Construction method of software development model adaptive to multiple hardware platforms

Abstract

The invention discloses a method for constructing a software development model adaptive to multiple hardware platforms, which comprises the following steps of: acquiring attribute information of the multiple hardware platforms, constructing a development component according to the attribute information, and generating a component database; obtaining a software development demand, and determining a target component in the component database according to the software development demand; combining the target components, and constructing a software development model according to a combination result; running the software development model to obtain running data; and correcting the software development model according to the operation data to obtain a corrected software development model. The method is convenient for adapting to multiple hardware platforms, avoids repeated software development, software testing and subsequent guarantee work, saves manpower and fund resources, effectively utilizes the existing components for combination, accurately grasps the combination process, ensures the accuracy of the established software development model, and can improve the software development efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of software development, and particularly to a method for constructing a software development model adaptable to multiple hardware platforms. Background Art

[0002] With the continuous development of network technology and software technology, higher and higher requirements are put forward for software development. In the current software development field, restricted by different hardware platforms (processor types, memory, etc.) and the design ideas of various hardware product providers, whenever the hardware platform is updated or the hardware product provider is changed, the software needs to be re-developed accordingly; when there are multiple competing hardware manufacturing units for a certain product, the software needs to be repeatedly developed according to the hardware design schemes of each manufacturer to implement the same software functions, which brings repetitive software development, software testing and subsequent guarantee work, resulting in a waste of a large amount of human and financial resources, low software development efficiency, inability to reasonably utilize existing components, and unreasonable combinations in the process of component combination. Summary of the Invention

[0003] The present invention aims to solve at least one of the technical problems in the above technologies to some extent. For this purpose, the object of the present invention is to provide a method for constructing a software development model adaptable to multiple hardware platforms, which is convenient for adapting to multiple hardware platforms, avoiding repetitive software development, software testing and subsequent guarantee work, saving human and financial resources, effectively using existing components for combination, accurately grasping the combination process, ensuring the accuracy of the established software development model, and improving software development efficiency.

[0004] To achieve the above object, an embodiment of the present invention provides a method for constructing a software development model adaptable to multiple hardware platforms, including:

[0005] Obtaining attribute information of multiple hardware platforms, constructing development components according to the attribute information, and generating a component database;

[0006] Obtaining software development requirements, and determining target components in the component database according to the software development requirements;

[0007] Combining the target components, and constructing a software development model according to the combination result;

[0008] Run the software development model to obtain operation data; evaluate the operation data to determine parameters to be corrected; obtain the types of the parameters to be corrected; when it is determined that the type of the parameter to be corrected is the first type, obtain the weight coefficients of each target component for building the software development model when building the software development model; determine adjustment parameters according to the parameter to be corrected, adjust the weight coefficients according to the adjustment parameters, and obtain a corrected software development model according to the adjustment result; when it is determined that the type of the parameter to be corrected is the second type, determine new target components according to the parameter to be corrected, and combine the new target components into the software development model to obtain a corrected software development model.

[0009] According to some embodiments of the present invention, obtaining attribute information of multiple hardware platforms, constructing development components according to the attribute information, and generating a component database, including:

[0010] Obtain the attribute information of multiple hardware platforms; the attribute information includes processor architecture, main frequency, number of cores, memory size, storage type, peripheral interface, operating system support, and power consumption;

[0011] Obtain data sources according to the attribute information, preprocess the data sources, determine the data for constructing development components and mark them as target data; wherein, the data sources include underlying driver data sources, middleware data sources, and application framework data sources;

[0012] Obtain the attribute information of the target data;

[0013] Perform clustering analysis on the target data according to the attribute information to obtain several classification sets;

[0014] Determine the level information of each constituent element in the same classification set, and use the constituent elements of the same level as a data group;

[0015] Construct development components according to the data group, and store the development components according to categories and levels to obtain a component database.

[0016] According to some embodiments of the present invention, the combining the target components and constructing a software development model according to the combination result includes:

[0017] Establish the running relationship between each target component;

[0018] Determine the combination order according to the running relationship;

[0019] In the process of combining the target components according to the combination order, determine the combination node of the first target component and the second target component, and merge the code of the first target component and the code of the second target component according to the combination node to obtain a first merged component. Based on the same method, merge the first merged component with the third target component to obtain a second merged component, combine the target components in sequence, and construct a software development model according to the combination result.

[0020] According to some embodiments of the present invention, it further includes: encrypting the corrected software development model based on the RSA algorithm, obtaining an encrypted software development model and storing it.

[0021] According to some embodiments of the present invention, in the process of combining the target components, it further includes:

[0022] Statistically analyze the duration of each combination between the target components, determine whether it is greater than a preset duration, determine the combination process with a duration greater than the preset duration, and perform optimization processing on the combination process to obtain an optimization method and store it.

[0023] According to some embodiments of the present invention, the preprocessing includes data cleaning.

[0024] According to some embodiments of the present invention, data cleaning of the data source includes:

[0025] Segment the data source based on the business type to obtain a number of data blocks;

[0026] Input the number of data blocks into a pre-trained feature extraction model respectively, and output the feature values of the data blocks in a number of spatial dimensions;

[0027] Calculate the average value of the feature values of the number of data blocks in the same spatial dimension;

[0028] Compare the feature value of each spatial dimension of the data block with the average value of the corresponding spatial dimension respectively, and determine the covariance matrix of the data block according to the comparison result;

[0029] Analyze the covariance matrix of the data block to determine the target data block, perform dimensionality reduction processing on the target data block to obtain the hash value of the target data block;

[0030] Remove the target data block corresponding to a hash value greater than the preset hash value to obtain cleaned data.

[0031] According to some embodiments of the present invention, it further includes:

[0032] During the process of combining the target components, collect the data generated during the combination process as the first data; the first data includes combination data and attribute data related to the combination data.

[0033] Parse the attribute data to obtain a parsing result.

[0034] According to the user-predefined data processing rules and the parsing result, delete the attribute data of the edge type in the first data, obtain the second data and store it.

[0035] According to some embodiments of the present invention, determining the target components in the component database according to the software development requirements includes:

[0036] Obtain sample software development requirements, perform clustering analysis on the sample software development requirements to obtain several classification sets, train the constructed initial neural network model based on each classification set until the loss function corresponding to the initial neural network model converges, and use the trained initial neural network as the advanced neural network model.

[0037] Determine the clustering centers in each classification set respectively, determine the distances between the clustering centers, and establish an identification tree; there are multiple nodes corresponding to the identification tree, and fill several advanced neural network models into each node, and fill one advanced neural network model on each node.

[0038] Calculate the distances between the software development requirements and each clustering center, screen out the advanced neural network model corresponding to the clustering center with the minimum distance as the target advanced neural network model; input the software development requirements into the target advanced neural network model to obtain several semantic recognition results of the software development requirements.

[0039] Divide the software development requirements to obtain several first sub-data, and fill them into the first nodes of the preset first data matrix respectively.

[0040] Divide each semantic recognition result to obtain several second sub-data, and fill them into the second nodes of the preset second data matrix respectively.

[0041] Establish the corresponding relationship between the first node and the second node, fuse the first sub-data on the first node and the second sub-data on the second node, determine the fusion rate of each fusion node, and calculate the average fusion rate.

[0042] Based on several semantic recognition results, obtain several average fusion rates, and use the semantic recognition result corresponding to the maximum average fusion rate as the target semantic recognition result.

[0043] Determine the target component in the component database according to the target semantic recognition result.

[0044] The present invention proposes a method for constructing a software development model adapted to multiple hardware platforms. Run the constructed software development model to obtain operating parameters, and perform model correction according to the operating parameters to ensure the accuracy of the finally obtained corrected software development model. During the correction process, different correction strategies are adopted based on different types of correction parameters to achieve rapid correction and ensure the accuracy of the correction. It is convenient to adapt to multiple hardware platforms, avoiding repeated software development, software testing and subsequent guarantee work, saving human and financial resources. Effectively utilize existing components for combination, accurately grasp the combination process, and ensure the accuracy of the established software development model, which can improve software development efficiency, reduce the workload of R & D personnel, and shorten the working time. Effectively utilize the attribute information of multiple hardware platforms to construct development components, and construct a software development model according to software development requirements. At the same time, through the evaluation of operation data and the correction of the model, the software development model is continuously optimized to make it more adaptable to the target hardware platform, meet user needs, not only improve the efficiency and quality of software development, but also enhance the flexibility and scalability of the software.

[0045] Other features and advantages of the present invention will be described in the following specification, and part of them will become obvious from the specification, or be understood by implementing the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structures specifically pointed out in the written specification and the drawings.

[0046] The technical solutions of the present invention will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0047] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the drawings:

[0048] Figure 1 is a flowchart of a method for constructing a software development model adapted to multiple hardware platforms according to an embodiment of the present invention;

[0049] Figure 2 is a flowchart of generating a component database according to an embodiment of the present invention. Detailed Embodiments

[0050] The following describes the preferred embodiments of the present invention with reference to the drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0051] Such as Figure 1As shown in the figure, an embodiment of the present invention proposes a method for constructing a software development model adapted to multiple hardware platforms, including steps S1 - S4:

[0052] S1. Obtain the attribute information of multiple hardware platforms, construct development components according to the attribute information, and generate a component database;

[0053] S2. Obtain the software development requirements, and determine target components in the component database according to the software development requirements;

[0054] S3. Combine the target components, and construct a software development model according to the combination result;

[0055] S4. Run the software development model to obtain operation data; evaluate the operation data to determine the parameters to be corrected; obtain the types of the parameters to be corrected; when it is determined that the type of the parameter to be corrected is the first type, obtain the weight coefficients of each target component for constructing the software development model when constructing the software development model; determine adjustment parameters according to the parameters to be corrected, adjust the weight coefficients according to the adjustment parameters, and obtain a corrected software development model according to the adjustment result; when it is determined that the type of the parameter to be corrected is the second type, determine new target components according to the parameters to be corrected, and combine the new target components into the software development model to obtain a corrected software development model.

[0056] Working principle of the above technical solution: Obtain the attribute information of multiple hardware platforms, design and construct development components adapted thereto, organize all the constructed development components into a database, and record the key information such as the functions, performance, and compatibility of each component. The software development requirements include the application scenarios, functions, application categories, and levels of the software. Clearly define the specific software development requirements, including functional requirements, performance requirements, compatibility requirements, etc. According to the software development requirements, search for and determine the required target components in the component database. Combine the determined target components according to the software architecture and functional requirements to form a preliminary software development model. In the development environment, use the combined target components to construct the software development model and conduct preliminary function and performance tests. Run the software development model on the target hardware platform, collect operation data, including performance metrics, error logs, user feedback, etc. Analyze and evaluate the collected operation data to determine the problems and parameters to be corrected in the model. According to the evaluation results, classify the parameters to be corrected into two categories: The first category is the parameters that can be corrected by adjusting the weights of existing components (such as performance tuning); the second category is the parameters that require the introduction of new components to solve (such as missing functions). The first type is the type of parameters indicating that the operation process of the software development model is complete, but the operation accuracy is lower than the preset operation accuracy; for the first category of parameters, calculate the weight coefficients of each target component when constructing the software development model, and determine the adjustment parameters according to the parameters to be corrected, and adjust the weight coefficients to obtain a corrected software development model. The second type represents the type of parameters corresponding to the incomplete operation process of the software development model. For the second category of parameters, determine the target components that need to be added according to the parameters to be corrected, combine these new components into the software development model, and conduct necessary adjustments and optimizations to obtain a corrected software development model.

[0057] Beneficial effects of the above technical solution: Run the constructed software development model to obtain operation parameters, and correct the model according to the operation parameters to ensure the accuracy of the finally obtained corrected software development model. During the correction process, different correction strategies are adopted based on the different types of correction parameters to achieve rapid correction and ensure the accuracy of the correction. Facilitate adaptation to multiple hardware platforms, avoid repeated software development, software testing, and subsequent guarantee work, and save human and financial resources. Effectively utilize the existing components for combination, accurately grasp the combination process, ensure the accuracy of the established software development model, can improve software development efficiency, reduce the workload of R & D personnel, and shorten the working time. Effectively utilize the attribute information of multiple hardware platforms to construct development components and construct a software development model according to software development requirements. At the same time, through the evaluation of operation data and the correction of the model, continuously optimize the software development model to make it more adaptable to the target hardware platform, meet user needs, not only improve the efficiency and quality of software development, but also enhance the flexibility and scalability of the software.

[0058] As shown Figure 2 in the figure, according to some embodiments of the present invention, obtaining attribute information of multiple hardware platforms, constructing development components according to the attribute information, and generating a component database, including steps S11 - S16:

[0059] S11. Obtain the attribute information of multiple hardware platforms; the attribute information includes processor architecture, main frequency, number of cores, memory size, storage type, peripheral interface, operating system support, and power consumption;

[0060] S12. Obtain data sources according to the attribute information, preprocess the data sources, and determine the data for constructing development components and mark them as target data; wherein, the data sources include underlying driver data sources, middleware data sources, and application framework data sources;

[0061] S13. Obtain the attribute information of the target data;

[0062] S14. Perform clustering analysis on the target data according to the attribute information to obtain several classification sets;

[0063] S15. Determine the level information of each constituent element in the same classification set, and use the constituent elements of the same level as a data group;

[0064] S16. Construct development components according to the data groups, and store the development components according to categories and levels to obtain a component database.

[0065] Working principle of the above technical solution: Attribute information: Processor architecture: such as x86, ARM, RISC-V, etc., which determines the software compilation method and execution efficiency. Main frequency: The clock frequency of the processor, which affects the processing speed. Number of cores: The number of cores of the processor, which affects the parallel processing ability. Memory size: The capacity of the system memory, which affects the software operation efficiency and the amount of data that can be loaded. Storage type: such as HDD, SSD, NVMe, etc., which affects the data read / write speed and storage capacity. Peripheral interfaces: such as USB, HDMI, Thunderbolt, etc., which affect the connection and communication with external devices. Operating system support: The type of operating system supported by the hardware platform, such as Windows, Linux, macOS, etc. Power consumption: The energy consumption of the hardware platform, which affects the battery life and energy efficiency. According to the collected attribute information, data for constructing development components is obtained from multiple data sources. These data sources include: Underlying driver data source: Provides drivers and interfaces for direct interaction with hardware, ensuring that the software can correctly identify and control the hardware. Middleware data source: Provides a cross-platform and cross-hardware abstraction layer, simplifies the software development process, and improves the portability and reusability of code. Application framework data source: Provides high-level components and tools required for building application programs, such as UI frameworks, database access layers, etc. During the preprocessing process, it is necessary to carefully analyze the data sources to determine which data is necessary for constructing development components. These target data should be able to accurately reflect the attribute information of the hardware platform and meet the software development requirements. Once the target data is determined, it needs to be marked so that it can be easily identified and used during the subsequent development process. The marking can include information such as the type, source, and use of the data. By obtaining the attribute information of multiple hardware platforms, and obtaining and preprocessing data from the underlying driver data source, middleware data source, and application framework data source, the target data for constructing development components is determined and marked. This provides a solid foundation for subsequent development work and helps to build an efficient, compatible, and easy-to-maintain software development environment. Obtain the attribute information of the target data, and the attribute information includes categories. Perform clustering analysis on the target data according to the attribute information to obtain several classification sets; determine the level information of each constituent element in the same classification set, and use the constituent elements of the same level as a data group; construct development components according to the data group, and store the development components according to categories and levels to obtain a component database

[0066] Beneficial effects of the above technical solution: Establish different categories and determine different levels of development components in the same category, realizing the effective management of development components, and facilitating the quick determination of target components according to development requirements in subsequent steps.

[0067] According to some embodiments of the present invention, the combining of the target components and constructing a software development model according to the combination result includes:

[0068] Establish the running relationship between each target component;

[0069] Determine the combination order according to the running relationship;

[0070] During the process of combining the target components according to the combination order, determine the combination node of the first target component and the second target component, and merge the code of the first target component and the code of the second target component according to the combination node to obtain a first merged component. Based on the same method, merge the first merged component with the third target component to obtain a second merged component, and combine the target components in sequence, and construct a software development model according to the combination result.

[0071] The working principle of the above technical solution: Establish the running relationship between each target component; determine the combination order according to the running relationship; during the process of combining the target components according to the combination order, determine the combination node of the first target component and the second target component, and merge the code of the first target component and the code of the second target component according to the combination node to obtain a first merged component. Based on the same method, merge the first merged component with the third target component to obtain a second merged component, and combine the target components in sequence, and construct a software development model according to the combination result.

[0072] The beneficial effect of the above technical solution: Combining the target components according to the combination order is convenient for ensuring the logic of constructing the software development model. Based on sequentially merging the target components, the orderly management of constructing the software development model is realized, which is convenient for improving the software development efficiency.

[0073] According to some embodiments of the present invention, after constructing the software development model according to the combination result, it further includes:

[0074] Encrypt the corrected software development model based on the RSA algorithm to obtain an encrypted software development model and store it.

[0075] The beneficial effect of the above technical solution: Improve the security of the corrected software development model.

[0076] According to some embodiments of the present invention, during the process of combining the target components, it further includes:

[0077] Statistically calculate the duration of each combination between the target components, and determine whether it is greater than a preset duration, determine the combination process with a duration greater than the preset duration, optimize the combination process to obtain an optimization method and store it.

[0078] Working principle of the above technical solution: Statistically analyze the duration of each combination between target components, determine whether it is greater than a preset duration, identify the combination process with a duration greater than the preset duration, perform optimization processing on the combination process, obtain the optimization method and store it.

[0079] Beneficial effects of the above technical solution: Facilitate the optimization processing of the combination process with a duration greater than the preset duration, obtain the optimization method, facilitate the subsequent optimization of the combination process, and improve the efficiency of subsequent software development.

[0080] According to some embodiments of the present invention, the preprocessing includes data cleaning.

[0081] According to some embodiments of the present invention, data cleaning of the data source includes:

[0082] Segment the data source based on the business type to obtain a number of data blocks;

[0083] Input the number of data blocks into a pre-trained feature extraction model respectively, and output the feature values of the data blocks in a number of spatial dimensions;

[0084] Calculate the average value of the feature values of the number of data blocks in the same spatial dimension;

[0085] Compare the feature values of the data blocks in each spatial dimension with the average value of the corresponding spatial dimension respectively, and determine the covariance matrix of the data blocks according to the comparison results;

[0086] Analyze the covariance matrix of the data blocks, determine the target data block, perform dimensionality reduction processing on the target data block to obtain the hash value of the target data block;

[0087] Eliminate the target data blocks corresponding to hash values greater than the preset hash value to obtain the cleaned data.

[0088] Working principle of the above technical solution: Segment the data source based on the business type to obtain a number of data blocks; Input the number of data blocks into a pre-trained feature extraction model respectively, and output the feature values of the data blocks in a number of spatial dimensions; Calculate the average value of the feature values of the number of data blocks in the same spatial dimension; Compare the feature values of the data blocks in each spatial dimension with the average value of the corresponding spatial dimension respectively, and determine the covariance matrix of the data blocks according to the comparison results; Analyze the covariance matrix of the data blocks, determine the target data block, perform dimensionality reduction processing on the target data block to obtain the hash value of the target data block; Eliminate the target data blocks corresponding to hash values greater than the preset hash value to obtain the cleaned data.

[0089] Beneficial effects of the above technical solution: Dimensionality reduction processing is performed on the target data blocks in the data source, which is convenient for reducing the size of the data, improving the efficiency of data cleaning, and ensuring the accuracy of data cleaning.

[0090] According to some embodiments of the present invention, it further includes:

[0091] During the process of combining the target components, collect the data generated during the combination process as the first data; the first data includes combined data and attribute data related to the combined data;

[0092] Analyze the attribute data to obtain an analysis result;

[0093] According to the user-preset data processing rules and the analysis result, delete the attribute data of the edge type in the first data, obtain the second data and store it.

[0094] Working principle of the above technical solution: During the process of combining the target components, collect the data generated during the combination process as the first data; the first data includes combined data and attribute data related to the combined data; analyze the attribute data to obtain an analysis result; according to the user-preset data processing rules and the analysis result, delete the attribute data of the edge type in the first data, obtain the second data and store it. The preset data processing rules are the data that the user specifies to save according to requirements.

[0095] Beneficial effects of the above technical solution: It is convenient to reduce the data that the user does not need, save the storage space of the memory, and make the second data more accurate.

[0096] In one embodiment, storing the encryption software development model in the memory includes:

[0097] Calculate the distribution balance degree of the data already stored in the memory, and determine whether it is within the preset distribution balance degree range;

[0098]

[0099] where S is the distribution balance degree of the data already stored in the memory; N is the number of storage nodes included in the memory; w i is the size of the data already stored in the i-th storage node; Q is the total amount of the data already stored;

[0100] When it is determined that the distribution balance degree is not within the preset distribution balance degree range, sort the amounts of the data already stored in each storage node from small to large, and select the first preset number of storage nodes;

[0101] Divide the encrypted software development model into equal parts according to a preset quantity, and store it in the first preset quantity of storage nodes.

[0102] The working principle and beneficial effects of the above technical solution: Calculate the distribution balance degree of the data already stored in the memory, and determine whether it is within the preset distribution balance degree range; when it is determined that the distribution balance degree is not within the preset distribution balance degree range, sort according to the amount of the data already stored in each storage node from small to large, and select the first preset quantity of storage nodes; divide the encrypted software development model into equal parts according to a preset quantity, and store it in the first preset quantity of storage nodes. When storing data in the memory each time, calculate the balance degree in the memory, and at the same time facilitate accurately determining the corresponding storage node, and store the data in the corresponding storage node, which can play a role in adjusting the distribution balance degree of the memory, avoid the situation that some storage nodes are overloaded and some nodes are in an idle state for a long time, and facilitate improving the service life of the memory.

[0103] According to some embodiments of the present invention, determining a target component in the component database according to the software development requirement includes:

[0104] Obtain a sample software development requirement, perform clustering analysis on the sample software development requirement to obtain a number of classification sets, train the constructed initial neural network model based on each classification set until the loss function corresponding to the initial neural network model converges, and use the trained initial neural network as an advanced neural network model;

[0105] Determine the clustering centers in each classification set respectively, determine the distances between the clustering centers, and establish an identification tree; there are multiple nodes corresponding to the identification tree, and fill a number of advanced neural network models into each node, and fill one advanced neural network model on each node;

[0106] Calculate the distances between the software development requirement and each clustering center, screen out the advanced neural network model corresponding to the clustering center with the minimum distance as the target advanced neural network model; input the software development requirement into the target advanced neural network model to obtain a number of semantic recognition results of the software development requirement;

[0107] Divide the software development requirement to obtain a number of first sub-data, and fill them into the first nodes of a preset first data matrix respectively;

[0108] Divide each semantic recognition result to obtain a number of second sub-data, and fill them into the second nodes of a preset second data matrix respectively;

[0109] Establish the correspondence between the first node and the second node, fuse the first sub-data on the first node and the second sub-data on the second node, determine the fusion rate of each fusion node, and calculate the average fusion rate;

[0110] Based on a number of semantic recognition results, obtain a number of average fusion rates, and use the semantic recognition result corresponding to the maximum average fusion rate as the target semantic recognition result;

[0111] Determine the target component in the component database according to the target semantic recognition result.

[0112] The working principle of the above technical solution: Obtain the sample software development requirements, perform clustering analysis on the sample software development requirements to obtain a number of classification sets, train the constructed initial neural network model based on each classification set until the loss function corresponding to the initial neural network model converges, and use the initial neural network after training as the advanced neural network model; respectively determine the clustering centers in each classification set, determine the distances between the clustering centers, and establish a recognition tree; there are multiple nodes corresponding to the recognition tree, and fill a number of advanced neural network models into each node, with one advanced neural network model filled in each node; calculate the distances between the software development requirements and each clustering center, and select the advanced neural network model corresponding to the clustering center with the minimum distance as the target advanced neural network model; input the software development requirements into the target advanced neural network model to obtain a number of semantic recognition results of the software development requirements; divide the software development requirements to obtain a number of first sub-data, and fill them into the first nodes of the preset first data matrix respectively; divide each semantic recognition result to obtain a number of second sub-data, and fill them into the second nodes of the preset second data matrix respectively; establish the correspondence between the first node and the second node, fuse the first sub-data on the first node and the second sub-data on the second node, determine the fusion rate of each fusion node, and calculate the average fusion rate; based on a number of semantic recognition results, obtain a number of average fusion rates, and use the semantic recognition result corresponding to the maximum average fusion rate as the target semantic recognition result; determine the target component in the component database according to the target semantic recognition result. The matrix structures of the first data matrix and the second data matrix are the same.

[0113] Advantages of the above technical solution: Cluster analysis is performed based on the software development requirements of multiple samples to determine multiple advanced neural network models and establish an identification tree, which facilitates semantic recognition for different types of software development requirements. Based on the target advanced neural network model, several semantic recognition results of the software development requirements are obtained. To further improve the accuracy of semantic recognition, the first sub-data on the first node is fused with the second sub-data on the second node. The implementation principle is that the first data and the second sub-data are based on the same principle. The smoother the fusion process, that is, the higher the fusion efficiency, the more accurate the parsing result. Therefore, based on several semantic recognition results, several average fusion rates are obtained, and the overall parsing result is represented by the average fusion rate. The larger the average fusion rate, the more accurate the corresponding semantic recognition result. The semantic recognition result corresponding to the largest average fusion rate is used as the target semantic recognition result; the target component is determined in the component database according to the target semantic recognition result, which improves the accuracy of semantic analysis of software development requirements and also improves the accuracy of determining the target component.

[0114] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these changes and modifications.

Claims

1. A method for constructing a software development model adapted to multiple hardware platforms, characterized in that: include: Obtain the attribute information of multiple hardware platforms, build development components based on the attribute information, and generate a component database; Acquire software development requirements, and determine target components in the component database according to the software development requirements; Combining the target components and constructing a software development model according to the combination result; Running the software development model to obtain running data; Evaluate the operating data to determine the parameters to be corrected; obtain the type of the parameters to be corrected; when it is determined that the type of the parameters to be corrected is the first type, obtain the weight coefficient of each target component for constructing the software development model when the target component constructs the software development model; determine adjustment parameters according to the parameters to be corrected, adjust the weight coefficient according to the adjustment parameters, and obtain a corrected software development model according to the adjustment results; when it is determined that the type of the parameters to be corrected is the second type, determine a new target component according to the parameters to be corrected, combine the new target component into the software development model, and obtain a corrected software development model.

2. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 1, characterized in that: Obtain the attribute information of multiple hardware platforms, build development components based on the attribute information, and generate a component database, including: Obtaining attribute information of multiple hardware platforms; the attribute information includes processor architecture, main frequency, number of cores, memory size, storage type, peripheral interface, operating system support and power consumption; Acquire a data source according to the attribute information, pre-process the data source, determine the data used to construct the development component and mark it as target data; wherein the data source includes an underlying driver data source, a middleware data source and an application framework data source; Acquiring attribute information of the target data; Performing cluster analysis on the target data according to the attribute information to obtain a plurality of classification sets; Determine the level information of each component element in the same classification set, and regard the component elements of the same level as a data group; Development components are constructed according to the data group, and the development components are stored according to categories and levels to obtain a component database.

3. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 1, characterized in that: The step of combining the target components and constructing a software development model according to the combination result includes: Establish the operational relationship between each target component; Determine a combination order according to the operation relationship; In the process of combining the target components according to the combination order, the combination node of the first target component and the second target component is determined, and the code of the first target component and the code of the second target component are merged according to the combination node to obtain a first merged component. Based on the same method, the first merged component and the third target component are merged to obtain a second merged component. The target components are combined in sequence, and a software development model is constructed according to the combination results.

4. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 1, characterized in that: Also includes: The modified software development model is encrypted based on the RSA algorithm to obtain the encrypted software development model and store it.

5. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 1, characterized in that: The process of combining the target components also includes: The duration of each combination between target components is counted, and it is determined whether it is greater than a preset duration, and the combination process whose duration is greater than the preset duration is determined, and the combination process is optimized to obtain and store the optimization method.

6. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 2, characterized in that: The preprocessing includes data cleaning.

7. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 6, characterized in that: Performing data cleaning on the data source includes: Segmenting the data source based on business type to obtain a plurality of data blocks; Inputting the plurality of data blocks into pre-trained feature extraction models respectively, and outputting feature values ​​of the data blocks in a plurality of spatial dimensions; Calculate the average value of the eigenvalues ​​of several data blocks in the same spatial dimension; The eigenvalues ​​of the data block in each spatial dimension are compared with the average value of the corresponding spatial dimension, and the covariance matrix of the data block is determined according to the comparison results; Analyzing the covariance matrix of the data block to determine the target data block, performing dimensionality reduction processing on the target data block to obtain a hash value of the target data block; The target data blocks whose hash values ​​are greater than the preset hash values ​​are removed to obtain cleaned data.

8. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 1, characterized in that: Also includes: In the process of combining the target components, data generated in the combination process is collected as first data; The first data includes combined data and attribute data related to the combined data; Parsing the attribute data to obtain a parsing result; According to the data processing rules preset by the user and the analysis result, the attribute data of the edge type in the first data is deleted to obtain the second data and store it.

9. The method for constructing a software development model adapted to multiple hardware platforms as claimed in claim 1, characterized in that: Determining a target component in the component database according to the software development requirements includes: Obtain sample software development requirements, perform cluster analysis on the sample software development requirements to obtain several classification sets, train the constructed initial neural network model based on each classification set until the loss function corresponding to the initial neural network model converges, and use the trained initial neural network as an advanced neural network model; Determine the cluster centers in each classification set respectively, determine the distances between the cluster centers, and establish a recognition tree; the recognition tree corresponds to a plurality of nodes, and fills a plurality of advanced neural network models into each node, with one advanced neural network model being filled into each node; Calculating the distance between the software development requirements and each cluster center, selecting the advanced neural network model corresponding to the cluster center corresponding to the minimum distance as the target advanced neural network model; inputting the software development requirements into the target advanced neural network model to obtain several semantic recognition results of the software development requirements; The software development requirements are divided to obtain a plurality of first sub-data, and the first sub-data are filled into the first nodes of the preset first data matrix respectively; Segment each semantic recognition result to obtain a plurality of second sub-data, and fill them into the second nodes of the preset second data matrix respectively; Establishing a corresponding relationship between the first node and the second node, fusing the first sub-data on the first node with the second sub-data on the second node, determining a fusion rate of each fusion node, and calculating an average fusion rate; Based on several semantic recognition results, several average fusion rates are obtained, and the semantic recognition result corresponding to the largest average fusion rate is taken as the target semantic recognition result; The target component is determined in the component database according to the target semantic recognition result.

Images