nodes of the graph, and the capital flows, collaboration relationships, and project dependencies between the nodes as edges of the graph.

[0022] S22. Assign initial feature vectors to each node. The initial feature vectors include the financial data, sales data, and production data information of the node. The initial feature vector of each node is expressed as:

[0023] v i =[f i1 ,f i2 ,…,f in ;

[0024] where v i is the initial feature vector of the i-th node, and f in represents the n-th feature of the node, and n is the total number of features.

[0025] S23. Organize the initial feature vectors of all nodes into a node feature matrix, with each row representing the feature vector of a node. The node feature matrix is:

[0026] V=[v1,v2,…,v m ;

[0027] Among them, V represents the node feature matrix, m is the total number of nodes, and v i is the preliminary feature vector of the i-th node;

[0028] S24. Construct the original adjacency matrix of the graph according to the capital flow, collaboration relationship, and project dependency information between nodes. The original adjacency matrix is expressed as A = [a ij , and a ij represents the relationship strength between node i and node j. If there is a direct association between node i and node j, then a ij is a non-zero value, otherwise it is zero.

[0029] Optionally, the S3 includes the following steps:

[0030] S31. In the operation of the graph convolutional layer, perform weighted aggregation on the node feature matrix through the adjacency matrix ;

[0031] S32. Use the graph convolutional layer for iteration to gradually extract the long-term dependency relationship between nodes and the interaction information between different departments and resources. Update the feature representation of the nodes through each convolutional operation to generate a high-order feature representation containing the deep dependency relationship between the nodes and their neighboring nodes:

[0032]

[0033] Among them, H (k) is the node feature matrix at the k-th iteration, H (k+1) is the node feature matrix at the (k + 1)-th iteration, is the feature vector of the j-th node at the k-th iteration, is the adjacency matrix, W (k) is the weight matrix at the k-th iteration, N(i) is the set of neighboring nodes of the i-th node, and α i is the weighting coefficient between node i and node j;

[0034] S33. Finally, generate the graph embedding representation H (K) of each node:

[0035]

[0036] Among them, is the feature vector of the i-th node after K iterations, and m is the total number of nodes.

[0037] Optionally, the S31 includes the following steps:

[0038] S311. The calculation method of the adjacency matrix is:

[0039]

[0040] Among them, A is the original adjacency matrix of the graph, and D is the degree matrix of the graph. D ii is the degree of node i, and D -1 / 2 is the inverse square root of the degree matrix.

[0041] Optionally, the S4 includes the following steps:

[0042] S41. Based on the graph embedding representation generated by the graph convolutional network, use the adaptive simulated annealing algorithm for budget optimization. The multi-objective optimization function for the budget optimization is:

[0043]

[0044] Among them, is the loss function representing the budget execution accuracy, is the loss function representing the cost control efficiency, is the loss function representing the resource utilization rate, is the loss function representing the minimization of budget risk. λ1, λ2, λ3, and λ4 are the weight coefficients of each objective, indicating the relative importance of each objective in the optimization;

[0045] S42. In the adaptive simulated annealing algorithm, perform global search through the temperature adjustment mechanism. The formula for temperature update is:

[0046] T k+1 = T k ·β (k) ;

[0047] Among them, T k is the temperature of the k-th iteration, and β (k) is the temperature decay factor, satisfying 0 < β (k) < 1, and β (k) has self-adaptability to balance the exploration and exploitation processes;

[0048] S43. Use the standard Metropolis criterion of the simulated annealing algorithm to calculate the probability of accepting a new solution:

[0049]

[0050] Among them, ΔE is the energy difference between the current solution and the new solution, exp(.) is the natural exponential function, T k is the current temperature, and λ5 is the adjustment factor used to control the influence of the square term of the energy difference on the probability distribution and enhance the global search ability of the algorithm;

[0051] S44. During each iteration of the simulated annealing, the update of the graph embedding representation is dynamically adjusted according to the energy difference calculation and temperature regulation mechanism of the adaptive simulated annealing algorithm, and gradually converges to the optimal solution through iterative optimization;

[0052] S45. During the entire optimization process, dynamically adjust the weight coefficients λ1, λ2, λ3, and λ4 of each objective, and adaptively update the weights of each objective according to the real-time feedback of the optimization effect and the change of the objective function to ensure that different objectives can achieve an effective balance during the multi-objective optimization process;

[0053] S46. Through the iterative optimization of the adaptive simulated annealing algorithm, output the budget allocation plan.

[0054] Optionally, S44 includes the following steps:

[0055] S441. During each iteration, the calculation formula for updating the graph embedding representation is:

[0056]

[0057] where, is the feature vector of the i-th node at the k-th iteration, is the feature vector of the j-th node at the k-th iteration, ΔE i is the energy difference of node i, T k is the temperature at the k-th iteration, α i is the learning rate coefficient, exp(.) is the natural exponential function, γ is the coefficient controlling the temperature change, w ij is the relationship weight between the i-th node and its neighbor node j, and N(i) is the set of neighborhood nodes of the i-th node.

[0058] Optionally, S5 includes the following steps:

[0059] S51. Based on the budget allocation plan optimized by the adaptive simulated annealing algorithm, monitor the actual expenditure of each department and project of the enterprise in real time, and calculate the difference between the actual expenditure and the budget;

[0060] S52. According to the real-time monitoring data, feedback the expenditure difference to the budget execution system, and adjust the budget allocation through the dynamic adjustment mechanism;

[0061] S53. Through the integrated real-time monitoring system, continuously track the actual expenditure and budget difference during the budget execution process. If the expenditure difference exceeds the set threshold, start the early warning mechanism;

[0062] S54. According to the feedback budget difference and real-time monitoring data, optimize the adjustment strategy during the budget execution process, and dynamically adjust the budget allocation strategy in combination with the specific situations of each department and project.

[0063] Optionally, S6 includes the following steps:

[0064] S61. Through the integrated visual decision support system, real-time display the cost data, budget execution status, and optimization suggestion information of each department and project of the enterprise;

[0065] S62. The optimization suggestion information is fed back to the management layer in real time through the decision support system;

[0066] S63. During the entire budget execution and adjustment process, the system supports dynamic query and multi-dimensional display.

[0067] Optionally, S7 includes the following steps:

[0068] S71. During the budget execution process, the system continuously monitors the actual execution results of each department and project, and calculates the difference between the actual expenditure and the optimized budget through the real-time data analysis and feedback mechanism;

[0069] S72. According to the difference, the system further adjusts the budget allocation strategy through the adaptive algorithm in combination with external environment changes and internal changes;

[0070] S73. The system feeds back the execution results and adjustment suggestions to the relevant management personnel in real time.

[0071] The beneficial effects of the present invention are:

[0072] (1) By combining the graph convolutional network and the adaptive simulated annealing algorithm, the present invention provides a new solution for enterprise cost optimization and budget management. Through the modeling of the complex relationships among various departments, projects, and resources within the enterprise by the graph convolutional network, it can comprehensively capture the resource dependencies and collaboration relationships among each node, making the budget allocation more accurate and reasonable. Compared with traditional single budget optimization methods, the present invention can more effectively optimize the budget allocation in multi-dimensional cost control and resource allocation, avoiding problems such as resource waste and uneven capital allocation caused by traditional methods ignoring internal complex relationships.

[0073] (2) By constructing a multi-objective optimization function and comprehensively considering multiple factors such as budget accuracy, cost control efficiency, resource utilization rate, and budget execution risk, the present invention achieves a balance among various optimization objectives. During the budget execution process, the system can monitor the difference between the actual expenditure and the budget in real time and perform dynamic adjustment based on the adaptive simulated annealing algorithm. Through the temperature attenuation and energy difference calculation mechanisms in the simulated annealing algorithm, it ensures the global search ability during the budget optimization process and avoids the trap of local optimal solutions, thereby achieving the optimization of budget allocation.

[0074] (3) In the process of budget optimization, the present invention introduces a real-time feedback mechanism. Based on the graph embedding representation generated by the graph convolutional network and combined with the adaptive simulated annealing algorithm, it ensures the real-time performance and flexibility of budget execution. By monitoring the deviation metrics of budget execution and dynamically adjusting budget allocation, the system can quickly respond and adjust the budget plan according to factors such as changes in the internal structure of the enterprise and changes in the market environment, ensuring that the financial management of the enterprise is always in an optimal state in a complex and dynamic environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the accompanying drawings:

[0076] Figure 1 is a flowchart of a method for enterprise cost optimization and budget management based on big data analysis proposed by the present invention;

[0077] Figure 2 is a flowchart of generating node graph embedding representation in a method for enterprise cost optimization and budget management based on big data analysis proposed by the present invention;

[0078] Figure 3 is a flowchart of updating the graph embedding representation through the adaptive simulated annealing algorithm in a method for enterprise cost optimization and budget management based on big data analysis proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0079] Now, the present invention will be further described in detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0080] Refer to Figures 1 - 3 , a method for enterprise cost optimization and budget management based on big data analysis, includes the following steps:

[0081] S1. Automatically collect real-time data of each department through various management systems of the enterprise and perform preprocessing;

[0082] S2. Construct a graph structure model of enterprise costs and budgets, assign preliminary feature vectors to each node, and construct a node feature matrix based on the preliminary feature vectors of each node;

[0083] S3. Use the graph convolutional network to process the node feature matrix through multi-layer convolutional operations, capture the long-term dependence relationships between nodes and the resource allocation information between departments, further fuse the interactions between nodes, and generate the graph embedding representation of the nodes;

[0084] S4. Based on the graph embedding representation generated by the graph convolutional network, use the adaptive simulated annealing algorithm to optimize the budget, define the multi-objective optimization function for budget optimization, and perform global search through the temperature adjustment mechanism in simulated annealing, dynamically adjusting the temperature drop speed and step size, and output the budget allocation plan;

[0085] S5. Based on the budget allocation plan, automatically adjust the budgets and resource allocations of each department and project in the enterprise, and conduct real-time monitoring of the budget execution process, and use the feedback mechanism to dynamically adjust the budget;

[0086] S6. Through the integrated visual decision support system, real-time display the cost data, budget execution status, and optimization suggestion information of the enterprise;

[0087] S7. During the budget execution process, the system continuously optimizes the cost management and budget control of the enterprise by continuously tracking the actual execution results, external environment changes, and internal adjustments.

[0088] By automatically collecting the data of various management systems of the enterprise and performing standardized processing, the present invention can ensure the data consistency and integrity of each department in the enterprise. Compared with the traditional method of manually collecting and processing data, the automated data collection and preprocessing significantly improve the data accuracy and processing efficiency. The standardized data provides a more reliable basis for subsequent analysis and decision-making, enabling the enterprise to accurately grasp the operation status of each department and project, and thus providing accurate input data for budget optimization and cost control.

[0089] In this embodiment, S1 includes the following steps:

[0090] S11. Automatically collect the real-time data of each department from various management systems of the enterprise;

[0091] S12. Perform preprocessing on the collected data, and the preprocessing steps include removing missing data, filling in missing values, and removing outliers;

[0092] S13. Use data cleaning technology to detect and correct outliers in the data, and perform standardized processing.

[0093] By constructing a graph structure model of the enterprise cost and budget, and using the graph convolutional network to process the nodes, the present invention can comprehensively and accurately analyze the complex relationships between each department, resource, and project within the enterprise. The graph convolutional network can effectively capture the dependency relationships between each node, thus providing a more real and profound perspective for subsequent budget optimization. Compared with the traditional linear budget model, the present invention can capture the long-distance node dependencies based on the graph structure, improving the accuracy and comprehensiveness of budget analysis, and ensuring the rationality and efficiency of resource allocation between departments.

[0094] In this embodiment, S2 includes the following steps:

[0095] S21. Construct a graph structure model of enterprise costs and budgets, taking each department, resource, and project as nodes of the graph, and taking the capital flow, collaboration relationship, and project dependency between the nodes as the edges of the graph;

[0096] S22. Assign preliminary feature vectors to each node. The preliminary feature vectors include the financial data, sales data, and production data information of the node. The preliminary feature vector of each node is expressed as:

[0097] v i =[f i1 ,f i2 ,…,f in ;

[0098] where v i is the preliminary feature vector of the i-th node, and f in represents the n-th feature of this node, and n is the total number of features;

[0099] S23. Organize the preliminary feature vectors of all nodes into a node feature matrix, where each row represents the feature vector of a node. The node feature matrix is:

[0100] V=[v1,v2,…,v m ;

[0101] where V represents the node feature matrix, m is the total number of nodes, and v i is the preliminary feature vector of the i-th node;

[0102] S24. According to the capital flow, collaboration relationship, and project dependency information between the nodes, construct the original adjacency matrix of the graph. The original adjacency matrix is expressed as A=[a ij , and a ij represents the relationship strength between node i and node j. If there is a direct association between node i and node j, then a ij is a non-zero value, otherwise it is zero.

[0103] In the present invention, the node features are processed through multi-layer convolution operations of the graph convolutional network, so that the features of each department, resource, and project can be fully integrated and optimized in a multi-layer network. This process enables the feature vector of each node to include not only its own attribute information but also the information of neighboring nodes, thereby obtaining a more comprehensive feature representation. This multi-layer processing improves the model's ability to capture complex dependency relationships, ensures that the collaborative effects of various resources of the enterprise can be fully considered during the budget optimization process, and avoids the defect of insufficient processing of resource relationships in traditional methods.

[0104] In this embodiment, S3 includes the following steps:

[0105] S31. In the operation of the graph convolutional layer, the node feature matrix is weighted and aggregated through the adjacency matrix ;

[0106] S32. Use the graph convolutional layer for iteration to gradually extract the long-term dependencies between nodes and the interaction information across departments and resources. Update the feature representation of nodes through each convolutional operation to generate a high-order feature representation containing the deep dependencies between nodes and their neighboring nodes:

[0107]

[0108] where, H (k) is the node feature matrix at the k-th iteration, H (k+1) is the node feature matrix at the (k + 1)-th iteration, is the feature vector of the j-th node at the k-th iteration, is the adjacency matrix, W (k) is the weight matrix at the k-th iteration, N(i) is the set of neighboring nodes of the i-th node, and α i is the weighting coefficient between node i and node j;

[0109] S33. Finally, generate the graph embedding representation H (K) for each node:

[0110]

[0111] where, is the feature vector of the i-th node after K iterations, and m is the total number of nodes.

[0112] In this embodiment, S31 includes the following steps:

[0113] S311. The calculation method of the adjacency matrix is as follows:

[0114]

[0115] where, A is the original adjacency matrix of the graph, D is the degree matrix of the graph, D ii = Σ j a ij , D ii is the degree of node i, and D -1 / 2 is the inverse square root of the degree matrix.

[0116] The present invention combines the graph embedding representation generated by the graph convolutional network with the adaptive simulated annealing algorithm for budget optimization, which can simultaneously solve the optimization problems of multiple objectives such as budget execution accuracy, cost control efficiency, and resource utilization rate. Through the temperature adjustment mechanism and global search ability of the simulated annealing algorithm, the present invention avoids the defect that traditional optimization algorithms are prone to falling into local optimal solutions. The way of adaptively adjusting the temperature and search strategy enables the optimization process to flexibly respond to different environments and demand changes, ensuring that global optimal budget allocation can always be achieved in a complex enterprise budget environment.

[0117] In this embodiment, S4 includes the following steps:

[0118] S41. Based on the graph embedding representation generated by the graph convolutional network, use the adaptive simulated annealing algorithm for budget optimization. The multi-objective optimization function for the budget optimization is:

[0119]

[0120] where, is the loss function representing the budget execution accuracy, is the loss function representing the cost control efficiency, is the loss function representing the resource utilization rate, is the loss function representing the budget risk minimization. λ1, λ2, λ3, and λ4 are the weight coefficients of each objective, representing the relative importance of each objective in the optimization;

[0121] S42. In the adaptive simulated annealing algorithm, perform global search through the temperature adjustment mechanism. The formula for temperature update is:

[0122] T k+1 = T k ·β (k) ; [[ID=3�]]

[0123] where, T k is the temperature of the k-th iteration, and β (k) is the temperature decay factor, satisfying 0 < β (k) < 1, and β (k) has self-adaptability to balance the exploration and exploitation processes;

[0124] S43. Use the standard Metropolis criterion of the simulated annealing algorithm to calculate the probability of accepting a new solution:

[0125]

[0126] where, ΔE is the energy difference between the current solution and the new solution, exp(.) is the natural exponential function, and T kis the current temperature, and λ5 is an adjustment factor used to control the influence of the square term of the energy difference on the probability distribution, enhancing the global search ability of the algorithm;

[0127] S44. During each iteration of simulated annealing, the update of the graph embedding representation is dynamically adjusted according to the energy difference calculation and temperature adjustment mechanism of the adaptive simulated annealing algorithm, and gradually converges to the optimal solution through iterative optimization;

[0128] S45. During the entire optimization process, dynamically adjust the weight coefficients λ1, λ2, λ3, and λ4 of each objective, and adaptively update the weights of each objective according to the real-time feedback of the optimization effect and the change of the objective function to ensure that different objectives can achieve an effective balance during the multi-objective optimization process;

[0129] S46. Through the iterative optimization of the adaptive simulated annealing algorithm, output the budget allocation plan.

[0130] In this embodiment, S44 includes the following steps:

[0131] S441. During each iteration, the calculation formula for updating the graph embedding representation is:

[0132]

[0133] where, is the feature vector of the i-th node at the k-th iteration, is the feature vector of the j-th node at the k-th iteration, ΔE i is the energy difference of node i, T k is the temperature at the k-th iteration, α i is the learning rate coefficient, exp(.) is the natural exponential function, γ is the coefficient controlling the temperature change, w ij is the relationship weight between the i-th node and the neighbor node j, and N(i) is the set of neighborhood nodes of the i-th node.

[0134] Through the real-time monitoring and feedback mechanism, the present invention can dynamically adjust the budget allocation according to the difference between the actual expenditure and the budget. This real-time adjustment mechanism enables enterprises to flexibly respond to emergencies and internal changes during the budget execution process, avoiding the deficiencies of lack of flexibility and adaptability in traditional budget management methods. Through the dynamic adjustment based on the budget difference, enterprises can achieve the optimal allocation of resources during the budget execution process, minimize resource waste to the greatest extent, and improve the accuracy and efficiency of budget execution.

[0135] In this embodiment, S5 includes the following steps:

[0136] S51. A budget allocation plan optimized based on the adaptive simulated annealing algorithm is used to monitor the actual expenditures of each department and project in the enterprise in real time, and calculate the difference between the actual expenditure and the budget.

[0137] S52. According to the real-time monitoring data, the expenditure difference is fed back to the budget execution system, and the budget allocation is adjusted through a dynamic adjustment mechanism.

[0138] S53. Through the integrated real-time monitoring system, continuously track the actual expenditure and budget difference during the budget execution process. If the expenditure difference exceeds the set threshold, the early warning mechanism is activated.

[0139] S54. According to the feedback budget difference and real-time monitoring data, optimize the adjustment strategy during the budget execution process, and dynamically adjust the budget allocation strategy in combination with the specific situations of each department and project.

[0140] Through the integrated visual decision support system of the present invention, it is possible to display in real time the cost data, budget execution status, and optimization suggestions of each department and project in the enterprise. This function enables the management to intuitively understand the dynamic situation of budget execution and make timely adjustments and decisions based on real-time data. Compared with the traditional budget management based on static reports, the decision support system of the present invention provides a more flexible and efficient decision-making basis, helps to improve the overall management efficiency of the enterprise, and ensures that every decision in the budget optimization process is based on the latest real-time data.

[0141] In this embodiment, S6 includes the following steps:

[0142] S61. Through the integrated visual decision support system, display in real time the cost data, budget execution status, and optimization suggestion information of each department and project in the enterprise.

[0143] S62. The optimization suggestion information is fed back to the management in real time through the decision support system.

[0144] S63. During the entire budget execution and adjustment process, the system supports dynamic query and multi-dimensional display.

[0145] The real-time monitoring and feedback mechanism of the present invention, combined with the calculation of budget deviation measurement, ensures the continuous optimization of the enterprise during the budget execution process. Through the dynamic feedback and adjustment of budget differences, the system can timely discover and solve problems in budget execution, avoiding the neglect of budget execution deviations in traditional methods. This real-time adaptability and flexibility enable the enterprise to quickly respond to and adjust the budget plan in a changing market environment, ensuring that budget execution always meets the actual needs and goals of the enterprise, thereby maximizing the effectiveness of financial management and the competitiveness of the enterprise.

[0146] In this embodiment, S7 includes the following steps:

[0147] S71. During the budget execution process, the system continuously monitors the actual execution results of each department and project, and through the real-time data analysis and feedback mechanism, calculates the difference between the actual expenditure and the optimized budget;

[0148] S72. According to the difference, the system further adjusts the budget allocation strategy through an adaptive algorithm combined with external environmental changes and internal changes;

[0149] S73. The system feeds back the execution results and adjustment suggestions to the relevant management personnel in real time.

[0150] Example:

[0151] In a large manufacturing enterprise, the enterprise has multiple production bases and multiple departments globally, including the finance department, sales department, procurement department, production department, etc. The types of resource allocation involved are numerous, and the budget execution scope is wide. In the past budget preparation and execution process, the enterprise used traditional budget methods based on historical data. This method is not only difficult to handle the complex dependency relationships between different departments, but also unable to respond to changes in the market and the enterprise's internal environment in real time, resulting in frequent serious deviations during the budget execution process. For example, the budget of the sales department often exceeds its actual expenditure, while the budget of the production department is seriously insufficient, causing production line stagnation and sales order delays.

[0152] To better address these challenges, the enterprise decided to adopt the method provided by the present invention based on the graph convolutional network and the adaptive simulated annealing algorithm. In the actual application process, first, each department of the enterprise is modeled. Each department and resource are used as nodes in the graph, and the capital flow, collaboration relationships, and project dependencies between departments are used as the edges of the graph. The graph convolutional network is used to extract the complex relationships between these departments and generate preliminary feature vectors for each department. These feature vectors are used as input data for budget optimization through the adaptive simulated annealing algorithm.

[0153] First, the enterprise automatically collects real-time data of each department through an integrated management system, including financial data, sales data, inventory data, production data, and employee salary data, etc. After these data are standardized, they are input into the graph structure model. Based on these standardized data, a graph structure model within the enterprise is constructed, and the graph convolutional network is applied to extract the relationships between nodes. For example, there are complex dependency relationships between the expenditures of the finance department and the resource procurement of the procurement department, the raw material use of the production department, etc. The graph convolutional network can capture the mutual dependencies between each department and other departments by performing multi-level processing on the node relationships between these departments.

[0154] Next, based on the graph embedding representation and node features generated by the graph convolutional network, the adaptive simulated annealing algorithm is used for budget optimization. The simulated annealing algorithm adjusts the budget allocation of each department step by step to achieve the optimization goal. During the optimization process, the algorithm avoids the situation of being easily trapped in local optimal solutions in traditional budget optimization methods through the temperature decay mechanism and energy difference calculation, ensuring the global optimality of the budget configuration. At the same time, the optimized multi-objective function includes multiple objectives such as budget execution accuracy, cost control efficiency, resource utilization rate, and risk minimization, making the budget allocation not only more accurate numerically but also able to comprehensively consider the resource synergy among various departments and projects.

[0155] In the actual application process, the enterprise collected and analyzed the budget execution situation in the past three months and found obvious problems with traditional budget methods. During this period, the budget overrun of the sales department was approximately 1.5 million yuan, while the budget gap of the production department reached 1.2 million yuan, resulting in serious production stagnation and order delays. By implementing the method of the present invention, the enterprise optimized the budget within three months, adjusted the budget allocation of each department, and accurately identified the dependency relationships among departments through the graph convolutional network, and further applied the adaptive simulated annealing algorithm for real-time optimization.

[0156] Within three months after optimization, the enterprise achieved remarkable results in budget execution. The budget overrun rate of the sales department decreased from 15% to 4%, and the budget gap of the production department decreased from 12% to 2%. By adjusting the resource allocation among departments, the enterprise successfully avoided production line stagnation and reduced operation interruptions caused by budget shortages. In addition, the resource allocation of the procurement department was reasonably optimized, resulting in an 8% increase in inventory turnover rate and a 5% increase in overall operation efficiency.

[0157] The following is a specific data comparison table before and after implementation:

[0158] Table 1 Comparison Table of Enterprise Budget Optimization Before and After

[0159]

[0160] Through the comprehensive application of the graph convolutional network and the adaptive simulated annealing algorithm, the enterprise not only successfully optimized the budget allocation but also ensured the reasonable allocation of various resources and the accurate execution of the budget through real-time monitoring and dynamic adjustment mechanisms. The optimized results not only solved the problem of budget deviation but also improved the overall operation efficiency, bringing more sustainable financial management results to the enterprise.

[0161] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

Claims

1. An enterprise cost optimization and budget management method based on big data analysis, characterized in that, It includes the following steps: S1. Automatically collect real-time data from various management systems of the enterprise and perform preprocessing; S2. Construct a graph structure model of the enterprise's cost and budget, assign initial feature vectors to each node, and construct a node feature matrix based on the initial feature vectors of each node; S3. Use a graph convolutional network to process the node feature matrix through multi-layer convolutional operations, capture the long-term dependencies between nodes and the resource allocation information across departments, further integrate the interactions between nodes, and generate a graph embedding representation of the nodes; S4. Based on the graph embedding representation generated by the graph convolutional network, use an adaptive simulated annealing algorithm for budget optimization, define a multi-objective optimization function for budget optimization, and perform a global search through the temperature adjustment mechanism in simulated annealing, dynamically adjust the temperature drop speed and step size, and output a budget allocation plan; S5. Based on the budget allocation plan, automatically adjust the budgets and resource allocations of each department and project of the enterprise, and monitor the budget execution process in real time, and use the feedback mechanism to dynamically adjust the budget; S6. Through an integrated visual decision support system, display the enterprise's cost data, budget execution status, and optimization suggestion information in real time; S7. During the budget execution process, the system continuously optimizes the enterprise's cost management and budget control by continuously tracking the actual execution results, external environment changes, and internal adjustments.

2. The enterprise cost optimization and budget management method based on big data analysis according to claim 1, characterized in that The S1 includes the following steps: S11. Automatically collect real-time data from various management systems of the enterprise; S12. Perform preprocessing on the collected data, and the preprocessing steps include removing missing data, filling in missing values, and removing outliers; S13. Use data cleaning technology to detect and correct outliers in the data and perform standardization processing.

3. The enterprise cost optimization and budget management method based on big data analysis according to claim 1, characterized in that, The S2 includes the following steps: S21. Construct a graph structure model of the enterprise's cost and budget, use each department, resource, and project as a node of the graph, and use the capital flow, collaboration relationship, and project dependency between nodes as the edges of the graph; S22. Assign a preliminary feature vector v to each node i , where the preliminary feature vector includes the financial data, sales data, and production data information of the node; S23. Organize the initial feature vectors of all nodes into a node feature matrix V, and each row represents the feature vector of a node; S24. Construct an original adjacency matrix of a graph based on the capital flow, collaboration relationship, and project dependency information between nodes. The original adjacency matrix is represented as A = [a ij , where a ij represents the relationship strength between node i and node j. If there is a direct association between node i and node j, then a ij is a non-zero value; otherwise, it is zero.

4. A method for enterprise cost optimization and budget management based on big data analysis according to claim 1, characterized in that, The S3 includes the following steps: S31. In the operation of the graph convolutional layer, weighted aggregation of the node feature matrix is performed through the adjacency matrix ; S32. Use graph convolutional layers for iteration, gradually extract the long-term dependencies between nodes and the interaction information across departments and resources, update the feature representation of nodes through each convolutional operation, and generate a high-order feature representation containing the deep dependencies between nodes and their neighboring nodes; Among them, H (k) is the node feature matrix at the k-th iteration, and H (k+1) is the node feature matrix at the (k + 1)-th iteration. is the feature vector of the j-th node at the k-th iteration, is the adjacency matrix, and W (k) is the weight matrix at the k-th iteration. N(i) is the set of neighbor nodes of the i-th node, and α i is the weighting coefficient between node i and node j. S33. Finally, generate the graph embedding representation H of each node (K) .

5. A method for enterprise cost optimization and budget management based on big data analysis according to claim 4, characterized in that, The S31 includes the following steps: S311. The adjacency matrix is calculated as follows: Among them, A is the original adjacency matrix of the graph, D is the degree matrix of the graph, D ii = ∑ j a ij , D ii is the degree of node i, D -1 / 2 is the inverse square root of the degree matrix.

6. A method for enterprise cost optimization and budget management based on big data analysis according to claim 1, characterized in that, The S4 includes the following steps: S41. Based on the graph embedding representation generated by the graph convolutional network, use an adaptive simulated annealing algorithm for budget optimization, and the multi-objective optimization function for budget optimization is: Among them, represents the loss function of budget execution accuracy, represents the loss function of cost control efficiency, represents the loss function of resource utilization rate, represents the loss function of minimizing budget risk, and λ1, λ2, λ3, and λ4 are the weight coefficients of each objective, indicating the relative importance of each objective in optimization; S42. In the adaptive simulated annealing algorithm, perform a global search through the temperature adjustment mechanism, and the formula for temperature update is: T k+1 = T k ·β (k) ; where, T k is the temperature of the k-th iteration, and β (k) is the temperature decay factor, satisfying 0 < β (k) < 1; S43. Calculate the probability P(ΔE) of accepting a new solution using the standard Metropolis criterion of the simulated annealing algorithm; S44. In each iteration process of simulated annealing, the update of the graph embedding representation is dynamically adjusted according to the energy difference calculation and temperature adjustment mechanism of the adaptive simulated annealing algorithm, and gradually converges to the optimal solution through iterative optimization; S45. During the entire optimization process, dynamically adjust the weight coefficients λ1, λ2, λ3, and λ4 of each objective, and adaptively update the weights of each objective according to the optimized effect feedback in real time and the changes in the objective function. S46. Through the iterative optimization of the adaptive simulated annealing algorithm, output the budget allocation plan.

7. A method for enterprise cost optimization and budget management based on big data analysis according to claim 6, characterized in that, The S44 includes the following steps: S441. During each iteration process, the calculation formula for updating the graph embedding representation is: Among them, is the feature vector of the $i$-th node at the $k$-th iteration, is the feature vector of the $j$-th node at the $k$-th iteration, and $\Delta E$ i is the energy difference of node $i$, $T$ k is the temperature at the $k$-th iteration, $\alpha$ i is the learning rate coefficient, $\exp(.)$ is the natural exponential function, $\gamma$ is the coefficient controlling the temperature change, and $w$ ij is the relationship weight between the $i$-th node and the neighbor node $j$, and $N(i)$ is the set of neighborhood nodes of the $i$-th node.

8. A method for enterprise cost optimization and budget management based on big data analysis according to claim 1, characterized in that, The S5 includes the following steps: S51. Based on the budget allocation plan optimized by the adaptive simulated annealing algorithm, monitor the actual expenditure of each department and project of the enterprise in real time, and calculate the difference between the actual expenditure and the budget. S52. According to the real-time monitoring data, feedback the expenditure difference to the budget execution system, and adjust the budget allocation through the dynamic adjustment mechanism. S53. Through the integrated real-time monitoring system, continuously track the actual expenditure and budget difference during the budget execution process. If the expenditure difference exceeds the set threshold, start the warning mechanism. S54. According to the feedback budget difference and real-time monitoring data, optimize the adjustment strategy during the budget execution process, and dynamically adjust the budget allocation strategy in combination with the specific situations of each department and project.

9. A method for enterprise cost optimization and budget management based on big data analysis according to claim 1, characterized in that The S6 includes the following steps: S61. Through the integrated visual decision support system, display the cost data, budget execution situation, and optimization suggestion information of each department and project of the enterprise in real time. S62. The optimization suggestion information is fed back to the management layer in real time through the decision support system. S63. During the entire budget execution and adjustment process, the system supports dynamic query and multi-dimensional display.

10. A method for enterprise cost optimization and budget management based on big data analysis according to claim 1, characterized in that, The S7 includes the following steps: S71. During the budget execution process, the system continuously monitors the actual execution results of each department and project, and calculates the difference between the actual expenditure and the optimized budget through the real-time data analysis and feedback mechanism. S72. According to the difference, the system further adjusts the budget allocation strategy through the adaptive algorithm in combination with the external environment changes and internal changes. S73. The system feeds back the execution results and adjustment suggestions to the relevant management personnel in real time.

Images