Water conservancy project budget management and control system based on contract quantity-price linkage

Abstract

The invention discloses a water conservancy project budget management and control system based on contract quantity-price linkage, and the system comprises a contract data collection and preprocessing module which is used for inputting original contract data of a water conservancy project and constructing a preprocessed contract data set; the budget reasoning model construction module is used for constructing a budget reasoning model; the contract change real-time monitoring module is used for collecting information in real time and generating a structured contract change data set; the budget reasoning execution module is used for outputting a preliminary budget reasoning result; the budget model optimization module is used for optimizing the budget reasoning model by adopting a cuckoo search algorithm; the management and control strategy generation module is used for generating a budget management and control strategy; and the control strategy execution module is used for executing the budget control strategy. According to the method, a budget reasoning and control process combining dynamic Bayesian network modeling and a cuckoo search optimization algorithm is adopted, and dynamic generation and execution of a budget management and control strategy are achieved.

Description

Technical Field

[0001] The present invention relates to the field of engineering project management and intelligent decision-making technology, and in particular to a water conservancy project budget management and control system based on contract quantity-price linkage. Background Art

[0002] With the continuous expansion of water conservancy project scale and the continuous improvement of project management refinement, the demand for intelligent and dynamic management and control of project budgets, a key component of controlling project costs, evaluating project economic benefits, and guiding project execution progress, is growing. In traditional water conservancy project construction, budget management typically relies on static approval and periodic manual adjustments. This approach has limited responsiveness to actual project volume, price fluctuations, and unforeseen events during contract execution, often leading to budget overruns, imbalanced funding allocation, and delayed risk warnings. This static budgeting mechanism no longer meets the current high-level requirements of project management for "dynamic perception, intelligent prediction, and real-time adjustment."

[0003] Currently, the industry has proposed a variety of supporting systems for project budget management, primarily including contract data management systems, project cost estimation platforms, and construction control platforms centered around workload and duration. Most of these systems rely on regular manual entry of contract change information and reflect budget utilization through spreadsheets and reports. Their underlying budget reasoning is often based on linear estimation, empirical rules, or traditional regression analysis. While this budget control approach is somewhat useful for conventional, small-scale projects, it lacks intelligence and responsiveness for modern water conservancy projects, which feature complex data, frequent changes, and densely populated nodes.

[0004] Traditional project budgeting methods mostly treat contract data as static input variables, performing a one-time estimate at the initial budget setting stage. This ignores the impact of project quantity fluctuations, unit price adjustments, and node payment distribution on the dynamic changes in the budget during contract execution. For example, if design changes or material price increases cause significant deviations in project quantity or unit price during the actual construction process within a contract execution cycle, traditional budgeting systems struggle to respond in real time and are often forced to make budget revisions only after periodic manual review. This approach not only has significant time lags but also provides virtually no early warning for budget risk control. Furthermore, traditional systems lack the ability to recognize structural semantic information in contracts, unable to automatically identify logical connections between clauses or construct the complex conditional dependency structures required for budget forecasting.

[0005] On the other hand, while some recent research has introduced data mining and machine learning technologies into the field of engineering budgeting, such as using models like support vector machines and decision trees to predict budget deviations or using neural networks to analyze budget execution trends, these methods are mostly trained on static data and lack the ability to model the dynamic feedback mechanism of "contract changes, budget reasoning, and budget updates." Furthermore, traditional forecasting models generally use fixed structures and static parameters, making them unable to adapt to new changes during project implementation. This leads to easily degraded model performance and severe forecast distortion.

[0006] It is particularly noteworthy that the current industry generally lacks a budget reasoning structure that can link elements such as contract quantities, unit prices, and payment nodes. Existing systems often treat project quantities and unit prices separately, ignoring the linkage effect of "quantity-price coupling" on budget consumption and surplus. In complex water conservancy projects, the nonlinear changes between contract quantities and prices often act together on the budget consumption curve. Existing methods are unable to effectively capture this linkage trend, making it difficult to achieve accurate predictions. In addition, for the identification and early warning of budget risks, traditional systems mostly use static judgment logic based on manually set thresholds. They are unable to adaptively set risk levels based on historical data, nor can they form a closed-loop strategy chain from risk identification to management and control execution.

[0007] Regarding the optimization problem of budget model parameters, although there are studies in the existing technology that use intelligent algorithms such as genetic algorithms and particle swarm optimization to tune the parameters of budget forecast models, most of them still remain at the level of adjusting the weights of neural networks or support vector parameters, and lack strategies for deep optimization of structured reasoning models (such as dynamic Bayesian networks). At the same time, the optimization objectives are mostly single prediction accuracy indicators, and multi-objective trade-off mechanisms such as "minimizing the budget adjustment error rate" and "maximizing the accuracy of budget risk warning" have not yet been introduced, resulting in the optimization effect being difficult to stably migrate in practical applications. In addition, existing algorithms mostly rely on traditional grid search or manual parameter adjustment methods, which cannot cope with the actual situation where the model parameter space is large and the coupling is complex. In particular, in scenarios where it is necessary to take into account both the global convergence speed and the local optimal accuracy, the existing optimization strategies appear to be inefficient and unstable.

[0008] In terms of budget execution and control, current research focuses primarily on visualizing budget execution results or comparing progress, lacking a systematic execution control mechanism. Existing systems are unable to automatically generate executable management and control strategies based on budget reasoning results, such as automatically identifying overspending nodes and setting freeze measures, identifying surplus nodes and recommending budget releases, and other actions. This gap creates a disconnect between budget reasoning results and actual execution behavior, preventing the formation of a complete closed loop of "prediction-recommendation-execution-feedback." Furthermore, the dynamic feedback capability for execution results is also weak. Systems typically lack the ability to reversely adjust budget reasoning model parameters based on execution results, and the model lacks self-learning and self-optimization capabilities, which in turn affects the timeliness and accuracy of budget reasoning results.

[0009] Therefore, how to provide a water conservancy project budget management and control system based on contract quantity and price linkage is an urgent problem that technical personnel in this field need to solve. Summary of the Invention

[0010] One objective of the present invention is to propose a water conservancy project budget management and control system based on the linkage between contract quantity and price. This system fully integrates dynamic Bayesian network modeling, a data-driven reasoning mechanism for contract changes, cuckoo search algorithm optimization, and budget execution control technology. It describes in detail how to implement an intelligent approach for the entire process, including budget reasoning model construction, real-time response to changes, intelligent optimization of model parameters, and generation and closed-loop control of budget adjustment plans. This system offers advantages such as timely budget response, accurate risk identification, closed-loop execution of management and control strategies, and adaptive model optimization. It can significantly improve the dynamics, precision, and intelligence of water conservancy project budget management.

[0011] A water conservancy project budget management and control system based on contract quantity-price linkage according to an embodiment of the present invention includes:

[0012] The contract data collection and preprocessing module is used to input the original contract data of water conservancy projects and construct a preprocessed contract data set;

[0013] A budget reasoning model building module is used to input the pre-processed contract data set into a dynamic Bayesian network to build a budget reasoning model;

[0014] The real-time contract change monitoring module is used to collect contract change information during the execution of water conservancy projects in real time and generate a structured contract change data set;

[0015] The budget reasoning execution module is used to input the structured contract change dataset into the budget reasoning model and output preliminary budget reasoning results;

[0016] The budget model optimization module is used to optimize the budget reasoning model based on the preliminary budget reasoning results using the cuckoo search algorithm;

[0017] A control strategy generation module is used to generate budget control strategies based on the optimized budget inference model;

[0018] The management and control policy execution module is used to receive budget management and control policies, dynamically update the budget execution status, and perform budget supplementation, surplus release, and high-risk node expenditure control operations.

[0019] Optionally, the feature is that the modules are implemented by the following method:

[0020] S1. Input the contract data of the water conservancy project, obtain the contract data set, preprocess the contract data set, and construct the preprocessed contract data set;

[0021] S2. Input the pre-processed contract data set into the dynamic Bayesian network, build a budget reasoning model, set functional nodes, establish conditional dependencies between functional nodes, initialize conditional probability table parameters, and set budget overspending risk warning thresholds;

[0022] S3: Real-time monitoring of contract change information in water conservancy projects, building a structured contract change dataset and inputting it into the budget reasoning model to output preliminary budget reasoning results;

[0023] S4. Based on the preliminary budget inference results, the cuckoo search algorithm is introduced, the search parameters and population are initialized, the optimization objective function is set, the population is updated using the Levy flight mechanism, and the optimized budget inference model is output;

[0024] S5. Perform budget reasoning based on the optimized budget reasoning model to generate a set of budget adjustment plans and a set of budget risk warning information, which are combined into a budget control strategy;

[0025] S6. Dynamically update the budget execution status according to the budget control strategy, and perform budget supplement, surplus release and high-risk node expenditure control operations.

[0026] Optionally, the pre-processed contract data set includes a project quantity field, a project quantity fluctuation field, a unit price formula, a unit price change rate field, a total price formula, a payment node field, and a payment node density field.

[0027] Optionally, the S2 specifically includes:

[0028] S21. Based on the structure of the dynamic Bayesian network, set the engineering quantity node, unit price node, budget consumption node, budget remaining node and risk indicator node to form a functional node set;

[0029] S22. Based on the pre-processed contract data set, the engineering quantity field and the engineering quantity fluctuation rate field are input into the engineering quantity node, the unit price formula and the unit price change rate field are input into the unit price node, the total price formula is input into the budget consumption node, and the payment node field and the payment node density field are input into the budget remaining node;

[0030] S23. Establish conditional dependency relationships between nodes, including establishing that the quantity node and the unit price node are parent nodes of the budget consumption node, that the budget consumption node is the parent node of the budget surplus node, and that the budget surplus node is the parent node of the risk indicator node;

[0031] S24. Configure a preliminary conditional probability table for each node, define the probability of each node taking a value when the parent node takes a different value, and use the maximum likelihood estimation method to preliminarily assign values to the conditional probability table parameters;

[0032] S25. Set a threshold value θ for the budget surplus. If the budget surplus is less than or equal to θ, the risk indicator is determined to be in a high-risk state. If the budget surplus is greater than θ, the risk indicator is determined to be in a low-risk state.

[0033] S25. Combining the functional node set, the conditional dependency relationship between the nodes, the preliminary conditional probability table configured for each node, and the threshold value of the budget remaining amount, to construct a budget reasoning model.

[0034] Optionally, the maximum likelihood estimation method includes, under the condition of given corresponding numerical values of engineering quantity nodes, unit price nodes and payment node density nodes, counting the frequency of budget consumption nodes taking each budget consumption amount to determine the conditional probability distribution of budget consumption nodes; under the condition of the budget consumption nodes taking a specific budget consumption amount, counting the frequency of budget surplus nodes taking each remaining budget amount to determine the conditional probability distribution of budget surplus nodes; under the condition of the budget surplus nodes taking a specific remaining budget amount, counting the frequency of risk indicator nodes taking each risk level to determine the conditional probability distribution of risk indicator nodes.

[0035] Optionally, the S3 specifically includes:

[0036] S31. Real-time collection of contract change information during the execution of water conservancy projects, including fields for changes in engineering quantities, fields for unit price adjustments, information on changes in payment nodes, and information on changes in key construction periods;

[0037] S32. Extract and format the contract change information to construct a structured contract change data set;

[0038] S33. Inputting the engineering quantity change field and the unit price adjustment field in the structured contract change data set into the engineering quantity node and the unit price node in the budget reasoning model respectively;

[0039] S34. Calculate the inference result of the budget consumption node based on the conditional dependency relationship between the engineering quantity node and the unit price node on the budget consumption node in the budget inference model:

[0040]

[0041] Among them, B′ i is the budget consumption forecast value of the i-th budget node after the change, i is the budget node index, n is the total number of engineering quantity and unit price items in the i-th budget node, j is the sequence number of different engineering quantity and unit price items in the i-th budget node, Q ij is the engineering quantity value of item j in the i-th budget node before the change, ΔQ ij is the change in the quantity of the jth item in the i-th budget node, P ij is the unit price of the jth item in the i-th budget node before the change, ΔP ij is the price change of the jth item in the i-th budget node;

[0042] S35. Based on the inference result of the budget consumption node, the inference result of the budget remaining node is calculated. The budget remaining value is the difference between the initial budget total amount and the budget consumption forecast value.

[0043] S36. Based on the inference result of the budget remaining node, the status of the risk indicator node is inferred:

[0044]

[0045] Among them, RiskLevel represents the budget risk indicator status, B r represents the budget remaining forecast value, θ represents the threshold of the budget remaining amount, High represents a high-risk state, and Low represents a low-risk state;

[0046] S37. Based on the reasoning output results of the budget consumption node, the budget remaining node, and the risk indicator node, a preliminary budget reasoning result is formed.

[0047] Optionally, the S4 specifically includes:

[0048] S41. Randomly generate a set of initial candidate solutions as population individuals of the cuckoo search algorithm to form the initial population set P (0) Each candidate solution includes a set of node conditional probability table parameters and a budget risk warning threshold, setting the population size N, the maximum number of iterations G, and the discovery probability p a and step size control coefficient α;

[0049] S42. For each candidate solution, set a dual-objective optimization function for the budget execution error rate and the risk warning accuracy:

[0050]

[0051] Where Fitness(x) represents the fitness value of the candidate solution x, x represents a candidate solution in the cuckoo search algorithm, w1 is the weighted coefficient of the budget adjustment error term, m is the number of budget nodes that need to be predicted in the budget inference model, and i is the index number of the budget prediction node. is the predicted budget consumption amount of the i-th budget node based on the current candidate solution x, is the actual budget consumption amount recorded at the i-th budget node during the execution of the actual project, w2 is the weighted coefficient of the risk warning accuracy term, TP is the number of records where the budget reasoning model correctly predicts the occurrence of budget overrun risks, TN is the number of records where the budget reasoning model correctly predicts that the budget is normal and has no risks, FP is the number of records where the budget reasoning model incorrectly predicts that there are budget overrun risks, and FN is the number of records where the budget reasoning model misses the budget overrun risks.

[0052] S43. For each candidate solution, perform perturbation according to the Levy flight formula to generate a new candidate solution and obtain the newly generated population P. (t+1) :

[0053]

[0054] in, represents the i-th candidate solution after updating in the t+1-th iteration, represents the i-th candidate solution in the t-th iteration, i is the candidate solution number, t is the iteration number of the current cuckoo search, α is the step size control coefficient in the cuckoo search algorithm, Levy(λ) represents the random step size vector obeying the Levy distribution, and λ is the Levy distribution exponent parameter;

[0055] S44, the newly generated population P (t+1) Substitute each individual into the optimization objective function, calculate the corresponding fitness value, and combine the discovery probability to find the probability p a Screen and replace individuals in the population to generate an updated population P (t+1′) ;

[0056] S45, repeating steps S43 and S44 until the maximum number of iterations G is reached, and selecting the candidate solution with the best final fitness;

[0057] S46. Apply the candidate solution with the best fitness to the budget reasoning model, update the conditional probability table parameters and the threshold of the budget remaining amount, and form an optimized budget reasoning model.

[0058] Optionally, the S5 specifically includes:

[0059] S51. Input the structured contract change dataset constructed in real time into the optimized budget reasoning model, and output the budget consumption forecast value, budget remaining forecast value, and forecast risk indicator status of each node;

[0060] S52, based on the budget consumption forecast value, construct a budget adjustment difference vector ΔB={ΔB1, ΔB2,…, ΔB N}, where ΔB i The difference between the budget consumption forecast value and the original budget value;

[0061] S53. Combining the adjustment difference of each budget node with the predicted risk indicator status to form a budget adjustment strategy tuple, and constructing a budget adjustment plan set based on the budget adjustment strategy tuple;

[0062] S54. Determine the budget overspending risk level based on the budget risk indicator status and construct a risk warning information set;

[0063] S55. The budget adjustment plan set and the risk warning information set are combined and output as a budget control strategy.

[0064] Optionally, constructing a set of budget adjustment schemes includes proposing budget adjustment suggestions and generating a set of budget adjustment schemes;

[0065] The budget adjustment suggestion is made to compare the budget consumption forecast value B of each budget node i Compared with the original budget like Identify the overspending node and propose additional budget suggestions. Identify savings points and propose budget adjustment and release suggestions;

[0066] The generating budget adjustment scheme set summarizes the node number, original budget setting value, budget adjustment difference, budget risk indicator status, recommended adjustment type and recommended adjustment amount of each node to form a budget adjustment scheme set.

[0067] Optionally, the construction of the risk warning information set includes determining the budget overrun risk level. If the predicted risk status of the budget node is a high-risk state, the warning information construction is triggered. At each high-risk node, a warning record is generated. The warning record includes the node number, budget remaining forecast value, overrun probability, risk level and recommended risk response measures. All warning records are combined into a risk warning information set.

[0068] The beneficial effects of the present invention are:

[0069] The present invention constructs a water conservancy project budget management and control system based on the linkage between contract quantity and price, establishing an intelligent closed loop for the entire process from contract data collection and preprocessing, budget reasoning model construction, real-time contract change-driven budget reasoning, to budget model parameter optimization, adjustment plan generation, and budget strategy execution. This achieves a deep integration and dynamic response of water conservancy project budget forecasting, adjustment, and risk control. The present invention introduces a dynamic Bayesian network as the core budget reasoning model, systematically modeling the conditional dependency relationship between project quantity, unit price, budget consumption, budget surplus, and risk status. This breaks through the problem of traditional budget static estimation models ignoring data linkage, and can dynamically infer budget node status based on contract data, identifying potential overspending risks in advance.

[0070] In particular, the present invention monitors contract change events during the execution of water conservancy projects in real time, inputs the changed engineering quantities and unit price data into the budget reasoning model, and automatically infers the trend of budget consumption changes, significantly improving the system's ability to respond to sudden changes in the construction process, and ensuring that the budget forecast is real-time and accurate. At the same time, the present invention constructs an optimization function with the dual objectives of minimizing the budget error rate and maximizing the risk warning accuracy, and uses the cuckoo search algorithm and its Levy flight mechanism to optimize the conditional probability table parameters and risk thresholds in the budget reasoning model, thereby improving the model's prediction ability and generalization performance. This optimization process does not rely on manual parameter adjustment, has good convergence speed and global search capabilities, and can quickly obtain a high-performance budget reasoning model under complex coupling of parameter space.

[0071] In terms of system architecture design, this invention achieves the linked output of budget adjustment plans and risk warning information, and further establishes a budget execution control mechanism, implementing budget freezes for high-risk nodes and recommending budget releases for low-risk surplus nodes, thus completing an intelligent closed-loop from budget reasoning to execution control. Furthermore, the budget execution status is synchronized in real time to the model feedback module, providing a data foundation for subsequent model structure adjustments and risk threshold adaptation, thus forming a four-dimensional budget closed-loop management mechanism of "reasoning-optimization-execution-feedback."

[0072] This system not only addresses existing issues such as coarse budget modeling, delayed adjustments, weak optimization capabilities, and execution gaps, but also improves the accuracy, efficiency, and intelligence of budget control, offering high scalability and engineering practicality. Compared to traditional budget systems based on rule bases or static regression models, this invention achieves substantial improvements in dynamic response capabilities, risk identification accuracy, and closed-loop control strategy execution, significantly enhancing the automation and intelligent control capabilities of water conservancy project budget management.

[0073] In summary, the present invention achieves a coordinated improvement in the prediction accuracy, risk control capability and execution closed-loop level of water conservancy project budget management and control by integrating contract data structure modeling, dynamic Bayesian reasoning mechanism, intelligent optimization algorithm and execution strategy generation logic. It has the beneficial effects of data-driven, model adaptation, accurate prediction, fast response and complete system closed-loop, providing practical technical support and theoretical basis for the intelligent budget management and control of complex engineering projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0075] Figure 1 This is a method flow chart of a water conservancy project budget control system based on contract quantity and price linkage proposed by the present invention;

[0076] Figure 2 This is a system diagram of a water conservancy project budget control system based on contract quantity and price linkage proposed by the present invention;

[0077] Figure 3 This is a cuckoo search optimization flow chart in a water conservancy project budget management and control system based on contract quantity and price linkage proposed by the present invention. DETAILED DESCRIPTION

[0078] The present invention will now be described in further detail with reference to the accompanying drawings, which are simplified schematic diagrams that illustrate the basic structure of the present invention in a schematic manner.

[0079] refer to Figure 1-3 , a water conservancy project budget control system based on contract quantity and price linkage, including:

[0080] The contract data collection and preprocessing module is used to input the original contract data of water conservancy projects and construct a preprocessed contract data set;

[0081] A budget reasoning model building module is used to input the pre-processed contract data set into a dynamic Bayesian network to build a budget reasoning model;

[0082] The real-time contract change monitoring module is used to collect contract change information during the execution of water conservancy projects in real time and generate a structured contract change data set;

[0083] The budget reasoning execution module is used to input the structured contract change dataset into the budget reasoning model and output preliminary budget reasoning results;

[0084] The budget model optimization module is used to optimize the budget reasoning model based on the preliminary budget reasoning results using the cuckoo search algorithm;

[0085] A control strategy generation module is used to generate budget control strategies based on the optimized budget inference model;

[0086] The management and control policy execution module is used to receive budget management and control policies, dynamically update the budget execution status, and perform budget supplementation, surplus release, and high-risk node expenditure control operations.

[0087] The present invention provides a water conservancy project budget management and control system based on the linkage between contract quantity and price, which integrates multiple functional modules such as contract data collection and preprocessing, budget reasoning model construction, real-time monitoring of contract changes, budget reasoning execution, budget model optimization, and management and control strategy generation and execution, thereby realizing dynamic, intelligent and closed-loop budget management. The system constructs a budget reasoning model through standardized analysis and dynamic modeling of water conservancy project contract data, accesses contract change events in real time, and automatically infers budget consumption and risk status. The cuckoo search algorithm is introduced to optimize model parameters and risk thresholds, thereby improving budget forecast accuracy and risk identification capabilities. Finally, a budget management and control strategy is generated and freezing, supplementing and saving release operations are executed. The system has the advantages of high prediction accuracy, fast response speed, accurate risk identification and strong strategy execution, which significantly improves the budget control efficiency and management intelligence level of water conservancy projects.

[0088] In this embodiment, the modules are connected by the following method:

[0089] S1. Input the contract data of the water conservancy project, obtain the contract data set, preprocess the contract data set, and construct the preprocessed contract data set;

[0090] S2. Input the pre-processed contract data set into the dynamic Bayesian network, build a budget reasoning model, set functional nodes, establish conditional dependencies between functional nodes, initialize conditional probability table parameters, and set budget overspending risk warning thresholds;

[0091] S3: Real-time monitoring of contract change information in water conservancy projects, building a structured contract change dataset and inputting it into the budget reasoning model to output preliminary budget reasoning results;

[0092] S4. Based on the preliminary budget inference results, the cuckoo search algorithm is introduced, the search parameters and population are initialized, the optimization objective function is set, the population is updated using the Levy flight mechanism, and the optimized budget inference model is output;

[0093] S5. Perform budget reasoning based on the optimized budget reasoning model to generate a set of budget adjustment plans and a set of budget risk warning information, which are combined into a budget control strategy;

[0094] S6. Dynamically update the budget execution status according to the budget control strategy, and perform budget supplement, surplus release and high-risk node expenditure control operations.

[0095] This invention provides a water conservancy project budget management and control method based on contract quantity-price linkage. Combining dynamic Bayesian network modeling with a cuckoo search optimization algorithm, this method achieves intelligent and closed-loop budget forecasting and risk management through multiple steps, including contract data collection and preprocessing, budget model construction, real-time contract change monitoring, inference execution, model optimization, and strategy generation and dynamic execution. The method first standardizes and preprocesses the raw contract data to construct a structured dataset. This input is a budget inference model, and functional nodes and dependencies are set. When contract changes occur, the model inference is driven in real time, outputting budget consumption forecasts and risk indicators. Model parameters are optimized using a cuckoo search algorithm, significantly improving forecast accuracy and risk identification accuracy. Finally, a budget adjustment plan and risk warning information are generated, and a budget management and control strategy is formulated to dynamically control budget execution status. This method boasts high forecast accuracy, fast response, strong adaptability, and high execution efficiency. It can effectively enhance the budget control capabilities and intelligent management level of complex water conservancy projects, and is suitable for the refined control needs of projects with fluctuating budgets and frequent changes.

[0096] In this embodiment, the pre-processed contract data set includes a project quantity field, a project quantity fluctuation rate field, a unit price formula, a unit price change rate field, a total price formula, a payment node field, and a payment node density field.

[0097] This paper constructs a preprocessed contract dataset containing fields for project quantity, project quantity volatility, unit price formula, unit price change rate, total price formula, payment node field, and payment node density field, enabling structured modeling and dynamic expression of contract quantity and price information. This dataset, used as input, significantly improves the budget reasoning model's responsiveness to actual changes and its forecast accuracy, providing high-quality foundational data support for subsequent budget optimization and risk control, and enhancing the system's adaptability and intelligence.

[0098] In this embodiment, S2 specifically includes:

[0099] S21. Based on the structure of the dynamic Bayesian network, set the engineering quantity node, unit price node, budget consumption node, budget remaining node and risk indicator node to form a functional node set;

[0100] S22. Based on the pre-processed contract data set, the engineering quantity field and the engineering quantity fluctuation rate field are input into the engineering quantity node, the unit price formula and the unit price change rate field are input into the unit price node, the total price formula is input into the budget consumption node, and the payment node field and the payment node density field are input into the budget remaining node;

[0101] S23. Establish conditional dependency relationships between nodes, including establishing that the quantity node and the unit price node are parent nodes of the budget consumption node, that the budget consumption node is the parent node of the budget surplus node, and that the budget surplus node is the parent node of the risk indicator node;

[0102] S24. Configure a preliminary conditional probability table for each node, define the probability of each node taking a value when the parent node takes a different value, and use the maximum likelihood estimation method to preliminarily assign values to the conditional probability table parameters;

[0103] S25. Set a threshold value θ for the budget surplus. If the budget surplus is less than or equal to θ, the risk indicator is determined to be in a high-risk state. If the budget surplus is greater than θ, the risk indicator is determined to be in a low-risk state.

[0104] S25. Combining the functional node set, the conditional dependency relationship between the nodes, the preliminary conditional probability table configured for each node, and the threshold value of the budget remaining amount, to construct a budget reasoning model.

[0105] The present invention constructs a budget reasoning model based on a dynamic Bayesian network, sets five functional nodes: engineering quantity, unit price, budget consumption, budget surplus and risk indicators, establishes a node input mapping relationship in combination with the key fields in the contract data, and clarifies the conditional dependency structure between nodes, thereby realizing the modeling expression of the budget quantity-price linkage relationship. The conditional probability table is preliminarily assigned using the maximum likelihood estimation method to ensure that the model has data-driven reasoning capabilities. The budget surplus amount threshold is introduced as a basis for risk judgment, so that the model has the function of risk status identification. The model can dynamically infer budget consumption and risk level based on contract data, has good scalability and real-time performance, provides core technical support for subsequent budget optimization and execution control, and significantly improves budget forecast accuracy and risk identification accuracy.

[0106] In this embodiment, the maximum likelihood estimation method includes, under the condition of given corresponding numerical values of engineering quantity nodes, unit price nodes and payment node density nodes, counting the frequency of budget consumption nodes taking each budget consumption amount to determine the conditional probability distribution of budget consumption nodes; under the condition of budget consumption nodes taking a specific budget consumption amount, counting the frequency of budget surplus nodes taking each remaining budget amount to determine the conditional probability distribution of budget surplus nodes; under the condition of budget surplus nodes taking a specific remaining budget amount, counting the frequency of risk indicator nodes taking each risk level to determine the conditional probability distribution of risk indicator nodes.

[0107] This paper uses maximum likelihood estimation to assign values to the conditional probability table in the budget inference model. Using historical data to calculate the frequency of child node values under specific parent node conditions, it establishes conditional probability distributions for budget consumption, budget surplus, and risk level. This method effectively enhances the authenticity of model parameters and data adaptability, significantly improving the accuracy and credibility of budget inference results and providing a stable data foundation for subsequent optimization and risk assessment.

[0108] In this embodiment, S3 specifically includes:

[0109] S31. Real-time collection of contract change information during the execution of water conservancy projects, including fields for changes in engineering quantities, fields for unit price adjustments, information on changes in payment nodes, and information on changes in key construction periods;

[0110] S32. Extract and format the contract change information to construct a structured contract change data set;

[0111] S33. Inputting the engineering quantity change field and the unit price adjustment field in the structured contract change data set into the engineering quantity node and the unit price node in the budget reasoning model respectively;

[0112] S34. Calculate the inference result of the budget consumption node based on the conditional dependency relationship between the engineering quantity node and the unit price node on the budget consumption node in the budget inference model:

[0113]

[0114] Among them, B′ i is the budget consumption forecast value of the i-th budget node after the change, i is the budget node index, n is the total number of engineering quantity and unit price items in the i-th budget node, j is the sequence number of different engineering quantity and unit price items in the i-th budget node, Q ij is the engineering quantity value of item j in the i-th budget node before the change, ΔQ ij is the change in the quantity of the jth item in the i-th budget node, P ij is the unit price of the jth item in the i-th budget node before the change, ΔP ij is the price change of the jth item in the i-th budget node;

[0115] S35. Based on the inference result of the budget consumption node, the inference result of the budget remaining node is calculated. The budget remaining value is the difference between the initial budget total amount and the budget consumption forecast value.

[0116] S36. Based on the inference result of the budget remaining node, the status of the risk indicator node is inferred:

[0117]

[0118] Among them, RiskLevel represents the budget risk indicator status, B r represents the budget remaining forecast value, θ represents the threshold of the budget remaining amount, High represents a high-risk state, and Low represents a low-risk state;

[0119] S37. Based on the reasoning output results of the budget consumption node, the budget remaining node, and the risk indicator node, a preliminary budget reasoning result is formed.

[0120] The present invention realizes the real-time collection and structured processing of change information such as engineering quantity, unit price, payment node and key construction period during the execution of water conservancy project by constructing a budget reasoning mechanism driven by contract changes, and inputs it into the budget reasoning model for dynamic budget forecasting. Combined with the conditional dependency of engineering quantity and unit price nodes on budget consumption nodes, the system can accurately calculate the budget consumption forecast value, and deduce the budget surplus and risk indicator status to form a complete preliminary budget reasoning result. The threshold judgment mechanism is introduced to automatically determine the risk level, which improves the sensitivity and recognition accuracy of overspending risk. This method realizes the dynamic and intelligent budget response, can reflect the impact of changes on the budget status in real time, significantly improves the timeliness of budget adjustment and the pre-emptive nature of risk control, and provides strong data support and reasoning basis for the financial coordination and cost management of engineering projects.

[0121] In this embodiment, the S4 specifically includes:

[0122] S41. Randomly generate a set of initial candidate solutions as population individuals of the cuckoo search algorithm to form the initial population set P (0) Each candidate solution includes a set of node conditional probability table parameters and a budget risk warning threshold, setting the population size N, the maximum number of iterations G, and the discovery probability p a and step size control coefficient α;

[0123] S42. For each candidate solution, set a dual-objective optimization function for the budget execution error rate and the risk warning accuracy:

[0124]

[0125] Where Fitness(x) represents the fitness value of the candidate solution x, x represents a candidate solution in the cuckoo search algorithm, w1 is the weighted coefficient of the budget adjustment error term, m is the number of budget nodes that need to be predicted in the budget inference model, and i is the index number of the budget prediction node. is the predicted budget consumption amount of the i-th budget node based on the current candidate solution x, is the actual budget consumption amount recorded at the i-th budget node during the execution of the actual project, w2 is the weighted coefficient of the risk warning accuracy term, TP is the number of records where the budget reasoning model correctly predicts the occurrence of budget overrun risks, TN is the number of records where the budget reasoning model correctly predicts that the budget is normal and has no risks, FP is the number of records where the budget reasoning model incorrectly predicts that there are budget overrun risks, and FN is the number of records where the budget reasoning model misses the budget overrun risks.

[0126] S43. For each candidate solution, perform perturbation according to the Levy flight formula to generate a new candidate solution and obtain the newly generated population P. (t+1) :

[0127]

[0128] in, represents the i-th candidate solution after updating in the t+1-th iteration, represents the i-th candidate solution in the t-th iteration, i is the candidate solution number, t is the iteration number of the current cuckoo search, α is the step size control coefficient in the cuckoo search algorithm, Levy(λ) represents the random step size vector obeying the Levy distribution, and λ is the Levy distribution exponent parameter;

[0129] S44, the newly generated population P (t+1) Substitute each individual into the optimization objective function, calculate the corresponding fitness value, and combine the discovery probability to find the probability p a Screen and replace individuals in the population to generate an updated population P (t+1′) ;

[0130] S45, repeating steps S43 and S44 until the maximum number of iterations G is reached, and selecting the candidate solution with the best final fitness;

[0131] S46. Apply the candidate solution with the best fitness to the budget reasoning model, update the conditional probability table parameters and the threshold of the budget remaining amount, and form an optimized budget reasoning model.

[0132] The present invention optimizes the parameters of the budget reasoning model by introducing the cuckoo search algorithm, constructs a multidimensional candidate solution population containing conditional probability table parameters and budget risk warning thresholds, and sets a dual-objective optimization function with the goal of minimizing the budget execution error rate and maximizing the risk warning accuracy, thereby achieving comprehensive optimization of the budget model performance. The Levy flight mechanism is used to perturb and iteratively update the candidate solutions, and the discovery probability strategy is combined to screen high-quality individuals, continuously evolve the population, and finally obtain the optimal model parameter solution. This optimization process can effectively search for the optimal solution in a high-dimensional complex parameter space, significantly improving the accuracy of budget forecasting and the robustness of risk identification. The optimized budget reasoning model has stronger data adaptability and generalization capabilities, and can stably output high-precision budget results in variable contract scenarios, enhancing the scientificity and intelligence of budget management.

[0133] In this embodiment, the S5 specifically includes:

[0134] S51. Input the structured contract change dataset constructed in real time into the optimized budget reasoning model, and output the budget consumption forecast value, budget remaining forecast value, and forecast risk indicator status of each node;

[0135] S52, based on the budget consumption forecast value, construct a budget adjustment difference vector ΔB={ΔB1, ΔB2,…, ΔB N}, where ΔB i The difference between the budget consumption forecast value and the original budget value;

[0136] S53. Combining the adjustment difference of each budget node with the predicted risk indicator status to form a budget adjustment strategy tuple, and constructing a budget adjustment plan set based on the budget adjustment strategy tuple;

[0137] S54. Determine the budget overspending risk level based on the budget risk indicator status and construct a risk warning information set;

[0138] S55. The budget adjustment plan set and the risk warning information set are combined and output as a budget control strategy.

[0139] The present invention inputs structured contract change data into an optimized budget reasoning model, outputs budget consumption, budget surplus and risk status prediction results, and then constructs a budget adjustment difference vector, and generates a budget adjustment strategy tuple in combination with the risk indicator status, and systematically forms a set of budget adjustment plans. At the same time, the system automatically determines the budget overrun level based on the risk status, constructs a risk warning information set, and merges it with the budget adjustment plan to output a budget management and control strategy, realizing the integrated linkage of budget response and risk control. This method has efficient, intelligent and structured budget adjustment capabilities, and can quickly generate difference compensation suggestions and risk warning measures in engineering scenarios with frequent budget changes, greatly improving the automation and real-time response capabilities of budget control, and ensuring the continuity of budget execution and the scientific use of funds.

[0140] In this embodiment, the building of the budget adjustment scheme set includes proposing a budget adjustment suggestion and generating a budget adjustment scheme set;

[0141] The budget adjustment suggestion is made to compare the budget consumption forecast value B of each budget node i Compared with the original budget like Identify the overspending node and propose additional budget suggestions. Identify savings points and propose budget adjustment and release suggestions;

[0142] The generating budget adjustment scheme set summarizes the node number, original budget setting value, budget adjustment difference, budget risk indicator status, recommended adjustment type and recommended adjustment amount of each node to form a budget adjustment scheme set.

[0143] This invention automatically identifies and categorizes budget node overruns and underruns by constructing a set of budget adjustment plans. The system compares budget consumption forecasts with original budget values, identifies overrun nodes, and proposes supplementary recommendations. It also identifies underrun nodes and proposes release recommendations. The system then aggregates the node number, original budget value, adjustment difference, risk status, and recommendation type to form a structured budget adjustment plan set. This method improves the intelligence and standardization of the budget adjustment process, enhancing the targeted nature and efficiency of budget decision-making.

[0144] In this embodiment, the construction of the risk warning information set includes determining the budget overrun risk level. If the predicted risk status of the budget node is a high-risk state, the warning information construction is triggered. At each high-risk node, a warning record is generated. The warning record includes the node number, budget remaining forecast value, overrun probability, risk level and recommended risk response measures. All warning records are combined into a risk warning information set.

[0145] By constructing a risk warning information set, this invention enables automatic identification and early warning response to high-risk budget nodes. The system determines the budget overrun risk level and generates early warning records for high-risk nodes, including the node number, predicted budget remaining value, overrun probability, risk level, and response recommendations. These records are then aggregated into a risk warning information set. This method significantly improves the real-time and targeted nature of budget risk monitoring, provides managers with clear early warning evidence and recommended measures, and enhances project budget security management and control capabilities.

[0146] Example 1:

[0147] To verify the feasibility of this invention, we applied it to a large-scale water conservancy infrastructure expansion project. With a total budget exceeding 200 million yuan, the project encompassed multiple sub-projects, including dam reinforcement, pump station reconstruction, gate control system upgrades, and water transmission line reinforcement and repair. The project involved a long construction period, numerous contract management items, and frequent changes. This represented a typical example of a water conservancy project prone to budgetary uncontrol and delayed funding responses.

[0148] In the early stages of project implementation, traditional budget control methods mainly relied on periodic manual summaries and financial audits, lacking a systematic budget forecasting model and real-time risk warning mechanism. This resulted in an average budget adjustment response time of more than 4 days. Budgets at some nodes were not adjusted in a timely manner due to changes, resulting in periodic funding breakpoints, which seriously affected the construction progress.

[0149] The system, deployed within the project's construction management platform, first analyzes the original contract structure using the contract data collection and preprocessing module, combining the bidding contracts, billing data, and supplementary agreements for each sub-bid. The system extracts key content, such as quantity fields, unit price formulas, payment node conditions, clause change conditions, and risk clauses, and then standardizes them to generate a structured, preprocessed contract dataset covering over 60 contracts and a total of more than 300 budget nodes.

[0150] The system then uses the budget reasoning model construction module to construct a budget reasoning model based on a dynamic Bayesian network. It defines five functional nodes: project quantity, unit price, budget consumption, budget surplus, and budget risk. It then initializes and assigns values to the conditional probability table based on statistical data from historical projects. Once the model is complete, it connects to the real-time contract change data stream, including project quantity adjustment records, unit price fluctuation information, and node change approval forms. The contract change monitoring module identifies these data and triggers reasoning updates.

[0151] For example, in a slope support project in a certain construction section, due to changes in geological conditions, the design unit adjusted the support thickness and structural form, resulting in the original project volume being increased from 3,800 cubic meters to 4,600 cubic meters. After the system automatically captured the change event, the budget reasoning execution module re-reasoned the corresponding node, and the budget consumption forecast was increased from 646,000 yuan to 782,000 yuan. The system identified that the budget remaining of the node was less than 15% of the contract set threshold, automatically marked it as a high-risk status and pushed it to the budget risk list.

[0152] Next, the system initiated the Cuckoo Search algorithm optimization process through the budget model optimization module. The optimization objectives were to minimize the budget error rate and maximize the risk warning accuracy. The algorithm employed a Levy flight mechanism to search within the parameter space. After 20 iterations, the average prediction error rate of the optimized budget inference model decreased from the initial 9.3% to 3.4%, while the risk warning accuracy increased to over 92%.

[0153] Based on the optimization model, the system generates budget adjustment plans and risk strategy recommendations. For example, when a construction delay on an irrigation and drainage project resulted in a compressed phase payment schedule, the system recommended releasing 700,000 yuan in surplus funds from low-risk phases and transferring them to newly identified high-risk phases to ensure capital flow continuity. The resulting structured adjustment plan is directly transmitted to the budget control strategy execution module, enabling the coordinated execution of various strategies, including freezing, supplementing, and releasing.

[0154] After deployment, the project identified 38 contract changes during the budget execution cycle, generated 18 dynamic budget adjustment recommendations, and successfully avoided four potential budget breakpoint risk events. After implementation, the system reduced average budget response time from 4.3 days to less than 1.2 days, and the monthly funding disbursement margin was reduced from ±8.1% to within ±2.6%, significantly improving the project's funding efficiency and risk control capabilities.

[0155] To demonstrate the budget control effect of the system of the present invention at typical nodes, the following is a comparison of the prediction deviation and risk identification results before and after optimization of some nodes:

[0156] Table 1 Comparison of prediction deviation and risk judgment before and after optimization of typical budget nodes

[0157]

[0158]

[0159] As can be seen from the above table, the present invention has achieved significant optimization in multiple key business indicators of water conservancy project budget forecasting and risk control, especially in budget error control, risk level determination accuracy, and budget management and control responsiveness.

[0160] Specifically, in terms of budget forecast error, the system of the present invention effectively improves the forecast accuracy of budget nodes by introducing dynamic Bayesian network models and cuckoo search optimization algorithms. Taking the typical nodes in the table as an example, the budget error rate before optimization is generally between 6% and 13%, among which the error rate of the "slope support project" node is as high as 12.9%, and the "concrete foundation reinforcement" node also reaches 11.1%, which poses a great interference to budget execution control. However, after the model is optimized by the system of the present invention, the forecast error is significantly reduced, and the error of multiple nodes is controlled within 3%. The nodes such as "water pipe installation" are even reduced to below 1.9%. The overall budget forecast error is reduced by an average of more than 60%, which significantly improves the accuracy and stability of fund use.

[0161] In terms of budget risk level determination, the risk identification mechanism based on dynamic reasoning of the present invention also shows stronger sensitivity and judgment ability. Before optimization, there were many cases where the risk level determination was ambiguous. For example, the "slope support project" and "concrete foundation reinforcement" nodes were determined to be medium risk before optimization. However, after optimization, based on the comparison between the budget remaining value and the threshold, the system accurately identified them as high risk and intervened in control in advance. After optimization, the "construction temporary drainage system" node was downgraded from high risk to low risk, reflecting that the system can not only identify risks, but also dynamically correct excessive warnings, thereby enhancing the accuracy and sensitivity of risk identification.

[0162] In terms of actual budget control effects, the budget adjustment suggestions and risk warning strategies generated by the present invention are used to drive the automated implementation of the budget execution module, including operations such as budget additions, freezing, and releasing savings. Through node-level differential value analysis and risk status identification, the strategy adjustments proposed by the system effectively alleviated the pressure of fund scheduling in actual projects. Data shows that after the implementation of the solution of the present invention, the monthly fund allocation error was significantly narrowed from the original ±8.1% to ±2.6%, and the budget adjustment response time was shortened from an average of 4.3 days to less than 1.2 days, and the overall budget control efficiency was improved by more than 70%.

[0163] Furthermore, system operational data shows that within a complete monitoring cycle, the system successfully identified 38 contract changes and dynamically generated 18 budget adjustment suggestions, accurately avoiding four potential construction interruptions at key points due to budget shortfalls. The accuracy of risk assessment increased to over 92%, further demonstrating the system's adaptability and practicality in environments with frequent budget fluctuations and complex project changes.

[0164] Comprehensive analysis shows that this method, by constructing a quantity-price-linked budget inference structure, employing a dynamic Bayesian modeling mechanism, and introducing an optimization algorithm for model parameter adjustment, not only improves the accuracy and real-time performance of budget forecasts but also achieves an intelligent closed-loop execution of budget strategies. Compared to traditional budget management methods that rely on static templates and manual judgment, this method achieves systematic improvements in forecast accuracy, risk identification, response speed, and execution efficiency, validating its feasibility and advancement in application to complex water conservancy projects.

[0165] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A water conservancy project budget management and control system based on contract quantity and price linkage, including: The contract data collection and preprocessing module is used to input the original contract data of water conservancy projects and construct a preprocessed contract data set; A budget reasoning model building module is used to input the pre-processed contract data set into a dynamic Bayesian network to build a budget reasoning model; The real-time contract change monitoring module is used to collect contract change information during the execution of water conservancy projects in real time and generate a structured contract change data set; The budget reasoning execution module is used to input the structured contract change dataset into the budget reasoning model and output preliminary budget reasoning results; The budget model optimization module is used to optimize the budget reasoning model based on the preliminary budget reasoning results using the cuckoo search algorithm; A control strategy generation module is used to generate budget control strategies based on the optimized budget inference model; The management and control policy execution module is used to receive budget management and control policies, dynamically update the budget execution status, and perform budget supplementation, surplus release, and high-risk node expenditure control operations.

2. A water conservancy project budget management and control system based on contract quantity and price linkage according to claim 1, characterized in that: The modules are implemented as follows: S1. Input the contract data of the water conservancy project, obtain the contract data set, preprocess the contract data set, and construct the preprocessed contract data set; S2. Input the pre-processed contract data set into the dynamic Bayesian network, build a budget reasoning model, set functional nodes, establish conditional dependencies between functional nodes, initialize conditional probability table parameters, and set budget overspending risk warning thresholds; S3: Real-time monitoring of contract change information in water conservancy projects, building a structured contract change dataset and inputting it into the budget reasoning model to output preliminary budget reasoning results; S4. Based on the preliminary budget inference results, the cuckoo search algorithm is introduced, the search parameters and population are initialized, the optimization objective function is set, the population is updated using the Levy flight mechanism, and the optimized budget inference model is output; S5. Perform budget reasoning based on the optimized budget reasoning model to generate a set of budget adjustment plans and a set of budget risk warning information, which are combined into a budget control strategy; S6. Dynamically update the budget execution status according to the budget control strategy, and perform budget supplement, surplus release and high-risk node expenditure control operations.

3. A water conservancy project budget control system based on contract quantity and price linkage according to claim 2, characterized in that: The pre-processed contract data set includes a project quantity field, a project quantity fluctuation rate field, a unit price formula, a unit price change rate field, a total price formula, a payment node field, and a payment node density field.

4. A water conservancy project budget control system based on contract quantity and price linkage according to claim 2, characterized in that: The S2 specifically includes: S21. Based on the structure of the dynamic Bayesian network, set the engineering quantity node, unit price node, budget consumption node, budget remaining node and risk indicator node to form a functional node set; S22. Based on the pre-processed contract data set, the engineering quantity field and the engineering quantity fluctuation rate field are input into the engineering quantity node, the unit price formula and the unit price change rate field are input into the unit price node, the total price formula is input into the budget consumption node, and the payment node field and the payment node density field are input into the budget remaining node; S23. Establish conditional dependency relationships between nodes, including establishing that the quantity node and the unit price node are parent nodes of the budget consumption node, that the budget consumption node is the parent node of the budget surplus node, and that the budget surplus node is the parent node of the risk indicator node; S24. Configure a preliminary conditional probability table for each node, define the probability of each node taking a value when the parent node takes a different value, and use the maximum likelihood estimation method to preliminarily assign values to the conditional probability table parameters; S25. Set a threshold value θ for the budget surplus. If the budget surplus is less than or equal to θ, the risk indicator is determined to be in a high-risk state. If the budget surplus is greater than θ, the risk indicator is determined to be in a low-risk state. S25. Combining the functional node set, the conditional dependency relationship between the nodes, the preliminary conditional probability table configured for each node, and the threshold value of the budget remaining amount, to construct a budget reasoning model.

5. A water conservancy project budget control system based on contract quantity and price linkage according to claim 4, characterized in that: The maximum likelihood estimation method includes, under the condition that the corresponding values of the engineering quantity node, the unit price node and the payment node density node are given, counting the frequency of the budget consumption node taking each budget consumption amount to determine the conditional probability distribution of the budget consumption node; under the condition that the budget consumption node takes a specific budget consumption amount, counting the frequency of the budget surplus node taking each remaining budget amount to determine the conditional probability distribution of the budget surplus node; Under the condition that the budget remaining node takes a specific remaining budget amount, the frequency of each risk level at the risk indicator node is statistically analyzed to determine the conditional probability distribution of the risk indicator node.

6. A water conservancy project budget control system based on contract quantity and price linkage according to claim 2, characterized in that: The S3 specifically includes: S31. Real-time collection of contract change information during the execution of water conservancy projects, including fields for changes in engineering quantities, fields for unit price adjustments, information on changes in payment nodes, and information on changes in key construction periods; S32. Extract and format the contract change information to construct a structured contract change data set; S33. Inputting the engineering quantity change field and the unit price adjustment field in the structured contract change data set into the engineering quantity node and the unit price node in the budget reasoning model respectively; S34. Calculate the inference result of the budget consumption node based on the conditional dependency relationship between the engineering quantity node and the unit price node on the budget consumption node in the budget inference model: Among them, B′ i is the budget consumption forecast value of the i-th budget node after the change, i is the budget node index, n is the total number of engineering quantity and unit price items in the i-th budget node, j is the sequence number of different engineering quantity and unit price items in the i-th budget node, Q ij is the engineering quantity value of item j in the i-th budget node before the change, ΔQ ij is the change in the quantity of the jth item in the i-th budget node, P ij is the unit price of the jth item in the i-th budget node before the change, ΔP ij is the price change of the jth item in the i-th budget node; S35. Based on the inference result of the budget consumption node, the inference result of the budget remaining node is calculated. The budget remaining value is the difference between the initial budget total amount and the budget consumption forecast value. S36. Based on the inference result of the budget remaining node, the status of the risk indicator node is inferred: Among them, RiskLevel represents the budget risk indicator status, B r represents the budget remaining forecast value, θ represents the threshold of the budget remaining amount, High represents a high-risk state, and Low represents a low-risk state; S37. Based on the reasoning output results of the budget consumption node, the budget remaining node, and the risk indicator node, a preliminary budget reasoning result is formed.

7. A water conservancy project budget control system based on contract quantity and price linkage according to claim 2, characterized in that: The S4 specifically includes: S41. Randomly generate a set of initial candidate solutions as population individuals of the cuckoo search algorithm to form the initial population set P (0) Each candidate solution includes a set of node conditional probability table parameters and a budget risk warning threshold, setting the population size N, the maximum number of iterations G, and the discovery probability p a and step size control coefficient α; S42. For each candidate solution, set a dual-objective optimization function for the budget execution error rate and the risk warning accuracy: Where Fitness(x) represents the fitness value of the candidate solution x, x represents a candidate solution in the cuckoo search algorithm, w1 is the weighted coefficient of the budget adjustment error term, m is the number of budget nodes that need to be predicted in the budget inference model, and i is the index number of the budget prediction node. is the predicted budget consumption amount of the i-th budget node based on the current candidate solution x, is the actual budget consumption amount recorded at the i-th budget node during the execution of the actual project, w2 is the weighted coefficient of the risk warning accuracy term, TP is the number of records where the budget reasoning model correctly predicts the occurrence of budget overrun risks, TN is the number of records where the budget reasoning model correctly predicts that the budget is normal and has no risks, FP is the number of records where the budget reasoning model incorrectly predicts that there are budget overrun risks, and FN is the number of records where the budget reasoning model misses the budget overrun risks. S43. For each candidate solution, perform perturbation according to the Levy flight formula to generate a new candidate solution and obtain the newly generated population P. (t+1) : in, represents the i-th candidate solution after updating in the t+1-th iteration, represents the i-th candidate solution in the t-th iteration, i is the candidate solution number, t is the iteration number of the current cuckoo search, α is the step size control coefficient in the cuckoo search algorithm, Levy(λ) represents the random step size vector obeying the Levy distribution, and λ is the Levy distribution exponent parameter; S44, the newly generated population P (t+1) Substitute each individual into the optimization objective function, calculate the corresponding fitness value, and combine the discovery probability to find the probability p a Screen and replace individuals in the population to generate an updated population P (t+1) ; S45, repeating steps S43 and S44 until the maximum number of iterations G is reached, and selecting the candidate solution with the best final fitness; S46. Apply the candidate solution with the best fitness to the budget reasoning model, update the conditional probability table parameters and the threshold of the budget remaining amount, and form an optimized budget reasoning model.

8. The water conservancy project budget control system based on contract quantity and price linkage according to claim 2 is characterized in that: The S5 specifically includes: S51. Input the structured contract change dataset constructed in real time into the optimized budget reasoning model, and output the budget consumption forecast value, budget remaining forecast value, and forecast risk indicator status of each node; S52, based on the budget consumption forecast value, construct a budget adjustment difference vector ΔB={ΔB1, ΔB2,…, ΔB N }, where ΔB i The difference between the budget consumption forecast value and the original budget value; S53. Combining the adjustment difference of each budget node with the predicted risk indicator status to form a budget adjustment strategy tuple, and constructing a budget adjustment plan set based on the budget adjustment strategy tuple; S54. Determine the budget overspending risk level based on the budget risk indicator status and construct a risk warning information set; S55. The budget adjustment plan set and the risk warning information set are combined and output as a budget control strategy.

9. A water conservancy project budget control system based on contract quantity and price linkage according to claim 8, characterized in that: The constructing of the budget adjustment scheme set includes proposing budget adjustment suggestions and generating the budget adjustment scheme set; The budget adjustment suggestion is made to compare the budget consumption forecast value B of each budget node i Compared with the original budget like Identify the overspending node and propose additional budget suggestions. Identify savings points and propose budget adjustment and release suggestions; The generating budget adjustment scheme set summarizes the node number, original budget setting value, budget adjustment difference, budget risk indicator status, recommended adjustment type and recommended adjustment amount of each node to form a budget adjustment scheme set.

10. A water conservancy project budget control system based on contract quantity and price linkage according to claim 8, characterized in that: The construction of the risk warning information set includes determining the budget overspending risk level. If the predicted risk status of the budget node is a high-risk state, the warning information construction is triggered. At each high-risk node, a warning record is generated. The warning record includes the node number, the predicted budget remaining value, the overspending probability, the risk level and the recommended risk response measures. All warning records are combined into a risk warning information set.

Images