A new energy power engineering cost intelligent prediction and optimization method and system
By combining deep coding networks and integrated cost prediction models, the problems of data flow delays and insufficient reliability of prediction models in new energy power engineering projects are solved, realizing intelligent prediction and optimization of engineering costs and improving the adaptability and efficiency of decision-making.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUANENG JIAXIANG POWER GENERATION CO LTD
- Filing Date
- 2026-03-19
- Publication Date
- 2026-07-14
AI Technical Summary
In existing technologies for new energy power engineering projects, the lack of real-time multi-source dynamic data, comprehensive features, and adaptability of models in engineering cost management leads to delays in data flow and knowledge feedback. Furthermore, the prediction models lack quantitative indicators of the reliability of the estimation results, making it difficult to effectively weigh the expected benefits and potential risks of the scheme.
A deep coding network is used to uniformly encode multi-source project data. Combined with an integrated cost prediction model, the predicted cost value and its uncertainty are output. The optimal construction scheme is obtained through a multi-objective evolutionary algorithm. At the same time, incremental data is used to update the model parameters to achieve dynamic synergy between prediction and optimization.
It improves the adaptability and efficiency of cost forecasting for new energy power projects, provides a reliability assessment of forecast results, helps decision-makers understand the potential fluctuation range, and reduces the forecast uncertainty of similar schemes.
Smart Images
Figure CN122390814A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power engineering construction technology, and in particular to a method and system for intelligent prediction and optimization of the cost of new energy power projects. Background Technology
[0002] Construction cost is a crucial aspect of investment decision-making for new energy power projects. Currently, construction cost management largely relies on historical statistical data, expert experience, and static models for analysis and estimation. However, existing methods have limitations in terms of data real-time performance, feature comprehensiveness, and model adaptability when dealing with multi-dimensional dynamic factors such as market price fluctuations, differences in project geographical environment, and policy adjustments.
[0003] On the one hand, data collection, feature engineering, cost prediction, and scheme optimization are usually relatively independent processes, leading to delays in data flow and knowledge feedback, making it difficult to form an efficient collaborative decision-making process. On the other hand, most prediction models only provide a single cost estimate, lacking quantitative indicators of the reliability of the estimate itself, making it impossible to effectively weigh the expected benefits of the scheme against potential uncertainties and risks in subsequent optimization. Summary of the Invention
[0004] This invention provides a method and system for intelligent prediction and optimization of the cost of new energy power projects, which solves the problems of insufficient multi-source dynamic data fusion, quantitative evaluation of the reliability of prediction results, and dynamic coordination between the prediction and optimization stages in the existing technology.
[0005] On the one hand, this invention provides a method for intelligent prediction and optimization of the cost of new energy power projects, which includes: The multi-source project data of multiple sub-schemes in new energy power engineering projects are obtained, and the multi-source project data of the sub-schemes are input into a pre-trained deep coding network to obtain a unified coding vector group. An integrated cost prediction model is established. The unified coding vector group is input into the integrated cost prediction model, and the predicted cost value of the corresponding sub-scheme and the prediction uncertainty of the new energy power project are output. The overall cost forecast is determined based on the cost forecast of the sub-schemes. The optimal construction scheme is obtained by optimizing the overall cost forecast based on the forecast uncertainty. Candidate construction schemes are extracted during the optimization iteration process. Incremental data is determined based on the candidate construction schemes. The parameters of the deep coding network and the integrated cost prediction model are updated based on the incremental data.
[0006] On the other hand, the present invention also provides an intelligent prediction and optimization system for the cost of new energy power projects, comprising: The system comprises four modules: an encoding module, a prediction module, and an update module. The encoding module acquires multi-source project data from multiple sub-schemes within a new energy power project, inputs this data into a pre-trained deep encoding network, and generates a unified encoding vector set. The prediction module establishes an integrated cost prediction model, inputting the unified encoding vector set into the model and outputting the predicted cost value for each sub-scheme and the prediction uncertainty of the new energy power project. The optimization module determines the overall cost prediction value based on the predicted cost values of the sub-schemes, optimizes the overall cost prediction value based on the prediction uncertainty, and obtains the optimal construction scheme. The update module extracts candidate construction schemes during the optimization iteration process, determines incremental data based on these schemes, and updates the parameters of the deep encoding network and the integrated cost prediction model based on the incremental data.
[0007] This invention provides an intelligent prediction and optimization method and system for the cost of new energy power projects. By mapping information such as temporal fluctuations, spatial constraints, and clause rules in project data to the same representation space, it provides integrated multi-dimensional feature inputs for subsequent cost prediction. The integrated cost prediction model outputs dual information including predicted values and their uncertainties, providing decision-makers with a reference for understanding the potential fluctuation range of the prediction results. Uncertainty is introduced into the search mechanism of a multi-objective evolutionary algorithm, improving the guidance and efficiency of optimization in complex solution spaces. Simultaneously, candidate solutions and their cost estimates generated during the search process are used as incremental data to fine-tune the parameters of the encoding network and some sub-prediction models. This allows the model to adaptively learn according to the solution region focused on in the current optimization process, potentially reducing the prediction uncertainty of similar solutions in subsequent iterations. Through this mechanism, a dynamic feedback and adjustment relationship is established between the prediction model and the optimization search, which helps improve the adaptability of the entire method in solving specific project cost optimization problems. Attached Figure Description
[0008] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0009] Figure 1 This is a flowchart illustrating an intelligent prediction and optimization method for the cost of new energy power projects provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of an intelligent prediction and optimization system for the cost of new energy power projects provided in an embodiment of the present invention. Detailed Implementation
[0010] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.
[0011] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0012] like Figure 1 As shown, this embodiment of the invention provides a method for intelligent prediction and optimization of the cost of new energy power projects, including: S101: Obtain multi-source project data of multiple sub-schemes in new energy power engineering projects, input the multi-source project data of sub-schemes into a pre-trained deep coding network, and obtain a unified coding vector group; In this embodiment, multiple preliminary schemes corresponding to each process in a new energy power engineering project are collected to obtain sub-schemes. By performing unified encoding processing on the multi-source project data corresponding to the sub-schemes, a unified encoding vector group is obtained.
[0013] In some embodiments of this application, the multi-source project data of the sub-scheme is input into a pre-trained deep coding network to obtain a unified coding vector set, including: extracting market price data from the project data, segmenting the market price data by multiple time scales (day, week, month) to obtain a first feature set; extracting geospatial data from the project data, determining the statistical characteristics of terrain slope and the proportion of ecologically sensitive areas based on the geospatial data to obtain a second feature set; extracting policy text data from the project data, extracting triplet information from the policy text data according to a preset triplet template to obtain a third feature set; standardizing the first feature set, the second feature set, and the third feature set respectively, and inputting the standardized first feature set, the second feature set, and the third feature set into the pre-trained deep coding network to output a unified coding vector set.
[0014] In this embodiment, the multi-source project data includes market price data, geospatial data, and policy text data. The first feature group, the second feature group, and the third feature group corresponding to the market price data, geospatial data, and policy text data are obtained respectively and standardized. The standardized first feature group, the second feature group, and the third feature group are fused into a unified, low-dimensional, semantically rich unified encoding vector group through a pre-trained deep encoding network.
[0015] In some embodiments of this application, the deep coding network includes: a linear projection layer for projecting a first feature group and a second feature group onto a unified dimension to obtain a first projected feature group and a second projected feature group; an attention fusion module for outputting a geo-price fusion vector based on the first projected feature group and the second projected feature group; and a policy-aware gating layer for receiving a third feature group and outputting a policy gating vector with the same dimension as the geo-price fusion vector.
[0016] In some embodiments of this application, the training loss function of the deep coding network is,
[0017] in, For the number of samples, Let Euclidean distance be the unified encoding vector between the i-th sub-scheme and the j-th sub-scheme. Let i be the actual cost of the i-th sub-scheme. Let the actual cost of the j-th sub-scheme be , is a standardization constant.
[0018] In this embodiment, a training loss function is established to make the vectors as close as possible to obtain an accurate unified encoding vector.
[0019] S102, Establish an integrated cost prediction model, input the unified coding vector group into the integrated cost prediction model, and output the cost prediction value of the corresponding sub-scheme and the prediction uncertainty of the new energy power project. In some embodiments of this application, establishing an integrated cost prediction model includes: Obtain the historical training set of the project, perform weighted sampling on the historical training set to obtain multiple training data subsets; train the corresponding cost prediction sub-model based on each training data subset, and combine all cost prediction sub-models to obtain the integrated cost prediction model.
[0020] In this embodiment, a unified coding vector group of a sub-scheme is input into the integrated cost prediction model, and multiple cost prediction sub-models within it output the corresponding cost prediction values and the prediction uncertainty of the new energy power project.
[0021] In some embodiments of this application, outputting the prediction uncertainty of a new energy power engineering project includes: determining the prediction uncertainty according to an uncertainty calculation formula, wherein the uncertainty calculation formula is as follows:
[0022] in, To predict uncertainty, This represents the maximum output value of the cost prediction sub-model. This represents the minimum output value of the cost prediction sub-model. This represents the average of the output values from the cost prediction sub-model. This is the stability coefficient.
[0023] In this embodiment, the prediction uncertainty is determined by the uncertainty calculation formula. The greater the uncertainty, the greater the discrepancy in the cost prediction of the sub-scheme within the integrated model, and the lower the prediction reliability.
[0024] S103. Determine the comprehensive cost forecast value based on the cost forecast value of the sub-schemes, and optimize the comprehensive cost forecast value based on the forecast uncertainty to obtain the optimal construction scheme. In some embodiments of this application, optimization is performed based on the predicted comprehensive cost value with the prediction uncertainty as the objective, including: establishing an objective function based on the predicted comprehensive cost value, solving the objective function based on a non-dominated sorting genetic algorithm; adjusting the exploration range of the mutation operation based on the prediction uncertainty to obtain a Pareto optimal solution set, and determining the optimal construction scheme based on the Pareto optimal solution set.
[0025] In some embodiments of this application, adjusting the exploration range of the mutation operation based on the prediction uncertainty includes: obtaining a preset uncertainty threshold, determining whether the prediction uncertainty is greater than the preset uncertainty threshold; if the prediction uncertainty is greater than the preset uncertainty threshold, then using a preset mutation amplitude as the exploration range of the mutation operation; if the prediction uncertainty is less than or equal to the preset uncertainty threshold, then using twice the preset mutation amplitude as the exploration range of the mutation operation.
[0026] In this embodiment, when the prediction uncertainty is less than or equal to the preset uncertainty threshold, a larger variable time is provided to the algorithm to enhance the exploration range.
[0027] S104: Extract candidate construction schemes during the optimization iteration process, determine incremental data based on the candidate construction schemes, and update the parameters of the deep coding network and the integrated cost prediction model based on the incremental data.
[0028] In some embodiments of this application, incremental data is determined based on candidate construction schemes, and parameters of the deep coding network and the integrated cost prediction model are updated based on the incremental data. This includes: using the unified coding vector group of the candidate construction schemes and the corresponding cost estimates as incremental data, and updating the parameters of the policy awareness gating layer in the deep coding network and the parameters of the cost prediction sub-model with the largest prediction error in the integrated cost prediction model based on the incremental data.
[0029] In this embodiment, after each generation of evolutionary iterations, elite solutions are uniformly selected from the non-dominant frontier of the current population based on the distribution of objective function values. For each elite solution, a cost estimate is generated using a rapid verification tool based on a detailed bill of quantities. The parameters of the cost prediction sub-model are then updated based on the deviation between the cost estimate and the model prediction.
[0030] This invention also provides an intelligent prediction and optimization system for the cost of new energy power projects, comprising: The system comprises four modules: an encoding module, a prediction module, and an update module. The encoding module acquires multi-source project data from multiple sub-schemes within a new energy power project, inputs this data into a pre-trained deep encoding network, and generates a unified encoding vector set. The prediction module establishes an integrated cost prediction model, inputting the unified encoding vector set into the model and outputting the predicted cost value for each sub-scheme and the prediction uncertainty of the new energy power project. The optimization module determines the overall cost prediction value based on the predicted cost values of the sub-schemes, optimizes the overall cost prediction value based on the prediction uncertainty, and obtains the optimal construction scheme. The update module extracts candidate construction schemes during the optimization iteration process, determines incremental data based on these schemes, and updates the parameters of the deep encoding network and the integrated cost prediction model based on the incremental data.
[0031] It should be noted that all relevant information that may be involved in the various embodiments of the present invention is processed in strict accordance with the requirements of laws and regulations, following the principles of legality, legitimacy, and necessity, based on the reasonable purpose of the business scenario, and is information that users actively provide or generate during the use of the product / service, as well as information obtained with user authorization.
[0032] The information processed by this invention may vary depending on the specific product / service scenario and should be based on the specific scenario in which the user uses the product / service. This may involve user account information, device information, or other related information. This invention will treat the relevant information and its processing with the utmost diligence.
[0033] This invention places great emphasis on the security of relevant information and has adopted reasonable and feasible security protection measures that comply with industry standards to protect user information and prevent unauthorized access, public disclosure, use, modification, damage or loss of relevant information.
[0034] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.
[0035] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.
[0036] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for intelligent prediction and optimization of new energy power engineering costs, characterized in that, include: The multi-source project data of multiple sub-schemes in new energy power engineering projects are obtained, and the multi-source project data of the sub-schemes are input into a pre-trained deep coding network to obtain a unified coding vector group. An integrated cost prediction model is established. The unified coding vector group is input into the integrated cost prediction model, and the predicted cost value of the corresponding sub-scheme and the prediction uncertainty of the new energy power project are output. The overall cost forecast is determined based on the cost forecast of the sub-schemes. The optimal construction scheme is obtained by optimizing the overall cost forecast based on the forecast uncertainty. Candidate construction schemes are extracted during the optimization iteration process. Incremental data is determined based on the candidate construction schemes. The parameters of the deep coding network and the integrated cost prediction model are updated based on the incremental data.
2. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 1, characterized in that, The multi-source project data from the sub-scheme is input into a pre-trained deep coding network to obtain a unified coding vector set, including: Market price data is extracted from project data, and the market price data is segmented into multiple time scales by day, week, and month to obtain the first feature group; Geospatial data is extracted from project data, and the statistical characteristics of terrain slope and the proportion of ecologically sensitive areas are determined based on the geospatial data to obtain the second feature group. Extract policy text data from project data, and extract triple information from policy text data according to a preset triple template to obtain the third feature group; The first feature group, the second feature group, and the third feature group are standardized respectively. The standardized first feature group, the second feature group, and the third feature group are input into the pre-trained deep encoding network, and a unified encoding vector group is output.
3. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 2, characterized in that, The deep coding network includes: A linear projection layer is used to project the first feature group and the second feature group onto a unified dimension to obtain the first projected feature group and the second projected feature group. The attention fusion module is used to output a geo-price fusion vector based on the first projection feature group and the second projection feature group; The policy-aware gating layer receives the third feature group and outputs a policy gating vector with the same dimension as the geo-price fusion vector.
4. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 2, characterized in that, The training loss function of the deep coding network is, in, For the project's sampling quantity, Let Euclidean distance be the uniform encoding vector between the i-th sample item and the j-th sample item. The actual cost of the i-th sample item. For the actual cost of the j-th sample item, is a standardized constant.
5. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 1, characterized in that, The establishment of the integrated cost prediction model includes: Obtain the project's historical training set, and perform weighted sampling on the historical training set to obtain multiple training data subsets; Based on each subset of training data, a corresponding cost prediction sub-model is trained. All cost prediction sub-models are then combined to obtain an integrated cost prediction model.
6. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 5, characterized in that, Output the prediction uncertainty of new energy power engineering projects, including: The prediction uncertainty is determined according to the uncertainty calculation formula, which is as follows: in, To predict uncertainty, This represents the maximum output value of the cost prediction sub-model. This represents the minimum output value of the cost prediction sub-model. This represents the average of the output values from the cost prediction sub-model. This is the stability coefficient.
7. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 1, characterized in that, Optimization is performed based on the overall cost forecast, taking into account the uncertainty of the forecast, including: An objective function is established based on the comprehensive cost prediction, and the objective function is solved using a non-dominated sorting genetic algorithm. The exploration range of the variation operation is adjusted according to the prediction uncertainty to obtain the Pareto optimal solution set, and the optimal construction scheme is determined based on the Pareto optimal solution set.
8. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 7, characterized in that, The scope of the mutation operation is adjusted based on the prediction uncertainty, including: Obtain a preset uncertainty threshold and determine whether the prediction uncertainty is greater than the preset uncertainty threshold; If the prediction uncertainty is greater than the preset uncertainty threshold, the preset variation range will be used as the exploration range for the variation operation. If the prediction uncertainty is less than or equal to the preset uncertainty threshold, then twice the preset variation amplitude will be used as the exploration range for the variation operation.
9. The intelligent prediction and optimization method for the cost of new energy power projects according to claim 1, characterized in that, Incremental data is determined based on candidate construction schemes, and the parameters of the deep coding network and integrated cost prediction model are updated based on the incremental data, including: The unified coding vector group of candidate construction schemes and the corresponding cost estimates are used as incremental data. Based on the incremental data, the parameters of the policy awareness gating layer in the deep coding network and the parameters of the cost prediction sub-model with the largest prediction error in the integrated cost prediction model are updated.
10. A smart prediction and optimization system for the cost of new energy power projects, characterized in that, include: The encoding module is used to acquire multi-source project data of multiple sub-schemes in new energy power engineering projects, and input the multi-source project data of the sub-schemes into a pre-trained deep encoding network to obtain a unified encoding vector group. The prediction module is used to establish an integrated cost prediction model. It inputs a unified coding vector group into the integrated cost prediction model and outputs the predicted cost value of the corresponding sub-scheme and the prediction uncertainty of the new energy power project. The optimization module is used to determine the comprehensive cost forecast based on the cost forecast of the sub-schemes, and to optimize the comprehensive cost forecast based on the forecast uncertainty to obtain the optimal construction scheme. The update module is used to extract candidate construction schemes during the optimization iteration process, determine incremental data based on the candidate construction schemes, and update the parameters of the deep coding network and the integrated cost prediction model based on the incremental data.