Government affair industry budget allocation method based on large model and related device

Abstract

The invention discloses a government affair industry budget allocation method based on a large model and a related device. The method comprises the steps that budget allocation associated data is acquired and preprocessed; extracting current feature data from the budget allocation associated data; generating a current budget allocation scheme according to an allocation principle and an allocation algorithm by using the current feature data and a preset framework; performing multi-objective optimization on the current budget allocation scheme through a multi-objective optimization model to obtain a current optimized budget allocation scheme; wherein the multi-objective optimization model comprises a radial basis function network and a deep reinforcement learning network; evaluating whether the current optimization budget allocation scheme meets a project target or not; if not, generating a new current budget allocation scheme, and returning to perform multi-objective optimization; if yes, collecting execution data of the scheme in real time; calling an evaluation model to perform evaluation by utilizing the execution data to obtain a current execution evaluation result; and adjusting the current optimization budget allocation scheme based on the current execution evaluation result.

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular to a budget allocation method for the government affairs industry based on a large model and related devices. Background Art

[0002] In the process of government work, it is usually necessary to allocate budgets to various projects or fields. Since the budget is relatively fixed, how to reasonably allocate the budget determines whether the desired goals can be achieved, so it is crucial.

[0003] The current budget allocation method is mainly for staff to collect relevant data according to requirements, and then analyze the data based on experience and budget allocation needs to obtain data analysis results. Then, based on the allocation strategy or rules, relying on their own experience, budget allocation is carried out to obtain an allocation plan that is considered to meet the expected goals.

[0004] Although the requirements, strategies and restrictions set can avoid unreasonable budgets to a certain extent and ensure that budget allocations meet the requirements, the data analysis and allocation processes still rely on manual experience, so the accuracy of data analysis and allocation cannot be guaranteed, and the accuracy of the generated allocation plan cannot be guaranteed, and the efficiency of this method is relatively low. Summary of the invention

[0005] Based on the above-mentioned deficiencies of the prior art, the present application provides a budget allocation method and related devices for the government industry based on a large model to solve the problems of low accuracy and efficiency of the prior art.

[0006] In order to achieve the above objectives, this application provides the following technical solutions:

[0007] The first aspect of the present application provides a budget allocation method for the government affairs industry based on a large model, comprising:

[0008] Obtain budget allocation related data;

[0009] Preprocessing each item of budget allocation related data respectively;

[0010] Extracting features from each item of budget allocation related data to obtain current feature data;

[0011] Using the current feature data and the preset framework, generating a current budget allocation plan according to the preset allocation principle and allocation algorithm;

[0012] Performing multi-objective optimization on the current budget allocation scheme through a multi-objective optimization model to obtain a current optimized budget allocation scheme; wherein the multi-objective optimization model is obtained by combining a radial basis function network and a deep reinforcement learning network;

[0013] Evaluate whether the current optimized budget allocation plan meets the project objectives;

[0014] If the current optimized budget allocation scheme does not meet the project goal, the current optimized preset allocation scheme is redesigned to generate a new current budget allocation scheme, and the method returns to execute the multi-objective optimization of the current budget allocation scheme through the multi-objective optimization model to obtain the current optimized budget allocation scheme;

[0015] If the current optimized budget allocation scheme meets the project goal, then the execution data of the current optimized budget allocation scheme is collected in real time;

[0016] Calling multiple evaluation models, using the execution data of the current optimized budget allocation scheme to perform evaluation, and obtaining a current execution evaluation result;

[0017] The current optimized budget allocation scheme is adjusted based on the current execution evaluation result.

[0018] Optionally, in the above-mentioned large-model-based government industry budget allocation method, the preprocessing of each item of the budget allocation related data respectively includes:

[0019] The data modes of each item of the budget allocation associated data are respectively identified, and the budget allocation associated data are preprocessed using a preprocessing method corresponding to the data mode of the budget allocation associated data.

[0020] Optionally, in the above-mentioned large-model-based government industry budget allocation method, the feature extraction of each item of the budget allocation related data to obtain current feature data includes:

[0021] Identify the patterns of each item of the budget allocation associated data, and use the extraction method corresponding to the pattern of each item of the budget allocation associated data to extract features of each item of the budget allocation associated data to obtain current feature information;

[0022] Using the relationship between the various pieces of current feature information, construct a current data structure diagram;

[0023] Analyze the current data structure graph using a graph neural network to obtain feature relationship information;

[0024] The current feature information and the feature relationship information are combined into the current feature data.

[0025] Optionally, in the above-mentioned large-model-based government affairs industry budget allocation method, the current feature data and the preset framework are used to generate the current budget allocation plan according to the preset allocation principle and allocation algorithm, including:

[0026] Predict future budget needs and trends based on the time series data composed of the current feature data using the Prophet model;

[0027] Based on the future budget requirements and trends, the allocation algorithm is called to calculate the budget allocation of each project that meets the allocation principle according to the preset framework;

[0028] A current budget allocation plan is generated according to the budget allocation of each of the projects.

[0029] Optionally, in the above-mentioned large model-based government affairs industry budget allocation method, before the real-time collection of the execution data of the current optimized budget allocation scheme, it also includes:

[0030] Performing a risk assessment on the current optimized budget allocation scheme to obtain a risk level of the current optimized budget allocation scheme;

[0031] Determining whether the risk level of the current optimized budget allocation scheme exceeds a risk threshold;

[0032] If it is determined that the risk level of the current optimized budget allocation scheme exceeds the risk threshold, returning to execute the redesign of the current optimized preset allocation scheme;

[0033] If it is determined that the risk level of the current optimized budget allocation plan exceeds the risk threshold, the current optimized budget allocation plan is subject to an approval process;

[0034] Determine whether the current optimized budget allocation plan has been approved;

[0035] If it is determined that the current optimized budget allocation scheme has passed the approval, the current optimized budget allocation scheme is displayed, and the execution data of the current optimized budget allocation scheme is collected in real time;

[0036] If it is determined that the current optimized budget allocation plan has not passed the approval, the process returns to the step of redesigning the current optimized preset allocation plan.

[0037] Optionally, in the above-mentioned large-model-based government affairs industry budget allocation method, calling multiple evaluation models and using the execution data of the current optimized budget allocation scheme to perform evaluation to obtain the current execution evaluation result includes:

[0038] Performing real-time processing and analysis on the real-time data in the execution data of the current optimized budget allocation scheme through the edge computing node, and performing batch processing on the batch data in the execution data of the current optimized budget allocation scheme through the central data center to obtain effect evaluation indicators;

[0039] The causal inference model, the statistical model and the machine learning evaluation model are respectively called, and the effect evaluation index is used to perform evaluation to obtain the current execution evaluation result.

[0040] Optionally, in the above-mentioned large-model-based government affairs industry budget allocation method, adjusting the current optimized budget allocation scheme based on the current execution evaluation result includes:

[0041] Based on the current execution evaluation result, a solution optimization suggestion is generated through the optimization model of federated learning, and the current optimization budget allocation solution is adjusted using the solution optimization suggestion.

[0042] The second aspect of the present application provides a budget allocation device for the government affairs industry based on a large model, comprising:

[0043] A data acquisition unit, used to acquire budget allocation related data;

[0044] A preprocessing unit, used for preprocessing each item of the budget allocation associated data;

[0045] A feature extraction unit, used to extract features from each item of the budget allocation associated data to obtain current feature data;

[0046] A scheme generating unit, used to generate a current budget allocation scheme according to a preset allocation principle and allocation algorithm by using the current feature data and a preset framework;

[0047] An optimization unit, configured to perform multi-objective optimization on the current budget allocation scheme through a multi-objective optimization model to obtain a current optimized budget allocation scheme; wherein the multi-objective optimization model is obtained by combining a radial basis function network and a deep reinforcement learning network;

[0048] A target evaluation unit, used to evaluate whether the current optimized budget allocation scheme meets the project target;

[0049] A scheme updating unit, configured to redesign the current optimized preset allocation scheme when the current optimized budget allocation scheme does not meet the project goal, generate a new current budget allocation scheme, and return it to the optimization unit;

[0050] A data collection unit, configured to collect execution data of the current optimized budget allocation scheme in real time when the current optimized budget allocation scheme meets the project goal;

[0051] A result evaluation unit, used to call multiple evaluation models, use the execution data of the current optimized budget allocation scheme to perform evaluation, and obtain a current execution evaluation result;

[0052] An adjusting unit is used to adjust the current optimized budget allocation scheme based on the current execution evaluation result.

[0053] Optionally, in the above-mentioned device for allocating budget for the government affairs industry based on a large model, the pre-processing unit includes:

[0054] The preprocessing subunit is used to respectively identify the data mode of each item of the budget allocation associated data, and preprocess the budget allocation associated data by using the preprocessing method corresponding to the data mode of the budget allocation associated data.

[0055] Optionally, in the above-mentioned device for allocating budget for the government affairs industry based on a large model, the feature extraction unit includes:

[0056] A first extraction unit is used to identify the pattern of each item of the budget allocation associated data, and use the extraction method corresponding to the pattern of each item of the budget allocation associated data to extract features of each item of the budget allocation associated data to obtain current feature information;

[0057] A construction unit, used to construct a current data structure diagram by using the relationship between the various items of current feature information;

[0058] A second extraction unit is used to analyze the current data structure graph using a graph neural network to obtain feature relationship information;

[0059] A combining unit is used to combine the current feature information and the feature relationship information into the current feature data.

[0060] Optionally, in the above-mentioned big model-based government affairs industry budget allocation device, the solution generation unit includes:

[0061] A prediction unit, used for predicting future budget requirements and trends based on the time series data composed of the current feature data through a Prophet model;

[0062] A budget calculation unit, configured to calculate the budget allocation of each project that complies with the allocation principle based on the future budget demand and trend and calling the allocation algorithm according to the preset framework;

[0063] The allocation scheme generating unit is used to generate a current budget allocation scheme according to the budget allocation of each of the projects.

[0064] Optionally, in the above-mentioned government affairs industry budget allocation device based on a large model, it also includes:

[0065] A risk assessment unit, configured to perform risk assessment on the current optimized budget allocation scheme to obtain a risk level of the current optimized budget allocation scheme;

[0066] A first judgment unit is used to judge whether the risk level of the current optimized budget allocation scheme exceeds the risk threshold; wherein, if it is judged that the risk level of the current optimized budget allocation scheme exceeds the risk threshold, the scheme is returned to the scheme updating unit;

[0067] An approval unit, configured to, upon determining that a risk level of the current optimized budget allocation scheme exceeds a risk threshold, conduct an approval process for the current optimized budget allocation scheme;

[0068] A second judgment unit is used to judge whether the current optimized budget allocation plan has been approved;

[0069] A display unit, configured to display the current optimized budget allocation plan and return the plan to the data collection unit when it is determined that the current optimized budget allocation plan has passed the approval;

[0070] If it is determined that the current optimized budget allocation plan has not passed the approval, the method returns to the plan updating unit.

[0071] Optionally, in the above-mentioned big model-based government affairs industry budget allocation device, the result evaluation unit includes:

[0072] An indicator processing unit is used to perform real-time processing and analysis on real-time data in the execution data of the current optimized budget allocation scheme through an edge computing node, and to perform batch processing on batch data in the execution data of the current optimized budget allocation scheme through a central data center to obtain an effect evaluation indicator;

[0073] The output unit is used to call the causal inference model, the statistical model and the machine learning evaluation model respectively, use the effect evaluation index to perform evaluation, and obtain the current execution evaluation result.

[0074] Optionally, in the above-mentioned government affairs industry budget allocation device based on a large model, the adjustment unit includes:

[0075] The adjustment subunit is used to generate a solution optimization suggestion based on the current execution evaluation result through the optimization model of federated learning, and use the solution optimization suggestion to adjust the current optimization budget allocation plan.

[0076] Optionally, in the above-mentioned government affairs industry budget allocation device based on a large model, it also includes:

[0077] A configuration optimization unit is used to optimize each model and strategy based on the execution evaluation result and the execution data of the current optimization budget allocation scheme.

[0078] A third aspect of the present application provides an electronic device, including:

[0079] Memory and processor;

[0080] Wherein, the memory is used to store programs;

[0081] The processor is used to execute the program, and when the program is executed, it is specifically used to implement the government affairs industry budget allocation method based on a large model as described in any one of the above items.

[0082] In a fourth aspect, the present application provides a computer storage medium for storing a computer program. When the computer program is executed by a processor, it is used to implement the big model-based government affairs industry budget allocation method as described in any one of the above.

[0083] The present application provides a budget allocation method for the government industry based on a large model, which obtains budget allocation related data and pre-processes each of the budget allocation related data. Then, feature extraction is performed on each of the budget allocation related data to obtain current feature data, thereby realizing automatic processing and analysis of the data. Then, the current feature data and the preset framework are used to generate the current budget allocation plan according to the preset allocation principle and allocation algorithm, thereby automatically generating the allocation plan. In addition, the current budget allocation plan is multi-objectively optimized by a multi-objective optimization model to obtain the current optimized budget allocation plan; wherein the multi-objective optimization model is obtained by combining a radial basis function network and a deep reinforcement learning network, thereby further improving the accuracy of the generated allocation plan. After accommodation, the current optimized budget allocation plan is evaluated to see whether it meets the project goal, so as to ensure that the optimized plan can meet the goal. If the current optimized budget allocation plan does not meet the project goal, the current optimized preset allocation plan is redesigned to generate a new current budget allocation plan, and the multi-objective optimization of the current budget allocation plan by the multi-objective optimization model is returned to execute to obtain the current optimized budget allocation plan through continuous adjustment. If the current optimized budget allocation plan meets the project goal, the execution data of the current optimized budget allocation plan is collected in real time. Multiple evaluation models are called, and the execution data of the current optimized budget allocation plan is used for evaluation to obtain the current execution evaluation result, so as to adjust the current optimized budget allocation plan based on the current execution evaluation result, so that during the execution of the plan, the adaptive adjustment can be made in real time to always maintain the accuracy of the plan, thereby realizing a budget allocation method that can ensure accuracy, and because it does not require manual work, it also effectively improves the efficiency of budget allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0085] Figure 1 A flowchart of a budget allocation method for the government affairs industry based on a large model provided in an embodiment of the present application;

[0086] Figure 2 A flow chart of a feature extraction method provided in an embodiment of the present application;

[0087] Figure 3A flowchart of a method for generating a current budget allocation scheme provided in an embodiment of the present application;

[0088] Figure 4 A flowchart of a scheme execution evaluation method provided in an embodiment of the present application;

[0089] Figure 5 A schematic diagram of the architecture of a budget allocation device for the government affairs industry based on a large model provided in an embodiment of the present application;

[0090] Figure 6 A schematic diagram of the architecture of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0091] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0092] In this application, relational terms such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0093] The present application embodiment provides a budget allocation method for the government affairs industry based on a large model, such as Figure 1 As shown, the following steps are included:

[0094] S101: Obtain budget allocation related data.

[0095] The budget allocation-related data refers to data associated with budget allocation, i.e., data that affects budget allocation. Specifically, the budget allocation-related data can be collected from multiple channels such as internal databases, public records, financial reports, and online resources. The collected budget allocation-related data may include, but are not limited to, revenue and expenditure data, project execution reports, policy documents, economic indicators, etc. In addition, the collected allocation-related data may include historical data and current data.

[0096] Optionally, the data source of the data to be collected can be identified based on the data source channel, format, content and accessibility, and then the data can be collected in a corresponding manner for the data source. Specifically, the data can be collected by using GraphQL API, application interface, data exchange platform, etc., and the real-time reception and update of data can also be achieved through the WebSocket protocol.

[0097] S102: Pre-process each item of budget allocation related data respectively.

[0098] In order to enable the network to learn and characterize data features more accurately, thereby ensuring the accuracy of the solution, in the embodiment of the present application, it is necessary to pre-process the data associated with each budget allocation, including but not limited to data deduplication, data format, data standardization, etc.

[0099] Optionally, in another embodiment of the present application, a specific implementation of step S102 includes:

[0100] The data patterns of each item of budget allocation associated data are respectively identified, and the budget allocation associated data are preprocessed using a preprocessing method corresponding to the data pattern of the budget allocation associated data.

[0101] In order to achieve better processing results, in the embodiments of the present application, corresponding preprocessing methods are set for data of different data modes, so that the data mode can be identified first, and then the data can be processed in a corresponding manner. For example, for numerical data, the preprocessing method used may include standardization or normalization to eliminate the influence of different dimensions. For categorical data, label encoding is used to convert category features, etc.

[0102] Optionally, AutoML technology can be used to automatically identify data patterns and select the best preprocessing method for processing to improve the automation and intelligence level of data processing.

[0103] S103: Extract features from each budget allocation related data to obtain current feature data.

[0104] It should be noted that since the collected data may include unnecessary data, and in order to facilitate the model to analyze the data, in an embodiment of the present application, a deep learning network is used to extract features of various budget allocation related data to obtain current feature data.

[0105] Optionally, after obtaining the current feature data, it can be stored in a corresponding manner according to the mode type of the current feature data. For example, structured data can be stored in a database, while unstructured data can be stored through a file system.

[0106] Optionally, in another embodiment of the present application, a specific implementation of step S103 is as follows: Figure 2 As shown, including:

[0107] S201 , identifying patterns of various budget allocation related data, and using extraction methods corresponding to the patterns of various budget allocation related data to extract features of various budget allocation related data, to obtain current feature information.

[0108] Since the characteristics of data of different pattern types are different, the features to be extracted are also different. Therefore, in the embodiment of the present application, it is also necessary to first identify the pattern of the budget allocation related data, and then extract features in a corresponding manner according to the pattern. For example, data mining can be used to identify and calculate numerical features; for time series features, Fourier transform can be used to extract; for text, TF-IDF can be used to extract features in text data.

[0109] S202: Utilize the relationship between each item of current feature information to construct a current data structure diagram.

[0110] Since there is a certain relationship between the feature information, the feature information will affect each other, thereby affecting the budget allocation. Therefore, in the embodiment of the present application, by constructing the current data structure diagram, not only can the direct relationship between the current feature information be clarified, but also the indirect relationship between the current feature information can be further analyzed in depth based on the current data structure diagram. Therefore, firstly, the current row data structure diagram is constructed according to the direct relationship between the current feature information.

[0111] S203. Analyze the current data structure graph using a graph neural network to obtain feature relationship information.

[0112] S204: Combining the current feature information and feature relationship information into current feature data.

[0113] S104: Generate a current budget allocation plan using current feature data and a preset framework according to preset allocation principles and allocation algorithms.

[0114] The preset framework is a basic framework for budget allocation, which sets the objects of allocation, such as various projects or fields, and also includes various priorities and restrictions. Therefore, the framework can determine which projects need to be allocated in advance and what priorities and restrictions need to be allocated. The allocation principle is an overall principle of budget allocation, which can be set according to specific needs.

[0115] Therefore, based on the current feature data and the preset framework, the allocation algorithm can be used to analyze the budget required for each project, the possible benefits, etc., and under the allocation principle, the budget allocation for each project can be determined and the current budget allocation plan can be generated.

[0116] Optionally, in another embodiment of the present application, a specific implementation of step S104 is as follows: Figure 3 As shown, including:

[0117] S301. Predict future budget needs and trends based on the time series data composed of current feature data through the Prophet model.

[0118] Specifically, by forming time series data from current feature data, its changes over time can be analyzed to predict future budget requirements, that is, to know the budget requirements for the project and changes in the requirements, so as to determine the minimum budget that needs to be allocated to the project.

[0119] S302: Based on future budget requirements and trends, the allocation algorithm is called to calculate the budget allocation of each project that complies with the allocation principle according to the preset framework.

[0120] Optionally, the allocation algorithm may be specifically called based on future budget requirements and trends, and priority may be given to projects with higher priorities according to information such as priorities and constraints in a preset framework, and budgets that meet requirements and changing trends may be allocated under the constraints.

[0121] S303: Generate a current budget allocation plan based on the budget allocation of each project.

[0122] S105. Perform multi-objective optimization on the current budget allocation plan through a multi-objective optimization model to obtain a current optimized budget allocation plan.

[0123] Among them, the multi-objective optimization model is obtained by combining the radial basis function network and the deep reinforcement learning network.

[0124] It should be noted that since the initially generated allocation plan is only a plan that meets the basic requirements and is not necessarily a better method, and subsequent plans may change during execution, in the embodiment of the present application, it is also necessary to perform multi-objective optimization on the allocation plan to optimize the budget allocation of each project.

[0125] In the embodiment of the present application, deep learning (DRL) technology is used to achieve multi-objective optimization. In order to solve the problem of exploration and utilization balance in deep learning, the optimization effect is better, so as to obtain the most accurate allocation plan, a radial basis function network (RBFN) is also combined in the embodiment of the present application to work with the value function or policy network. The exploration signal provided by RBFN can affect the behavior selection of the agent. By using RBFN to guide the exploration process, the agent can more effectively discover potential high-return areas in the environment while avoiding falling into local optimality, so as to obtain the optimal solution.

[0126] Among them, the radial basis function network serves as the exploration module of the agent, where each RBF corresponds to a potential area in the state space. The center point can represent a known interesting or unknown state, and the width parameter determines the locality or extensiveness of the exploration. In addition, the position of the RBFN center point is dynamically adjusted according to the learning progress of the agent and the feedback from the environment. For example, when the agent obtains a high reward in a certain area, the center point can be moved to cover the adjacent area. Similarly, its width parameter can also be adaptively adjusted, that is, the width parameter can be adjusted according to the exploration needs of the agent. Specifically, in the early stage, a larger width is used to encourage extensive exploration, and as the learning deepens, the width is gradually reduced for more refined exploration.

[0127] S106. Evaluate whether the current optimized budget allocation plan meets the project objectives.

[0128] It should be noted that the main purpose of the budget allocation plan is to meet the required goals, and the plan has been optimized for multiple goals through the network, that is, changes have occurred. Therefore, in order to ensure that it can meet the goals before execution, it is necessary to first evaluate whether the current optimized budget allocation plan can meet the project goals according to the current situation.

[0129] If the current optimized budget allocation scheme does not meet the project goal, then step S107 is executed. If the current optimized budget allocation scheme meets the project goal, then step S108 is executed.

[0130] S107: Redesign the current optimized preset allocation plan to generate a new current budget allocation plan.

[0131] Optionally, the current optimized preset allocation scheme may be redesigned according to the specific situation of the evaluation, thereby generating a new current budget allocation scheme.

[0132] Since a new solution has been regenerated at this time, the optimization effect may no longer exist, so after executing step S107, it is necessary to return to execute step S105.

[0133] S108. Collect execution data of the current optimized budget allocation plan in real time.

[0134] Since the allocation plan has been optimized and can meet the project goals, the current optimized budget allocation plan can be executed. Considering that the situation may change during the execution process, in order to make adaptive adjustments to the allocation plan in a timely manner, the execution data of the current optimized budget allocation plan will be collected in real time.

[0135] Optionally, in order to enable the user to understand the current optimized budget allocation plan, the current optimized budget allocation plan and related data may be visualized through augmented reality (AR) technology, thereby assisting the user in making a decision.

[0136] Optionally, in another embodiment of the present application, before executing step S108, the scheme may be further approved and risk assessed. The scheme is executed only after it passes the approval and assessment. A method for scheme risk assessment and approval provided in an embodiment of the present application includes:

[0137] Perform a risk assessment on the current optimized budget allocation plan to obtain the risk level of the current optimized budget allocation plan, and determine whether the risk level of the current optimized budget allocation plan exceeds a risk threshold.

[0138] If it is determined that the risk level of the current optimized budget allocation scheme exceeds the risk threshold, the process returns to step S107.

[0139] If it is determined that the risk level of the current optimized budget allocation plan exceeds the risk threshold, the current optimized budget allocation plan is put through an approval process, and it is determined whether the current optimized budget allocation plan is approved.

[0140] Specifically, the current optimized budget allocation plan may be distributed to corresponding approval personnel for approval according to the approval process, and the approval results fed back by the approval personnel are received.

[0141] If it is determined that the current optimized budget allocation plan has passed the approval, the current optimized budget allocation plan is displayed and step S108 is executed.

[0142] If it is determined that the current optimized budget allocation plan has not passed the approval, the process returns to step S107.

[0143] S109: Call multiple evaluation models, use the execution data of the current optimized budget allocation plan to perform evaluation, and obtain the current execution evaluation result.

[0144] In order to evaluate the execution of the current optimized budget allocation scheme from multiple aspects and to adjust it from multiple aspects to maintain the accuracy of the scheme, in an embodiment of the present application, the execution data of the current optimized budget allocation scheme is input into the evaluation models corresponding to each aspect, and the current execution evaluation result is obtained through each evaluation model.

[0145] Optionally, in another embodiment of the present application, a specific implementation of step S109 is as follows: Figure 4 As shown, including:

[0146] S401. Perform real-time processing and analysis on the real-time data in the execution data of the current optimized budget allocation scheme through the edge computing node, and perform batch processing on the batch data in the execution data of the current optimized budget allocation scheme through the central data center to obtain effect evaluation indicators.

[0147] It should be noted that since there will be real-time data and some non-real-time batch data in the execution data, they need to be processed separately. In addition, in order for the model to achieve the desired results, it is necessary to define effect evaluation indicators, and then use the execution data to process the effect evaluation indicators and provide them to the model for evaluation.

[0148] S402: respectively call the causal inference model, the statistical model, and the machine learning evaluation model, and use the effect evaluation index to perform evaluation to obtain the current execution evaluation result.

[0149] S110: Adjust the current optimization budget allocation plan based on the current execution evaluation result.

[0150] The current execution evaluation result reflects the execution status of the current optimized budget allocation plan, such as certain budget deficiencies or other aspects. Therefore, at this time, based on the current execution evaluation result, the current optimized budget allocation plan can be appropriately adjusted to maintain the accuracy of the current optimized budget allocation plan.

[0151] Optionally, in another embodiment of the present application, a specific implementation of step S110 includes:

[0152] Based on the current execution evaluation results, the optimization model of federated learning is used to generate solution optimization suggestions, and the solution optimization suggestions are used to adjust the current optimization budget allocation plan.

[0153] It should be noted that in order to effectively obtain various data, the plan is adjusted to ensure that the adjusted plan is more in line with the actual situation. Therefore, in the embodiment of the present application, federated learning is used to learn various knowledge, generate an optimization model, and use it to adjust the plan.

[0154] Optionally, in another embodiment of the present application, after executing step S110, the following may be further included:

[0155] Optimize each model and strategy based on the execution evaluation results and the execution data of the current optimized budget allocation plan.

[0156] Since the execution evaluation results and the execution data of the current optimized budget allocation plan reflect the specific circumstances of the plan generated by the model and strategy in this plan, these data can be used to optimize the model and strategy in a targeted manner to further improve the accuracy of the produced plan.

[0157] The embodiment of the present application provides a budget allocation method for the government industry based on a large model, obtains budget allocation related data, and pre-processes each of the budget allocation related data. Then, feature extraction is performed on each of the budget allocation related data to obtain current feature data, thereby realizing automatic processing and analysis of the data. Then, the current feature data and the preset framework are used to generate the current budget allocation plan according to the preset allocation principle and allocation algorithm, thereby automatically generating the allocation plan. In addition, the current budget allocation plan is multi-objectively optimized by a multi-objective optimization model to obtain the current optimized budget allocation plan; wherein the multi-objective optimization model is obtained by combining a radial basis function network and a deep reinforcement learning network, thereby further improving the accuracy of the generated allocation plan. After accommodation, the current optimized budget allocation plan is evaluated to see whether it meets the project goal, so as to ensure that the optimized plan can meet the goal. If the current optimized budget allocation plan does not meet the project goal, the current optimized preset allocation plan is redesigned to generate a new current budget allocation plan, and the multi-objective optimization of the current budget allocation plan by the multi-objective optimization model is returned to execute to obtain the current optimized budget allocation plan through continuous adjustment. If the current optimized budget allocation plan meets the project goal, the execution data of the current optimized budget allocation plan is collected in real time. Multiple evaluation models are called, and the execution data of the current optimized budget allocation plan is used for evaluation to obtain the current execution evaluation result, so as to adjust the current optimized budget allocation plan based on the current execution evaluation result, so that during the execution of the plan, the adaptive adjustment can be made in real time to always maintain the accuracy of the plan, thereby realizing a budget allocation method that can ensure accuracy, and because it does not require manual work, it also effectively improves the efficiency of budget allocation.

[0158] Another embodiment of the present application provides a budget allocation device for the government affairs industry based on a large model, such as Figure 5 As shown, including:

[0159] The data acquisition unit 501 is used to acquire budget allocation related data.

[0160] The preprocessing unit 502 is used to preprocess each item of budget allocation related data respectively.

[0161] The feature extraction unit 503 is used to extract features from each budget allocation related data to obtain current feature data.

[0162] The solution generation unit 504 is used to generate a current budget allocation solution according to a preset allocation principle and allocation algorithm by using the current feature data and a preset framework.

[0163] The optimization unit 505 is used to perform multi-objective optimization on the current budget allocation scheme through a multi-objective optimization model to obtain the current optimized budget allocation scheme. The multi-objective optimization model is obtained by combining a radial basis function network and a deep reinforcement learning network.

[0164] The target evaluation unit 506 is used to evaluate whether the current optimized budget allocation scheme meets the project target.

[0165] The scheme updating unit 507 is used to redesign the current optimized preset allocation scheme when the current optimized budget allocation scheme does not meet the project goal, generate a new current budget allocation scheme, and return it to the optimizing unit 505.

[0166] The data collection unit 508 is used to collect the execution data of the current optimized budget allocation scheme in real time when the current optimized budget allocation scheme meets the project goal.

[0167] The result evaluation unit 509 is used to call multiple evaluation models, use the execution data of the current optimized budget allocation solution to perform evaluation, and obtain the current execution evaluation result.

[0168] The adjusting unit 510 is configured to adjust the current optimized budget allocation scheme based on the current execution evaluation result.

[0169] Optionally, in the apparatus for allocating budget for the government affairs industry based on a large model provided in another embodiment of the present application, the preprocessing unit includes:

[0170] The preprocessing subunit is used to respectively identify the data patterns of each item of budget allocation associated data, and preprocess the budget allocation associated data using a preprocessing method corresponding to the data pattern of the budget allocation associated data.

[0171] Optionally, in the apparatus for allocating budget for the government affairs industry based on a large model provided in another embodiment of the present application, the feature extraction unit includes:

[0172] The first extraction unit is used to identify the pattern of each item of budget allocation related data, and extract features from each item of budget allocation related data using an extraction method corresponding to the pattern of each item of budget allocation related data to obtain current feature information.

[0173] The construction unit is used to construct the current data structure diagram by utilizing the relationship between each item of current feature information.

[0174] The second extraction unit is used to analyze the current data structure graph using a graph neural network to obtain feature relationship information.

[0175] The combining unit is used to combine the current feature information and feature relationship information into the current feature data.

[0176] Optionally, in the apparatus for allocating budget for the government affairs industry based on a large model provided in another embodiment of the present application, the solution generating unit includes:

[0177] The forecasting unit is used to forecast future budget requirements and trends based on the time series data composed of current feature data through the Prophet model.

[0178] The budget calculation unit is used to call the allocation algorithm based on future budget needs and trends and calculate the budget allocation of each project in accordance with the allocation principle according to the preset framework.

[0179] The allocation plan generating unit is used to generate the current budget allocation plan according to the budget allocation of each project.

[0180] Optionally, in the big model-based government affairs industry budget allocation device provided in another embodiment of the present application, the device further includes:

[0181] The risk assessment unit is used to perform risk assessment on the current optimized budget allocation plan to obtain the risk level of the current optimized budget allocation plan.

[0182] The first judgment unit is used to judge whether the risk level of the current optimized budget allocation scheme exceeds the risk threshold. If it is judged that the risk level of the current optimized budget allocation scheme exceeds the risk threshold, the scheme is returned to the scheme update unit.

[0183] The approval unit is used to carry out the approval process for the current optimized budget allocation plan when it is determined that the risk level of the current optimized budget allocation plan exceeds the risk threshold.

[0184] The second judgment unit is used to judge whether the current optimized budget allocation plan has been approved.

[0185] The display unit is used to display the current optimized budget allocation plan and return it to the data collection unit when it is determined that the current optimized budget allocation plan has passed the approval.

[0186] If it is determined that the current optimized budget allocation plan has not passed the approval, the plan updating unit is returned.

[0187] Optionally, in the apparatus for allocating budget for the government affairs industry based on a large model provided in another embodiment of the present application, the result evaluation unit includes:

[0188] The indicator processing unit is used to perform real-time processing and analysis on the real-time data in the execution data of the current optimized budget allocation plan through the edge computing node, and to perform batch processing on the batch data in the execution data of the current optimized budget allocation plan through the central data center to obtain effect evaluation indicators.

[0189] The output unit is used to call the causal inference model, the statistical model, and the machine learning evaluation model respectively, and use the effect evaluation index to evaluate and obtain the current execution evaluation result.

[0190] Optionally, in the apparatus for allocating budget for the government affairs industry based on a large model provided in another embodiment of the present application, the adjusting unit includes:

[0191] The adjustment subunit is used to generate solution optimization suggestions based on the current execution evaluation results through the optimization model of federated learning, and use the solution optimization suggestions to adjust the current optimization budget allocation plan.

[0192] Optionally, in the big model-based government affairs industry budget allocation device provided in another embodiment of the present application, the device further includes:

[0193] The configuration optimization unit is used to optimize each model and strategy based on the execution evaluation results and the execution data of the current optimization budget allocation plan.

[0194] It should be noted that the specific working process of each unit provided in the above embodiments of the present application can refer to the implementation process of the corresponding steps provided in the above method embodiments, and will not be repeated here.

[0195] Another embodiment of the present application provides an electronic device, such as Figure 6 As shown, including:

[0196] Memory 601 and processor 602 .

[0197] The memory 601 is used to store programs.

[0198] The processor 602 is used to execute the program stored in the memory 601. When the program is executed, it is specifically used to implement the government affairs industry budget allocation method based on a large model as provided in any one of the above embodiments.

[0199] Another embodiment of the present application provides a computer storage medium for storing a computer program. When the computer program is executed by a processor, it is used to implement the big model-based government industry budget allocation method provided in any of the above embodiments.

[0200] Computer storage media includes permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0201] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0202] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A budget allocation method for government affairs industry based on a large model, characterized in that: include: Obtain budget allocation related data; Preprocessing each item of budget allocation related data respectively; Extracting features from each item of budget allocation related data to obtain current feature data; Using the current feature data and the preset framework, generating a current budget allocation plan according to the preset allocation principle and allocation algorithm; Performing multi-objective optimization on the current budget allocation scheme through a multi-objective optimization model to obtain a current optimized budget allocation scheme; wherein the multi-objective optimization model is obtained by combining a radial basis function network and a deep reinforcement learning network; Evaluate whether the current optimized budget allocation plan meets the project objectives; If the current optimized budget allocation scheme does not meet the project goal, the current optimized preset allocation scheme is redesigned to generate a new current budget allocation scheme, and the method returns to execute the multi-objective optimization of the current budget allocation scheme through the multi-objective optimization model to obtain the current optimized budget allocation scheme; If the current optimized budget allocation scheme meets the project goal, then the execution data of the current optimized budget allocation scheme is collected in real time; Calling multiple evaluation models, using the execution data of the current optimized budget allocation scheme to perform evaluation, and obtaining a current execution evaluation result; The current optimized budget allocation scheme is adjusted based on the current execution evaluation result.

2. The method according to claim 1, characterized in that The preprocessing of each item of the budget allocation associated data respectively includes: The data modes of each item of the budget allocation associated data are respectively identified, and the budget allocation associated data are preprocessed using a preprocessing method corresponding to the data mode of the budget allocation associated data.

3. The method according to claim 1, characterized in that The feature extraction of each item of the budget allocation associated data to obtain current feature data includes: Identify the patterns of each item of the budget allocation associated data, and use the extraction method corresponding to the pattern of each item of the budget allocation associated data to extract features of each item of the budget allocation associated data to obtain current feature information; Using the relationship between the various pieces of current feature information, construct a current data structure diagram; Analyze the current data structure graph using a graph neural network to obtain feature relationship information; The current feature information and the feature relationship information are combined into the current feature data.

4. The method according to claim 1, characterized in that The using the current feature data and the preset framework to generate the current budget allocation scheme according to the preset allocation principle and allocation algorithm includes: Predict future budget needs and trends based on the time series data composed of the current feature data using the Prophet model; Based on the future budget requirements and trends, the allocation algorithm is called to calculate the budget allocation of each project that meets the allocation principle according to the preset framework; A current budget allocation plan is generated according to the budget allocation of each of the projects.

5. The method according to claim 1, characterized in that Before the real-time collection of the execution data of the current optimized budget allocation scheme, the method further includes: Performing a risk assessment on the current optimized budget allocation scheme to obtain a risk level of the current optimized budget allocation scheme; Determining whether the risk level of the current optimized budget allocation scheme exceeds a risk threshold; If it is determined that the risk level of the current optimized budget allocation scheme exceeds the risk threshold, returning to execute the redesign of the current optimized preset allocation scheme; If it is determined that the risk level of the current optimized budget allocation plan exceeds the risk threshold, the current optimized budget allocation plan is subject to an approval process; Determine whether the current optimized budget allocation plan has been approved; If it is determined that the current optimized budget allocation scheme has passed the approval, the current optimized budget allocation scheme is displayed, and the execution data of the current optimized budget allocation scheme is collected in real time; If it is determined that the current optimized budget allocation plan has not passed the approval, the process returns to the step of redesigning the current optimized preset allocation plan.

6. The method according to claim 1, characterized in that The calling of multiple evaluation models and using the execution data of the current optimized budget allocation scheme to perform evaluation to obtain the current execution evaluation result includes: Performing real-time processing and analysis on the real-time data in the execution data of the current optimized budget allocation scheme through the edge computing node, and performing batch processing on the batch data in the execution data of the current optimized budget allocation scheme through the central data center to obtain effect evaluation indicators; The causal inference model, the statistical model and the machine learning evaluation model are respectively called, and the effect evaluation index is used to perform evaluation to obtain the current execution evaluation result.

7. The method according to claim 1, characterized in that The adjusting the current optimization budget allocation scheme based on the current execution evaluation result includes: Based on the current execution evaluation result, a solution optimization suggestion is generated through the optimization model of federated learning, and the current optimization budget allocation solution is adjusted using the solution optimization suggestion.

8. A budget allocation device for government affairs industry based on a large model, characterized in that: include: A data acquisition unit, used to acquire budget allocation related data; A preprocessing unit, used for preprocessing each item of the budget allocation associated data; A feature extraction unit, used to extract features from each item of the budget allocation associated data to obtain current feature data; A scheme generating unit, used to generate a current budget allocation scheme according to a preset allocation principle and allocation algorithm by using the current feature data and a preset framework; An optimization unit, configured to perform multi-objective optimization on the current budget allocation scheme through a multi-objective optimization model to obtain a current optimized budget allocation scheme; wherein the multi-objective optimization model is obtained by combining a radial basis function network and a deep reinforcement learning network; A target evaluation unit, used to evaluate whether the current optimized budget allocation scheme meets the project target; A scheme updating unit, configured to redesign the current optimized preset allocation scheme when the current optimized budget allocation scheme does not meet the project goal, generate a new current budget allocation scheme, and return it to the optimization unit; A data collection unit, configured to collect execution data of the current optimized budget allocation scheme in real time when the current optimized budget allocation scheme meets the project goal; A result evaluation unit, used to call multiple evaluation models, use the execution data of the current optimized budget allocation scheme to perform evaluation, and obtain a current execution evaluation result; An adjusting unit is used to adjust the current optimized budget allocation scheme based on the current execution evaluation result.

9. An electronic device, characterized in that: include: Memory and processor; Wherein, the memory is used to store programs; The processor is used to execute the program, and when the program is executed, it is specifically used to implement the government affairs industry budget allocation method based on a large model as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: Used to store computer programs, which, when executed by a processor, are used to implement the large model-based government affairs industry budget allocation method as described in any one of claims 1 to 7.

Images