Engineering cost management method based on artificial intelligence

Abstract

The invention provides a project cost management method based on artificial intelligence, and relates to the technical field of project cost, and the method comprises the following steps: 1, data mining and analysis: extracting valuable information from a large amount of historical project cost data through building a data mining model, and analyzing the mined data through a data analysis tool; step 2, automatic measurement and pricing; step 3, constructing a cost prediction model; step 4, risk assessment and early warning; step 5, project progress and cost collaborative management; step 6, intelligent cost control; and 7, intelligent supplier management. By constructing the data mining model and using the neural network algorithm to cooperate with the association rule mining algorithm, deep mining can be effectively performed on the engineering cost data, and meanwhile, the data analysis tool is used to analyze the data, so that the workload of data collection is reduced, and the efficiency of engineering cost management is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering cost, and in particular to an engineering cost management method based on artificial intelligence. Background Art

[0002] The project cost refers to the total cost of a certain project construction, that is, the total one-time cost of the project's planned fixed asset reproduction, the formation of corresponding intangible assets and the laying of working capital. Obviously, this meaning is defined from the perspective of the investor-owner. After the investor selects a project, he must make a decision through project evaluation, and then conduct a series of investment management activities such as design bidding, engineering bidding, and completion acceptance. All the expenses paid in the investment activities form fixed assets and intangible assets. All these expenses constitute the project cost. In this sense, the project cost is the project investment cost, and the construction project cost is the fixed asset investment of the construction project. The project cost is one of the main bases for evaluating the rationality of the total investment and sub-investment and the investment benefits. When evaluating the rationality of land prices, building installation products and equipment prices, it is necessary to use project cost data. When evaluating the loan repayment ability, profitability and macro-benefits of construction projects, the project cost can also be used.

[0003] At present, the management of project cost is mainly through the construction of project cost model to predict the total cost of the project and the cost of each stage and each sub-project based on project data such as building area, structure type, decoration standard, and construction period data. Although the project cost can be managed accurately, a large amount of manpower is needed in the early stage to collect and analyze the data that affects the project cost, and then input the analyzed data into the model. In this process, the collection and analysis of data has the greatest impact on the result of project cost, and it will also seriously reduce the efficiency of project cost management. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the shortcomings of the prior art, the present invention provides an artificial intelligence-based project cost management method, which solves the current management of project cost, which is mainly to predict the total cost of the project and the cost of each stage and each sub-project based on project data such as building area, structure type, decoration standard, and construction period data by constructing a project cost model. Although the project cost can be managed accurately, a large amount of manpower is needed in the early stage to collect and analyze the data affecting the project cost, and then input the analyzed data into the model. In this process, the collection and analysis of data have the greatest impact on the result of the project cost, and will also seriously reduce the efficiency of project cost management.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: an engineering cost management method based on artificial intelligence, comprising the following steps:

[0008] Step 1: Data mining and analysis: extract valuable information from a large amount of historical engineering cost data by establishing a data mining model, and use data analysis tools to analyze the mined data;

[0009] Step 2: Automatic quantity calculation and pricing, which uses AI-driven image recognition and natural language processing technology to automatically calculate and price the engineering quantity;

[0010] Step 3: Cost prediction model: construct a project cost prediction model based on machine learning algorithms such as linear regression, decision tree and neural network;

[0011] Step 4: Risk assessment and early warning, using artificial intelligence technology to identify, assess and warn project cost risks;

[0012] Step 5: Collaborative management of project progress and cost. Use artificial intelligence technology to establish a collaborative management model for project progress and cost, and realize dynamic correlation and mutual influence analysis between the two.

[0013] Step 6: Intelligent cost control, using artificial intelligence to achieve real-time monitoring and dynamic control of costs during project implementation;

[0014] Step 7: Intelligent supplier management: Analyze the supplier's historical data through artificial intelligence algorithms, evaluate the supplier's reputation, price level, delivery time and quality performance, and establish a supplier evaluation model.

[0015] Furthermore, the construction of the data mining model in step 1 includes the following steps:

[0016] S1: Build a web crawler program to automatically crawl web page content on the Internet according to certain rules by simulating browser behavior to collect data;

[0017] S2: Using the data collected by the web crawler program, use the data mining algorithm to further mine and collect the cost data;

[0018] S3: The collected cost data is combined with the autonomous learning model in machine learning to further improve and update the deeply mined cost data.

[0019] Furthermore, the data mining algorithm includes a neural network algorithm and an association rule mining algorithm. The association rule mining algorithm utilizes an a priori property, that is, all non-empty subsets of frequent item sets must also be frequent item sets. First, it starts from a single item and finds all frequent 1-item sets that meet the minimum support. Then, candidate 2-item sets are generated by combining frequent 1-item sets, and the support of candidate 2-item sets is calculated to screen out frequent 2-item sets. This process is repeated until no new frequent item sets can be generated.

[0020] Furthermore, the neural network algorithm needs to construct a neuron model, which is composed of multiple neurons. Each neuron receives a set of input signals, performs weighted summation on them, and converts them through a nonlinear activation function to generate an output signal. The formula is expressed as:

[0021]

[0022] in is the ith input, is the corresponding weight, is the bias term, is the activation function.

[0023] Furthermore, when establishing the neuron model, it is first necessary to select a neural network architecture, then determine the model parameters, and finally perform model training through the set parameters.

[0024] Furthermore, the neural network architecture includes a multi-layer perceptron architecture for processing general prediction and classification problems such as tabular data and having strong nonlinear fitting capabilities, a convolutional neural network architecture for performing well in processing data with a grid structure, such as images and audio, and capable of automatically extracting local features, and a recurrent neural network architecture for processing sequence data, such as time series prediction and natural language processing tasks, and capable of capturing long-term dependencies in the data.

[0025] Furthermore, the model parameters include determining the number of layers and neurons, selecting activation functions and other parameters. The number of layers and neurons determines the difficulty of processing the problem. The activation functions include Sigma, ReLU, and tanh. The other parameters include determining the learning rate, the number of iterations, and the batch size.

[0026] Furthermore, the model training uses the training set data to train the neural network model, calculates the predicted output through forward propagation, and then calculates the loss function, such as mean square error and cross entropy loss, based on the difference between the predicted output and the true label, and then uses the back propagation algorithm to calculate the gradient of the loss function to the model parameters, and updates the parameters of the model according to the gradient to minimize the loss function.

[0027] Furthermore, after the neuron model is established, it is necessary to evaluate and optimize the model. First, it is necessary to select evaluation indicators, then evaluate the model performance, and finally optimize the model.

[0028] Furthermore, firstly, according to the goal of construction cost management, appropriate evaluation indicators are selected to measure the performance of the model, such as mean square error, mean absolute error, accuracy and recall. Secondly, the trained model is evaluated using the test set, and the corresponding evaluation indicator values ​​are calculated to understand the generalization ability of the model on unseen data. Finally, according to the evaluation results, corresponding optimization measures are taken, such as adjusting the hyperparameters of the model, increasing training data, improving feature engineering and adopting regularization methods to improve the performance and generalization ability of the model.

[0029] (III) Beneficial effects

[0030] The present invention provides an engineering cost management method based on artificial intelligence, which has the following beneficial effects:

[0031] 1. In this solution, by constructing a data mining model and using a neural network algorithm in conjunction with an association rule mining algorithm, the engineering cost data can be effectively mined in depth. At the same time, data analysis tools are used to analyze the data, which helps reduce the workload of collecting data and thereby improve the efficiency of engineering cost management.

[0032] 2. In this scheme, by using a web crawler program in conjunction with a data mining algorithm in a data mining model, the effect of autonomous data mining can be effectively formed, which is conducive to improving the effect of artificial intelligence in the process of engineering cost management. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flow chart of the engineering cost management method based on artificial intelligence proposed by the present invention;

[0034] Figure 2 This is a schematic diagram of the data mining model of the engineering cost management method based on artificial intelligence proposed in the present invention. DETAILED DESCRIPTION

[0035] 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] Embodiment 1:

[0037] like Figure 1-2As shown, the embodiment of the present invention provides an engineering cost management method based on artificial intelligence, comprising the following steps:

[0038] Step 1: Data mining and analysis: extract valuable information from a large amount of historical engineering cost data by establishing a data mining model, and use data analysis tools to analyze the mined data;

[0039] The data analysis tool is Tableau, which can connect to a variety of data sources with its powerful customization capabilities and intuitive drag-and-drop interface, and display the data in the form of various exquisite charts and interactive dashboards, supporting the creation of complex visualization effects and data;

[0040] The construction of a data mining model consists of the following steps:

[0041] S1: Build a web crawler program to automatically crawl web page content on the Internet according to certain rules by simulating browser behavior to collect data. By using the Scrapy framework in Python, you can easily write a crawler program to collect data on a specific website. For example, if you want to collect information about building materials on a certain network platform, you can write corresponding crawler rules to let it automatically traverse the product page and extract data such as building material name, price, and comments;

[0042] S2: Using the data collected by the web crawler program, the data mining algorithm is used to further mine and collect the cost data. The data mining algorithm includes the neural network algorithm. The neural network algorithm needs to build a neuron model. The neuron model is composed of multiple neurons. Each neuron receives a set of input signals, performs weighted summation on them, and converts them through a nonlinear activation function to generate an output signal. The formula is expressed as:

[0043]

[0044] in is the ith input, is the corresponding weight, is the bias term, is the activation function;

[0045] S3: The collected cost data is combined with the autonomous learning model in machine learning to further improve and update the deeply mined cost data. Taking the image classification task as an example, assuming that we have a large number of unlabeled cost data images, the active learning algorithm will first use a small number of labeled images to train an initial model, and then select some of the most informative unlabeled images based on the uncertainty of the model or other strategies, and let construction engineering experts label them. The labeled images are then added to the training set, and the model is retrained. This process is repeated until the model achieves satisfactory performance.

[0046] When building a neuron model, you first need to select a neural network architecture, then determine the model parameters, and finally train the model using the set parameters. The neural network architecture includes a multi-layer perceptron architecture that is used to process general prediction and classification problems such as tabular data and has strong nonlinear fitting capabilities, a convolutional neural network architecture that performs well in processing grid-structured data such as images and audio and can automatically extract local features, and a recurrent neural network architecture that is suitable for processing sequence data such as time series prediction and natural language processing tasks and can capture long-term dependencies in data.

[0047] Integrate the trained neural network model with the company's existing engineering cost management system to achieve seamless data flow and automatic model calling. The model can be embedded in the business process through development interfaces, middleware, etc. to ensure that it can be effectively applied in actual work. Apply neural network-based solutions in actual engineering cost management projects to predict project costs, conduct risk assessments, and control costs. At the same time, establish a monitoring mechanism to track the operating effects of the model in real time, and promptly discover and solve possible problems, such as untimely data updates and excessive model prediction deviations. With the continuous accumulation of new data and changes in business needs, regularly update and optimize the model to maintain its effectiveness and adaptability in engineering cost management, continuously improve data collection and preprocessing processes, optimize feature engineering and model architecture, and improve the performance and value of the solution.

[0048] Model parameters include determining the number of layers and neurons, selecting activation functions and other parameters. The number of layers and neurons determines the difficulty of processing the problem. For simple problems, fewer hidden layers and neurons may be sufficient; for complex problems, you may need to increase the number of hidden layers and neurons to improve the expressiveness of the model. Generally speaking, you can start with a smaller number of layers and neurons, gradually increase the complexity and observe the changes in model performance;

[0049] The activation functions include Sigma ID, ReLU, and tanh. The ReLU function performs well in most cases and can effectively alleviate the gradient vanishing problem. It needs to be selected and tested according to the specific problem.

[0050] Other parameters include determining the learning rate, number of iterations, and batch size. The learning rate controls the step size of the model parameter update. Too large or too small a learning rate may affect the convergence of the model. The number of iterations determines the number of rounds of model training. Too many iterations may lead to overfitting, while too few iterations may result in the model not fully learning the characteristics of the data. The batch size affects the number of samples for each parameter update. Generally, a power of 2 is selected as the batch size, such as 32, 64, or 128.

[0051] Model training uses training set data to train the neural network model, calculates the predicted output through forward propagation, and then calculates the loss function based on the difference between the predicted output and the true label, such as mean square error and cross entropy loss. Then, the back propagation algorithm is used to calculate the gradient of the loss function to the model parameters, and the parameters of the model are updated according to the gradient to minimize the loss function. During the training process, optimizers are usually used to adjust the step size of parameter updates. Common optimizers include stochastic gradient descent, Adagrad, Adam, RMSProp, and Adam. The Adam optimizer has better performance and convergence speed in most cases and is one of the more commonly used optimizers.

[0052] After the neuron model is established, it is necessary to evaluate and optimize the model. First, according to the goal of engineering cost management, select appropriate evaluation indicators to measure the performance of the model, such as mean square error, mean absolute error, accuracy and recall. For cost prediction problems, MSE and MAE are commonly used evaluation indicators that can reflect the error between the predicted value and the true value. Secondly, use the test set to evaluate the trained model and calculate the corresponding evaluation index value to understand the generalization ability of the model on unseen data. If the model performance does not meet the expected requirements, the model needs to be optimized and adjusted. Finally, according to the evaluation results, take corresponding optimization measures, such as adjusting the model's hyperparameters, increasing training data, improving feature engineering and using regularization methods to improve the performance and generalization ability of the model.

[0053] Association rule mining algorithm: The association rule mining algorithm uses a priori property that all non-empty subsets of frequent item sets must also be frequent item sets. First, it starts with a single item and finds all frequent 1-item sets that meet the minimum support. Then, it generates candidate 2-item sets by combining frequent 1-item sets, and then calculates the support of candidate 2-item sets to filter out frequent 2-item sets. This process is repeated until no new frequent item sets can be generated.

[0054] Transactions and Itemsets in Association Rule Mining Algorithms In association rule algorithms, data is usually viewed as a series of transactions, each transaction contains several items, and the collection of all items is called an itemset. For example, in a shopping basket data set, a shopping behavior is a transaction, and each item in the shopping basket is an item.

[0055] In the association rule mining algorithm, support is used to measure the frequency of an item set appearing in a data set; it is the ratio of the number of transactions that contain the item set to the total number of transactions. For example, the support of the item set {A, B} = the number of transactions containing {A, B} / the total number of transactions. Support can help us filter out frequently appearing item sets.

[0056] In the association rule mining algorithm, confidence is used to measure the strength of the association rule. For the association rule A->B, its confidence is the ratio of the number of transactions containing A and B to the number of transactions containing A. For example, confidence (A->B) = number of transactions containing {A, B} / number of transactions containing A, which indicates the probability of B appearing when A appears.

[0057] Lift is another indicator to measure the effectiveness of association rules in the association rule mining algorithm. It is the ratio of confidence to support of B. A lift greater than 1 means that the appearance of A has a promoting effect on the appearance of B, that is, there is a positive correlation between A and B; a lift equal to 1 means that A and B are independent of each other; a lift less than 1 means that the appearance of A has an inhibitory effect on the appearance of B.

[0058] The neurons in a neural network can perform calculations simultaneously, greatly improving the processing speed and being able to process large amounts of data at the same time. Information is not stored in a specific location, but is distributed in the connection weights of the entire network, making the network more fault-tolerant and robust. At the same time, it can automatically discover rules and patterns from the data and adapt to new data by constantly adjusting weights and biases. It does not require manual intervention to define too many rules and can handle complex nonlinear relationships. For problems that are difficult to describe with traditional linear models, neural networks can better model and predict.

[0059] Step 2: Automatic quantity calculation and pricing. This is done by using AI-driven image recognition and natural language processing technologies to automatically calculate and price the project quantities. For example, by automatically extracting project quantity information through image recognition of construction drawings, the workload and errors of manual quantity calculation can be reduced. Natural language processing technology can be used to analyze and understand texts such as bidding documents and contract documents, automatically extract pricing-related terms and requirements, and improve the accuracy and efficiency of pricing.

[0060] Step 3: Cost prediction model, based on machine learning algorithms such as linear regression, decision tree and neural network, to build a project cost prediction model. These models can predict the total cost of the project and the cost of each stage and each sub-project according to the characteristic parameters of the project, such as building area, structure type, decoration standard and construction period. Then, by continuously inputting new project data, the model is trained and optimized to improve the accuracy and reliability of the prediction, helping cost management personnel to make reasonable cost estimates and cost planning in the early stage of the project.

[0061] Step 4: Risk assessment and early warning. Use artificial intelligence technology to identify, assess and warn project cost risks. By analyzing historical data and project characteristics, identify risk factors that may lead to cost overruns, such as market price fluctuations, design changes and changes in construction conditions. Use artificial intelligence technology to establish a risk assessment model to quantitatively assess the probability and impact of risks. Based on the assessment results, issue early warning signals to remind cost management personnel to take timely measures to respond to risks and reduce cost risks.

[0062] Step 5: Collaborative management of project progress and cost. With the help of artificial intelligence technology, a collaborative management model of project progress and cost is established to achieve dynamic correlation and mutual influence analysis between the two. Through real-time monitoring of project progress, the impact of progress deviation on cost and the reaction of cost change on progress are predicted. According to the collaborative management model, a reasonable progress plan and cost control plan are formulated to ensure that the project can achieve effective control of cost targets while ensuring progress.

[0063] Step 6: Intelligent cost control. During the project implementation process, AI is used to achieve real-time monitoring and dynamic control of costs. Real-time data from the construction site is collected through the Internet of Things technology, such as material consumption, equipment usage, and personnel attendance, and compared with the cost budget for analysis. When the actual cost deviates from the budget, the system automatically issues a reminder and finds out the cause of the cost deviation through data analysis, providing decision-making support for cost management personnel so that adjustment measures can be taken in a timely manner to ensure that project costs are controlled within a reasonable range.

[0064] Step 7: Intelligent supplier management: Use artificial intelligence algorithms to analyze suppliers’ historical data, evaluate suppliers’ reputation, price level, delivery time and overall quality performance, and establish a supplier evaluation model. Then, during the project procurement phase, based on project requirements and the evaluation model, it intelligently recommends suitable suppliers to help cost management personnel select cost-effective and reputable suppliers, thereby reducing procurement costs and risks.

[0065] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. The project cost management method based on artificial intelligence is characterized by: The following steps are involved: Step 1: Data mining and analysis: extract valuable information from a large amount of historical engineering cost data by establishing a data mining model, and use data analysis tools to analyze the mined data; Step 2: Automatic quantity calculation and pricing, which uses AI-driven image recognition and natural language processing technology to automatically calculate and price the engineering quantity; Step 3: Cost prediction model: construct a project cost prediction model based on machine learning algorithms such as linear regression, decision tree and neural network; Step 4: Risk assessment and early warning, using artificial intelligence technology to identify, assess and warn project cost risks; Step 5: Collaborative management of project progress and cost. Use artificial intelligence technology to establish a collaborative management model for project progress and cost, and realize dynamic correlation and mutual influence analysis between the two. Step 6: Intelligent cost control, using artificial intelligence to achieve real-time monitoring and dynamic control of costs during project implementation; Step 7: Intelligent supplier management: Analyze the supplier's historical data through artificial intelligence algorithms, evaluate the supplier's reputation, price level, delivery time and quality performance, and establish a supplier evaluation model.

2. The construction cost management method based on artificial intelligence according to claim 1 is characterized in that: The construction of the data mining model in step 1 includes the following steps: S1: Build a web crawler program to automatically crawl web page content on the Internet according to certain rules by simulating browser behavior to collect data; S2: Using the data collected by the web crawler program, use the data mining algorithm to further mine and collect the cost data; S3: The collected cost data is combined with the autonomous learning model in machine learning to further improve and update the deeply mined cost data.

3. The construction cost management method based on artificial intelligence according to claim 2 is characterized in that: The data mining algorithm includes a neural network algorithm and an association rule mining algorithm. The association rule mining algorithm utilizes a priori property, that is, all non-empty subsets of frequent item sets must also be frequent item sets. First, it starts from a single item and finds all frequent 1-item sets that meet the minimum support. Then, candidate 2-item sets are generated by combining frequent 1-item sets. The support of candidate 2-item sets is calculated to screen out frequent 2-item sets. This process is repeated until no new frequent item sets can be generated.

4. The construction cost management method based on artificial intelligence according to claim 3 is characterized in that: The neural network algorithm needs to build a neuron model, which is composed of multiple neurons. Each neuron receives a set of input signals, performs weighted summation on them, and converts them through a nonlinear activation function to generate an output signal. The formula is expressed as: in It is Inputs, is the corresponding weight, is the bias term, is the activation function.

5. The construction cost management method based on artificial intelligence according to claim 4 is characterized in that: When establishing the neuron model, it is first necessary to select a neural network architecture, then determine the model parameters, and finally perform model training using the set parameters.

6. The construction cost management method based on artificial intelligence according to claim 5 is characterized in that: The neural network architecture includes a multi-layer perceptron architecture with strong nonlinear fitting capabilities for processing general prediction and classification problems such as tabular data, a convolutional neural network architecture that performs well in processing data with grid structures such as images and audio and can automatically extract local features, and a recurrent neural network architecture that is suitable for processing sequence data such as time series prediction and natural language processing tasks and can capture long-term dependencies in the data.

7. The construction cost management method based on artificial intelligence according to claim 5 is characterized in that: The model parameters include determining the number of layers and neurons, selecting activation functions and other parameters. The number of layers and neurons determines the difficulty of processing the problem. The activation functions include Sigmoid, ReLU, and tanh. The other parameters include determining the learning rate, number of iterations, and batch size.

8. The construction cost management method based on artificial intelligence according to claim 5 is characterized by: The model training uses the training set data to train the neural network model, calculates the predicted output through forward propagation, and then calculates the loss function based on the difference between the predicted output and the true label, such as mean square error and cross entropy loss, and then uses the back propagation algorithm to calculate the gradient of the loss function to the model parameters, and updates the model parameters according to the gradient to minimize the loss function.

9. The construction cost management method based on artificial intelligence according to claim 5 is characterized by: After the neuron model is established, it is necessary to evaluate and optimize the model. First, the evaluation index needs to be selected, then the model performance needs to be evaluated, and finally the model needs to be optimized.

10. The construction cost management method based on artificial intelligence according to claim 9 is characterized in that: First, according to the goal of construction cost management, select appropriate evaluation indicators to measure the performance of the model, such as mean square error, mean absolute error, accuracy and recall. Secondly, use the test set to evaluate the trained model and calculate the corresponding evaluation indicator value to understand the generalization ability of the model on unseen data. Finally, according to the evaluation results, take corresponding optimization measures, such as adjusting the model's hyperparameters, increasing training data, improving feature engineering and using regularization methods to improve the performance and generalization ability of the model.

Images