Power distribution network project cost risk early warning method and system

Abstract

The invention discloses a power distribution network project cost risk early warning method and system, and the method comprises the steps: obtaining project data and a design cost scheme of a target power distribution network project, carrying out the extraction of key index items of each sub-project from the project data, and forming a feature vector corresponding to each sub-project; and outputting a predicted cost scheme corresponding to the feature vector of each sub-project by using the cost prediction model corresponding to each sub-project, thereby obtaining a predicted cost chain of the target power distribution network project, and finally carrying out risk early warning on the cost of the power distribution network project based on the deviation between the design cost scheme and the predicted cost chain. According to the method, calculation is carried out through the data-driven prediction model, and abnormity or non-standardization in the design cost scheme can be recognized.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution network engineering cost, and in particular relates to a risk early warning method and system for distribution network engineering cost. Background Art

[0002] With the rapid development of modern distribution networks, project cost has gradually become a key factor affecting project feasibility, progress and quality. Risk warning of distribution network project cost has always been an important issue in the power industry. Distribution network project cost involves various sub-project environments from design, material procurement to construction, acceptance, etc. The cost of each sub-project needs to be carefully calculated and controlled. The existing project cost is generally completed by cost engineers based on engineering design data, and a large project must be completed by multiple cost engineers.

[0003] However, due to the large number and complexity of the contents involved in distribution network projects, cost engineers are prone to overlook certain details during manual review. For example, during a long review process, fatigue and distraction may lead to the omission of irregular items, potential errors or deviations. Manual review usually relies on personal experience to find anomalies, but in the face of a large amount of data and information, manual review is inefficient, especially when the cost rules / standards of different sub-projects are different. The overall cost plan may not be fully and accurately evaluated due to the inaccurate cost plan of the sub-project, thus affecting the cost control of the overall project. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the present invention proposes a risk early warning method and system for distribution network engineering cost.

[0005] The technical solution of the present invention is as follows:

[0006] A risk early warning method for the construction cost of a distribution network project, wherein the distribution network project includes a plurality of sub-projects, each of which has a pre-built cost prediction model, and the method includes:

[0007] Obtain project data and design cost plan for target distribution network project;

[0008] Extracting key indicator items of each sub-project from the project data to form a feature vector corresponding to each sub-project;

[0009] The cost prediction model corresponding to each sub-project is used to output a predicted cost scheme corresponding to the characteristic vector of each sub-project, thereby obtaining a predicted cost chain of the target distribution network project;

[0010] Based on the deviation between the design cost scheme and the predicted cost chain, a risk warning of the distribution network project cost is carried out.

[0011] Furthermore, the construction method of the cost prediction model is as follows:

[0012] Obtain project data and actual cost plans for historical distribution network projects;

[0013] Extract key indicators corresponding to each sub-project in the project data of historical distribution network projects and obtain the actual cost plan corresponding to each sub-project;

[0014] The key indicators corresponding to each sub-project are used as the input of the cost prediction model, and the actual cost plan corresponding to each sub-project is used as the output of the cost prediction model. The cost prediction model of each sub-project is trained separately to complete the construction of the cost prediction model.

[0015] Furthermore, the key indicator items are extracted by the following method:

[0016] Divide the project data of the distribution network project into sub-project data corresponding to the sub-projects to form sample data for each sub-project;

[0017] Perform principal component analysis on the sample data of each sub-project to obtain the principal component data of each sub-project;

[0018] Extracting a plurality of principal component items of each sub-project from the principal component data according to a preset principal component threshold, thereby generating a load matrix of each sub-project;

[0019] Based on the load matrix of each sub-project, the key indicators of each sub-project are obtained.

[0020] Furthermore, the specific method of obtaining the key indicator items of each sub-project based on the load matrix of each sub-project includes:

[0021] Taking the absolute value of each element of the load matrix to form a degree influence matrix;

[0022] Obtain the row index corresponding to the maximum value of each column of the degree influence matrix;

[0023] The index item corresponding to each row index is used as the key index item of the sub-project.

[0024] Furthermore, the cost prediction model adopts a combined model based on a BP neural network model, a support vector machine model and a random forest model.

[0025] Furthermore, the expression of the cost prediction model is:

[0026] C final =w1*C1+w2*C2+w3*C3

[0027] Among them, C finalis the cost prediction value output by the combined model; C1 is the cost prediction value of the BP neural network model; C2 is the cost prediction value of the support vector machine model; C3 is the cost prediction value of the random forest model; w1, w2 and w3 are the weight coefficients of the BP neural network model, support vector machine model and random forest model respectively.

[0028] Furthermore, the specific method for performing risk warning of the distribution network project cost based on the deviation between the design cost scheme and the predicted cost chain includes:

[0029] The deviation rate between the design cost scheme and the predicted cost chain is obtained, and the deviation rate is compared with a preset deviation rate threshold. If the deviation rate is greater than or equal to the deviation rate threshold, the design cost scheme of each sub-project is obtained based on the project data and the design cost scheme of the target distribution network project, and compared with the corresponding predicted cost scheme of each sub-project, and the top K sub-projects with the largest difference between the design cost scheme and the predicted cost scheme are screened out for risk warning.

[0030] A risk warning system for the construction cost of a distribution network project, wherein the distribution network project includes a plurality of sub-projects, each of which has a pre-built cost prediction model, and the system includes a data acquisition module, a feature vector extraction module, a predicted cost chain acquisition module, and a risk warning module;

[0031] The data acquisition module is used to acquire the project data and design cost plan of the target distribution network project;

[0032] The feature vector extraction module is used to extract key indicator items of each sub-project from the project data to form a feature vector corresponding to each sub-project;

[0033] The predicted cost chain acquisition module is used to use the cost prediction model corresponding to each sub-project to output a predicted cost solution corresponding to the feature vector of each sub-project, thereby acquiring the predicted cost chain of the target distribution network project;

[0034] The risk warning module is used to provide risk warning for the distribution network project cost based on the deviation between the design cost scheme and the predicted cost chain.

[0035] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to execute any of the methods described above.

[0036] A computer-readable storage medium stores a computer program, wherein the computer program implements the steps of any of the above methods when executed by a processor.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] The present invention proposes a risk warning method and system for the cost of a distribution network project. The method obtains the project data and design cost plan of a target distribution network project, extracts the key indicators of each sub-project from the project data, forms a feature vector corresponding to each sub-project, and then uses the cost prediction model corresponding to each sub-project to output the predicted cost plan corresponding to the feature vector of each sub-project, thereby obtaining the predicted cost chain of the target distribution network project, and finally performs a risk warning for the cost of the distribution network project based on the deviation between the design cost plan and the predicted cost chain. The method calculates through a data-driven prediction model, can identify anomalies or irregularities in the design cost plan, and can screen out the top K sub-projects with a large difference between the design cost plan and the predicted cost plan according to the design cost plan of the sub-project and compare it with the predicted cost plan of the corresponding sub-project, and quickly discover and locate potential problems, thereby improving the review efficiency and accuracy of the cost plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 Flow chart of the risk warning method for distribution network project cost in the embodiment. DETAILED DESCRIPTION

[0040] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent modifications of the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0041] Embodiment 1:

[0042] A risk early warning method for the cost of a distribution network project of the present invention, the distribution network project includes a plurality of sub-projects, each of which has a pre-built cost prediction model, the method comprising:

[0043] S1. Obtain the project data and design cost plan of the target distribution network project;

[0044] S2. Extract key indicators of each sub-project from the project data to form a feature vector corresponding to each sub-project;

[0045] S3, using the cost prediction model corresponding to each sub-project to output a predicted cost solution corresponding to the feature vector of each sub-project, thereby obtaining a predicted cost chain of the target distribution network project;

[0046] S4. Provide risk warning for distribution network project cost based on the deviation between the design cost plan and the predicted cost chain.

[0047] Embodiment 2:

[0048] This embodiment is further designed on the basis of the first embodiment in that the construction method of the cost prediction model in this embodiment is as follows:

[0049] Obtain project data and actual cost plans for historical distribution network projects;

[0050] Extract key indicators corresponding to each sub-project in the project data of historical distribution network projects and obtain the actual cost plan corresponding to each sub-project;

[0051] The key indicators corresponding to each sub-project are used as the input of the cost prediction model, and the actual cost plan corresponding to each sub-project is used as the output of the cost prediction model. The cost prediction model of each sub-project is trained separately to complete the construction of the cost prediction model.

[0052] Furthermore, the cost prediction model is stored in the form of a model library, and the specific characteristics of different sub-projects are adapted by selecting models that match the project type. This model selection mechanism increases the scope of application of the method and ensures that different projects can obtain standardized cost assessments. This process can be automated to speed up cost assessment.

[0053] Furthermore, the sub-projects of the distribution network project include substation projects, overhead line projects, etc. Among them, the indicators of the substation project may include rated voltage, substation type, number of main transformers in this period, main transformer capacity, main transformer unit price, number of high-voltage side outgoing lines in this period, number of low-voltage side outgoing lines in this period, high-voltage side distribution type, number of high-voltage side circuit breakers, high-voltage side circuit breaker unit price, low-voltage side distribution type, number of low-voltage side circuit breakers, low-voltage side circuit breaker unit price, number of low-voltage capacitors, low-voltage capacitor unit price, secondary equipment unit price, intelligent related equipment unit price, site leveling costs, foundation treatment costs, station access road costs, altitude, topography. The indicators of the overhead line project may include conductor cross-sectional area, conductor quality, conductor price, tower material price, number of tower bases, substation project, tower material quality, earthwork volume, concrete volume, steel quality, and comprehensive terrain coefficient.

[0054] Furthermore, the method of the present invention extracts key indicators based on historical data, which can effectively reduce the prediction deviation caused by redundant or missing indicators. The model reduces unnecessary data interference and improves the accuracy of estimation by focusing on the core factors affecting the cost. At the same time, prediction models are constructed for different types of sub-projects to provide customized prediction support for different sub-project types, thereby meeting the cost assessment needs of different sub-projects.

[0055] Embodiment three:

[0056] This embodiment is further designed on the basis of the second embodiment in that the key indicator items in this embodiment are extracted by the following method:

[0057] Divide the project data of the distribution network project into sub-project data corresponding to the sub-projects to form sample data for each sub-project;

[0058] Perform principal component analysis on the sample data of each sub-project to obtain the principal component data of each sub-project;

[0059] Extracting multiple principal component items of each sub-project from the principal component data according to a preset principal component threshold, thereby generating a load matrix for each sub-project;

[0060] Based on the load matrix of each sub-project, the key indicators of each sub-project are obtained.

[0061] Furthermore, principal component analysis can reduce the data dimension, retain only the main factors that have a significant impact on the cost, reduce redundant information, and simplify the complexity of the model. This helps to improve the training efficiency and prediction performance of the model.

[0062] The following is an example of a process of performing principal component analysis on sample data to obtain principal component data (eigenvalues ​​and eigenvectors of sample data), which specifically includes:

[0063] Step 1: Standardize the sample data;

[0064] Step 2: Calculate the covariance matrix of the sample data;

[0065] Step 3: Perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalues ​​(λ1, λ2, ..., λ n ,λ1>λ2>...>λ n ,) and eigenvectors (V1, V2, ..., V n );

[0066] According to the preset threshold, multiple principal component items (k eigenvalues) are extracted from the principal component data, including:

[0067] Step 1: Calculate the sum of eigenvalues.

[0068] Step 2: gradually calculate the cumulative variance contribution rate of the eigenvalues ​​until the cumulative variance contribution rate is greater than the preset threshold t, and obtain k eigenvalues, k satisfying:

[0069] Based on multiple principal component terms, the loading matrix is ​​generated including:

[0070] According to k and the eigenvectors corresponding to the eigenvalues, the load matrix L = [V1, V2, ..., Vk ], whose size is n×k.

[0071] Embodiment 4:

[0072] This embodiment is further designed on the basis of the third embodiment in that, in this embodiment, based on the load matrix of each sub-project, a specific method for obtaining the key indicator items of each sub-project includes:

[0073] Take the absolute value of each element of the load matrix to form a degree influence matrix;

[0074] Get the row index corresponding to the maximum value of each column of the degree of influence matrix;

[0075] The index item corresponding to each row index is used as the key index item of the sub-project.

[0076] Furthermore, the process of taking the absolute value and selecting the maximum index can quickly identify the feature item with the highest contribution to each principal component, thereby simplifying the selection of key indicator items. Selecting the feature item with the largest contribution to the principal component allows the model to make decisions on the core data, reducing dependence on noise and secondary features, and enhancing the stability of the model and its generalization ability on new data.

[0077] Embodiment five:

[0078] This embodiment is further designed on the basis of the first embodiment in that the cost prediction model in this embodiment adopts a combined model composed of a BP neural network model, a support vector machine model and a random forest model.

[0079] Embodiment six:

[0080] This embodiment is further designed on the basis of the fifth embodiment in that the expression of the cost prediction model in this embodiment is:

[0081] C final =w1*C1+w2*C2+w3*C3

[0082]

[0083] Among them, C final is the cost prediction value output by the combined model; C1 is the cost prediction value of the BP neural network model; C2 is the cost prediction value of the support vector machine model; C3 is the cost prediction value of the random forest model; w1, w2 and w3 are the weight coefficients of the BP neural network model, the support vector machine model and the random forest model respectively; w i is the weight coefficient of the i-th model; R i is the root mean square error of the ith model; R j is the root mean square error of the jth model.

[0084] Furthermore, a single model may be sensitive to certain data characteristics and may be at risk of overfitting or underfitting. The combined model can effectively alleviate this problem by introducing the prediction results of multiple algorithms and enhance the stability and generalization ability of the model on different data sets. The combined model can adjust the contribution of each model to the final prediction result through a preset weight coefficient and flexibly optimize the combination scheme. This weighting mechanism can dynamically assign weights to each model based on its performance on the validation set, making the prediction results more in line with the actual situation.

[0085] At the same time, by calculating the weight coefficient through the root mean square error, the influence of each model in the combined model can be adaptively adjusted. Models with smaller errors receive higher weights, which helps improve the overall prediction effect of the combined model. Compared with simple averaging or manually setting weights, weight coefficients are automatically generated through a data-driven method, reducing the subjectivity and uncertainty of manually setting weights. This automated processing method improves the scientificity and objectivity of weight allocation, making the model more consistent with the characteristics of actual data.

[0086] Embodiment seven:

[0087] This embodiment is further designed on the basis of the first embodiment in that the specific method for risk warning of the distribution network project cost based on the deviation between the design cost scheme and the predicted cost chain in this embodiment includes:

[0088] The deviation rate between the design cost scheme and the predicted cost chain is obtained, and the deviation rate is compared with the preset deviation rate threshold. If the deviation rate is greater than or equal to the deviation rate threshold, the design cost scheme of each sub-project is obtained based on the project data and design cost scheme of the target distribution network project, and compared with the corresponding predicted cost scheme of each sub-project, and the top K sub-projects with the largest difference between the design cost scheme and the predicted cost scheme are screened out for risk warning.

[0089] Furthermore, the present invention can systematically evaluate the cost distribution of the entire distribution network project through cost chain analysis, reduce reliance on manual experience, save time and manpower required for cost evaluation, and achieve efficient management. It is convenient to comprehensively compare and optimize the cost plans of each sub-project, which helps to discover high-cost links and take corresponding measures.

[0090] At the same time, by calculating the deviation rate, the difference between the actual cost and the design cost of the target sub-project can be monitored in real time. If the predicted cost deviates too much from the design cost, potential project cost overruns or plan deviations can be discovered in time, which helps to quickly take corrective measures and avoid unnecessary waste. By setting the deviation rate threshold (which can be set to 10%), early warning information is automatically generated, reducing the workload of manual monitoring and judgment. This automated early warning system can accurately point out which sub-projects may have problems, thereby improving management efficiency. By identifying cost deviations in a timely manner, project risks can be better predicted and avoided, especially in large-scale distribution network projects, where cost deviations may affect the cost control and schedule of the overall project. This method provides an effective tool to manage and control risks and ensure the smooth progress of the project.

[0091] Embodiment eight:

[0092] A risk early warning system for the construction cost of a distribution network project of the present invention, the distribution network project includes a plurality of sub-projects, each of which has a pre-built cost prediction model, characterized in that the system includes a data acquisition module, a feature vector extraction module, a predicted cost chain acquisition module and a risk early warning module;

[0093] Data acquisition module, used to obtain project data and design cost plan of target distribution network project;

[0094] The feature vector extraction module is used to extract key indicators of each sub-project from the project data to form a feature vector corresponding to each sub-project;

[0095] A forecast cost chain acquisition module is used to use the cost forecast model corresponding to each sub-project to output a forecast cost solution corresponding to the feature vector of each sub-project, thereby acquiring the forecast cost chain of the target distribution network project;

[0096] The risk warning module is used to provide risk warning for distribution network project costs based on the deviation between the design cost plan and the predicted cost chain.

[0097] Embodiment nine:

[0098] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor is used to call and run the computer program stored in the memory to execute the method of any of the above embodiments.

[0099] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any of the above embodiments are implemented.

[0100] The above description is only a specific implementation of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a technician familiar with the technical field within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.

Claims

1. A risk early warning method for the construction cost of a distribution network project, wherein the distribution network project includes a plurality of sub-projects, each of which has a pre-built cost prediction model, characterized in that: The method comprises: Obtain project data and design cost plan of target distribution network project; Extracting key indicator items of each sub-project from the project data to form a feature vector corresponding to each sub-project; The cost prediction model corresponding to each sub-project is used to output a predicted cost scheme corresponding to the characteristic vector of each sub-project, thereby obtaining a predicted cost chain of the target distribution network project; Based on the deviation between the design cost scheme and the predicted cost chain, a risk warning of the distribution network project cost is carried out.

2. The method for early warning of the cost of power distribution network engineering according to claim 1, characterized in that: The construction method of the cost prediction model is as follows: Obtain project data and actual cost plans for historical distribution network projects; Extract key indicators corresponding to each sub-project in the project data of historical distribution network projects and obtain the actual cost plan corresponding to each sub-project; The key indicators corresponding to each sub-project are used as the input of the cost prediction model, and the actual cost plan corresponding to each sub-project is used as the output of the cost prediction model. The cost prediction model of each sub-project is trained separately to complete the construction of the cost prediction model.

3. The risk early warning method for distribution network engineering cost according to claim 2 is characterized by: The key indicator items are extracted by the following method: Divide the project data of the distribution network project into sub-project data corresponding to the sub-projects to form sample data for each sub-project; Perform principal component analysis on the sample data of each sub-project to obtain the principal component data of each sub-project; Extracting a plurality of principal component items of each sub-project from the principal component data according to a preset principal component threshold, thereby generating a load matrix of each sub-project; Based on the load matrix of each sub-project, the key indicators of each sub-project are obtained.

4. The risk early warning method for distribution network engineering cost according to claim 3 is characterized by: The specific method of obtaining the key indicator items of each sub-project based on the load matrix of each sub-project includes: Taking the absolute value of each element of the load matrix to form a degree influence matrix; Obtain the row index corresponding to the maximum value of each column of the degree influence matrix; The index item corresponding to each row index is used as the key index item of the sub-project.

5. The risk early warning method for distribution network engineering cost according to claim 1 is characterized by: The construction cost prediction model adopts a combined model based on a BP neural network model, a support vector machine model and a random forest model.

6. The method for early warning of the cost of power distribution network engineering according to claim 5, characterized in that: The expression of the cost prediction model is: C final =w1*C1+w2*C2+w3*C3 Among them, C final is the cost prediction value output by the combined model; C1 is the cost prediction value of the BP neural network model; C2 is the cost prediction value of the support vector machine model; C3 is the cost prediction value of the random forest model; w1, w2 and w3 are the weight coefficients of the BP neural network model, the support vector machine model and the random forest model respectively; w i is the weight coefficient of the i-th model; R i is the root mean square error of the ith model; R j is the root mean square error of the jth model.

7. The risk early warning method for distribution network engineering cost according to claim 1 is characterized by: The specific method for performing risk warning of the distribution network project cost based on the deviation between the design cost scheme and the predicted cost chain includes: The deviation rate between the design cost scheme and the predicted cost chain is obtained, and the deviation rate is compared with a preset deviation rate threshold. If the deviation rate is greater than or equal to the deviation rate threshold, the design cost scheme of each sub-project is obtained based on the project data and the design cost scheme of the target distribution network project, and compared with the corresponding predicted cost scheme of each sub-project, and the top K sub-projects with the largest difference between the design cost scheme and the predicted cost scheme are screened out for risk warning.

8. A risk early warning system for the construction cost of a distribution network project, wherein the distribution network project includes a plurality of sub-projects, each of which has a pre-built cost prediction model, characterized in that: The system includes a data acquisition module, a feature vector extraction module, a predicted cost chain acquisition module and a risk warning module; The data acquisition module is used to acquire the project data and design cost plan of the target distribution network project; The feature vector extraction module is used to extract key indicator items of each sub-project from the project data to form a feature vector corresponding to each sub-project; The predicted cost chain acquisition module is used to use the cost prediction model corresponding to each sub-project to output a predicted cost solution corresponding to the feature vector of each sub-project, thereby acquiring the predicted cost chain of the target distribution network project; The risk warning module is used to provide risk warning for the distribution network project cost based on the deviation between the design cost scheme and the predicted cost chain.

9. An electronic device, characterized in that: The electronic device comprises a memory and a processor, wherein the memory stores a computer program, and the processor is configured to call and run the computer program stored in the memory to execute the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method as claimed in any one of claims 1 to 7 are implemented.

Images