Power transmission and transformation project settlement probability exceeding early warning method and device based on attention mechanism

Abstract

The invention relates to an attention mechanism-based power transmission and transformation project settlement probability exceeding early warning method and device, and the method comprises the following steps: obtaining first settlement data of a power transmission and transformation project, and building a first settlement data set; performing abnormal value and missing value processing on the first settlement data set in sequence to obtain a second standard data set; performing weight analysis on the second standard data set to generate attention weight data, and performing attention weighting to obtain a third weighted data set; constructing a settlement cost prediction model by taking the settlement cost as a target on the basis of the third weighted data set based on a deep learning model; inputting the real-time first settlement data into a settlement cost prediction model, and calculating an estimated settlement cost of the power transmission and transformation project; and comparing the pre-estimated settlement cost with the project reply budget to obtain a fourth comparison result, and automatically triggering a corresponding over-budget early warning. According to the invention, timely mastering and warning of project override conditions are realized, the project management efficiency is improved, and override risks are reduced.

Description

Technical Field

[0001] The present invention relates to an attention mechanism-based over-estimate early warning method and device for power transmission and transformation project settlement, belonging to the technical field of engineering project management. Background Art

[0002] In recent years, the State Grid headquarters has strengthened the control over the settlement of power transmission and transformation projects, and clearly stipulated that the settlement should not exceed the budget in principle. All provincial and municipal units also use whether the settlement exceeds the budget as an indicator and as a basis for assessment.

[0003] The over-estimate of the settlement of power transmission and transformation projects refers to the situation that the final settlement cost exceeds the originally approved budget amount during the actual settlement process of the power transmission and transformation project. The over-estimate of the settlement can be perceived in advance and certain preventive measures can be taken. In the existing technical solutions, the data of the approved budget, change visa, land acquisition fee, material price, etc. are mainly collected manually, and the estimated settlement cost is estimated based on experience, and then compared with the budget. However, the existing technology mainly relies on manual experience to estimate the settlement cost. The experience level and understanding of different personnel may vary, which affects the consistency and accuracy of the settlement results. Manual data collection and calculation can only be carried out according to the cycle, and it is impossible to realize the real-time perception and feedback of the over-estimate of the settlement; when manually calculating the estimated settlement cost, there is a lack of unified standards and specifications, resulting in low traceability of the calculation amount and difficulty in tracking and verifying the calculation process. Since each project involves a large number of materials and cost items, it is difficult for manual calculation to accurately grasp the impact of market changes on their prices, and estimation errors are prone to occur.

[0004] The existing technology fails to make full use of automation technologies, such as machine learning or artificial intelligence, and cannot realize automatic perception and prediction of settlement overestimation, which increases the complexity and workload of manual operations. The AM attention mechanism is a method in deep learning to highlight important information by assigning different weights. This mechanism was originally introduced for natural language processing tasks, but its ideas can also be applied to other fields, such as image processing, speech recognition, and engineering cost prediction; the basic idea of ​​the attention mechanism is to simulate humans who selectively focus on certain important parts when processing large amounts of information. By assigning different weights to different parts of the input data, the model's attention to important information can be enhanced, thereby improving the accuracy and efficiency of the prediction. Therefore, the present invention proposes a method and device for early warning of overestimation of settlement of power transmission and transformation projects based on an attention mechanism. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a method and device for early warning of over-budget settlement of power transmission and transformation projects based on an attention mechanism, which can not only solve the problem that it is difficult for managers to collect estimated settlement data manually calculated and the accuracy is difficult to ensure, but also avoid the problem that managers are belatedly aware of the over-budget settlement.

[0006] The technical solution adopted by the present invention to solve the technical problem is:

[0007] In a first aspect, an embodiment of the present invention provides a method for early warning of over-estimated settlement of a power transmission and transformation project based on an attention mechanism, comprising the following steps:

[0008] Step S1, obtaining first settlement data of the power transmission and transformation project, and updating and recording in real time to establish a first settlement data set, wherein the first settlement data includes the project approval budget, change visa cost, site construction fee, equipment and material prices, bid discount coefficient, market material price, bid quotation and tender control price, and the first settlement data set includes historical data of each settlement time point of the power transmission and transformation project;

[0009] Step S2, firstly, performing outlier processing on the first settlement data set by using a box plot, and then performing missing value processing on the first settlement data set by using a mean filling method, and using the first settlement data after the outlier processing and the missing value processing as the second standard data set;

[0010] Step S3, based on the AM attention mechanism, performing weight analysis on the second standard data set to generate attention weight data, and performing attention weighting on the second standard data set to obtain a third weighted data set;

[0011] Step S4, constructing a settlement fee prediction model based on the third weighted data set based on the deep learning model and taking the settlement fee as the target;

[0012] Step S5, inputting the real-time first settlement data into the settlement cost prediction model to calculate the estimated settlement cost of the power transmission and transformation project;

[0013] Step S6, comparing the estimated settlement cost with the approved budget estimate of the project to obtain a fourth comparison result, and automatically triggering a corresponding over-budget warning according to the fourth comparison result.

[0014] As a possible implementation of this embodiment, in step S2, performing outlier processing on the first settlement data set through a box plot includes the following steps:

[0015] sorting the first settlement data set, and calculating the first quartile and the third quartile of the first settlement data set;

[0016] The interquartile range was calculated by calculating the difference between the third quartile and the first quartile;

[0017] The lower and upper boundaries of outliers were determined using 1.5 times the interquartile range;

[0018] The values ​​in the first settlement data that are lower than the lower boundary or higher than the upper boundary are regarded as outliers and are deleted.

[0019] As a possible implementation of this embodiment, in step S2, the missing value processing of the first settlement data set by using the mean filling method includes the following steps:

[0020] representing the first settlement data set in the form of a two-dimensional table;

[0021] For each column of data, count the number of missing values;

[0022] For each column with missing values, calculate its mean and fill the missing values ​​with the mean.

[0023] As a possible implementation of this embodiment, step S3 includes the following steps:

[0024] Mapping the number of the second standard data as rows and the data type as columns into a second standard data matrix;

[0025] Perform a linear transformation on the second standard data matrix to obtain the query matrix, key matrix, and value matrix:

[0026] Z=S*F Z +v Z

[0027] X=S*F X +v X

[0028] C=S*F C +v C

[0029] Where Z is the query matrix, X is the key matrix, C is the value matrix, S is the second standard data matrix, and F Z is a shape of (d, d k )’s query weight matrix, F X is a shape of (d, d k )’s key weight matrix, F C is a shape of (d, d v ) value weight matrix, v Z 、v X 、v C is the bias term, d k and d v is the dimension after linear transformation;

[0030] The attention score is calculated based on the dot product of the query matrix and the key matrix:

[0031]

[0032] Where α is the attention score;

[0033] Normalize the attention score to get the attention weight:

[0034]

[0035] In the formula, zyl εε is the attention weight, softmax(w i ) is the softmax function, w i is any function parameter;

[0036] The third weighted data is obtained by weighted summing of the value matrix according to the attention weight:

[0037]

[0038] Wherein, β is the third weighted data.

[0039] As a possible implementation of this embodiment, step S4 includes the following steps:

[0040] Performing time series processing on the third weighted data in the third weighted data set and arranging the data into a time series format;

[0041] Capturing the time dependency of each type of data in the sorted third weighted data;

[0042] Map the time dependency to the preset calculation formula for estimated settlement costs of each project type to obtain the settlement cost prediction model:

[0043] H=(K JZ +P JZ )+(K AZ +P AZ )+δ+(K QT +ε)

[0044] G=(K BT +SF S )+SF JG +(K QT +ε)+P JK

[0045] Where H is the estimated settlement cost of the substation project or cable project, K JZ is the converted value of construction engineering cost, P JZ For the change certification of construction projects, K AZ is the converted value of installation engineering cost, P AZ is the change visa for the installation project, δ is the increase in equipment purchase cost, K QT is the conversion value of other costs, ε is the cost change correction factor, G is the estimated settlement cost of the overhead line project, K BT is the converted value of the main engineering cost, SF SSF is the floating coefficient of equipment materials without main materials supplied by Party A. JG P is the floating coefficient of the main material cost supplied by Party A, JK To apply for change certification for overhead line projects; the types of projects described include substation projects, cable projects and overhead line projects.

[0046] As a possible implementation of this embodiment, the capturing of the time dependency of each type of data of the sorted third weighted data includes the following steps:

[0047] Obtain input data, input the input data of time step t into a preset processing unit, the preset processing unit includes an input gate, a forget gate and an output gate, the input data includes the third weighted data, the estimated settlement fee of the previous time step and the memory cell state of the previous time step;

[0048] The input gate calculates the introduced coefficients of the control input data;

[0049] The retention coefficient of the memory cell state at the previous time step is controlled by calculating the forget gate;

[0050] Calculate the candidate memory cell states for the current input data;

[0051] The memory cell state at the current time step is calculated by combining the introduction coefficient, the retention coefficient and the candidate memory cell state;

[0052] The estimated influence coefficient of the memory cell state on the estimated settlement fee is calculated by the output gate;

[0053] The estimated settlement fee of the current time step is calculated by combining the estimated impact coefficient, the estimated settlement fee of the previous time step, and the memory cell state of the current time step;

[0054] The input gate calculation formula is as follows:

[0055] Y t =μ(M SR *[ω t-1 , X t ])

[0056] The calculation formula of the forget gate is as follows:

[0057] BL t =μ(M YW *[ω t-1 , X t ])

[0058] The candidate memory cell state calculation formula is as follows:

[0059]

[0060] The memory cell state calculation formula is as follows:

[0061]

[0062] The output gate calculation formula is as follows:

[0063] YS t =μ(M SC *[ω t-1 , X t ])

[0064] The calculation formula for the estimated settlement fee is as follows:

[0065] ω t =YS t *tanh(B t )

[0066] In the formula, YR t is the input data introduction coefficient, μ() is the sigmoid activation function, M SR is the weight matrix of the input gate, ω t-1 is the estimated settlement fee at time step t-1, X t is the third weighted data, BL t is the retention coefficient of the memory cell state at the previous time step, M YW is the weight matrix of the forget gate, is the candidate memory cell state, tanh() is the tanh activation function, M HX is the weight matrix for selecting the memory cell state, B t is the memory cell state at time step t, B t-1 The state of the memory cell at the t-1 time step is the state of the memory cell at the previous time step, YS t is the estimated impact coefficient of the memory cell status on the estimated settlement fee, M SC is the weight matrix of the output gate, ω t is the estimated settlement fee at time step t.

[0067] As a possible implementation of this embodiment, in step S6, the fourth comparison result includes a slightly over-generalized result, a moderately over-generalized result, and a severely over-generalized result;

[0068] The automatically triggering of the corresponding over-generalization warning according to the fourth comparison result comprises the following steps:

[0069] Compare the estimated settlement cost with the approved budget of the project. When the amount of the excess is less than 2% of the approved budget of the project and the excess amount is less than 1 million, the estimated settlement excess will be marked as a minor excess and a first-level warning will be issued.

[0070] When the amount of the over-budget exceeds 10% of the approved budget of the project or the over-budget amount exceeds RMB 5 million, the estimated settlement over-budget will be marked as a severe over-budget result and a Level 3 warning will be issued;

[0071] When the estimated settlement excess does not fall into the category of slight excess or severe excess, the estimated settlement excess will be marked as moderate excess and a level 2 warning will be issued.

[0072] In a second aspect, an embodiment of the present invention provides an attention mechanism-based power transmission and transformation project settlement over-estimate early warning device, comprising:

[0073] A first settlement data set establishment module is used to obtain first settlement data of the power transmission and transformation project, and update and record in real time to establish a first settlement data set, wherein the first settlement data includes the project approval budget, change certification cost, site construction fee, equipment and material prices, bid discount coefficient, market material price, bid quotation and tender control price, and the first settlement data set includes historical data of each settlement time point of the power transmission and transformation project;

[0074] An outlier and missing value processing module, used to first perform outlier processing on the first settlement data set through a box plot, then perform missing value processing on the first settlement data set through a mean filling method, and use the first settlement data after outlier processing and missing value processing as a second standard data set;

[0075] An attention weight generation module is used to perform weight analysis on the second standard data set based on the AM attention mechanism to generate attention weight data, and perform attention weighting on the second standard data set to obtain a third weighted data set;

[0076] A settlement fee prediction model building module is used to build a settlement fee prediction model based on the third weighted data set based on the deep learning model and taking the settlement fee as the target;

[0077] An estimated settlement cost calculation module, used to input the real-time first settlement data into a settlement cost prediction model to calculate the estimated settlement cost of the power transmission and transformation project;

[0078] The project settlement over-budget warning module is used to compare the estimated settlement cost with the project approved budget to obtain a fourth comparison result, and automatically trigger a corresponding over-budget warning according to the fourth comparison result.

[0079] In a third aspect, an embodiment of the present invention provides a computer device, comprising a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory through the bus, and the processor executes the machine-readable instructions to perform any of the steps of the above-mentioned power transmission and transformation project settlement over-estimate early warning method based on the attention mechanism.

[0080] In a fourth aspect, an embodiment of the present invention provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned attention-based power transmission and transformation project settlement over-estimate early warning methods are executed.

[0081] The beneficial effects of the technical solution of the embodiment of the present invention are as follows:

[0082] The technical solution of the embodiment of the present invention is a method for warning of over-budget in power transmission and transformation project settlement based on an attention mechanism. By acquiring and updating data in real time, it realizes timely grasp and warning of project over-budget situations, thereby improving the efficiency of project management. By automatically identifying over-budget risk factors and comprehensively evaluating risks, it can help management units to promptly discover and solve project over-budget problems, thereby reducing the occurrence of over-budget risks.

[0083] The present invention realizes the real-time calculation of engineering overestimation, improves the real-time computing capability of the system, and adapts to the real-time requirements of modern engineering management; by automatically identifying overestimation risk factor items and overestimation early warning triggering functions, it reduces the need for manual intervention and improves management efficiency and work accuracy.

[0084] The present invention can timely grasp the situation of project overruns, which helps the management unit to take timely measures to avoid additional expenses and save unnecessary costs. The present invention provides the management unit with real-time project overruns, which helps them to more accurately evaluate and control risks in investment decisions, thereby improving the accuracy and benefits of investment decisions.

[0085] The present invention not only solves the problem that it is difficult for managers to collect estimated settlement data manually and ensure accuracy, but also avoids the problem that managers are belatedly aware of budget settlement over-control.

[0086] A power transmission and transformation project settlement over-estimate warning device based on an attention mechanism according to the technical solution of an embodiment of the present invention has the same beneficial effects as a power transmission and transformation project settlement over-estimate warning method based on an attention mechanism according to the technical solution of an embodiment of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0087] Figure 1 It is a flow chart of a method for early warning of over-estimated settlement of power transmission and transformation projects based on an attention mechanism according to an exemplary embodiment;

[0088] Figure 2 It is a structural schematic diagram of an attention mechanism-based power transmission and transformation project settlement over-estimate early warning device according to an exemplary embodiment. DETAILED DESCRIPTION

[0089] In order to more clearly illustrate the technical features of the solution of the present invention, the present invention is described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0090] During the construction process, based on the project approval budget, it automatically obtains and aggregates other costs such as land acquisition fees, changes in construction costs caused by change visas, and intelligently calculates the price changes of equipment and materials during the preparation period and the current price changes, and the amount of impact on construction costs based on market price conditions. The above data is fed into the settlement over-budget risk prediction model to calculate the estimated settlement costs in real time, and warn of the estimated settlement over-budget projects, and output an over-budget warning report.

[0091] like Figure 1 As shown, an embodiment of the present invention provides a method for early warning of over-estimated settlement of power transmission and transformation projects based on an attention mechanism, comprising the following steps:

[0092] Step S1, obtaining first settlement data of the power transmission and transformation project, and updating and recording in real time to establish a first settlement data set, wherein the first settlement data includes the project approval budget, change visa cost, site construction fee, equipment and material prices, bid discount coefficient, market material price, bid quotation and tender control price, and the first settlement data set includes historical data of each settlement time point of the power transmission and transformation project;

[0093] Step S2, firstly, performing outlier processing on the first settlement data set by using a box plot, and then performing missing value processing on the first settlement data set by using a mean filling method, and using the first settlement data after the outlier processing and the missing value processing as the second standard data set;

[0094] Step S3, based on the AM attention mechanism, performing weight analysis on the second standard data set to generate attention weight data, and performing attention weighting on the second standard data set to obtain a third weighted data set;

[0095] Step S4, constructing a settlement fee prediction model based on the third weighted data set based on the deep learning model and taking the settlement fee as the target;

[0096] Step S5, inputting the real-time first settlement data into the settlement cost prediction model to calculate the estimated settlement cost of the power transmission and transformation project;

[0097] Step S6, comparing the estimated settlement cost with the approved budget estimate of the project to obtain a fourth comparison result, and automatically triggering a corresponding over-budget warning according to the fourth comparison result.

[0098] As a possible implementation of this embodiment, in step S2, performing outlier processing on the first settlement data set through a box plot includes the following steps:

[0099] sorting the first settlement data set, and calculating the first quartile and the third quartile of the first settlement data set;

[0100] The interquartile range was calculated by calculating the difference between the third quartile and the first quartile;

[0101] The lower and upper boundaries of outliers were determined using 1.5 times the interquartile range;

[0102] The values ​​in the first settlement data that are lower than the lower boundary or higher than the upper boundary are regarded as outliers and are deleted.

[0103] As a possible implementation of this embodiment, in step S2, the missing value processing of the first settlement data set by using the mean filling method includes the following steps:

[0104] representing the first settlement data set in the form of a two-dimensional table;

[0105] For each column of data, count the number of missing values;

[0106] For each column with missing values, calculate its mean and fill the missing values ​​with the mean.

[0107] As a possible implementation of this embodiment, step S3 includes the following steps:

[0108] Mapping the number of the second standard data as rows and the data type as columns into a second standard data matrix;

[0109] Perform a linear transformation on the second standard data matrix to obtain the query matrix, key matrix, and value matrix:

[0110] Z=S*F Z +v Z

[0111] X=S*F X +v X

[0112] C=S*F C +v C

[0113] Where Z is the query matrix, X is the key matrix, C is the value matrix, S is the second standard data matrix, and F Z is a shape of (d, d k )’s query weight matrix, F X is a shape of (d, d k )’s key weight matrix, F C is a shape of (d, d v ) value weight matrix, v Z 、v X 、v Cis the bias term, d k and d v is the dimension after linear transformation;

[0114] The attention score is calculated based on the dot product of the query matrix and the key matrix:

[0115]

[0116] Where α is the attention score;

[0117] Normalize the attention score to get the attention weight:

[0118]

[0119] In the formula, zyl εε is the attention weight, softmax(w i ) is the softmax function, wi is any function parameter;

[0120] The third weighted data is obtained by weighted summing of the value matrix according to the attention weight:

[0121]

[0122] Wherein, β is the third weighted data.

[0123] As a possible implementation of this embodiment, step S4 includes the following steps:

[0124] Performing time series processing on the third weighted data in the third weighted data set and arranging the data into a time series format;

[0125] Capturing the time dependency of each type of data in the sorted third weighted data;

[0126] Map the time dependency to the preset calculation formula for estimated settlement costs of each project type to obtain the settlement cost prediction model:

[0127] H=(K JZ +P JZ )+(K AZ +P AZ )+δ+(K QT +ε)

[0128] G=(K BT +SF S )+SF JG +(K QT +ε)+P JK

[0129] Where H is the estimated settlement cost of the substation project or cable project, K JZis the converted value of construction engineering cost, P JZ For the change certification of construction projects, K AZ is the converted value of installation engineering cost, P AZ is the change visa for the installation project, δ is the increase in equipment purchase cost, K QT is the conversion value of other costs, ε is the cost change correction factor, G is the estimated settlement cost of the overhead line project, K BT is the converted value of the main engineering cost, SF S SF is the floating coefficient of equipment materials without main materials supplied by Party A. JG P is the floating coefficient of the main material cost supplied by Party A, JK To apply for change certification for overhead line projects; the types of projects described include substation projects, cable projects and overhead line projects.

[0130] As a possible implementation of this embodiment, the capturing of the time dependency of each type of data of the sorted third weighted data includes the following steps:

[0131] Obtain input data, input the input data of time step t into a preset processing unit, the preset processing unit includes an input gate, a forget gate and an output gate, the input data includes the third weighted data, the estimated settlement fee of the previous time step and the memory cell state of the previous time step;

[0132] The input gate calculates the introduced coefficients of the control input data;

[0133] The retention coefficient of the memory cell state at the previous time step is controlled by calculating the forget gate;

[0134] Calculate the candidate memory cell states for the current input data;

[0135] The memory cell state at the current time step is calculated by combining the introduction coefficient, the retention coefficient and the candidate memory cell state;

[0136] The estimated influence coefficient of the memory cell state on the estimated settlement fee is calculated by the output gate;

[0137] The estimated settlement fee of the current time step is calculated by combining the estimated impact coefficient, the estimated settlement fee of the previous time step, and the memory cell state of the current time step;

[0138] The input gate calculation formula is as follows:

[0139] Y t =μ(M SR *[ω t-1 , X t ])

[0140] The calculation formula of the forget gate is as follows:

[0141] BL t =μ(M YW *[ω t-1 , X t ])

[0142] The candidate memory cell state calculation formula is as follows:

[0143]

[0144] The memory cell state calculation formula is as follows:

[0145]

[0146] The output gate calculation formula is as follows:

[0147] YS t =μ(M SC *[ω t-1 , X t ])

[0148] The calculation formula for the estimated settlement fee is as follows:

[0149] ω t =YS t *tanh(B t )

[0150] In the formula, YR t is the input data introduction coefficient, μ() is the sigmoid activation function, M SR is the weight matrix of the input gate, ω t-1 is the estimated settlement fee at time step t-1, X t is the third weighted data, BL t is the retention coefficient of the memory cell state at the previous time step, M YW is the weight matrix of the forget gate, is the candidate memory cell state, tanh() is the tanh activation function, M HX is the weight matrix for selecting the memory cell state, B t is the memory cell state at time step t, B t-1 The state of the memory cell at the t-1 time step is the state of the memory cell at the previous time step, YS t is the estimated impact coefficient of the memory cell status on the estimated settlement fee, M SC is the weight matrix of the output gate, ω t is the estimated settlement fee at time step t.

[0151] The specific mathematical expression of the sigmoid activation function is as follows:

[0152]

[0153] In the formula, e is the base of the natural logarithm, k is the input value; the graph of the sigmoid function presents an S-shaped curve, so it is also called the S-type function;

[0154] The mathematical expression of the tanh activation function is as follows:

[0155]

[0156] The graph of the tanh activation function also shows an S-shaped curve, similar to the sigmoid function, but its output range is wider, between -1 and 1.

[0157] As a possible implementation of this embodiment, in step S6, the fourth comparison result includes a slightly over-generalized result, a moderately over-generalized result, and a severely over-generalized result;

[0158] The automatically triggering of the corresponding over-generalization warning according to the fourth comparison result comprises the following steps:

[0159] Compare the estimated settlement cost with the approved budget of the project. When the amount of the excess is less than 2% of the approved budget of the project and the excess amount is less than 1 million, the estimated settlement excess will be marked as a minor excess and a first-level warning will be issued.

[0160] When the amount of the over-budget exceeds 10% of the approved budget of the project or the over-budget amount exceeds RMB 5 million, the estimated settlement over-budget will be marked as a severe over-budget result and a Level 3 warning will be issued;

[0161] When the estimated settlement excess does not fall into the category of slight excess or severe excess, the estimated settlement excess will be marked as moderate excess and a level 2 warning will be issued.

[0162] According to the actual situation, through the real-time calculation and early warning functions of this embodiment, the processing time of managers for engineering overruns can be shortened from several days to several hours or even minutes, greatly improving work efficiency and response speed. The real-time calculation and automatic identification and prediction functions of the present invention help to improve the modernization level and technical capabilities of engineering management, making engineering management more intelligent and efficient. The present invention reduces the additional expenses caused by overruns, which can save millions or even tens of millions of yuan in costs.

[0163] like Figure 2 As shown, an embodiment of the present invention provides an attention mechanism-based power transmission and transformation project settlement over-estimate early warning device, comprising:

[0164] A first settlement data set establishment module is used to obtain first settlement data of the power transmission and transformation project, and update and record in real time to establish a first settlement data set, wherein the first settlement data includes the project approval budget, change certification cost, site construction fee, equipment and material prices, bid discount coefficient, market material price, bid quotation and tender control price, and the first settlement data set includes historical data of each settlement time point of the power transmission and transformation project;

[0165] An outlier and missing value processing module, used to first perform outlier processing on the first settlement data set through a box plot, then perform missing value processing on the first settlement data set through a mean filling method, and use the first settlement data after outlier processing and missing value processing as a second standard data set;

[0166] An attention weight generation module is used to perform weight analysis on the second standard data set based on the AM attention mechanism to generate attention weight data, and perform attention weighting on the second standard data set to obtain a third weighted data set;

[0167] A settlement fee prediction model building module is used to build a settlement fee prediction model based on the third weighted data set based on the deep learning model and taking the settlement fee as the target;

[0168] An estimated settlement cost calculation module, used to input the real-time first settlement data into a settlement cost prediction model to calculate the estimated settlement cost of the power transmission and transformation project;

[0169] The project settlement over-budget warning module is used to compare the estimated settlement cost with the project approved budget to obtain a fourth comparison result, and automatically trigger a corresponding over-budget warning according to the fourth comparison result.

[0170] The specific process of using the power transmission and transformation project settlement over-budget warning device based on the AM attention mechanism provided by an embodiment of the present invention to perform a power transmission and transformation project settlement over-budget warning is as follows.

[0171] 1. Obtain the first settlement data of the power transmission and transformation project, update and record it in real time, and establish the first settlement data set.

[0172] The first settlement data includes the project approval estimate, change visa cost, site construction fee, equipment and material prices, winning bid discount coefficient, market material price, bid quotation and tender control price; the first settlement data set includes the first settlement data at each historical time point.

[0173] The basic data required for the present invention are shown in Table 1.

[0174] Table 1: Basic data

[0175]

[0176] The present invention realizes the real-time acquisition and update of data such as change visas, site construction fees, equipment material information prices, etc.; through real-time data update, the latest data information can be obtained in time, and is no longer limited by periodic data collection, thus solving the problem of poor real-time performance in the prior art.

[0177] 2. The first settlement data set is processed for outliers through box plots, and the missing values ​​are processed through mean filling method to obtain the second standard data set.

[0178] Outliers are processed through box plots, as follows:

[0179] Sort the first settlement data set, calculate the first quartile and the third quartile of the first settlement data set; calculate the difference between the third quartile and the first quartile as the interquartile range; use 1.5 times the interquartile range to determine the lower and upper boundaries of the outliers; define the values ​​in the first settlement data that are below the lower boundary or above the upper boundary as outliers; and delete the outliers.

[0180] The missing values ​​are processed by the mean filling method, specifically:

[0181] The first settlement data set is represented in the form of a two-dimensional table; for each column of data, the number of missing values ​​is counted; for each column with missing values, its mean is calculated, and the missing values ​​are filled with the mean.

[0182] The first settlement data after outlier processing and missing value processing is used as the second standard data set.

[0183] 3. Based on the AM attention mechanism, a weight analysis is performed on the second standard data set to generate attention weight data, and the second standard data set is attention-weighted to obtain the third weighted data set.

[0184] The number of second standard data is taken as rows and the data type is taken as columns to map to a second standard data matrix; the second standard data matrix is ​​linearly transformed to obtain a query matrix, a key matrix and a value matrix; the attention score is calculated according to the dot product of the query matrix and the key matrix; the attention score is normalized to obtain the attention weight; and the value matrix is ​​weighted and summed according to the attention weight to obtain the third weighted data.

[0185] Perform linear transformation on the data matrix to obtain the query matrix, key matrix, and value matrix. The specific calculation formula is as follows:

[0186] Z=S*F Z +v Z

[0187] X=S*F X +v X

[0188] C=S*FC +v C

[0189] In this embodiment, the specific calculation formula of the attention score is as follows:

[0190]

[0191] In this embodiment, the specific calculation formula of the attention weight is as follows:

[0192]

[0193] In this embodiment, the specific calculation formula of the third weighted data is as follows:

[0194]

[0195] The softmax function in this embodiment is:

[0196]

[0197] Where Z is the query matrix, X is the key matrix, C is the value matrix, S is the second standard data matrix, and F Z is a shape of (d, d k )’s query weight matrix, F X is a shape of (d, d k )’s key weight matrix, F C is a shape of (d, d v ) value weight matrix, v Z 、v X 、v C is the bias term, d k and d v is the dimension after linear transformation, α Score for attention, zyl εε is the attention weight; β is the third weighted data of a specific data type, softmax(w i ) is the softmax function, w i Any function parameter.

[0198] Fourth, the third weighted data set is based on the deep learning model to build a settlement cost prediction model with settlement costs as the target.

[0199] The third weighted data in the third weighted data set is processed into a time series format; the time dependency of each type of data in the third weighted data after the arrangement is captured; the time dependency is mapped to the preset estimated settlement cost calculation formula of each project type to obtain a settlement cost prediction model. The various project types include substation projects, cable projects, and overhead line projects.

[0200] The specific calculation formula for the estimated settlement cost of substation engineering or cable engineering is as follows:

[0201] H=(K JZ +P JZ )+(K AZ +P AZ )+δ+(K QT +ε)

[0202] The specific calculation formula for the estimated settlement cost of overhead line projects is as follows:

[0203] G=(K BT +SF S )+SF JG +(K QT +ε)+P JK

[0204] It should be noted that the converted value of each project cost = each project cost * (bid quotation / bidding control price) * 100%;

[0206] Where H is the estimated settlement cost of the substation project or cable project, K JZ is the converted value of construction engineering cost, P JZ For the change certification of construction projects, K AZ is the converted value of installation engineering cost, P AZ is the change visa for the installation project, δ is the increase in equipment purchase cost, K QT is the conversion value of other costs, ε is the cost change correction factor, G is the estimated settlement cost of the overhead line project, K BT is the converted value of the main engineering cost, SF S SF is the floating coefficient of equipment materials without main materials supplied by Party A. JG P is the floating coefficient of the main material cost supplied by Party A, JK To provide certification for changes in overhead line projects.

[0207] The specific process of capturing the time dependency of each type of data of the sorted third weighted data is as follows:

[0208] Obtain input data, and input the input data of t time steps into a preset processing unit, the preset processing unit including an input gate, a forget gate and an output gate, the input data including a third weighted data, an estimated settlement fee of a previous time step and a memory cell state of a previous time step; calculate an introduction coefficient for controlling the input data through the input gate; calculate a retention coefficient for controlling the memory cell state of the previous time step through the forget gate; calculate candidate memory cell states for the current input data; calculate the memory cell state of the current time step by combining the introduction coefficient, the retention coefficient and the candidate memory cell states; calculate an estimated influence coefficient of the memory cell state on the estimated settlement fee through the output gate; calculate the estimated settlement fee of the current time step by combining the estimated influence coefficient, the estimated settlement fee of the previous time step and the memory cell state of the current time step.

[0209] The input gate calculation formula is as follows:

[0210] Y t =μ(M SR *[ω t-1 , X t ])

[0211] The calculation formula of the forget gate is as follows:

[0212] BL t =μ(M YW *[ω t-1 , X t ])

[0213] The candidate memory cell state calculation formula is as follows:

[0214]

[0215] The memory cell state calculation formula is as follows:

[0216]

[0217] The output gate calculation formula is as follows:

[0218] YS t =μ(M SC *[ω t-1 , X t ])

[0219] The calculation formula for the estimated settlement fee is as follows:

[0220] ω t =YS t *tanh(B t )

[0221] In the formula, YR tis the input data introduction coefficient, μ() is the sigmoid activation function, M SR is the weight matrix of the input gate, ω t-1 is the estimated settlement fee at time step t-1, X t is the third weighted data, BL t is the retention coefficient of the memory cell state at the previous time step, M YW is the weight matrix of the forget gate, is the candidate memory cell state, tanh() is the tanh activation function, M HX is the weight matrix for selecting the memory cell state, B t is the memory cell state at time step t, B t-1 The state of the memory cell at the t-1 time step is the state of the memory cell at the previous time step, YS t is the estimated impact coefficient of the memory cell status on the estimated settlement fee, M SC is the weight matrix of the output gate, ω t is the estimated settlement fee at time step t.

[0222] The specific mathematical expression of the sigmoid activation function is as follows:

[0223]

[0224] In the formula, e is the base of the natural logarithm, k is the input value; the graph of the sigmoid function presents an S-shaped curve, so it is also called the S-type function;

[0225] The mathematical expression of the tanh activation function is as follows:

[0226]

[0227] The graph of the tanh activation function also shows an S-shaped curve, similar to the sigmoid function, but its output range is wider, between -1 and 1.

[0228] 5. Input the real-time first settlement data into the settlement cost prediction model to calculate the estimated settlement cost of the power transmission and transformation project.

[0229] 1. Estimated settlement cost of substation engineering / cable engineering = ((converted value of construction engineering cost + change certificate) + (converted value of installation engineering cost + change certificate) + (increase coefficient of equipment purchase cost) + (converted value of other costs + cost changes));

[0230] 2. Estimated settlement cost of overhead line project = (conversion value of main project cost (excluding main materials supplied by Party A) and equipment material floating coefficient) + main materials supplied by Party A floating coefficient + (other costs + changes in site construction costs) + change visa fees;

[0231] The converted value of each sub-item cost of the project = (the bid quotation of each sub-item cost / the bidding control price of each sub-item cost) * 100%.

[0232] 6. Compare the estimated settlement costs with the approved budget estimate of the project to obtain a fourth comparison result, and automatically trigger a corresponding over-budget warning based on the fourth comparison result.

[0233] The fourth comparison result includes a slightly over-estimated result, a moderately over-estimated result, and a severely over-estimated result.

[0234] The corresponding over-estimate warning is automatically triggered according to the fourth comparison result, as follows:

[0235] The estimated settlement costs will be compared with the approved project budget. When the amount of the excess is less than 2% of the approved project budget and the excess amount is less than RMB 1 million, the estimated settlement excess will be marked as a minor excess and a first-level warning will be issued; when the amount of the excess is greater than 10% of the approved project budget or the excess amount is greater than RMB 5 million, the estimated settlement excess will be marked as a severe excess and a third-level warning will be issued; when the estimated settlement excess does not fall into the category of minor excess or severe excess, the estimated settlement excess will be marked as a moderate excess and a second-level warning will be issued.

[0236] Combined with the data acquired in real time, the present invention can calculate the settlement costs of the project in real time and compare and analyze it with the budget estimate. Once the budget is exceeded, the system will immediately issue an alarm to help the management unit grasp the situation of the project exceeding the budget at the first time, and realize the function of triggering the over-budget early warning. By calculating in real time whether the project exceeds the budget, the present invention can automatically identify the over-budget risk factor items, avoid the shortcomings and subjectivity of manual experience, and improve the accuracy of identifying the over-budget risk. The present invention can not only identify the over-budget risk factor items, but also comprehensively evaluate the risk of cost over-budget. By comprehensively considering various data and factors, a more comprehensive and accurate project over-budget risk assessment is provided to the management unit.

[0237] According to the proportion and amount of project overruns, project overrun warnings are divided into three levels, as shown in Table 2.

[0238] Table 2: Warning indicators and levels

[0239]

[0240]

[0241] The effects that the present invention can achieve are as follows:

[0242] 1. Social effects:

[0243] Improve project management efficiency: Real-time data acquisition and updating enables timely understanding and warning of project overruns, which helps improve project management efficiency.

[0244] Reduce over-estimate risks: By automatically identifying over-estimate risk factors and comprehensively assessing risks, it can help management units to promptly discover and resolve project over-estimate problems, thereby reducing the occurrence of over-estimate risks.

[0245] 2. Economic effect:

[0246] Cost saving: timely understanding of project overruns will help management units take timely measures to avoid additional expenses and save unnecessary costs.

[0247] Improve the accuracy of investment decisions: The present invention provides management units with real-time project overrun conditions, which helps them more accurately assess and control risks in investment decisions, thereby improving the accuracy and efficiency of investment decisions.

[0248] 3. Technical effects:

[0249] Improved real-time computing capability: The present invention realizes real-time computing of engineering over-estimation, improves the real-time computing capability of the system, and meets the real-time requirements of modern engineering management.

[0250] Automated identification and early warning function: By automatically identifying excessive risk factors and triggering excessive early warnings, the need for manual intervention is reduced, and management efficiency and work accuracy are improved.

[0251] A computer device provided in an embodiment of the present invention includes a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor. When the device is running, the processor communicates with the memory through the bus, and the processor executes the machine-readable instructions to perform any of the steps of the above-mentioned power transmission and transformation project settlement over-estimate early warning method based on the attention mechanism.

[0252] Specifically, the above-mentioned memory and processor can be general-purpose memory and processor, which are not specifically limited here. When the processor runs the computer program stored in the memory, it can execute the above-mentioned power transmission and transformation project settlement over-estimate early warning method based on the attention mechanism.

[0253] Those skilled in the art will appreciate that the structure of the computer device does not constitute a limitation on the computer device, and may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or arrange the components differently.

[0254] In some embodiments, the computer device may also include a touch screen that can be used to display a graphical user interface (e.g., a startup interface of an application) and receive user operations on the graphical user interface (e.g., startup operations for an application). The specific touch screen may include a display panel and a touch panel. The display panel may be configured in the form of LCD (Liquid Crystal Display), OLED (Organic Light-Emitting Diode), etc. The touch panel may collect the user's contact or non-contact operations on or near it, and generate pre-set operation instructions, for example, the user uses any suitable object such as a finger or stylus to operate on or near the touch panel. In addition, the touch panel may include two parts: a touch detection device and a touch controller. The touch detection device detects the user's touch position and posture, detects the signal brought by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into information that the processor can process, and then sends it to the processor, and can receive and execute commands sent by the processor. In addition, the touch panel can be implemented by various types such as resistive, capacitive, infrared and surface acoustic wave, and any technology developed in the future can also be used to implement the touch panel. Further, the touch panel can cover the display panel, and the user can operate on or near the touch panel covered on the display panel according to the graphical user interface displayed on the display panel. After the touch panel detects the operation on or near it, it is transmitted to the processor to determine the user input, and then the processor provides corresponding visual output on the display panel in response to the user input. In addition, the touch panel and the display panel can be implemented as two independent components or integrated.

[0255] Corresponding to the method for starting the above-mentioned application, an embodiment of the present invention further provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods for over-estimated early warning of power transmission and transformation project settlement based on the attention mechanism are executed.

[0256] The application startup device provided in the embodiment of the present application can be specific hardware on the device or software or firmware installed on the device. The device provided in the embodiment of the present application, its implementation principle and the technical effect produced are the same as those in the aforementioned method embodiment. For the sake of brief description, the parts not mentioned in the device embodiment can refer to the corresponding contents in the aforementioned method embodiment. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can all refer to the corresponding processes in the aforementioned method embodiment, and will not be repeated here.

[0257] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0258] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0259] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0260] In addition, each functional module in the embodiments provided in the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0261] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0262] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0263] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0264] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for early warning of over-estimated settlement of power transmission and transformation projects based on attention mechanism, characterized in that: The steps include: Step S1, obtaining first settlement data of the power transmission and transformation project, and updating and recording in real time to establish a first settlement data set, wherein the first settlement data includes the project approval budget, change visa cost, site construction fee, equipment and material prices, bid discount coefficient, market material price, bid quotation and tender control price, and the first settlement data set includes historical data of each settlement time point of the power transmission and transformation project; Step S2, firstly, performing outlier processing on the first settlement data set by using a box plot, and then performing missing value processing on the first settlement data set by using a mean filling method, and using the first settlement data after the outlier processing and the missing value processing as the second standard data set; Step S3, based on the AM attention mechanism, performing weight analysis on the second standard data set to generate attention weight data, and performing attention weighting on the second standard data set to obtain a third weighted data set; Step S4, constructing a settlement fee prediction model based on the third weighted data set based on the deep learning model and taking the settlement fee as the target; Step S5, inputting the real-time first settlement data into the settlement cost prediction model to calculate the estimated settlement cost of the power transmission and transformation project; Step S6, comparing the estimated settlement cost with the approved budget estimate of the project to obtain a fourth comparison result, and automatically triggering a corresponding over-budget warning according to the fourth comparison result.

2. The method for early warning of over-estimated settlement of power transmission and transformation projects based on attention mechanism according to claim 1 is characterized in that: In step S2, the outlier processing of the first settlement data set by using a box plot includes the following steps: sorting the first settlement data set, and calculating the first quartile and the third quartile of the first settlement data set; The interquartile range was calculated by calculating the difference between the third quartile and the first quartile; The lower and upper boundaries of outliers were determined using 1.5 times the interquartile range; The values ​​in the first settlement data that are lower than the lower boundary or higher than the upper boundary are regarded as outliers and are deleted.

3. The method for early warning of over-estimated settlement of power transmission and transformation projects based on attention mechanism according to claim 1 is characterized in that: In step S2, the missing value processing is performed on the first settlement data set by using the mean filling method, including the following steps: representing the first settlement data set in the form of a two-dimensional table; For each column of data, count the number of missing values; For each column with missing values, calculate its mean and fill the missing values ​​with the mean.

4. The method for early warning of over-estimated settlement of power transmission and transformation projects based on attention mechanism according to claim 1 is characterized in that: The step S3 comprises the following steps: Mapping the number of the second standard data as rows and the data type as columns into a second standard data matrix; Perform a linear transformation on the second standard data matrix to obtain the query matrix, key matrix, and value matrix: Z=S*F Z +v Z X=S*F X +v X C=S*F C +v C Where Z is the query matrix, X is the key matrix, C is the value matrix, S is the second standard data matrix, and F Z is a shape of (d, d k )’s query weight matrix, F X is a shape of (d, d k )’s key weight matrix, F C is a shape of (d, d v ) value weight matrix, v Z 、v X 、v C is the bias term, d k and d v is the dimension after linear transformation; The attention score is calculated based on the dot product of the query matrix and the key matrix: Where α is the attention score; Normalize the attention score to get the attention weight: In the formula, zyl εε is the attention weight, softmax(w i ) is the softmax function, wi is any function parameter; The third weighted data is obtained by weighted summing of the value matrix according to the attention weight: Wherein, β is the third weighted data.

5. The method for early warning of over-estimated settlement of power transmission and transformation projects based on attention mechanism according to claim 1 is characterized in that: The step S4 comprises the following steps: Performing time series processing on the third weighted data in the third weighted data set and arranging the data into a time series format; Capturing the time dependency of each type of data in the sorted third weighted data; Map the time dependency to the preset calculation formula for estimated settlement costs of each project type to obtain the settlement cost prediction model: H=(K JZ +P JZ )+(K AZ +P AZ )+δ+(K QT +e) G=(K BT +SF S )+SF JG +(K QT +ε)+P JK Where H is the estimated settlement cost of the substation project or cable project, K JZ is the converted value of construction engineering cost, P JZ For the change certification of construction projects, K AZ is the converted value of installation engineering cost, P AZ is the change visa for the installation project, δ is the increase in equipment purchase cost, K QT is the conversion value of other costs, ε is the cost change correction factor, G is the estimated settlement cost of the overhead line project, K BT is the converted value of the main engineering cost, SF S SF is the floating coefficient of equipment materials without main materials supplied by Party A. JG P is the floating coefficient of the main material cost supplied by Party A, JK To apply for change certification for overhead line projects; the types of projects described include substation projects, cable projects and overhead line projects.

6. The method for early warning of over-estimated settlement of power transmission and transformation projects based on attention mechanism according to claim 5 is characterized in that: The capturing of the time dependency of each type of data of the sorted third weighted data comprises the following steps: Obtain input data, input the input data of time step t into a preset processing unit, the preset processing unit includes an input gate, a forget gate, and an output gate, the input data includes the third weighted data, the estimated settlement fee of the previous time step, and the memory cell state of the previous time step; The input gate calculates the introduced coefficients of the control input data; The retention coefficient of the memory cell state at the previous time step is controlled by calculating the forget gate; Calculate the candidate memory cell states for the current input data; The memory cell state at the current time step is calculated by combining the introduction coefficient, the retention coefficient and the candidate memory cell state; The estimated influence coefficient of the memory cell state on the estimated settlement fee is calculated by the output gate; The estimated settlement fee of the current time step is calculated by combining the estimated impact coefficient, the estimated settlement fee of the previous time step, and the memory cell state of the current time step.

7. The method for early warning of over-estimated settlement of power transmission and transformation projects based on attention mechanism according to any one of claims 1 to 6, characterized in that: In step S6, the fourth comparison result includes a slightly over-generalized result, a moderately over-generalized result, and a severely over-generalized result; The automatically triggering of the corresponding over-generalization warning according to the fourth comparison result comprises the following steps: Compare the estimated settlement cost with the approved budget of the project. When the amount of the excess is less than 2% of the approved budget of the project and the excess amount is less than 1 million, the estimated settlement excess will be marked as a minor excess and a first-level warning will be issued. When the amount of the over-budget exceeds 10% of the approved budget of the project or the over-budget amount exceeds RMB 5 million, the estimated settlement over-budget will be marked as a severe over-budget result and a Level 3 warning will be issued; When the estimated settlement excess does not fall into the category of slight excess or severe excess, the estimated settlement excess will be marked as moderate excess and a level 2 warning will be issued.

8. A power transmission and transformation project settlement over-estimate early warning device based on attention mechanism, characterized in that: include: A first settlement data set establishment module is used to obtain first settlement data of the power transmission and transformation project, and update and record in real time to establish a first settlement data set, wherein the first settlement data includes the project approval budget, change certification cost, site construction fee, equipment and material prices, bid discount coefficient, market material price, bid quotation and tender control price, and the first settlement data set includes historical data of each settlement time point of the power transmission and transformation project; An outlier and missing value processing module, used to first perform outlier processing on the first settlement data set through a box plot, then perform missing value processing on the first settlement data set through a mean filling method, and use the first settlement data after outlier processing and missing value processing as a second standard data set; An attention weight generation module, used to perform weight analysis on the second standard data set based on the AM attention mechanism, generate attention weight data, and perform attention weighting on the second standard data set to obtain a third weighted data set; A settlement fee prediction model building module is used to build a settlement fee prediction model based on the third weighted data set based on the deep learning model and taking the settlement fee as the target; An estimated settlement cost calculation module, used to input the real-time first settlement data into a settlement cost prediction model to calculate the estimated settlement cost of the power transmission and transformation project; The project settlement over-budget warning module is used to compare the estimated settlement cost with the project approved budget to obtain a fourth comparison result, and automatically trigger a corresponding over-budget warning according to the fourth comparison result.

9. A computer device, characterized in that: It includes a processor, a memory and a bus, the memory stores machine-readable instructions executable by the processor, when the computer device is running, the processor and the memory communicate through the bus, and the processor executes the machine-readable instructions to perform the steps of the over-estimate early warning method for power transmission and transformation project settlement based on the attention mechanism as described in any one of claims 1-7.

10. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by a processor, executes the steps of the attention mechanism-based over-estimate early warning method for power transmission and transformation project settlement as described in any one of claims 1 to 7.

Images