Project performance evaluation method and system based on artificial intelligence

Abstract

The invention discloses a project performance evaluation method and system based on artificial intelligence, and relates to the technical field of big data. Dividing the total construction project according to the construction project to obtain a plurality of sub-projects, and constructing each sub-project to obtain a plurality of topological graphs; obtaining a key index of each sub-project, and adding the key index as a feature tag into the topological graph to obtain a feature topological graph; performing feature extraction on the feature topological graph through a first preset model to obtain a feature vector set, and inputting the feature vector set into a second preset model for evaluation to obtain a project score; a total construction project is divided to obtain a plurality of sub-projects, and a topological graph is constructed, so that clear display of an internal structure of the project is realized, key indexes are characterized and added into the topological graph, quantitative description is provided for project characteristics, a model is used for feature extraction and evaluation, and objectivity and consistency of evaluation results are ensured.

Description

Technical Field

[0001] The present invention belongs to the field of big data technology, and specifically relates to a project performance evaluation method and system based on artificial intelligence. Background Art

[0002] The traditional human approach to performance evaluation of natural resource financial projects faces many challenges, especially in engineering construction projects, such as disaster management, village relocation and other facility construction. Due to the large amount of calculations and the proneness to errors, coupled with the subjectivity of human evaluation, these factors may lead to inaccurate evaluation results. In order to optimize the performance evaluation process, the industry has begun to explore more scientific and objective evaluation methods.

[0003] Construction projects often include multiple sub-projects. The existing project performance evaluation is performed through human subjectivity or artificial intelligence models to evaluate the project as a whole, resulting in inaccurate and inefficient project performance evaluation. The lack of analysis of the relationship between sub-tasks makes this evaluation method limited and objective. Summary of the invention

[0004] The purpose of the present invention is to solve the problem and to propose a project performance evaluation method and system based on artificial intelligence.

[0005] In a first aspect of the implementation of the present invention, a project performance evaluation method based on artificial intelligence is first proposed, the method comprising: Dividing the total construction project into multiple sub-projects according to the construction projects, and constructing each sub-project to obtain multiple topological graphs; Obtain key indicators of each sub-project, and add the key indicators as feature tags to the topology map to obtain a feature topology map; the key indicators are cost data of the project in the planning stage and the construction stage; The first preset model is used to extract features from the feature topology map to obtain a feature vector set, and the feature vector set is input into the second preset model for evaluation to obtain a project score.

[0006] Optionally, before extracting features from the feature topology map using the first preset model to obtain a feature vector set, the method further includes: Obtain a historical feature topology map, input the historical feature topology map into a first preset model through an input layer to obtain an input matrix, and perform adaptive feature processing on the input matrix through an embedding layer and a convolution layer to obtain a feature map; The feature map is input into the pooling layer for downsampling to obtain a downsampled feature map, and the downsampled feature map is subjected to matrix multiplication by the fully connected layer to obtain the classification result; Output the classification result through the output layer, update the model through back propagation according to the classification result to obtain a model update parameter, and update the graph neural network model according to the model update parameter to obtain a first preset model; The feature topology map is input into the first preset model to extract features to obtain a feature vector set.

[0007] Optionally, adaptive feature processing is performed on the input matrix through an embedding layer and a convolution layer to obtain a feature map, including: Preprocessing the historical feature topology map to obtain a target topology map, and inputting the target topology map into the first preset model through the input layer for processing; The feature matrix is ​​processed by one-dimensional convolution to obtain a first feature matrix, the first feature matrix is ​​fused with the historical feature matrix to obtain a second feature matrix, and the second feature matrix is ​​input into the convolution layer for training; Mapping each node in the target topology graph to a low-dimensional vector space to obtain an embedding matrix, determining a transposed matrix of the embedding matrix according to the embedding matrix, and multiplying the embedding matrix and the transposed matrix to obtain a target matrix; Determine the adjacency matrix of the target topology graph, perform Hadamard basis operation on the target matrix and the adjacency matrix to obtain the target adjacency matrix, and input the target adjacency matrix into the convolution layer for training; The second feature matrix and the target adjacency matrix are input into the convolutional layer to obtain the feature map.

[0008] Optionally, before inputting the feature vector set into the second preset model for evaluation to obtain the project score, the method includes: Obtaining a preset scoring rule and a historical feature vector set, and inputting the preset scoring rule and the historical feature vector into a machine learning model for training; Using the output of the machine learning model as a model parameter, and iteratively updating the machine learning model according to the model parameter to obtain a second preset model; The feature topology map is input into the second preset model for evaluation to obtain the project score.

[0009] Optionally, the formulation of the preset scoring rules includes: Determine the first budget cost and actual cost corresponding to the actual workload of the sub-project at a certain moment, the second budget cost corresponding to the workload that should be completed according to the plan at a certain moment, the number of reworks and the total number of constructions of the sub-project based on the key indicators; A cost performance coefficient is calculated based on the first budget cost and the actual cost, a schedule performance coefficient is calculated based on the first budget cost and the second budget cost, a project rework coefficient is calculated based on the number of reworks and the total number of constructions, and a preset scoring rule is determined based on the cost performance coefficient, the schedule performance coefficient, and the project rework coefficient.

[0010] In a second aspect of the present invention, a project performance evaluation system based on artificial intelligence is proposed, including: a topology map construction module, a feature label adding module and a project performance evaluation module: The topology map construction module is used to divide the total construction project into multiple sub-projects according to the construction projects, and construct multiple topology maps for each sub-project; The feature tag adding module is used to obtain the key indicators of each sub-project, and add the key indicators as feature tags to the topology map to obtain the feature topology map; the key indicators are the cost data of the project in the planning stage and the construction stage; The project performance evaluation module is used to extract features from the feature topology map through a first preset model to obtain a feature vector set, and input the feature vector set into a second preset model for evaluation to obtain a project score.

[0011] Optionally, the system further includes: a feature map acquisition module, a downsampling processing module, a model updating module and a feature extraction module: The feature map acquisition module is used to acquire a historical feature topology map, input the historical feature topology map into the first preset model through the input layer to obtain an input matrix, and perform adaptive feature processing on the input matrix through the embedding layer and the convolution layer to obtain a feature map; The downsampling processing module is used to input the feature map into the pooling layer for downsampling processing to obtain a downsampled feature map, and perform matrix multiplication operation on the downsampled feature map through the fully connected layer to obtain a classification result; The model updating module is used to output the classification result through the output layer, update the model update parameters according to the classification result through back propagation, and update the graph neural network model according to the model update parameters to obtain the first preset model; The feature extraction module is used to input the feature topology map into the first preset model to perform feature extraction to obtain a feature vector set.

[0012] Optionally, the feature map acquisition module includes: a preprocessing module, a feature fusion module, a node embedding module, a matrix operation module and a convolutional layer input module: The preprocessing module is used to preprocess the historical feature topology map to obtain a target topology map, and input the target topology map into the first preset model for processing through the input layer; The feature fusion module is used to process the feature matrix through one-dimensional convolution to obtain a first feature matrix, fuse the first feature matrix with the historical feature matrix to obtain a second feature matrix, and input the second feature matrix into the convolution layer for training; The node embedding module is used to map each node in the target topological graph to a low-dimensional vector space to obtain an embedding matrix, determine a transposed matrix of the embedding matrix according to the embedding matrix, and multiply the embedding matrix and the transposed matrix to obtain a target matrix; The matrix operation module is used to determine the adjacency matrix of the target topological graph, perform Hadamard basis operation on the target matrix and the adjacency matrix to obtain the target adjacency matrix, and input the target adjacency matrix into the convolution layer for training; The convolution layer input module is used to input the second feature matrix and the target adjacency matrix into the convolution layer to obtain a feature map.

[0013] Optionally, the project performance evaluation module includes: a historical data training module, a model updating module and a project scoring module: The historical data training module is used to obtain preset scoring rules and historical feature vector sets, and input the preset scoring rules and historical feature vectors into the machine learning model for training; The model updating module is used to use the output of the machine learning model as a model parameter, and iteratively update the machine learning model according to the model parameter to obtain a second preset model; The project scoring module is used to input the feature topology map into the second preset model for evaluation to obtain a project score.

[0014] Optionally, the formulation of the preset scoring rule includes: a key indicator data determination module and a preset scoring rule determination module: The key indicator data determination module is used to determine the first budget cost and actual cost corresponding to the workload actually completed by the sub-project at a certain moment, the second budget cost corresponding to the workload that should be completed according to the plan at a certain moment, the number of rework times and the total number of construction times of the sub-project according to the key indicators; The preset scoring rule determination module is used to calculate the cost performance coefficient based on the first budget cost and the actual cost, calculate the schedule performance coefficient based on the first budget cost and the second budget cost, calculate the project rework coefficient based on the number of reworks and the total number of constructions, and determine the preset scoring rules based on the cost performance coefficient, the schedule performance coefficient and the project rework coefficient.

[0015] Beneficial effects of the present invention: The present invention proposes a project performance evaluation method based on artificial intelligence. The method divides a total construction project into multiple sub-projects according to construction projects, and constructs multiple topological maps for each sub-project; obtains key indicators of each sub-project, and adds the key indicators as feature labels to the topological map to obtain a feature topological map; extracts features from the feature topological map through a first preset model to obtain a feature vector set, and inputs the feature vector set into a second preset model for evaluation to obtain a project score; divides the total construction project into multiple sub-projects and constructs a topological map, thereby achieving a clear display of the internal structure of the project, characterizing the key indicators and adding them to the topological map, providing a quantitative description for the project characteristics, and using the model to extract and evaluate features, thereby ensuring the objectivity and consistency of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The present invention will be further described below in conjunction with the accompanying drawings.

[0017] Figure 1 A flowchart of a project performance evaluation method based on artificial intelligence is provided for an embodiment of the present invention; Figure 2 A framework diagram of a project performance evaluation system based on artificial intelligence is provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the present invention, the description of "first", "second", etc. is only used for descriptive purposes, and cannot be understood as indicating or implying its relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0019] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0020] The embodiment of the present invention provides a project performance evaluation method based on artificial intelligence. Figure 1 , Figure 1 A flowchart of a project performance evaluation method based on artificial intelligence provided by an embodiment of the present invention. The method comprises the following steps: S101, dividing the total construction project into multiple sub-projects according to the construction projects, and constructing each sub-project to obtain multiple topological graphs; S102, obtaining key indicators of each sub-project, and adding the key indicators as feature tags to the topology map to obtain a feature topology map; S103, extracting features from the feature topology map using the first preset model to obtain a feature vector set, and inputting the feature vector set into the second preset model for evaluation to obtain a project score; The key indicators are the cost data of the project during the planning and construction stages; An artificial intelligence-based project performance evaluation method provided in an embodiment of the present invention divides the total construction project into multiple sub-projects and constructs a topological map, thereby achieving a clear display of the internal structure of the project, characterizing key indicators and adding them to the topological map, providing a quantitative description of the project characteristics, and using the model to extract and evaluate features, thereby ensuring the objectivity and consistency of the evaluation results.

[0021] In one implementation, by dividing the facilities of the construction project (the construction project is one of the projects in the construction project), multiple sub-projects are obtained, and a topology map is further constructed, which helps to clearly show the correlation and structural relationship between the facilities within the project; the key indicators of each sub-project (such as cost data in the planning stage and the construction stage) are added to the topology map as feature labels to achieve a quantitative description of the project characteristics. This characterization processing method helps the model to more accurately capture the characteristics of the project, improve the ability to understand the complexity of the project, and provide a reliable data basis for subsequent project evaluation and prediction; the first preset model is used to extract features from the feature topology map to obtain a feature vector set, and these feature vectors are input into the second preset model for evaluation to obtain a project score. This process not only simplifies the complexity of project evaluation, but also improves the accuracy and objectivity of the evaluation.

[0022] In one implementation, the construction project is divided into multiple sub-projects according to the different stages and functional modules of the project construction. For example, it can be divided into sub-projects such as power generation facility construction, transmission facility construction, distribution facility construction, and substation construction. Each sub-project can be further refined into more specific tasks or modules, such as generator installation in power generation facility construction, tower base construction in transmission line construction, etc. The key indicators are added as feature labels to the topology map to obtain a feature topology map, that is, the key indicators are added as feature labels to the nodes in the topology map to obtain a feature topology map. The machine learning model is used as the basic model of the second preset model (project performance evaluation model), and the graph neural network model is the basic model of the first preset model.

[0023] In one implementation, the topology map is mainly used to display the internal structure and process of the project in the performance evaluation of the construction project. The topology map can be used to obtain the sequence, parallel relationship and interaction between each sub-project, task or module (for each sub-project, a topology map is established based on the dependencies and logical relationships between its internal tasks or modules); Node: In the topology map, the nodes usually represent the sub-projects, tasks or modules in the construction project. Each node contains some specific information, such as task name, start time, end time, resource requirements, etc. Edge: The edge represents the dependency and logical relationship between nodes. In the topology map, the edge is usually directional, pointing from one node to another, indicating that the former is a prerequisite or dependency of the latter. The existence of the edge can clarify the execution order and dependency between the construction tasks.

[0024] In one embodiment, before extracting features from the feature topology map using the first preset model to obtain a feature vector set, the method further includes: Obtain a historical feature topology map, input the historical feature topology map into a first preset model through an input layer to obtain an input matrix, and perform adaptive feature processing on the input matrix through an embedding layer and a convolution layer to obtain a feature map; The feature map is input into the pooling layer for downsampling to obtain a downsampled feature map, and the downsampled feature map is subjected to matrix multiplication by the fully connected layer to obtain the classification result; Output the classification result through the output layer, update the model through back propagation according to the classification result to obtain the model update parameter, and update the graph neural network model according to the model update parameter to obtain the first preset model; The feature topology map is input into the first preset model to extract features to obtain a feature vector set.

[0025] In one implementation, the original data (such as topology) is extracted through features, and the vectorized features become the input layer. The input layer directly passes the input attribute value to the next layer, and the output of the input layer is the vectorized representation of the original data, that is, a feature vector or input matrix; Convolution layer: The feature vector passed from the input layer is convolved with multiple trainable filters (convolution kernels), and the output of the convolution layer is multiple feature maps, each of which corresponds to a filter. These feature maps contain local feature information of the input data.

[0026] In one implementation, the pooling layer: downsamples the feature map output by the convolution layer to reduce the dimension of the feature map and reduce the computational complexity. Common pooling operations include maximum pooling and average pooling. The output of the pooling layer is the downsampled feature map, which retains the main feature information of the input data while reducing the dimension of the data; the fully connected layer: flattens the feature map output by the pooling layer into a one-dimensional vector, and performs matrix multiplication with the weight matrix of the fully connected layer, plus the bias term, and finally performs a nonlinear transformation through an activation function (such as ReLU, Sigmoid, etc.). The output of the fully connected layer is the result of classification or regression of the feature representation of the input data. In classification tasks, the output layer usually uses the Softmax function to calculate the probability distribution of each category; in regression tasks, the output layer usually has only one neuron, and its output value represents the predicted value for the input data; back propagation is an algorithm used in neural networks to optimize model parameters. It calculates the gradient of the loss function with respect to each parameter, and then uses these gradients to update the parameters to reduce the value of the loss function.

[0027] In one embodiment, adaptive feature processing is performed on the input matrix through an embedding layer and a convolution layer to obtain a feature map, including: Preprocessing the historical feature topology map to obtain a target topology map, and inputting the target topology map into the first preset model through the input layer for processing; The feature matrix is ​​processed by one-dimensional convolution to obtain a first feature matrix, the first feature matrix is ​​fused with the historical feature matrix to obtain a second feature matrix, and the second feature matrix is ​​input into the convolution layer for training; Mapping each node in the target topology graph to a low-dimensional vector space to obtain an embedding matrix, determining a transposed matrix of the embedding matrix according to the embedding matrix, and multiplying the embedding matrix and the transposed matrix to obtain a target matrix; Determine the adjacency matrix of the target topology graph, perform Hadamard basis operation on the target matrix and the adjacency matrix to obtain the target adjacency matrix, and input the target adjacency matrix into the convolution layer for training; The second feature matrix and the target adjacency matrix are input into the convolutional layer to obtain the feature map.

[0028] In one implementation, the first feature matrix and the historical feature matrix are fused to obtain the second feature matrix. The first feature matrix refers to a matrix containing node features, and the historical feature matrix refers to the initial state of the node features before any processing; each node in the target topology graph is mapped to a low-dimensional vector space to obtain an embedding matrix: node embedding is the process of mapping the nodes in the graph to a low-dimensional vector space. This vector space can capture the characteristics and structural information of the nodes. The two-dimensional node embedding matrix refers to each node being mapped to a vector in a two-dimensional space. The embedding matrix is ​​multiplied by its own transposed matrix to generate a new matrix. This new matrix can be used to represent the relationship between nodes or the characteristics of the nodes; Hadamard product is a matrix operation that multiplies the corresponding elements of two matrices of the same shape to generate a new matrix. In graph convolutional neural networks, Hadamard product is usually used to multiply the node embedding matrix with the adjacency matrix to generate a new adjacency matrix. This new adjacency matrix can reflect the connection relationship between nodes and node characteristics.

[0029] In one implementation, by preprocessing the historical feature topology map, noise can be removed, missing values ​​can be filled, data can be standardized, etc., thereby improving data quality; by using one-dimensional convolution to process the feature matrix, local features of the data can be efficiently extracted, especially when the data has a certain sequence or one-dimensional structure, this processing method can capture important feature information; fusing the newly extracted features with the historical features can integrate new and old information, enhance the expressiveness of the features, and help the model better understand and distinguish different data patterns; mapping the nodes to a low-dimensional vector space and constructing an embedding matrix can extract the potential relationships and similarities between the nodes; by combining the target matrix and the adjacency matrix, the structural information of the graph can be introduced to enhance the model's understanding and processing capabilities of the graph data; training the model in the convolutional layer can make full use of the powerful capabilities of graph convolution, extract deep features of graph data, and learn the complex relationships between nodes, thereby improving the model's prediction performance and generalization capabilities.

[0030] In one embodiment, before inputting the feature vector set into the second preset model for evaluation to obtain the project score, the method includes: Obtaining a preset scoring rule and a historical feature vector set, and inputting the preset scoring rule and the historical feature vector into a machine learning model for training; Using the output of the machine learning model as a model parameter, iteratively updating the machine learning model according to the model parameter to obtain a second preset model; The feature topology map is input into the second preset model for evaluation to obtain the project score.

[0031] In one implementation, when training a machine learning model, the actual data of the project needs to be input as a training set. The training data may include feature vectors of a topology map (such as the structure, process, and dependencies between tasks of the project) and preset scoring rules; Data preprocessing: preprocessing the collected project data, including data cleaning, normalization, and standardization to ensure data quality and consistency. Feature extraction: extract feature vectors from the topology map. These features can describe key information such as the structure and process of the project. At the same time, the preset scoring rules can also be extracted as one of the features. Model building: select appropriate machine learning algorithms, such as support vector machines, decision trees, and random forests, to build a project performance evaluation model. These algorithms can learn and train based on the input feature vectors and target variables (such as the comprehensive performance score of the project). Model training: input the preprocessed project data as a training set into the model for training. During the training process, the model will continuously adjust its parameters and structure to minimize the prediction error and improve the prediction accuracy. Model evaluation: evaluate the trained model using evaluation methods such as cross-validation and leave-one-out to check its generalization ability and prediction accuracy. If the model performs poorly, you can adjust feature selection, algorithm parameters, etc. to improve model performance.

[0032] In one implementation, a feature topology map is generated from the subprojects, which contains the association information and topological structure between the subprojects. The feature topology map can also provide additional contextual information for the machine learning model, helping the model to better understand the intrinsic structure and feature representation of the data. The historical feature vector set is diverse and rich. Using a diverse historical feature vector set for training can enable the model to learn the feature representation of different projects and subprojects, thereby enhancing the generalization ability of the model.

[0033] In one embodiment, step S103 includes: Determine the first budget cost and actual cost corresponding to the actual workload of the sub-project at a certain moment, the second budget cost corresponding to the workload that should be completed according to the plan at a certain moment, the number of reworks and the total number of constructions of the sub-project based on the key indicators; The cost performance coefficient is calculated based on the first budget cost and the actual cost, the schedule performance coefficient is calculated based on the first budget cost and the second budget cost, the project rework coefficient is calculated based on the number of reworks and the total number of constructions, and the preset scoring rules are determined based on the cost performance coefficient, the schedule performance coefficient and the project rework coefficient; In one implementation, the preset scoring rule may be: , where F is the project score, CPI is the cost performance coefficient, SPI is the schedule performance coefficient, and PR is the project rework coefficient; cost performance coefficient = first budget cost ÷ actual cost, the larger the CPI value, the higher the cost efficiency of the project; if CPI>1, it means that the actual cost is lower than the budget cost, if CPI<1, it means that the actual cost is higher than the budget cost. Schedule performance coefficient = first budget cost ÷ second budget cost, the larger the SPI value, the higher the schedule efficiency of the project; if SPI>1, it means that the project is ahead of schedule; if SPI<1, it means that the project is behind schedule. Project rework coefficient = number of reworks ÷ total number of constructions, the larger the PR, the more reworks there are, will increase when there is no rework =1, when other variables remain unchanged, F will increase, that is, the higher the project performance score.

[0034] In one implementation, a scoring rule is preset, for example: Cost Performance Index (CPI) = 1.2 (indicates that the actual cost is 20% lower than the budgeted cost) Schedule Performance Index (SPI) = 1.1 (indicates that the project schedule is 10% ahead of schedule) Project Rework Factor (PR) = 0.1 (indicates that the number of reworks accounts for 10% of the total construction times) Project scoring score ; Cost Performance Index (CPI) = 0.9 (indicates that the actual cost is 10% higher than the budgeted cost) Schedule Performance Index (SPI) = 0.8 (indicates that the project schedule is 20% behind schedule) Project Rework Factor (PR) = 0.3 (indicates that the number of reworks accounts for 30% of the total construction times) Project Score ; Cost Performance Index (CPI) = 1.0 (indicates that the actual cost is equal to the budgeted cost) Schedule Performance Index (SPI) = 1.0 (indicates that the project progress is consistent with the plan) Project Rework Factor (PR) = 0.0 (indicates no rework) Project Score .

[0035] In one implementation, scoring is performed using quantitative indicators (such as cost performance coefficient, schedule performance coefficient, and project rework coefficient), which reduces the influence of subjective factors and improves the objectivity of the evaluation. Embedding the preset scoring rules into the machine learning model can guide the model to learn an evaluation method that is more in line with the actual situation of the project. This helps to optimize the performance of the model and improve the accuracy and reliability of the model in project performance evaluation. When the machine learning model is trained, the automated evaluation of project performance can be achieved. This can not only improve the evaluation efficiency, but also reduce the influence of human intervention and subjective factors, making the evaluation results more objective and accurate.

[0036] Based on the same inventive concept, the embodiment of the present invention also provides a project performance evaluation system based on artificial intelligence. Figure 2 , Figure 2A schematic diagram of the structure of an artificial intelligence-based project performance evaluation system provided by an embodiment of the present invention includes: a topology map construction module, a feature label adding module and a project performance evaluation module: A topology map construction module is used to divide the total construction project into multiple sub-projects according to the construction project, and construct each sub-project to obtain multiple topology maps; The feature tag adding module is used to obtain the key indicators of each sub-project, and add the key indicators as feature tags to the topology map to obtain the feature topology map; the key indicators are the cost data of the project in the planning stage and the construction stage; The project performance evaluation module is used to extract features from the feature topology map through a first preset model to obtain a feature vector set, and input the feature vector set into a second preset model for evaluation to obtain a project score.

[0037] An artificial intelligence-based project performance evaluation system provided based on an embodiment of the present invention divides the total construction project into multiple sub-projects and constructs a topological map, thereby achieving a clear display of the internal structure of the project, characterizing key indicators and adding them to the topological map, providing a quantitative description of the project characteristics, and using the model to extract and evaluate features, thereby ensuring the objectivity and consistency of the evaluation results.

[0038] In one embodiment, the system further includes: a feature map acquisition module, a downsampling processing module, a model updating module and a feature extraction module: A feature map acquisition module is used to acquire a historical feature topology map, input the historical feature topology map into a first preset model through an input layer to obtain an input matrix, and perform adaptive feature processing on the input matrix through an embedding layer and a convolution layer to obtain a feature map; A downsampling processing module is used to input the feature map into the pooling layer for downsampling processing to obtain a downsampled feature map, and perform matrix multiplication operation on the downsampled feature map through the fully connected layer to obtain a classification result; A model updating module, used to output the classification result through the output layer, update the model through back propagation according to the classification result to obtain the model update parameter, and update the graph neural network model according to the model update parameter to obtain the first preset model; The feature extraction module is used to input the feature topology map into the first preset model to perform feature extraction to obtain a feature vector set.

[0039] In one embodiment, the feature map acquisition module includes: a preprocessing module, a feature fusion module, a node embedding module, a matrix operation module and a convolutional layer input module: A preprocessing module, used for preprocessing the historical feature topology map to obtain a target topology map, and inputting the target topology map into the first preset model for processing through the input layer; A feature fusion module, used to process the feature matrix through one-dimensional convolution to obtain a first feature matrix, fuse the first feature matrix with the historical feature matrix to obtain a second feature matrix, and input the second feature matrix into the convolution layer for training; A node embedding module is used to map each node in the target topology graph to a low-dimensional vector space to obtain an embedding matrix, determine a transposed matrix of the embedding matrix according to the embedding matrix, and multiply the embedding matrix and the transposed matrix to obtain a target matrix; The matrix operation module is used to determine the adjacency matrix of the target topology graph, perform Hadamard basis operation on the target matrix and the adjacency matrix to obtain the target adjacency matrix, and input the target adjacency matrix into the convolution layer for training; The convolution layer input module is used to input the second feature matrix and the target adjacency matrix into the convolution layer to obtain a feature map.

[0040] In one embodiment, the project performance evaluation module includes: a historical data training module, a model updating module and a project scoring module: A historical data training module is used to obtain preset scoring rules and historical feature vector sets, and input the preset scoring rules and historical feature vectors into a machine learning model for training; A model updating module, used to use the output of the machine learning model as a model parameter, and iteratively update the machine learning model according to the model parameter to obtain a second preset model; The project scoring module is used to input the feature topology map into the second preset model for evaluation to obtain a project score.

[0041] In one embodiment, the formulation of the preset scoring rules includes: a key indicator data determination module and a preset scoring rule determination module: The key indicator data determination module is used to determine the first budget cost and actual cost corresponding to the actual workload of the sub-project at a certain moment, the second budget cost corresponding to the workload that should be completed according to the plan at a certain moment, the number of rework times and the total number of construction times of the sub-project according to the key indicators; A preset scoring rule determination module is used to calculate a cost performance coefficient based on the first budget cost and the actual cost, calculate a schedule performance coefficient based on the first budget cost and the second budget cost, calculate a project rework coefficient based on the number of reworks and the total number of constructions, and determine a preset scoring rule based on the cost performance coefficient, the schedule performance coefficient and the project rework coefficient.

[0042] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A project performance evaluation method based on artificial intelligence, characterized in that: The method comprises: Dividing the total construction project into multiple sub-projects according to the construction projects, and constructing each sub-project to obtain multiple topological graphs; Obtain key indicators of each sub-project, and add the key indicators as feature tags to the topology map to obtain a feature topology map; the key indicators are cost data of the project in the planning stage and the construction stage; The first preset model is used to extract features from the feature topology map to obtain a feature vector set, and the feature vector set is input into the second preset model for evaluation to obtain a project score.

2. According to the project performance evaluation method based on artificial intelligence in claim 1, it is characterized in that: Before extracting features from the feature topology map using the first preset model to obtain a feature vector set, the method further includes: Obtain a historical feature topology map, input the historical feature topology map into a first preset model through an input layer to obtain an input matrix, and perform adaptive feature processing on the input matrix through an embedding layer and a convolution layer to obtain a feature map; The feature map is input into the pooling layer for downsampling to obtain a downsampled feature map, and the downsampled feature map is subjected to matrix multiplication by the fully connected layer to obtain the classification result; Output the classification result through the output layer, update the model through back propagation according to the classification result to obtain a model update parameter, and update the graph neural network model according to the model update parameter to obtain a first preset model; The feature topology map is input into the first preset model to extract features to obtain a feature vector set.

3. According to claim 2, a project performance evaluation method based on artificial intelligence is characterized in that: The input matrix is ​​adaptively processed through the embedding layer and the convolution layer to obtain the feature map, including: Preprocessing the historical feature topology map to obtain a target topology map, and inputting the target topology map into the first preset model through the input layer for processing; The feature matrix is ​​processed by one-dimensional convolution to obtain a first feature matrix, the first feature matrix is ​​fused with the historical feature matrix to obtain a second feature matrix, and the second feature matrix is ​​input into the convolution layer for training; Mapping each node in the target topology graph to a low-dimensional vector space to obtain an embedding matrix, determining a transposed matrix of the embedding matrix according to the embedding matrix, and multiplying the embedding matrix and the transposed matrix to obtain a target matrix; Determine the adjacency matrix of the target topology graph, perform Hadamard basis operation on the target matrix and the adjacency matrix to obtain the target adjacency matrix, and input the target adjacency matrix into the convolution layer for training; The second feature matrix and the target adjacency matrix are input into the convolutional layer to obtain the feature map.

4. According to the project performance evaluation method based on artificial intelligence in claim 1, it is characterized in that: Before inputting the feature vector set into the second preset model for evaluation to obtain the project score, the method includes: Obtaining a preset scoring rule and a historical feature vector set, and inputting the preset scoring rule and the historical feature vector into a machine learning model for training; Using the output of the machine learning model as a model parameter, and iteratively updating the machine learning model according to the model parameter to obtain a second preset model; The feature topology map is input into the second preset model for evaluation to obtain the project score.

5. The project performance evaluation method based on artificial intelligence according to claim 1 is characterized in that: The formulation of the preset scoring rules includes: Determine the first budget cost and actual cost corresponding to the actual workload of the sub-project at a certain moment, the second budget cost corresponding to the workload that should be completed according to the plan at a certain moment, the number of reworks and the total number of constructions of the sub-project based on the key indicators; A cost performance coefficient is calculated based on the first budget cost and the actual cost, a schedule performance coefficient is calculated based on the first budget cost and the second budget cost, a project rework coefficient is calculated based on the number of reworks and the total number of constructions, and a preset scoring rule is determined based on the cost performance coefficient, the schedule performance coefficient, and the project rework coefficient.

6. A project performance evaluation system based on artificial intelligence, characterized in that: The system includes: a topology map building module, a feature tag adding module and a project performance evaluation module: The topology map construction module is used to divide the total construction project into multiple sub-projects according to the construction projects, and construct multiple topology maps for each sub-project; The feature tag adding module is used to obtain the key indicators of each sub-project, and add the key indicators as feature tags to the topology map to obtain the feature topology map; the key indicators are the cost data of the project in the planning stage and the construction stage; The project performance evaluation module is used to extract features from the feature topology map through a first preset model to obtain a feature vector set, and input the feature vector set into a second preset model for evaluation to obtain a project score.

7. The project performance evaluation system based on artificial intelligence according to claim 6 is characterized in that: The system also includes: a feature map acquisition module, a downsampling processing module, a model updating module and a feature extraction module: The feature map acquisition module is used to acquire a historical feature topology map, input the historical feature topology map into the first preset model through the input layer to obtain an input matrix, and perform adaptive feature processing on the input matrix through the embedding layer and the convolution layer to obtain a feature map; The downsampling processing module is used to input the feature map into the pooling layer for downsampling processing to obtain a downsampled feature map, and perform matrix multiplication operation on the downsampled feature map through the fully connected layer to obtain a classification result; The model updating module is used to output the classification result through the output layer, update the model update parameters according to the classification result through back propagation, and update the graph neural network model according to the model update parameters to obtain the first preset model; The feature extraction module is used to input the feature topology map into the first preset model to perform feature extraction to obtain a feature vector set.

8. The project performance evaluation system based on artificial intelligence according to claim 7 is characterized in that: The feature map acquisition module includes: a preprocessing module, a feature fusion module, a node embedding module, a matrix operation module and a convolutional layer input module: The preprocessing module is used to preprocess the historical feature topology map to obtain a target topology map, and input the target topology map into the first preset model for processing through the input layer; The feature fusion module is used to process the feature matrix through one-dimensional convolution to obtain a first feature matrix, fuse the first feature matrix with the historical feature matrix to obtain a second feature matrix, and input the second feature matrix into the convolution layer for training; The node embedding module is used to map each node in the target topological graph to a low-dimensional vector space to obtain an embedding matrix, determine a transposed matrix of the embedding matrix according to the embedding matrix, and multiply the embedding matrix and the transposed matrix to obtain a target matrix; The matrix operation module is used to determine the adjacency matrix of the target topological graph, perform Hadamard basis operation on the target matrix and the adjacency matrix to obtain the target adjacency matrix, and input the target adjacency matrix into the convolution layer for training; The convolution layer input module is used to input the second feature matrix and the target adjacency matrix into the convolution layer to obtain a feature map.

9. The project performance evaluation system based on artificial intelligence according to claim 6 is characterized in that: The project performance evaluation module includes: a historical data training module, a model updating module and a project scoring module: The historical data training module is used to obtain preset scoring rules and historical feature vector sets, and input the preset scoring rules and historical feature vectors into the machine learning model for training; The model updating module is used to use the output of the machine learning model as a model parameter, and iteratively update the machine learning model according to the model parameter to obtain a second preset model; The project scoring module is used to input the feature topology map into the second preset model for evaluation to obtain a project score.

10. The project performance evaluation system based on artificial intelligence according to claim 6, characterized in that: The formulation of the preset scoring rules includes: a key indicator data determination module and a preset scoring rule determination module: The key indicator data determination module is used to determine the first budget cost and actual cost corresponding to the workload actually completed by the sub-project at a certain moment, the second budget cost corresponding to the workload that should be completed according to the plan at a certain moment, the number of rework times and the total number of construction times of the sub-project according to the key indicators; The preset scoring rule determination module is used to calculate the cost performance coefficient based on the first budget cost and the actual cost, calculate the schedule performance coefficient based on the first budget cost and the second budget cost, calculate the project rework coefficient based on the number of reworks and the total number of constructions, and determine the preset scoring rules based on the cost performance coefficient, the schedule performance coefficient and the project rework coefficient.

Images