Enterprise investment scale intelligent optimization system based on big data analysis

Abstract

The invention relates to the technical field of big data analysis and intelligent decision, and discloses an enterprise investment scale intelligent optimization system based on big data analysis. The system comprises the steps of constructing an initial investment scene model, injecting real-time market disturbance variables, simulating the cascade effect of the real-time market disturbance variables on an investment conduction path, and generating a dynamic market situation map to identify pressure nodes. And analyzing the situation map to generate a resource reconfiguration strategy, simulating a diffusion process of an abnormal state under the strategy based on a historical abnormal case library, and outputting a risk dip-dyeing early warning report defining a diffusion boundary and a penetration depth. And according to report reverse calibration strategy parameters, an optimization scheme is locked through iterative convergence. According to the scheme, chain reaction of market disturbance and networked diffusion of risks can be dynamically simulated, and the perspectiveness of investment decision and the risk prevention and control accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of big data analysis and intelligent decision-making technology, specifically to an intelligent optimization system for enterprise investment scale based on big data analysis. Background Technology

[0002] Currently, when making investment decisions, companies generally rely on statistical analysis models based on historical data or expert systems with pre-defined rules. These existing technologies typically assess the static impact of specific variables on investment returns through methods such as regression analysis, sensitivity testing, or multi-objective programming, and issue early warnings based on risk thresholds or probability models. The operational basis of such methods is to treat the market environment and internal resources of the enterprise as a relatively stable or linearly related set of factors.

[0003] These conventional technical solutions have shortcomings. They struggle to effectively characterize the chain reactions triggered by real-time external market disturbances within a company's complex investment network. Traditional models cannot simulate how an initial disturbance propagates and amplifies along multi-dimensional paths involving business, capital, and information, ultimately altering the structure and state of the overall investment environment, leading to decision-making lagging behind the dynamic market. At the risk assessment level, existing methods mostly output static risk levels or probabilities of occurrence, lacking the ability to dynamically extrapolate the specific diffusion paths, scope of impact, and depth of evolution once a risk event occurs. This limits risk warnings to the level of "whether it is possible," failing to answer the crucial questions of "how it will spread" and "where it will affect," resulting in resource reallocation strategies that may overlook implicit transmission channels and create blind spots in prevention and control.

[0004] Existing technical solutions are insufficient in simulating the cascading effects of real-time disturbances and dynamically extrapolating the spread of risk states when dealing with complex investment systems that are highly interconnected and dynamic. This limits the foresight, adaptability, and robustness of investment optimization solutions. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent optimization system for enterprise investment scale based on big data analysis, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides an intelligent optimization system for enterprise investment scale based on big data analysis, the system comprising:

[0007] The investment scenario construction module is used to receive multi-source enterprise operation data, define multi-dimensional influencing factors based on the received data, and construct an initial investment scenario model by analyzing the constraints and coupling relationships between the multi-dimensional influencing factors.

[0008] The dynamic feedback correction module is used to import the initial investment scenario model and inject real-time market disturbance variables into the calculation framework of the initial investment scenario model. By simulating the cascading effect of the real-time market disturbance variables on the investment transmission path, a dynamic market situation diagram containing fluctuation characteristics and pressure nodes is generated.

[0009] The resource topology adaptation module is used to parse the dynamic market situation map, identify the pressure nodes and resource overload areas, and re-plan the path and priority of resource flow based on the enterprise resource inventory map to form a resource reallocation strategy.

[0010] The abnormal state infiltration simulation module is used to load the resource reallocation strategy and simulate the diffusion process of abnormal states under the execution conditions of the strategy based on the historical abnormal case library. By defining the diffusion boundary and penetration depth, it outputs a risk infiltration early warning report.

[0011] The investment efficiency convergence module is used to reverse-calibrate the key parameters in the resource reallocation strategy based on the risk contamination early warning report, and make the investment output characteristics approach the preset convergence target in multiple rounds of iterative calibration, and finally lock the investment scale optimization scheme.

[0012] Preferably, the steps for constructing the initial investment scenario model are as follows:

[0013] Acquire multi-source enterprise operation data from enterprise operation databases, market intelligence sources, and internal monitoring systems, and perform time-series alignment and dimension normalization on the multi-source enterprise operation data to obtain a standardized enterprise operation data stream.

[0014] In the standardized enterprise operation data flow, association rules between data fields are preset according to the investment decision logic, and a set of indicators representing market capacity, supply chain resilience, technology iteration rate and policy compliance cost are extracted from the data flow based on the association rules.

[0015] Each indicator in the indicator set is defined as an independent influencing factor, and based on the data mining results of the historical decision case library, the constraint strength coefficient and coupling weight matrix between the influencing factors are quantified.

[0016] Based on the constraint strength coefficient and coupling weight matrix, the influencing factors are embedded into a hierarchical computing framework. Through signal transmission and aggregation operations at each layer within the framework, an initial investment scenario model reflecting the interaction relationship of all factors of corporate investment is constructed.

[0017] Preferably, the steps for generating the dynamic market situation map are as follows:

[0018] The computational framework of the initial investment scenario model is used as the basic simulation environment to continuously capture real-time market disturbance variables, including sudden index, raw material price anomaly rate and instantaneous exchange rate fluctuation, from external data interfaces.

[0019] The real-time market disturbance variables are injected into the simulation base environment in the form of pulse signals, triggering the recalculation of the state of each influencing factor in the simulation base environment, and tracking the transmission process of state changes to subsequent nodes along the preset investment transmission path;

[0020] Record the cascading effect data generated during the transmission process. The cascading effect data includes the state offset of each node, path conduction delay, and energy attenuation coefficient. Based on the cascading effect data, mark the sensitive areas of system stability decline as pressure nodes.

[0021] By combining the spatial distribution of the pressure nodes, the time series of state offsets, and the evolution trend of sensitive areas, a dynamic market situation map is generated that can visualize the characteristics of market fluctuations and structural pressure nodes.

[0022] Preferably, the steps for forming the resource reallocation strategy are as follows:

[0023] The dynamic market situation map is analyzed to identify the attribute type and overload degree of all pressure nodes in the map, and the resource overload area formed by resource competition or path congestion is delineated.

[0024] Retrieve a pre-constructed enterprise resource inventory map, which uses a network structure to mark the current location, quantity, and flow attributes of funds, production capacity, human resources, and technology reserves;

[0025] With the primary goal of alleviating the overload of the pressure nodes and clearing the overloaded areas of resources, path optimization calculations are performed on the enterprise resource inventory map to plan one or more backup flow paths for each type of resource.

[0026] Based on the estimated throughput efficiency, required activation cost, and contribution to the overall system stability of the backup flow paths, execution priorities are assigned to all planned paths, and all path planning and priority information are integrated to form a complete resource reallocation strategy.

[0027] Preferably, the output steps of the risk contamination early warning report are as follows:

[0028] The resource reconfiguration strategy is loaded as the new configuration baseline for system operation, and the historical anomaly case library containing historical anomaly events and their development trajectories is accessed.

[0029] Based on the anomaly propagation patterns abstracted from the historical anomaly case library, under the operating state defined by the new configuration baseline, simulate how an anomaly state that starts at a local node spreads along the resource network and information network.

[0030] During the simulation, the spread rate and impact range of the abnormal state are calculated in real time, the maximum possible spread boundary is defined, and the depth of its penetration into the core investment functional modules is assessed.

[0031] The diffusion boundary, penetration depth, key springboard nodes on the diffusion path, and estimated time to reach the critical risk threshold are integrated and packaged into a risk infiltration early warning report.

[0032] Preferably, the specific steps for locking in the investment scale optimization scheme are as follows:

[0033] Receive the risk infiltration early warning report and extract key risk parameters from it. The key risk parameters include the identifier of the module with the highest penetration depth and the shortest time to approach the critical threshold.

[0034] The key risk parameters are used as feedback signals and input to the parameter regulator of the resource reallocation strategy to perform reverse calibration on the resource allocation ratio and path response threshold of the corresponding module in the strategy.

[0035] After completing a round of reverse calibration, the calibrated resource reallocation strategy is rerun in the simulation environment, and its investment output characteristics, including return stability, risk buffer capacity, and scale growth rate, are evaluated.

[0036] The re-evaluated investment output characteristics are compared with the preset convergence target. If the convergence criterion is not met, a new round of calibration, operation and evaluation iteration is started until the investment output characteristics meet the preset convergence target.

[0037] The final investment scale optimization scheme is determined by locking all system configuration parameters and resource allocation schemes that ultimately meet the convergence target.

[0038] Preferably, the step of acquiring the standardized enterprise operation data stream further includes a pre-calibration process for the weights of influencing factors, specifically:

[0039] After completing the time alignment and dimensional normalization, the standardized enterprise operation data stream is scanned using a feature analysis model based on the attention mechanism.

[0040] The feature analysis model automatically identifies data segments with high volatility and predictive value in historical investment cycles, and assigns high initial weight coefficients to the underlying influencing factors corresponding to these data segments.

[0041] The standardized enterprise operation data stream carrying the initial weight coefficients will be output for weighted calculation when extracting the indicator set later.

[0042] Preferably, the step of quantifying the constraint strength coefficients and coupling weight matrices between the influencing factors specifically involves:

[0043] A large number of successful and unsuccessful decision examples are extracted from the historical decision case library. Each example contains a snapshot of the state of the indicator set at the time of the decision and the subsequent investment results.

[0044] The graph neural network is used to analyze the relationship pattern of the coordinated change of any two index states in the state snapshot, and the stable positive coordination or negative constraint relationship is quantified into constraint strength coefficient.

[0045] Simultaneously, the combined impact of multiple indicators changing together on investment results is analyzed, and this nonlinear impact of multi-factor interaction is quantified into a coupling weight matrix.

[0046] Preferably, the steps for constructing and updating the enterprise resource inventory map are as follows:

[0047] Establish a real-time data connection with the enterprise resource management system to continuously obtain updated information on fund account flow, equipment operation logs, project manpower deployment, and patent technology ledger;

[0048] Using the enterprise's organizational structure and business processes as a blueprint, the acquired updated information is mapped onto a virtual resource topology network, where network nodes represent resource storage units or consumption units, and edges represent resource scheduling or transfer relationships.

[0049] Periodically perform map snapshots to record the location, quantity, and status of resources on each time slice, forming a time-series version of the enterprise resource inventory map.

[0050] Preferably, the steps for defining and dynamically adjusting the convergence target are as follows:

[0051] During the system initialization phase, the management interface inputs the basic expected values ​​of investment return rate, risk level and capital turnover rate, and these basic expected values ​​are converted into quantifiable initial convergence targets.

[0052] During operation, the system continuously monitors the external macroeconomic prosperity index and the industry average performance level.

[0053] When external monitoring data continuously deviates from the preset benchmark range, the system automatically adjusts the parameters in the initial convergence target according to the built-in fitness function to generate a dynamic convergence target that adapts to the new environment, which is used to guide the subsequent iterative calibration process.

[0054] Compared with the prior art, the beneficial effects of the present invention are:

[0055] By introducing a technique that simulates the cascading effects of real-time market disturbance variables along the investment transmission path within a computational framework, the system surpasses traditional single-point impact assessments. A dynamic network reflecting the nonlinear and temporal correlations among multidimensional factors is constructed, transforming discrete disturbance events into their propagation trajectories and energy evolution processes throughout the investment system. The resulting dynamic market situation map includes volatility characteristics and critically identifies pressure nodes—key vulnerable points where disturbance energy converges or paths intersect. This allows decision-makers to intuitively understand the diffusion logic of disturbances from local to global perspectives, identify systemic weaknesses in advance, and thus implement more targeted interventions, rather than merely addressing the initial disturbance or adjusting based on uniform assumptions.

[0056] Based on a historical database of unusual cases, this technology simulates the dynamic propagation of risk events under specific resource allocation strategies, transforming the perspective of risk assessment. Risk events are abstracted as propagating agents that can move within the enterprise's resource and relationship networks, and their behavioral patterns are learned through case studies. In strategy simulation, the system can dynamically depict the potential impact range of risk events and the degree of impact in different regions. The output early warning reports thus possess spatial and intensity attributes, enabling resource strategies not only to assess static risks but also to pre-test whether they might form new risk transmission paths or exacerbate vulnerabilities in specific areas, achieving closed-loop verification of the strategy's own risk effects. Attached Figure Description

[0057] Figure 1 This is a timeline diagram of the intelligent optimization system for enterprise investment scale based on big data analysis described in this invention. Figure 2 A flowchart for building the initial investment scenario model; Figure 3 A flowchart generated for a dynamic market situation diagram; Figure 4 A bar chart analyzing the risk of abnormal diffusion in the enterprise's resource flow path; Figure 5 A grouped bar chart showing how corporate investment convergence targets dynamically adjust with the external economic environment. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Please see Figure 1 This invention provides an intelligent optimization system for enterprise investment scale based on big data analysis. The system includes: an investment scenario construction module responsible for accessing multi-source enterprise operational data, defining multi-dimensional influencing factors, and analyzing their constraints and coupling relationships to construct an initial investment scenario model; a dynamic feedback correction module using this initial model as a computational framework, injecting real-time market disturbance variables to simulate their cascading effects on the investment transmission path, generating a dynamic market situation map containing fluctuation characteristics and pressure nodes; a resource topology adaptation module analyzing the dynamic market situation map, identifying pressure nodes and resource overload areas, and replanning resource flow paths and priorities based on the enterprise resource stock map to form a resource reallocation strategy; an abnormal state infiltration simulation module loading this strategy and simulating the diffusion process of abnormal states based on a historical abnormal case library, outputting a risk infiltration early warning report by defining the diffusion boundary and penetration depth; and an investment efficiency convergence module performing reverse calibration on the key parameters of the resource reallocation strategy based on the early warning report, using multiple iterations to bring the investment output characteristics close to the preset convergence target, thereby locking in the final investment scale optimization scheme.

[0060] In one embodiment of the present invention, see [reference] Figure 2 The system acquires multi-source enterprise operational data from enterprise operation databases, market intelligence sources, and internal monitoring systems. This data undergoes time-series alignment and dimensional normalization to form a standardized enterprise operational data stream. During standardization, an attention-based feature analysis model scans the data stream. This model automatically identifies data segments with high volatility and predictive value in historical investment cycles and pre-assigns high initial weight coefficients to the underlying influencing factors corresponding to these segments. From the standardized data stream carrying these initial weight coefficients, a set of indicators representing market capacity, supply chain resilience, technology iteration rate, and policy compliance costs is extracted based on pre-defined association rules. Each indicator in the set is defined as an independent influencing factor. To quantify the relationships between influencing factors, a large number of decision-making instances are extracted from a historical decision case library. Each instance includes a snapshot of the indicator set's state and the investment result. Graph neural networks are used to analyze the relationship patterns of synergistic changes between any two indicators in the state snapshot, quantifying stable positive synergy or negative constraint relationships as constraint strength coefficients. Simultaneously, the system analyzes the composite impact patterns of multiple indicator changes on investment results, quantifying such nonlinear interactive effects as coupling weight matrices. Based on the quantified constraint strength coefficients and coupling weight matrices, each influencing factor is embedded into a hierarchical computational framework. Through signal transmission and aggregation operations at each layer within the framework, an initial investment scenario model reflecting the interaction of all elements of corporate investment is constructed.

[0061] In its implementation, an example of an intelligent optimization system for enterprise investment scale based on big data analysis involves the construction of an initial investment scenario model. Taking an enterprise investment decision-making process called "Alpha Manufacturing Company" as an example, the system's investment scenario construction module obtains quarterly sales data and capacity utilization logs from the enterprise resource planning system, industry growth rate reports and raw material price indices from an external market data platform, and environmental policy update records and corresponding renovation cost estimates from an internal compliance audit system. These constitute multi-source enterprise operational data. In the specific implementation, the multi-source enterprise operational data undergoes time-series alignment and dimensional normalization, unifying data from different sources to the same timestamp sequence. Furthermore, sales revenue, capacity utilization, price indices, and other values ​​are converted to a minimum-maximum normalization method. This creates a standardized data flow for enterprise operations within a given range.

[0062] In practice, the acquisition of standardized enterprise operation data streams includes a pre-calibration process for influencing factor weights. After completing time-series alignment and dimensional normalization, an attention-based feature analysis model is used to scan the standardized enterprise operation data streams. This model automatically identifies data segments with high volatility and predictive value in historical investment cycles. For example, the model might identify data segments in the past eight quarters where, whenever the "copper price index" rises by more than 10% for two consecutive weeks and is accompanied by a decline of more than 5% in the "port logistics efficiency index," the variance of the company's return on investment in the following quarter will significantly increase. In practice, the attention-based feature analysis model assigns higher initial weight coefficients to the underlying influencing factors corresponding to these identified data segments with high predictive value. For example, it assigns higher-than-average initial weights to the "raw material price sensitivity factor" and the "supply chain resilience factor." The standardized enterprise operation data stream carrying these initial weight coefficients is then output.

[0063] In practical implementation, within the standardized enterprise operation data stream carrying initial weight coefficients, association rules between data fields are pre-set according to investment decision-making logic. Based on these association rules, a set of indicators characterizing market capacity, supply chain resilience, technology iteration rate, and policy compliance costs are extracted from the data stream. For example, "annual sales growth rate of the target market" is extracted as a market capacity indicator, "average delivery cycle volatility of key suppliers" as a supply chain resilience indicator, "annual growth rate of patent applications for similar products" as a technology iteration rate indicator, and "increase in unit product cost due to new environmental regulations" as a policy compliance cost indicator. Each indicator within the indicator set is defined as an independent influencing factor.

[0064] In practice, the steps of quantifying the constraint strength coefficients and coupling weight matrices between influencing factors are specifically executed. A large number of successful and unsuccessful decision examples are extracted from a historical decision case library. Each example includes a snapshot of the indicator set at the time of decision and subsequent investment results. Graph neural networks are used to analyze the relationship patterns of the coordinated changes of any two indicator states in the snapshots. For example, analyzing the numerical combinations of "average delivery cycle volatility of key suppliers" and "annual sales growth rate of the target market" in hundreds of historical snapshots reveals that they often exhibit a stable negative correlation; that is, when delivery cycle volatility increases, sales growth tends to slow down. This stable negative constraint relationship is quantified as a constraint strength coefficient in the range [-1, 0]. It can be understood that the combined impact patterns of multiple indicators changing together on investment results can be analyzed simultaneously. For example, analyzing the nonlinear impact on the rate of return on investment when "technology iteration rate," "market capacity," and "policy compliance cost" are all within a specific numerical range is quantified as a coupling weight matrix. Each element W_ij in the matrix represents the joint impact weight of influencing factors i and j on the final result under a specific combination. An example of the quantization constraint strength coefficient and coupling weight matrix is ​​shown in the following formula:

[0065]

[0066] in: This represents the constraint strength coefficient between influence factors a and b. This represents the total number of historical decision instances. and These represent the changes in influence factors a and b relative to their baseline values ​​in the k-th instance, respectively. Used to calculate the correlation pattern between two changes. It is the normalized utility value of the investment outcome of the kth instance, used to assess the importance of the weighted association pattern.

[0067] In some embodiments, influencing factors are embedded into a hierarchical computational framework based on constraint strength coefficients and coupling weight matrices. Specifically, the first layer of the computational framework is the input layer, which receives the current values ​​of each influencing factor and its pre-calibrated initial weights; the second layer is the interaction layer, which calculates the mutual constraints or enhancement effects between factors based on the constraint strength coefficient matrix; and the third layer is the aggregation layer, which applies the coupling weight matrix to perform nonlinear weighted aggregation of the factor states processed by the interaction layer. It can be understood that through signal transmission and aggregation operations within each layer of the computational framework, an initial investment scenario model reflecting the interaction relationships of all investment elements in the "Alpha Manufacturing Company" is constructed. This model can output a quantitative score of investment feasibility under different combinations of influencing factors.

[0068] In one embodiment of the present invention, see [reference] Figure 3 The computational framework of the aforementioned initial investment scenario model is used as the simulation environment. Real-time market disturbance variables, including sudden index, raw material price anomaly rate, and instantaneous exchange rate fluctuation, are continuously captured from external data interfaces. These real-time market disturbance variables are injected into the simulation environment in the form of pulse signals, triggering the recalculation of the state of each influencing factor in the environment. The system tracks the transmission process of state changes to subsequent nodes along a preset investment transmission path. During this process, the generated cascading effect data is recorded, including the state offset of each node, path transmission delay, and energy attenuation coefficient. Based on these cascading effect data, sensitive areas of decreased system stability are marked as pressure nodes. Finally, by combining the spatial distribution of pressure nodes, the time series of state offsets, and the evolution trend of sensitive areas, a dynamic market situation map that can visualize the characteristics of market fluctuations and structural pressure nodes is generated.

[0069] In practical implementation, the dynamic feedback correction module uses the computational framework of the initial investment scenario model as the simulation foundation. For example, for "Alpha Manufacturing Company," the initial investment scenario model already includes a hierarchical computational structure encompassing market capacity, supply chain resilience, technology iteration rate, and policy compliance cost impact factors. In practice, it continuously captures real-time market disturbance variables from external data interfaces, including sudden indexes, raw material price anomalies, and instantaneous exchange rate fluctuations. For instance, it captures sudden indices related to major copper mines by accessing financial news APIs, captures the 15% daily copper price increase anomaly rate from commodity exchange data streams, and obtains the 2% instantaneous exchange rate fluctuation within a short period from foreign exchange trading platforms.

[0070] In practice, real-time market disturbance variables are injected into the simulation environment as pulse signals, triggering the recalculation of the state of each influencing factor in the simulation environment. Taking the real-time market disturbance variable "copper price fluctuation rate" as an example, the dynamic feedback correction module injects the value of this variable as a pulse signal with an intensity of 0.15 into the influencing factor node representing "raw material cost" in the initial investment scenario model. The injection of the pulse signal triggers the recalculation of the state of each influencing factor in the model. For example, if the state value of the "raw material cost" factor changes abruptly, this change is transmitted to the "production cost" factor node through the preset constraints in the calculation framework, thereby affecting the state values ​​of the "product pricing" factor and the "market demand" factor nodes. The system tracks the transmission process of state changes along the preset investment transmission path to subsequent nodes. For example, it tracks the sequence of changes in the values ​​of all nodes along the path from the "raw material cost" node to the "production cost" node, then to the "product gross profit margin" node, and finally to the "quarterly expected profit" node.

[0071] In practical implementation, cascading effect data generated during the transmission process is recorded. This data includes the state offset of each node, path propagation delay, and energy attenuation coefficient. Taking the simulation of "Alpha Manufacturing Company" as an example, the state offset of each node in the cascading effect data includes the "Product Gross Profit Margin" node decreasing from its initial value of 25% to 19%, a state offset of -6%. The path propagation delay in the cascading effect data includes the propagation of changes from the "Raw Material Cost" node to the "Quarterly Expected Profit" node, which takes three simulation time units in the computational framework. The energy attenuation coefficient in the cascading effect data includes an initial pulse signal strength of 0.15, which, when propagated to the final "Quarterly Expected Profit" node, is converted to a strength of 0.08, resulting in an energy attenuation coefficient of 0.53. Based on the cascading effect data, sensitive areas of decreased system stability are marked as stress nodes. For example, in the simulation, the state offsets of both the "Product Gross Profit Margin" node and the "Cash Flow Health" node exceeded the system's set stability threshold; these two nodes were marked as stress nodes.

[0072] In some embodiments, by integrating the spatial distribution of pressure nodes, the time series of state offsets, and the evolution trend of sensitive areas, a dynamic market situation map is generated that can visually represent market volatility characteristics and structural pressure nodes. The dynamic market situation map is a visualization layer that displays the computational framework network of the initial investment scenario model in topological form, with nodes representing influencing factors. Pressure nodes are highlighted in red, and the size of the node is proportional to the absolute value of its state offset. The time series data of the state offsets are transformed into a dynamic halo fluctuating around each pressure node, with the frequency of the halo fluctuating proportional to the rate of state change over the past five simulation time units. The evolution trend of sensitive areas is reflected by the color and thickness of the connecting lines in the map. The color of the connecting lines gradually changes from blue to yellow and finally to red, indicating an increase in tension on the transmission path. The thickness of the connecting lines is proportional to the reciprocal of the energy attenuation coefficient passing through the path.

[0073] In one embodiment of the invention, the generated dynamic market situation map is analyzed to identify the attribute types and overload levels of all pressure nodes in the map, and to delineate resource overload areas caused by resource competition or path congestion. A pre-constructed enterprise resource inventory map is retrieved. The construction and updating of this map rely on real-time data connection with the enterprise resource management system, continuously acquiring updated information on cash flow, equipment logs, manpower deployment, and technical ledgers. Using the enterprise organization and business processes as a blueprint, the information is mapped to a virtual resource topology network, and a time-series version of the map is generated periodically. This map uses a mesh structure to mark the current location, quantity, and flow attributes of funds, production capacity, manpower, and technical reserves. With the goal of alleviating pressure node overload and clearing resource overload areas, path optimization calculations are performed on the enterprise resource inventory map to plan one or more backup flow paths for each type of resource. Based on the estimated throughput, required activation cost, and contribution to overall stability of the backup paths, execution priorities are assigned to all planned paths. All path planning and priority information are integrated to form a complete resource reallocation strategy.

[0074] In practical implementation, the resource topology adaptation module performs graph analysis on the dynamic market situation map, identifying the attribute types and overload levels of all pressure nodes in the graph. For example, in the dynamic market situation map of "Alpha Manufacturing Company," the "Product Gross Profit Margin" node is identified as a financial pressure node with its state offset indicating a "high" overload level, while the "Production Line A Capacity Utilization Rate" node is identified as a capacity pressure node with a "medium" overload level. In practical implementation, the resource topology adaptation module also identifies resource overload areas caused by resource competition or path congestion. For example, in the network topology of the dynamic market situation map, the links connecting the three nodes "Raw Material Procurement," "Work-in-Process Inventory," and "Finished Product Shipment" are generally identified as dark red and continuously flashing; this triangular area is identified as a logistics resource overload area.

[0075] In practical implementation, the resource topology adaptation module retrieves a pre-built enterprise resource inventory map. This map, in a network structure, marks the current location, quantity, and flow attributes of funds, production capacity, human resources, and technology reserves. The construction and updating steps of the enterprise resource inventory map are as follows: A real-time data connection is established with the enterprise resource management system to continuously acquire updated information on fund account flows, equipment operation logs, project manpower deployment, and patent technology ledgers. Using the enterprise's organizational structure and business processes as a blueprint, the acquired updated information is mapped onto a virtual resource topology network, where network nodes represent resource storage or consumption units, and edges represent resource scheduling or transfer relationships. Periodic map snapshots are executed, recording the location, quantity, and status of resources in each time slice, forming a time-series version of the enterprise resource inventory map. For example, the enterprise resource inventory map of "Alpha Manufacturing Company" shows that the node "East China Warehouse" stores 1,000 units of finished product inventory, the node "Flexible Production Line B" currently has a capacity utilization rate of 60%, the node "R&D Project Team Y" has 15 engineers, and the edge connecting the "Central Funding Pool" node and the "New Production Line Procurement" node is marked with the attribute "Monthly allocation limit of 5 million".

[0076] In some embodiments, the primary objectives are to alleviate the overload of pressure nodes and clear resource overload areas, and path optimization calculations are performed on the enterprise resource inventory map. In specific implementations, one or more backup flow paths are planned for each type of resource. For example, to alleviate the overload of the "product gross profit margin" pressure node, a backup flow path is planned from the "central cash pool" node to the "supply chain optimization project" node, aiming to reduce raw material procurement costs through investment in supply chain optimization. Simultaneously, to clear logistics resource overload areas, a backup path for finished goods inventory allocation is planned from the "East China warehouse" node to the "South China distribution center" node, and a backup path for capacity reallocation is planned to utilize some idle capacity from the "flexible production line B" node to share the pressure on the "production line A" node.

[0077] It is understandable that execution priorities are assigned to all planned paths based on the estimated traffic efficiency, required activation cost, and contribution to the overall system stability of the alternative flow paths. The calculation of path priorities involves a comprehensive evaluation of multiple quantitative indicators. Optionally, the path priority assignment process can employ a multi-objective decision function, which is as follows:

[0078]

[0079] in: This represents the comprehensive priority evaluation value of the p-th alternative flow path. This represents the normalized value of the estimated traffic efficiency of the alternative flow path. This represents the normalized value of the required activation cost for the alternative flow path. This represents the estimated reduction in the average overload level at key pressure nodes in the dynamic market situation diagram after implementing the backup flow path. , and These are configurable weighting coefficients corresponding to the dimensions of efficiency, cost, and effectiveness, respectively. This can be understood as a comprehensive priority evaluation value. The higher the value, the higher the priority of the backup flow path being executed. For example, calculating the "activate flexible production line B" path... The value is 0.85, while the "finished goods inventory transfer" path... If the value is 0.72, the former is given a higher execution priority.

[0080] In some embodiments, all path planning and priority information are integrated to form a complete resource reallocation strategy. In specific implementations, the resource reallocation strategy is presented as a structured list of instructions. Each instruction specifies a resource scheduling action, source and target nodes, resource quantity or proportion, expected execution time window, and its calculated priority value. Optionally, the resource reallocation strategy also includes a visualization view overlaid with an enterprise resource inventory map and a dynamic market situation map. In this view, planned alternative flow paths are displayed as highlighted arrows on the resource topology, with the arrow color and thickness intuitively reflecting their priority. The resource topology adaptation module ultimately outputs this resource reallocation strategy, containing specific action instructions, priority ranking, and visual references, for subsequent modules to load and execute.

[0081] In one embodiment of the present invention, the aforementioned resource reallocation strategy is loaded and used as the new configuration baseline for system operation. Simultaneously, a historical anomaly case library storing historical anomaly events and their development trajectories is accessed. Based on the anomaly propagation pattern abstracted from the historical anomaly case library, under the operating state defined by the new configuration baseline, the simulation demonstrates how an anomaly state originating at a local node propagates along the resource network and information network. During the simulation, the propagation speed and impact range of the anomaly state are calculated in real time, its maximum potential propagation boundary is defined, and its penetration depth into core investment functional modules is assessed. The obtained propagation boundary, penetration depth, key springboard nodes on the propagation path, and estimated time to reach the critical risk threshold are integrated and packaged into a risk contamination early warning report.

[0082] In a specific implementation, an embodiment of an intelligent optimization system for enterprise investment scale based on big data analysis involves the output process of a risk contamination early warning report. The abnormal state contamination simulation module loads a resource reallocation strategy as a new configuration baseline for system operation. In the specific implementation, the abnormal state contamination simulation module also accesses a historical abnormal case library that stores historical abnormal events and their development trajectories. The historical abnormal case library stores structured records. For example, Case 1 records a "fire at a core supplier's factory" event, including the event occurrence time, the initial supply chain node number affected, and the subsequent weekly spread trajectory data of the impact on production plans, raw material inventory, and customer delivery delays. Case 2 records a "large-scale failure of the production execution system" event, including the failure outbreak node, system recovery time, and quantitative data on the impact on the order completion rate of each production line during the failure period.

[0083] In some embodiments, based on anomaly propagation patterns abstracted from a historical anomaly case library, under the operating conditions defined by the new configuration baseline, the simulation demonstrates how an anomaly state initially originating at a local node propagates along the resource and information networks. For example, simulating an anomaly state initially originating at the "Import Clearance Delay" node, the relevant propagation patterns abstracted from the historical anomaly case library indicate that such anomalies typically propagate to the "Raw Material Inventory" node with a 70% probability within three working days, and further propagate to the "Production Scheduling" node with a 40% probability within one week. During the simulation, the anomaly state contagion simulation module calculates the probability and intensity of the anomaly state propagating along the new path of "Activating Flexible Production Line B" according to the path planned by the resource reallocation strategy in the new configuration baseline.

[0084] In practical implementation, the spread rate and impact range of the abnormal state are calculated in real time during the simulation. The maximum potential spread boundary of the abnormal state is defined. The method for defining the maximum spread boundary is to trace all nodes affected by the abnormal state in the simulation and connect the outermost set of nodes to form a closed region. In practical implementation, the penetration depth of the abnormal state on the core investment functional modules is assessed. For example, the core investment functional module "New Production Line Commissioning Project" depends on three sub-modules: "Equipment Procurement," "Installation and Commissioning," and "Personnel Training." The impact ratio of the abnormal state on the functional integrity of these three sub-modules is simulated and calculated, and the average impact ratio is used as the penetration depth.

[0085] In some embodiments, the simulation of the spread of anomalies generates a series of key performance indicator (KPI) data that evolves over time, which can be tabulated for analysis. Referring to Table 1, the table shows the KPI data at five consecutive simulation time points after the anomaly begins to spread from the initial node during a single simulation run.

[0086] Table 1: Time Series Table of Key Indicators for Abnormal State Propagation Simulation

[0087]

[0088] In practical implementation, generating a risk infiltration early warning report requires a quantitative assessment of the risk of the diffusion process, with calculating the remaining time to reach the critical risk threshold being a crucial step. The critical risk threshold can be understood as a pre-set risk level by the system; for example, it is considered reached when the average penetration depth of core investment functional modules exceeds 50% or the number of affected nodes exceeds 30% of the total number of nodes. The formula for calculating the estimated remaining time to the critical risk threshold is as follows:

[0089]

[0090] in: This represents the estimated remaining time until the critical risk threshold at simulation time point t. This represents a numerical value indicating a critical risk threshold. For example, the average penetration depth threshold for core investment functional modules could be set at 50%. This represents the average penetration depth of the core investment functional modules actually calculated at the current simulation time point t. This represents the moving average of the rate of change in the average penetration depth of the core investment functional modules over a period of time prior to the current point in time. Optionally, key springboard nodes on the diffusion path refer to nodes that are frequently traversed by anomalous states in the simulation and connect to multiple network communities; for example, the "Central Logistics Scheduling" node might be identified as a key springboard node.

[0091] In practice, the diffusion boundary, penetration depth, key stepping stone nodes along the diffusion path, and the estimated time to reach the critical risk threshold are integrated and packaged into a risk contamination early warning report. This report uses a structured document format and includes an execution summary, a description of the initial simulation conditions, a time-series analysis of the diffusion process, a list of key stepping stone nodes, a penetration depth assessment of the core module, a risk timeline estimate, and detailed information similar to the table above. The abnormal state contamination simulation module ultimately outputs this risk contamination early warning report for further processing by the investment efficiency convergence module.

[0092] See Figure 4This is a bar chart analyzing the abnormal diffusion risk of enterprise resource flow paths, primarily used to compare the probability and intensity of abnormal diffusion across different resource paths. The original resource path has the highest diffusion probability (70%) and intensity (85%), indicating the most concentrated risk and weak risk resistance. The flexible production line B path has a lower diffusion probability (45%) than the original path, but its intensity (55%) remains relatively high, classifying it as a medium-risk path. The backup logistics path has a further decrease in both diffusion probability (30%) and intensity (40%), making its risk manageable. The emergency funding path has the lowest diffusion probability (15%) and intensity (20%), making it the most resilient path. This chart serves as a decision-making tool during the enterprise resource topology adaptation phase. By quantifying the risk differences between different paths, it can guide resource reallocation strategies and provide path priority for subsequent risk contagion simulations.

[0093] In one embodiment of the present invention, a risk contamination warning report is received from the abnormal state contamination simulation module, and key risk parameters are extracted from it, including the identifier of the module with the highest penetration depth and the shortest time to approach the critical threshold. These key risk parameters are used as feedback signals and input to the parameter regulator of the resource reallocation strategy to perform reverse calibration on the resource allocation ratio and path response threshold of the corresponding module in the strategy. After completing one round of reverse calibration, the calibrated resource reallocation strategy is rerun in the simulation environment, and its investment output characteristics are evaluated, including return stability, risk buffer capacity, and scale growth rate. The evaluated investment output characteristics are compared with the preset convergence target. The convergence target is converted from the basic expected value input by the management interface during system initialization. During system operation, the system continuously monitors the external macroeconomic prosperity index and industry average performance. When the external data continuously deviates from the benchmark range, the system automatically adjusts the convergence target parameters elastically according to the built-in fitness function to generate a dynamic convergence target. If the investment output characteristics do not meet the convergence criteria, a new round of calibration, operation, and evaluation iteration is started until it meets the current dynamic convergence target. The final investment scale optimization scheme is determined by locking all system configuration parameters and resource allocation schemes that ultimately meet the convergence target.

[0094] In practical implementation, the investment efficiency convergence module receives risk contamination warning reports from the abnormal state contamination simulation module. For example, it receives a report indicating that the penetration depth of the core module has reached 60% and the estimated remaining time to the critical risk threshold is two simulation time units. The investment efficiency convergence module extracts key risk parameters from the risk contamination warning report. The key risk parameters include the module with the highest penetration depth and the shortest time to approach the critical threshold. For example, the "equipment procurement sub-module of the new production line commissioning project" is extracted as the module with the highest penetration depth, and "two simulation time units" is extracted as the shortest time.

[0095] In practical implementation, the investment efficiency convergence module inputs key risk parameters as feedback signals to the parameter regulator of the resource reallocation strategy, and performs reverse calibration on the resource allocation ratio and path response threshold of the corresponding module in the resource reallocation strategy. In some embodiments, after completing a round of reverse calibration, the investment efficiency convergence module reruns the calibrated resource reallocation strategy in the simulation environment and evaluates its investment output characteristics. The investment output characteristics include return stability, risk buffer capacity and scale growth rate. For example, by running the simulation, a new set of evaluation values ​​is obtained: the return stability index increases from 0.65 to 0.72, the risk buffer capacity index increases from 500 units to 580 units, and the scale growth rate index remains at 12% annualized.

[0096] In practical implementation, the investment efficiency convergence module compares the re-evaluated investment output characteristics with the preset convergence targets. The definition and dynamic adjustment steps of the convergence targets are as follows: During the system initialization phase, the management interface inputs the basic expected values ​​for investment return rate, risk level, and capital turnover rate. For example, the input is an annualized investment return rate greater than 10%, a risk level lower than medium, and a capital turnover rate higher than the industry average of 20%. These basic expected values ​​are converted into quantifiable initial convergence targets through the system's built-in quantitative model. For example, they are converted into a return stability target value greater than 0.75, a risk buffer capacity target value greater than 600 units, and a scale growth rate target value greater than 11.5% annualized. Understandably, during operation, the system continuously monitors the external macroeconomic prosperity index and industry average performance. When the external monitoring data deviates from the preset benchmark range, the system automatically adjusts the parameters in the initial convergence target according to the built-in fitness function. For example, when the industry average growth rate is generally reduced to 8% annualized due to market contraction, the fitness function dynamically adjusts the growth rate target value from 11.5% annualized to 9.5% annualized, generating a dynamic convergence target adapted to the new environment to guide the subsequent iterative calibration process.

[0097] The investment efficiency convergence module compares the evaluated investment output characteristics with the current dynamic convergence target. If the investment output characteristics do not meet the convergence criteria, a new round of calibration, operation, and evaluation iterations is initiated until the investment output characteristics meet the preset convergence target. During the iteration process, the quantitative evaluation of the gap between the investment output characteristics and the convergence target can be achieved using a convergence function. The formula for calculating the convergence function is as follows:

[0098]

[0099] in: This indicates the convergence degree of the current iteration round. This indicates the quantity of investment output characteristics. This represents the current assessed value of the output characteristic of the i-th investment. This represents the dynamic convergence target value corresponding to the i-th investment output characteristic. It is a small constant to prevent division by zero. This can be understood as, when convergence... When the value is greater than or equal to the system's preset convergence threshold, such as 0.95, the investment output characteristics are deemed to meet the convergence criteria. Optionally, after each iteration, the system records the changes in calibration parameters, investment output characteristic evaluation values, and convergence, forming an iteration history log.

[0100] See Figure 5 This is a grouped bar chart showing the dynamic adjustment of corporate investment convergence targets with the external economic environment. Its core purpose is to demonstrate the flexible adaptation logic of investment targets under different macroeconomic environments. The chart data reveals the flexible adjustment pattern of convergence targets with the external environment. From a booming market to a significant market contraction, the risk buffer capacity target value gradually decreases from 600 to 550, and the scale growth rate target value decreases from 11.5% to 9.5%, reflecting the strategy of "reducing expansion expectations and moderately compressing risk buffer costs during market contraction." From a significant market contraction to market recovery, the risk buffer capacity target value rebounds to 620, and the scale growth rate target value rebounds to 11.0%, reflecting the strategy of "restoring expansion expectations and strengthening risk buffers during market recovery." This chart serves the dynamic calibration of targets during the investment efficiency convergence phase. By visualizing target adjustments under different environments, it can help the system quickly match reasonable investment expectations to the current market state, avoiding over-investment or conservative decisions caused by "target rigidity." It is one of the core adaptation mechanisms for intelligent optimization of corporate investment scale.

[0101] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0102] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An enterprise investment scale intelligent optimization system based on big data analysis, characterized in that, The system comprises: An investment scenario construction module for receiving multi-source enterprise operation data and defining multi-dimensional influence factors based on the received data, and constructing an initial investment scenario model by analyzing the constraint and coupling relationship between the multi-dimensional influence factors; A dynamic feedback correction module for importing the initial investment scenario model, injecting real-time market disturbance variables under the calculation framework of the initial investment scenario model, generating a dynamic market situation map containing fluctuation characteristics and stress nodes by simulating the cascading effect of the real-time market disturbance variables on the investment transmission path; A resource topology adaptation module for analyzing the dynamic market situation map, identifying stress nodes and resource overload areas therein, and re-planning the path and priority of resource flow according to the enterprise resource inventory map to form a resource reallocation strategy; An abnormal state dyeing simulation module for loading the resource reallocation strategy, simulating the diffusion process of abnormal state under the strategy execution condition based on the historical abnormal case library, and outputting a risk dyeing early warning report by defining the diffusion boundary and penetration depth; An investment efficiency convergence module for inversely calibrating key parameters in the resource reallocation strategy according to the risk dyeing early warning report, and making the investment output characteristics approach the preset convergence target in multiple rounds of iterative calibration to finally lock the investment scale optimization scheme. 2.The big data analysis based enterprise investment scale intelligent optimization system according to claim 1, characterized in that, The construction steps of the initial investment scenario model are specifically: Obtain multi-source enterprise operation data from enterprise operation database, market intelligence sources and internal monitoring system, perform time series alignment and dimensionless processing on the multi-source enterprise operation data to obtain standardized enterprise operation data stream; In the standardized enterprise operation data stream, according to the association rules between the data fields preset in the investment decision logic, the index set representing market capacity, supply chain elasticity, technology iteration rate and policy compliance cost is extracted from the data stream according to the association rules; Define each index in the index set as an independent influence factor, and based on the data mining results of the historical decision case library, quantify the constraint strength coefficients and coupling weight matrices existing between the influence factors; According to the constraint strength coefficients and coupling weight matrices, embed the influence factors into a calculation framework with hierarchical structure, and construct an initial investment scenario model reflecting the full-factor interaction relationship of enterprise investment by signal transmission and aggregation operation of each layer in the framework. 3.The big data analysis based enterprise investment scale intelligent optimization system according to claim 2, characterized in that, The generation steps of the dynamic market situation map are specifically: Take the calculation framework of the initial investment scenario model as the simulation basic environment, continuously capture real-time market disturbance variables including burst index, raw material price fluctuation rate and exchange rate instantaneous amplitude from external data interface; Inject the real-time market disturbance variables into the simulation basic environment in the form of pulse signal, trigger the state recalculation of each influence factor in the simulation basic environment, and track the transmission process of state changes along the preset investment transmission path to subsequent nodes; record the cascade effect data generated in the transmission process, the cascade effect data including node state offset, path conduction delay and energy attenuation coefficient, mark the sensitive area of system stability decline as a pressure node according to the cascade effect data; integrate the spatial distribution of the pressure node, the time sequence of the state offset and the evolution trend of the sensitive area to generate a dynamic market situation map that can visually present market fluctuation characteristics and structural pressure nodes. 4.The big data analysis based enterprise investment scale intelligent optimization system according to claim 3, characterized in that, The forming step of the resource reconfiguration strategy is specifically: perform atlas analysis on the dynamic market situation map to identify the attribute type and overload degree of all pressure nodes in the map, and to circumscribe the resource overload area formed due to resource competition or path blockage; retrieve the pre-constructed enterprise resource inventory atlas, which marks the current location, quantity and flow attribute of the fund, capacity, manpower and technology reserves in a network structure; as the primary goal of relieving the overload degree of the pressure node and unblocking the resource overload area, perform path optimization calculation on the enterprise resource inventory atlas to plan one or more backup flow paths for each type of resource; according to the estimated passing efficiency, required activation cost and contribution to global system stability of the backup flow path, assign execution priority to all planned paths, integrate all path planning and priority information to form a complete resource reconfiguration strategy. 5.The big data analysis based enterprise investment scale intelligent optimization system according to claim 4, characterized in that, The output step of the risk contamination early warning report is specifically: load the resource reconfiguration strategy as the new configuration baseline of system operation, and access the historical abnormal case library storing historical abnormal events and their development trajectories; based on the abnormal propagation pattern abstracted from the historical abnormal case library, simulate how an abnormal state initiated at a local node spreads along the resource network and information network under the running state defined by the new configuration baseline; calculate the diffusion speed and influence range of the abnormal state in real time during the simulation process, define the maximum diffusion boundary it can reach, and evaluate its penetration depth into the core investment function module; integrate and package the diffusion boundary, penetration depth, key hop node on the diffusion path and time estimation information reaching the critical risk threshold into a risk contamination early warning report. 6.The big data analysis based enterprise investment scale intelligent optimization system according to claim 5, characterized in that, The locking step of the investment scale optimization scheme is specifically: receive the risk contamination early warning report, and extract the key risk parameters including the module identification with the highest penetration depth and the shortest time approaching the critical threshold from the report; input the key risk parameters as feedback signals into the parameter adjuster of the resource reconfiguration strategy to reverse calibrate the resource allocation proportion and path response threshold of the corresponding module in the strategy; after completing a round of reverse calibration, re-run the calibrated resource reconfiguration strategy in a simulation environment and evaluate its investment output characteristics, including yield stability, risk buffer capacity and scale growth rate; compare the investment output characteristics obtained after re-evaluation with the preset convergence target, if the convergence standard is not met, start a new round of calibration, running and evaluation iteration until the investment output characteristics meet the preset convergence target; Lock the final investment scale optimization scheme as the final investment scale optimization scheme when the final convergence target is met.

7. The big data analytics based intelligent optimization system for enterprise investment sizing according to claim 2, wherein, The obtaining step of the standardized enterprise operation data stream further includes an influence factor weight pre-calibration process, specifically: After completing the timing alignment and dimensionless normalization, a feature analysis model based on an attention mechanism is used to scan the standardized enterprise operation data stream; The feature analysis model automatically identifies data bands with high volatility and high prediction value in the historical investment cycle, and assigns high initial weight coefficients to the underlying influence factors corresponding to these data bands; The standardized enterprise operation data stream carrying the initial weight coefficients is output for subsequent weighted calculation when extracting the indicator set. 8.The big data analysis based enterprise investment scale intelligent optimization system according to claim 2, characterized in that, The step of quantifying the constraint strength coefficients and coupling weight matrix between the influence factors is specifically: A large number of successful and failed decision-making instances are extracted from the historical decision-making case library, each instance including a state snapshot of the indicator set at the decision-making time and subsequent investment results; Graph neural networks are used to analyze the relationship patterns of the coordinated changes of any two indicators in the state snapshot, and the stable positive coordination or negative constraint relationship is quantified as a constraint strength coefficient; At the same time, the complex influence mode of multiple indicators changing together on investment results is analyzed, and the nonlinear influence of multi-factor interaction is quantified as a coupling weight matrix. 9.The big data analysis based enterprise investment scale intelligent optimization system according to claim 4, characterized in that, The construction and updating steps of the enterprise resource stock atlas are specifically: Establish real-time data connection with the enterprise resource management system, continuously obtain update information of fund account flow, equipment operation log, project manpower deployment and patent technology account; Using enterprise organizational structure and business process as a blueprint, the obtained update information is mapped to a virtual resource topology network, where network nodes represent resource storage units or consumption units, and edges represent resource scheduling or transfer relationships; Periodically perform atlas snapshots to record the location, quantity and state of resources at each time slice, forming an enterprise resource stock atlas with time sequence version. 10.The big data analysis based enterprise investment scale intelligent optimization system according to claim 6, wherein, The definition and dynamic adjustment steps of the convergence target are specifically: In the system initialization stage, the basic expected values of investment yield, risk level and capital turnover rate are input from the management interface, and these basic expected values are converted into quantifiable initial convergence targets; The system continuously monitors the external macroeconomic prosperity index and industry average performance level during operation; When the external monitoring data continuously deviates from the preset benchmark range, the system automatically adjusts the parameters in the initial convergence target according to the built-in fitness function, generates a dynamic convergence target that adapts to the new environment, and guides the subsequent iterative calibration process.

Images