Software project cost prediction method and system based on artificial intelligence

Abstract

The invention discloses a software project cost prediction method and system based on artificial intelligence. The method comprises project data collection, data optimization, project management cost prediction, project technology cost prediction and software project cost comprehensive prediction. The invention relates to the technical field of software project cost, in particular to a software project cost prediction method and system based on artificial intelligence. According to the scheme, prediction of two routes of project management and project technology is adopted, and weighted summation of the two predicted routes is adopted to predict the cost of a software project. A specific reference for software project cost prediction is provided more sufficiently and more clearly; by adopting a prediction method combining linear regression, convolution and an artificial neural network, the feasibility and accuracy of software project management prediction are improved; a prediction method based on the recurrent neural network is used for software project cost prediction, and a good data basis is provided for subsequent comprehensive prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of software project cost, and in particular to a software project cost prediction method and system based on artificial intelligence. Background Art

[0002] The AI-based software project cost prediction method uses AI technologies such as machine learning and deep learning to predict the cost of future software projects by analyzing historical project data, project characteristics, and related variables. The definition of this method includes steps such as data collection, feature engineering, model training, and evaluation. Its main function is to improve the accuracy and efficiency of cost prediction, reduce the risk of project budget overruns, optimize resource allocation, and help project managers make more informed decisions, thereby improving project success rates and corporate competitiveness.

[0003] However, in the existing software project cost forecast, there are existing traditional methods for software projects, which mainly carry out single numerical forecast. The traditional method is often based on pure cost numerical data for forecast, but this method easily ignores various complex project development cost factors, which leads to insufficient forecast results and processes, and reduces the technical problem of the referenceability of the forecast results; in the existing software project cost forecast, there is a technical problem that the project management cost forecast factors are complex, involving both numerical data and complex management element data, and the traditional simple model cannot meet the forecast needs; in the existing software project cost forecast, there is a technical problem that the data variability of the project technical forecast is large, thereby increasing the difficulty of the forecast and subsequent weighted analysis. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a software project cost prediction method and system based on artificial intelligence. In view of the fact that in the existing software project cost prediction, there are existing traditional methods for software projects, which mainly perform single numerical prediction. The traditional method is often based on pure cost numerical data for prediction, but this method easily ignores various complex project development cost factors, which leads to insufficient prediction results and processes, and reduces the technical problem of the referenceability of the prediction results. This scheme creatively adopts the prediction of two routes, project management and project technology, and then provides a more sufficient and clear specific reference for software project cost prediction through the weighted summation of the two prediction routes, thereby improving the overall reliability of the method; in view of the fact that in the existing software project cost prediction, there are complex project management cost prediction factors, which involve both numerical data and complex management element numbers. According to the technical problem that the traditional simple model cannot meet the needs of prediction, this solution creatively adopts a prediction method that combines linear regression, convolution and artificial neural network. It processes linear data through regression analysis, processes local features through convolution subnet and further optimizes the prediction effect of linear data, and finally uses artificial neural network to process nonlinear data and weight the results, which improves the feasibility and accuracy of software project management prediction. In view of the technical problem that in the existing software project cost forecast, there is a large variability in the data of project technical forecast, which increases the difficulty of prediction and subsequent weighted analysis, this solution creatively adopts a prediction method based on recursive neural network to predict software project cost, and reduces the data complexity and variability of project technical cost forecast through the loop layer, which improves the overall reference of software project cost forecast and provides a good data foundation for subsequent comprehensive forecast.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides a software project cost prediction method based on artificial intelligence, the method comprising the following steps:

[0006] Step S1: Project data collection;

[0007] Step S2: data optimization;

[0008] Step S3: Project management cost forecast;

[0009] Step S4: Project technical cost forecast;

[0010] Step S5: Comprehensive forecast of software project cost.

[0011] Further, in step S1, the project data collection is used to collect the original data set required for software project cost prediction, specifically, to obtain the original data set for software project cost prediction through data collection from project management records and software project development task management system;

[0012] The software project cost prediction original data set specifically includes internal data and external data;

[0013] The internal data specifically includes basic project information data, project scale data, project staffing data, project management information data, project technical data and project risk data;

[0014] The external data specifically refers to external public data sets used by the software project.

[0015] Further, in step S2, the data optimization is used to optimize the original data, specifically, based on the original data set of the software project cost prediction, perform data cleaning preprocessing and storage management optimization operations to obtain a project cost prediction optimized data set;

[0016] The project cost forecast optimization data set specifically includes optimization of project basic information data, optimization of project scale data, optimization of project personnel configuration data, optimization of project management information data, optimization of project technical data and optimization of project risk data;

[0017] The optimized project basic information data, optimized project scale data, optimized project personnel configuration data and optimized project management information data are used to perform project management cost forecasting;

[0018] The optimized project technical data and optimized project risk data are used to perform project technical cost forecasting.

[0019] Further, in step S3, the project management cost forecast is used to forecast the project management cost from the perspective of project plan execution and quality management cost, specifically, based on the project cost forecast optimization data set, a convolutional artificial neural network combined with a regression analysis model is used to forecast the project management cost, and project management cost forecast data is obtained;

[0020] The convolutional artificial neural network combined with the regression analysis model specifically includes a regression analysis subnet, a convolution subnet and an artificial neural subnet;

[0021] The regression analysis subnet is used to predict numerical data in project management cost prediction by linear regression and reduce the overall resource requirements for numerical data processing;

[0022] The convolutional subnetwork is used to extract relevant local features related to project management cost prediction;

[0023] The artificial neural subnetwork is used to process nonlinear data related to project management cost forecasting and improve the overall scalability of the model;

[0024] The step of using a convolutional artificial neural network combined with a regression analysis model to predict project management costs and obtain project management cost prediction data includes:

[0025] Step S31: construct a regression analysis subnet, specifically to establish a linear regression model and perform a preliminary numerical prediction of the project management cost, and obtain linear regression prediction cost data. The calculation formula is:

[0026] ;

[0027] In the formula, y R is the linear regression prediction cost data, is the initial intercept data, which is used to represent the minimum value of project management cost, n is the total number of input data, i is the input data index, is the characteristic coefficient of the i-th input data, x i is the i-th original input data, used to represent the data in the project cost forecast optimization dataset, is the linear regression error term;

[0028] Step S32: Construct a convolutional subnet, specifically construct a basic convolutional subnet including a convolutional layer and a pooling layer, perform feature extraction and dimensionality reduction through convolution and pooling operations, perform convolution feature extraction and cost prediction on the linear regression prediction cost data, and obtain project management convolution cost prediction data. The calculation formula is:

[0029] ;

[0030] In the formula, y c is the project management convolution cost prediction data, Maxpool(·) is the maximum pooling operation, ReLU(·) is the nonlinear activation function, J is the total number of convolution kernels, j is the convolution kernel index, K is the total number of data in the linear regression prediction cost data, k is the data index of the linear regression prediction cost data, and w jk is the convolution weight corresponding to the kth linear regression prediction cost data and the jth convolution kernel, x k is the kth linear regression prediction cost data, b j is the convolution bias term of the jth convolution kernel;

[0031] Step S33: construct an artificial neural subnet, specifically construct an artificial neural network including an input layer, a hidden layer and an output layer as the artificial neural subnet, and perform artificial neural subnet cost prediction based on the project cost prediction optimization data set to obtain artificial neural subnet cost prediction data, and the calculation formula is:

[0032] ;

[0033] In the formula, yA is the artificial neural subnet cost prediction data, M is the total number of hidden layers of the artificial neural subnet, m is the hidden layer index, and v m is the weight from the hidden layer to the output layer, ReLU(·) is the nonlinear activation function, n is the total number of input data, i is the input data index, and w im is the hidden layer weight corresponding to the i-th input data and the m-th hidden layer, x i is the i-th original input data, c is the output layer bias term;

[0034] Step S34: subnet output integration, specifically, outputting weighted sum of the linear regression prediction cost data, the project management convolution cost prediction data and the artificial neural subnet cost prediction data to obtain subnet output integrated prediction cost data, and the calculation formula is:

[0035] ;

[0036] Where Y is the integrated prediction cost data of the subnet output, is the linear regression prediction cost weight, y R is the linear regression prediction cost data, is the project management convolution cost weight, y c It is the project management convolution cost forecast data, is the artificial neural subnetwork cost prediction weight, y A is the artificial neural subnet cost prediction data;

[0037] Step S35: Project management cost prediction model training, specifically, by constructing the regression analysis subnet, the convolution subnet, the artificial neural subnet and the subnet output integration, the project management cost prediction model training is performed to obtain the management cost prediction model Model MoTP ;

[0038] Step S36: Project management cost prediction, specifically, optimizing the project cost prediction data set and using the management cost prediction model Model MoTP , conduct project management cost forecasting and obtain project management cost forecasting data;

[0039] The project management cost forecast data specifically includes the total management cost forecast value, the labor cost sub-item forecast value and the quality management cost forecast value.

[0040] Further, in step S4, the project technical cost forecast is used to forecast the project technical cost from the perspective of technical adaptability and updating, specifically, based on the project cost forecast optimization data set, a recursive neural network is used to forecast the project technical cost, and project technical cost forecast data is obtained, specifically including the following steps:

[0041] Step S41: constructing forecast sequence data, specifically inputting the optimized project technical data and optimized project risk data in the project cost forecast optimization data set as forecast sequence data;

[0042] Step S42: preprocessing the sequence data, specifically normalizing the predicted sequence data input to obtain technical cost prediction model training data;

[0043] Step S43: constructing a recursive neural network, specifically using a recursive algorithm to calculate the time step hidden state of the model training data, and setting the loop layer of the recursive neural network by calculating the time step hidden state;

[0044] The hidden state at that time step T The calculation formula is:

[0045] ;

[0046] In the formula, Hidden T is the time step hidden state corresponding to the training data of the technology cost prediction model, where T is the data index of the training data of the technology cost prediction model, ReLU(·) is the nonlinear activation function, and W H is the time step weight matrix, Hidden T-1 is the hidden state of the previous data index T-1 corresponding to the training data of the technology cost prediction model, W D is the training data weight matrix of the technology cost prediction model, is the matrix representation corresponding to the training data of the technology cost prediction model, b H is the hidden state bias vector;

[0047] Step S44: Technical cost prediction model training, specifically, performing model training based on the technical cost prediction model training data and the recursive neural network to obtain a technical cost prediction model Model PoTP ;

[0048] Step S45: Technical cost prediction, specifically, optimizing the project cost prediction data set and using the technical cost prediction model Model PoTP Carry out project technical cost forecast and obtain project technical cost forecast data;

[0049] The project technical cost forecast data specifically includes the total counting cost forecast value, tool cost forecast value, hardware cost forecast value, technology implementation cost forecast value and safety maintenance cost forecast value.

[0050] Furthermore, in step S5, the comprehensive forecast of software project cost is used to combine the project management cost and the project technical cost forecast to make a comprehensive forecast of the software project cost, specifically combining the project management cost forecast data and the project technical cost forecast data, and obtaining the comprehensive forecast data of software project cost by weighted summation of the forecast data.

[0051] The present invention provides an artificial intelligence-based software project cost prediction system, which includes a project data collection module, a data optimization module, a project management cost prediction module, a project technology cost prediction module and a software project cost comprehensive prediction module;

[0052] The project data collection module is used for collecting project data, obtaining a software project cost prediction original data set through project data collection, and sending the software project cost prediction original data set to the data optimization module;

[0053] The data optimization module is used for data optimization, and obtains a project cost forecast optimization data set through data optimization, and sends the project cost forecast optimization data set to the project management cost forecast module;

[0054] The project management cost prediction module is used for project management cost prediction, obtains project management cost prediction data through project management cost prediction, and sends the project management cost prediction data to the software project cost comprehensive prediction module;

[0055] The project technical cost forecasting module is used for project technical cost forecasting, obtains project technical cost forecasting data through project technical cost forecasting, and sends the project technical cost forecasting data to the software project cost comprehensive forecasting module;

[0056] The software project cost comprehensive prediction module is used for comprehensive prediction of software project cost, and obtains software project cost comprehensive prediction data through comprehensive prediction of software project cost.

[0057] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0058] (1) In the existing software project cost forecasting, there are existing traditional software project methods, which mainly perform single numerical forecasting. Traditional methods often make forecasts based on pure cost numerical data, but this method easily ignores various complex project development cost factors, which leads to insufficient forecast results and processes, reducing the technical problem of the referenceability of the forecast results. This solution creatively adopts the forecasting of two routes, project management and project technology, and then provides a more comprehensive and clear reference for software project cost forecasting through the weighted summation of the two forecasting routes, thereby improving the overall reliability of the method;

[0059] (2) In view of the technical problem that the factors of project management cost prediction are complex in the existing software project cost prediction, which involves both numerical data and complex management factor data, and the traditional simple model cannot meet the prediction needs, this solution creatively adopts a prediction method that combines linear regression, convolution and artificial neural network. It processes linear data through regression analysis, processes local features through convolution subnet and further optimizes the prediction effect of linear data. Finally, it uses artificial neural network to process nonlinear data and weight the results, which improves the feasibility and accuracy of software project management prediction.

[0060] (3) In view of the technical problem that the data variability of project technical forecasts in existing software project cost forecasts is large, which increases the difficulty of forecasting and subsequent weighted analysis, this solution creatively adopts a forecasting method based on recursive neural networks to forecast software project costs, and reduces the data complexity and variability of project technical cost forecasts through the loop layer, thereby improving the overall reference value of software project cost forecasts and providing a good data foundation for subsequent comprehensive forecasts. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 A flowchart of a software project cost prediction method based on artificial intelligence provided by the present invention;

[0062] Figure 2 A schematic diagram of a software project cost prediction system based on artificial intelligence provided by the present invention;

[0063] Figure 3 A schematic diagram of the process for data optimization in step S2;

[0064] Figure 4 A flowchart of project management cost forecasting in step S3;

[0065] Figure 5 This is a flow chart of the project technical cost forecast in step S4.

[0066] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0067] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0068] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0069] Example 1, see Figure 1 The present invention provides a software project cost prediction method based on artificial intelligence, which comprises the following steps:

[0070] Step S1: Project data collection;

[0071] Step S2: data optimization;

[0072] Step S3: Project management cost forecast;

[0073] Step S4: Project technical cost forecast;

[0074] Step S5: Comprehensive forecast of software project cost.

[0075] By performing the above operations, in the existing software project cost forecasting, there are existing traditional software project methods, which mainly perform single numerical prediction. The traditional method is often based on pure cost numerical data for prediction, but this method easily ignores various complex project development cost factors, which leads to insufficient prediction results and processes, reducing the technical problem of the referenceability of the prediction results. This solution creatively adopts the prediction of two routes, project management and project technology, and then provides a more sufficient and clear specific reference for software project cost prediction through the weighted summation of the two prediction routes, thereby improving the overall reliability of the method.

[0076] Example 2, see Figure 1 and Figure 2 In step S1, the project data collection is used to collect the original data set required for software project cost forecasting, specifically, to obtain the original data set for software project cost forecasting through data collection from project management records and software project development task management system;

[0077] The software project cost prediction original data set specifically includes internal data and external data;

[0078] The internal data specifically includes basic project information data, project scale data, project staffing data, project management information data, project technical data and project risk data;

[0079] The external data refers specifically to the external public data sets used by the software project;

[0080] The basic information data of the project, specifically including the project name, project period, project budget and actual expenditure records;

[0081] The project scale data specifically includes the number of lines of code, the number of functional modules, and the number of use cases;

[0082] The project staffing data specifically includes the number of project team members, project team experience level data, and project member division of labor data;

[0083] The project management information data specifically includes project development cycle data, project task allocation data, project task progress data and project development cycle planning data;

[0084] The project technical data specifically includes programming language data, framework data, library data and development tool record data;

[0085] The project risk data specifically includes risk control record data and risk management measures data;

[0086] The data collection is specifically carried out through automated data extraction, manual data collection and data integration storage.

[0087] Example 3, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the data optimization is used to optimize the original data. Specifically, based on the original data set of software project cost prediction, data cleaning preprocessing and storage management optimization operations are performed to obtain a project cost prediction optimization data set, including the following steps:

[0088] Step S21: missing value processing, specifically using random interpolation and mean filling methods to fill missing values ​​of the numerical data in the original data set of the software project cost prediction, to obtain filled optimized data;

[0089] Step S22: outlier processing, specifically using a Z-score statistical method to detect and process outliers in the padded optimized data to obtain outlier optimized data;

[0090] Step S23: data standardization, specifically using the Min-Max scaling method to standardize the abnormal optimization data to obtain standardized data;

[0091] Step S24: data conversion optimization, specifically, manually labeling and one-hot encoding conversion of non-numerical data in the original data set of the software project cost prediction to obtain format-optimized data;

[0092] Step S25: data storage management optimization, specifically, optimizing the data storage of the standardized data and the format-optimized data by designing a data storage database, and integrating the standardized data and the format-optimized data into a storage-optimized data set by data integration;

[0093] Step S26: data optimization, specifically, optimizing the original data set of the software project cost prediction through the missing value processing, the outlier processing, the data standardization, the data conversion optimization and the data storage management optimization to obtain a project cost prediction optimized data set;

[0094] The project cost forecast optimization data set specifically includes optimization of project basic information data, optimization of project scale data, optimization of project personnel configuration data, optimization of project management information data, optimization of project technical data and optimization of project risk data;

[0095] The optimized project basic information data, optimized project scale data, optimized project personnel configuration data and optimized project management information data are used to perform project management cost forecasting;

[0096] The optimized project technical data and optimized project risk data are used to perform project technical cost forecasting.

[0097] Example 4, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the project management cost forecast is used to forecast the project management cost from the perspective of project plan execution and quality management cost. Specifically, the project management cost forecast is conducted based on the project cost forecast optimization data set, using a convolutional artificial neural network combined with a regression analysis model to obtain project management cost forecast data;

[0098] The convolutional artificial neural network combined with the regression analysis model specifically includes a regression analysis subnet, a convolution subnet and an artificial neural subnet;

[0099] The regression analysis subnet is used to predict numerical data in project management cost prediction by linear regression and reduce the overall resource requirements for numerical data processing;

[0100] The convolutional subnetwork is used to extract relevant local features related to project management cost prediction;

[0101] The artificial neural subnetwork is used to process nonlinear data related to project management cost forecasting and improve the overall scalability of the model;

[0102] The step of using a convolutional artificial neural network combined with a regression analysis model to predict project management costs and obtain project management cost prediction data includes:

[0103] Step S31: construct a regression analysis subnet, specifically to establish a linear regression model and perform a preliminary numerical prediction of the project management cost, and obtain linear regression prediction cost data. The calculation formula is:

[0104] ;

[0105] In the formula, y R is the linear regression prediction cost data, is the initial intercept data, which is used to represent the minimum value of project management cost, n is the total number of input data, i is the input data index, is the characteristic coefficient of the i-th input data, x i is the i-th original input data, used to represent the data in the project cost forecast optimization dataset, is the linear regression error term;

[0106] Step S32: Construct a convolutional subnet, specifically construct a basic convolutional subnet including a convolutional layer and a pooling layer, perform feature extraction and dimensionality reduction through convolution and pooling operations, perform convolution feature extraction and cost prediction on the linear regression prediction cost data, and obtain project management convolution cost prediction data. The calculation formula is:

[0107] ;

[0108] In the formula, y c is the project management convolution cost prediction data, Maxpool(·) is the maximum pooling operation, ReLU(·) is the nonlinear activation function, J is the total number of convolution kernels, j is the convolution kernel index, K is the total number of data in the linear regression prediction cost data, k is the data index of the linear regression prediction cost data, and w jk is the convolution weight corresponding to the kth linear regression prediction cost data and the jth convolution kernel, x k is the kth linear regression prediction cost data, b j is the convolution bias term of the jth convolution kernel;

[0109] Step S33: construct an artificial neural subnet, specifically construct an artificial neural network including an input layer, a hidden layer and an output layer as the artificial neural subnet, and perform artificial neural subnet cost prediction based on the project cost prediction optimization data set to obtain artificial neural subnet cost prediction data, and the calculation formula is:

[0110] ;

[0111] In the formula, yA is the artificial neural subnet cost prediction data, M is the total number of hidden layers of the artificial neural subnet, m is the hidden layer index, and v m is the weight from the hidden layer to the output layer, ReLU(·) is the nonlinear activation function, n is the total number of input data, i is the input data index, and w im is the hidden layer weight corresponding to the i-th input data and the m-th hidden layer, x i is the i-th original input data, c is the output layer bias term;

[0112] Step S34: subnet output integration, specifically, outputting weighted sum of the linear regression prediction cost data, the project management convolution cost prediction data and the artificial neural subnet cost prediction data to obtain subnet output integrated prediction cost data, and the calculation formula is:

[0113] ;

[0114] Where Y is the integrated prediction cost data of the subnet output, is the linear regression prediction cost weight, y R is the linear regression prediction cost data, is the project management convolution cost weight, y c It is the project management convolution cost forecast data, is the artificial neural subnetwork cost prediction weight, y A is the artificial neural subnet cost prediction data;

[0115] Step S35: Project management cost prediction model training, specifically, by constructing the regression analysis subnet, the convolution subnet, the artificial neural subnet and the subnet output integration, the project management cost prediction model training is performed to obtain the management cost prediction model Model MoTP ;

[0116] Step S36: Project management cost prediction, specifically, optimizing the project cost prediction data set and using the management cost prediction model Model MoTP , conduct project management cost forecast and obtain project management cost forecast data;

[0117] The project management cost forecast data specifically includes the total management cost forecast value, the labor cost sub-item forecast value and the quality management cost forecast value.

[0118] By performing the above operations, in order to address the technical problem that in the existing software project cost forecast, the project management cost forecast factors are complex, involving both numerical data and complex management element data, and the traditional simple model cannot meet the forecast needs, this solution creatively adopts a forecasting method that combines linear regression, convolution and artificial neural network. It processes linear data through regression analysis, processes local features through convolution subnet and further optimizes the forecasting effect of linear data, and finally uses artificial neural network to process nonlinear data and weight the results, thereby improving the feasibility and accuracy of software project management forecasting.

[0119] Example 5, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S4, the project technical cost forecast is used to forecast the project technical cost from the perspective of technical adaptability and updating. Specifically, the project technical cost forecast is conducted using a recursive neural network based on the project cost forecast optimization data set to obtain project technical cost forecast data. Specifically, the following steps are included:

[0120] Step S41: constructing forecast sequence data, specifically inputting the optimized project technical data and optimized project risk data in the project cost forecast optimization data set as forecast sequence data;

[0121] Step S42: preprocessing the sequence data, specifically normalizing the predicted sequence data input to obtain technical cost prediction model training data;

[0122] Step S43: constructing a recursive neural network, specifically using a recursive algorithm to calculate the time step hidden state of the model training data, and setting the loop layer of the recursive neural network by calculating the time step hidden state;

[0123] The hidden state at that time step T The calculation formula is:

[0124] ;

[0125] In the formula, Hidden T is the time step hidden state corresponding to the training data of the technology cost prediction model, where T is the data index of the training data of the technology cost prediction model, ReLU(·) is the nonlinear activation function, and W H is the time step weight matrix, Hidden T-1 is the hidden state of the previous data index T-1 corresponding to the training data of the technology cost prediction model, W D is the training data weight matrix of the technology cost prediction model, is the matrix representation corresponding to the training data of the technology cost prediction model, bH is the hidden state bias vector;

[0126] Step S44: Technical cost prediction model training, specifically, performing model training based on the technical cost prediction model training data and the recursive neural network to obtain a technical cost prediction model Model PoTP ;

[0127] Step S45: Technical cost prediction, specifically, optimizing the project cost prediction data set and using the technical cost prediction model Model PoTP Carry out project technical cost forecast and obtain project technical cost forecast data;

[0128] The project technical cost forecast data specifically includes the total counting cost forecast value, tool cost forecast value, hardware cost forecast value, technology implementation cost forecast value and safety maintenance cost forecast value.

[0129] By performing the above operations, in order to address the technical problem that in the existing software project cost forecasts, the data variability of the project technical forecast is large, which increases the difficulty of the forecast and subsequent weighted analysis, this solution creatively adopts a forecasting method based on a recursive neural network to forecast the software project cost, and reduces the data complexity and variability of the project technical cost forecast through the loop layer, thereby improving the overall reference value of the software project cost forecast and providing a good data foundation for the subsequent comprehensive forecast.

[0130] Embodiment 7, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S5, the comprehensive forecast of software project cost is used to combine the project management cost and the project technical cost forecast to make a comprehensive forecast of software project cost. Specifically, the project management cost forecast data and the project technical cost forecast data are combined, and the weighted sum of the forecast data is taken to obtain the comprehensive forecast data of software project cost.

[0131] Embodiment 8, see Figure 1 and Figure 2 , this embodiment is based on the above embodiment, and the present invention provides an artificial intelligence-based software project cost prediction system, including a project data collection module, a data optimization module, a project management cost prediction module, a project technology cost prediction module and a software project cost comprehensive prediction module;

[0132] The project data collection module is used for collecting project data, obtaining a software project cost prediction original data set through project data collection, and sending the software project cost prediction original data set to the data optimization module;

[0133] The data optimization module is used for data optimization, and obtains a project cost forecast optimization data set through data optimization, and sends the project cost forecast optimization data set to the project management cost forecast module;

[0134] The project management cost prediction module is used for project management cost prediction, obtains project management cost prediction data through project management cost prediction, and sends the project management cost prediction data to the software project cost comprehensive prediction module;

[0135] The project technical cost forecasting module is used for project technical cost forecasting, obtains project technical cost forecasting data through project technical cost forecasting, and sends the project technical cost forecasting data to the software project cost comprehensive forecasting module;

[0136] The software project cost comprehensive prediction module is used for comprehensive prediction of software project cost, and obtains software project cost comprehensive prediction data through comprehensive prediction of software project cost.

[0137] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0138] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0139] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A software project cost prediction method based on artificial intelligence, characterized by: The method comprises the following steps: Step S1: Project data collection, obtaining the original data set for software project cost prediction; Step S2: data optimization, specifically, performing data cleaning preprocessing and storage management optimization operations based on the original data set of software project cost prediction to obtain a project cost prediction optimized data set; Step S3: Project management cost prediction, which is used to predict the project management cost from the perspective of project plan execution and quality management cost. Specifically, based on the project cost prediction optimization data set, a convolutional artificial neural network combined with a regression analysis model is used to predict the project management cost, so as to obtain project management cost prediction data; Step S4: Project technical cost prediction, which is used to predict the technical cost of the project from the perspective of technical adaptability and updating, specifically, based on the project cost prediction optimization data set, a recursive neural network is used to predict the project technical cost, and project technical cost prediction data is obtained; Step S5: Comprehensively predict the software project cost to obtain comprehensive software project cost prediction data.

2. The method for predicting software project cost based on artificial intelligence according to claim 1, characterized in that: In step S3, the convolutional artificial neural network combined with the regression analysis model specifically includes a regression analysis subnet, a convolution subnet and an artificial neural subnet; The regression analysis subnet is used to predict numerical data in project management cost prediction by linear regression and reduce the overall resource requirements for numerical data processing; The convolutional subnetwork is used to extract relevant local features related to project management cost prediction; The artificial neural subnetwork is used to process nonlinear data related to project management cost prediction and improve the overall scalability of the model.

3. The method for predicting software project cost based on artificial intelligence according to claim 2, characterized in that: In step S3, the step of using a convolutional artificial neural network combined with a regression analysis model to predict project management costs and obtain project management cost prediction data includes: Step S31: constructing a regression analysis subnet, specifically establishing a linear regression model and performing a preliminary numerical prediction of the project management cost to obtain linear regression prediction cost data; Step S32: constructing a convolutional subnet, specifically constructing a basic convolutional subnet including a convolutional layer and a pooling layer, and performing feature extraction and dimensionality reduction through convolution and pooling operations, performing convolutional feature extraction and cost prediction on the linear regression prediction cost data, and obtaining project management convolutional cost prediction data; Step S33: constructing an artificial neural subnet, specifically constructing an artificial neural network including an input layer, a hidden layer and an output layer as the artificial neural subnet, and performing artificial neural subnet cost prediction based on the project cost prediction optimization data set to obtain artificial neural subnet cost prediction data; Step S34: subnet output integration, specifically, performing weighted summation on the linear regression prediction cost data, the project management convolution cost prediction data and the artificial neural subnet cost prediction data to obtain subnet output integrated prediction cost data; Step S35: Project management cost prediction model training, specifically, by constructing the regression analysis subnet, the convolution subnet, the artificial neural subnet and the subnet output integration, the project management cost prediction model training is performed to obtain the management cost prediction model Model MoTP ; Step S36: Project management cost prediction, specifically, optimizing the project cost prediction data set and using the management cost prediction model Model MoTP , conduct project management cost forecast and obtain project management cost forecast data.

4. The method for predicting software project cost based on artificial intelligence according to claim 3, characterized in that: In step S4, the project technical cost forecast specifically includes the following steps: Step S41: construct prediction sequence data; Step S42: sequence data preprocessing; Step S43: constructing a recursive neural network; Step S44: Technical cost prediction model training; Step S45: Technical cost prediction.

5. The method for predicting software project cost based on artificial intelligence according to claim 4, characterized in that: In step S41, the prediction sequence data is constructed by inputting the optimized project technical data and optimized project risk data in the project cost prediction optimization data set as the prediction sequence data; In step S42, the sequence data is preprocessed, specifically, normalizing the predicted sequence data input to obtain technical cost prediction model training data; In step S43, the recursive neural network is constructed by using a recursive algorithm to calculate the time step hidden state of the model training data, and setting the recurrent layer of the recursive neural network by calculating the time step hidden state; In step S44, the technical cost prediction model is trained, specifically by performing model training based on the technical cost prediction model training data and the recursive neural network to obtain a technical cost prediction model Model PoTP ; In step S45, the technical cost prediction is specifically based on the project cost prediction optimization data set, using the technical cost prediction model Model PoTP Conduct project technical cost forecast and obtain project technical cost forecast data.

6. The method for predicting software project cost based on artificial intelligence according to claim 5, characterized in that: In step S5, the comprehensive forecast of software project cost is used to combine the project management cost and the project technical cost forecast to make a comprehensive forecast of the software project cost. Specifically, the project management cost forecast data and the project technical cost forecast data are combined, and the weighted sum of the forecast data is taken to obtain the comprehensive forecast data of the software project cost.

7. The method for predicting software project cost based on artificial intelligence according to claim 6, characterized in that: In step S1, the software project cost prediction original data set specifically includes internal data and external data; The internal data specifically includes basic project information data, project scale data, project staffing data, project management information data, project technical data and project risk data; The external data specifically refers to external public data sets used by the software project.

8. A software project cost prediction system based on artificial intelligence, used to implement a software project cost prediction method based on artificial intelligence as claimed in any one of claims 1 to 7, characterized in that: It includes project data collection module, data optimization module, project management cost forecasting module, project technical cost forecasting module and software project cost comprehensive forecasting module.

9. The artificial intelligence-based software project cost prediction system according to claim 8, characterized in that: The project data collection module is used for collecting project data, obtaining a software project cost prediction original data set through project data collection, and sending the software project cost prediction original data set to the data optimization module; The data optimization module is used for data optimization, and obtains a project cost forecast optimization data set through data optimization, and sends the project cost forecast optimization data set to the project management cost forecast module; The project management cost prediction module is used for project management cost prediction, obtains project management cost prediction data through project management cost prediction, and sends the project management cost prediction data to the software project cost comprehensive prediction module; The project technical cost forecasting module is used for project technical cost forecasting, obtains project technical cost forecasting data through project technical cost forecasting, and sends the project technical cost forecasting data to the software project cost comprehensive forecasting module; The software project cost comprehensive prediction module is used for comprehensive prediction of software project cost, and obtains software project cost comprehensive prediction data through comprehensive prediction of software project cost.

Images