Generalization processing protection method, device and system for cloud engineering quotation

Abstract

The application relates to the technical field of engineering quotation, in particular to a generalization processing protection method, device and system of a cloud engineering quotation system, which comprises the following steps: obtaining project list information of all project engineering which needs to be batched for engineering quotation and corresponding policy document information of the project engineering, composing an engineering data set, obtaining a quotation error of each quotation task, and recording a system response time of each quotation task; determining a dimension reduction performance coefficient of each quotation task; determining a resource scheduling coefficient of each quotation task; and performing hierarchical scheduling processing on different quotation tasks based on the resource scheduling coefficient. The application can realize dynamic priority scheduling, tilt supply of calculation resources to key quotation tasks by identifying urgent quotation task demands, improve the average response speed of the quotation tasks, and realize efficient parallel operation of complex quotation tasks under the premise of guaranteeing data security.

Description

Technical Field

[0001] This application relates to the field of engineering quotation technology, specifically to a generalized processing and protection method, device, and system for cloud-based engineering quotation. Background Technology

[0002] With the deepening of digital transformation in the engineering industry, cloud-based engineering quotation systems, through their elastic computing and centralized storage capabilities, can integrate dynamic market data and historical data, enabling real-time information updates and sharing, and solving the problem of information silos in traditional engineering cost estimation.

[0003] When existing cloud-based engineering quotation systems process multidimensional data from batch engineering projects, they lack the ability to generalize and integrate multimodal data. The strong coupling and high redundancy between features lead to an explosion in the dimensionality of the feature space. At the same time, the direct participation of raw data in calculations can easily expose sensitive information, resulting in excessive consumption of computing resources and delays in real-time quotation response. It is impossible to balance computational efficiency and feature interpretability while ensuring data security. Summary of the Invention

[0004] To address the aforementioned technical issues, a generalized processing protection method, apparatus, and system for cloud-based engineering quotations are provided to resolve existing problems.

[0005] The solution to the technical problem in this application is to provide a generalized processing protection method, apparatus, and system for cloud-based engineering quotations, including the following steps:

[0006] In a first aspect, embodiments of this application provide a generalized processing protection method for cloud-based engineering quotations, the method comprising the following steps:

[0007] By converting the data types of the project list information and the corresponding policy document information of all projects that require batch engineering quotations, an engineering dataset is formed. Engineering quotations are then performed for each project, the quotation error of each quotation task is calculated, and the system response time of each quotation task is recorded.

[0008] Dimensionality reduction analysis is performed on the engineering dataset to extract each principal component. The importance of different principal components corresponding to each bidding task and their contribution to the original data information are analyzed to determine the dimensionality reduction efficiency coefficient of each bidding task.

[0009] By analyzing the correlation between the system response time and dimensionality reduction efficiency coefficient of each bidding task, as well as the distribution density of dimensionality reduction efficiency coefficients among different bidding tasks, and combining the dimensionality reduction efficiency coefficients, the resource scheduling coefficient of each bidding task is determined.

[0010] Based on the resource scheduling coefficient, different pricing tasks are subject to hierarchical scheduling processing.

[0011] Preferably, the quotation error is obtained through the following process:

[0012] The cloud-based engineering quotation system predicts the engineering quotation for each project and obtains the total quotation amount for each project.

[0013] The formula for calculating the pricing error for each pricing task is as follows: .

[0014] Preferably, the step of performing dimensionality reduction analysis on the engineering dataset and extracting each principal component includes: using the PCA method to reduce the dimensionality of all data in the engineering dataset of each quotation task and extracting the eigenvectors and variance contribution rates of each principal component.

[0015] Preferably, the further measurement process of the importance is as follows: the bidding error of each bidding task and the feature vectors of all principal components are used as inputs to the random forest model to obtain the importance score of each principal component.

[0016] Preferably, the dimensionality reduction efficiency coefficient is the sum of the products of the importance scores and the variance contribution rates of all principal components in each quotation task.

[0017] Preferably, the relevant changes are further measured using regression coefficients, specifically as follows:

[0018] The system response time of each quotation task is combined with the dimensionality reduction efficiency coefficient to form a two-dimensional array;

[0019] Perform linear regression analysis on the two-dimensional arrays corresponding to all quotation tasks to obtain the regression coefficients.

[0020] Preferably, determining the resource scheduling coefficient for each bidding task includes:

[0021] Cluster the dimensionality reduction efficiency coefficients of all bidding tasks to obtain the local density corresponding to each bidding task; perform positive mapping on the dimensionality reduction efficiency coefficients;

[0022] The resource scheduling coefficient is the product of the positive mapping result, the regression coefficient, and the local density corresponding to each bidding task.

[0023] Preferably, the hierarchical scheduling process for different pricing tasks includes:

[0024] Based on the resource scheduling coefficients of all bidding tasks, a kernel density estimation curve is constructed. The resource scheduling coefficients corresponding to the cumulative distribution function reaching the preset first value and the preset second value are calculated through the kernel density estimation curve, and are denoted as the first segmentation threshold and the second segmentation threshold, where the preset second value is greater than the preset first value.

[0025] Based on the first and second segmentation thresholds, the multi-level feedback queue is divided into a high-priority queue, a medium-priority queue, and a low-priority queue.

[0026] Bidding tasks with different resource scheduling coefficients are assigned to the corresponding priority teams for resource scheduling.

[0027] Secondly, embodiments of this application also provide a generalization processing protection device for cloud engineering quotations. The device stores a computer program, which, when executed by a processor, implements the steps of any of the above-described generalization processing protection methods for cloud engineering quotations.

[0028] Thirdly, embodiments of this application also provide a generalized processing protection system for cloud engineering quotations, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described generalized processing protection methods for cloud engineering quotations.

[0029] This application has at least the following beneficial effects:

[0030] This application performs dimensionality reduction on the high-dimensional data of each bidding task, extracts principal components, analyzes the importance characteristics of different principal components and the amount of original information contained in the principal components, and calculates the dimensionality reduction efficiency coefficient for each bidding task. Its beneficial effect lies in considering the impact of the information contained in the data in the dimensionality-reduced low-dimensional space on the evaluation of the final project bid, reflecting the information interpretability of the high-dimensional data after dimensionality reduction. It can provide high information density and low redundancy information interpretability for project bids while eliminating the direct exposure of original sensitive data, balancing the dual needs of computational efficiency and privacy protection. The dimensionality reduction efficiency coefficients of all bidding tasks are clustered, grouping bidding tasks with similar dimensionality reduction efficiency into one category. Furthermore, the correlation between the dimensionality reduction efficiency coefficient and the system response time reflects tasks with higher dimensionality reduction efficiency. The impact of the bidding task coefficient on real-time response latency is analyzed. The resource scheduling coefficient for each bidding task is determined, which has the advantage of considering the dynamic balance between dimensionality reduction of high-dimensional data and resource competition, eliminating the monopoly of resources by extreme bidding tasks, and identifying high-value bidding tasks and resource-intensive tasks by integrating dimensionality reduction efficiency and cluster density characteristics, effectively preventing the monopolization of computing resources by extremely complex bidding tasks. Based on the resource scheduling coefficient, different bidding tasks are subject to hierarchical scheduling. This has the advantage of dividing the multi-level feedback queue into high, medium, and low priority queues, realizing dynamic priority scheduling, identifying urgent bidding task needs, allocating computing resources preferentially to key bidding tasks, improving the average response speed of bidding tasks, and achieving efficient parallel computation of complex bidding tasks while ensuring data security. Attached Figure Description

[0031] The following section provides a more detailed explanation of the generalized processing protection method for cloud engineering quotations in this application, with reference to the accompanying drawings.

[0032] Figure 1 A flowchart illustrating the steps of the generalized processing protection method for cloud-based engineering quotations provided in this application embodiment;

[0033] Figure 2 A flowchart illustrating the steps of a method for obtaining the resource scheduling coefficient for each bidding task provided in this application embodiment. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description, in conjunction with the accompanying drawings and implementation examples, provides a more comprehensive explanation of the generalized processing and protection method, apparatus, and system for cloud-based engineering quotations proposed in this application. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0036] Please see Figure 1 The diagram illustrates a flowchart of a cloud-based engineering quotation generalization processing protection method according to an embodiment of this application. The method includes the following steps:

[0037] Step 1: After converting the data types of the project list information and the corresponding policy document information of all projects that require batch engineering quotations, an engineering dataset is formed. Engineering quotations are then performed for each project, the quotation error of each quotation task is calculated, and the system response time of each quotation task is recorded.

[0038] Extract the project list information and corresponding policy documents for all projects requiring batch quoting from the cloud-based engineering quotation system. Based on this information, quotes are generated for all projects through the cloud-based engineering quotation system. The quotation error for each project is calculated by comparing the total quotation amount with the actual cost of that project, and the system response time for each quotation task is recorded. The formula for calculating the quotation error is as follows: .

[0039] It should be noted that the project list information includes, but is not limited to, material properties, construction parameters, and environmental variables. Material properties include categorical data such as steel grade and concrete impermeability grade; construction parameters include formwork turnover times and machinery shift combinations, such as 2 excavators and 3 mixers, which are numerical time-series data; environmental variables include a document describing the geological conditions of the project location and temperature and humidity data of the climate zone; secondly, policy document information includes environmental tax coefficient PDF text containing environmental tax adjustment coefficients for different materials or construction processes, as well as regional quota difference table data, describing the quota adjustments for labor costs, material costs, etc. in different regions.

[0040] Because the data types of the project list information and its policy information are inconsistent, all the acquired data is preprocessed, specifically as follows:

[0041] By mapping text descriptions, category labels, and discrete numerical values ​​to fixed-dimensional numerical vectors through hash encoding, semantic differences are eliminated and the consistency of the feature space is guaranteed. This process not only ensures the uniformity of the data, but also achieves data privacy protection through hash encoding, avoids the direct exposure of original sensitive data, and enhances data security protection.

[0042] For information data such as policy documents and geological reports, a global vectorization strategy based on word frequency statistics represents the content as a standardized bag-of-words vector, preserving the density of key information.

[0043] For numerical data, a combination of binning discretization and equal-frequency segmentation is used to convert it into discrete interval numbers aligned with the hash coding dimension, ensuring that the numerical distribution characteristics are preserved.

[0044] After all the data is transformed, it is standardized to form an engineering dataset. Through the above processing, not only is the data processability improved, but the security risks of sensitive data during transmission and storage are also reduced, thus ensuring data security.

[0045] In this embodiment, Z-score standardization is used for processing. Z-score standardization is a well-known technology and will not be described in detail here.

[0046] Therefore, we obtain the engineering dataset and bidding error for each bidding task, as well as the system response time for each bidding task.

[0047] Step 2: Perform dimensionality reduction analysis on the engineering dataset, extract each principal component, analyze the importance of different principal components corresponding to each bidding task, and their contribution to the original data information, and determine the dimensionality reduction efficiency coefficient of each bidding task.

[0048] Due to the strong coupling and high redundancy among the data in the engineering dataset, such as the strong positive correlation between concrete strength and cement grade and aggregate particle size, and the dynamic correlation between construction process parameters and environmental temperature and humidity, the feature space dimension explosion and information overlap can occur, resulting in problems such as excessive consumption of computing resources and delays in real-time quotation response. As a result, the collected data exhibits complex distribution characteristics with high noise and low interpretability.

[0049] Based on the above analysis, a dimensionality reduction analysis is performed on the engineering dataset, specifically as follows:

[0050] For each quotation task, dimensionality reduction is performed on all data in the engineering dataset, and the feature vectors and variance contribution rates of each principal component are extracted.

[0051] In this embodiment, Principal Component Analysis (PCA) is used for dimensionality reduction. PCA is a well-known technique and will not be described in detail here. The extraction of principal components with a cumulative variance contribution rate of 85% or higher is another implementation method that the implementer can set according to the actual situation.

[0052] It should be noted that by extracting principal components through dimensionality reduction, key business information can be preserved in the lower-dimensional space after dimensionality reduction, and the method can be generalized to multi-dimensional and dynamically changing pricing scenarios. Furthermore, the PCA method extracts the orthogonal projection direction with the largest variance from massive heterogeneous data. By compressing redundant dimensions and eliminating multicollinearity interference, high-dimensional features are mapped to the lower-dimensional space. This reduces the resource load of real-time computing in the cloud system and reduces the risk of sensitive raw data being directly exposed through feature condensation, thereby indirectly enhancing data security protection.

[0053] Furthermore, by using the eigenvectors of each principal component, the significance of each principal component in predicting pricing errors is evaluated, specifically as follows:

[0054] The pricing error of each pricing task and the feature vectors of all principal components are used as inputs to the random forest model to obtain the importance score of each principal component.

[0055] It should be noted that the process of training a random forest model by collecting a large amount of engineering datasets from historical projects, and then performing feature importance analysis using the random forest model, is a well-known technique and will not be elaborated here. Secondly, by quantifying the impact weights of orthogonalized comprehensive features on pricing errors through random forest, key driving factors are extracted from the privacy-preserving generalized feature space. This avoids the redundancy of the original data and the risk of sensitive information being exposed, and provides an interpretable basis for dynamically optimizing pricing strategies.

[0056] Furthermore, based on the variance contribution rate and the importance score, the dimensionality reduction efficiency coefficient is calculated, specifically as follows:

[0057] The sum of the products of the importance scores and the variance contribution rates of all principal components in each pricing task is used as the dimensionality reduction efficiency coefficient for each pricing task.

[0058] It should be noted that the variance contribution rate reflects the explanatory power of the principal component on the original high-dimensional data. The larger the variance contribution rate, the more original information features the principal component retains, indicating that the principal component contains more information affecting the project quotation. The importance score reflects the correlation or causal relationship between the principal component and the quotation error, quantifying the principal component's predictive power on the quotation error. The larger the importance score, the stronger the predictive power of the principal component on the quotation error. The dimensionality reduction efficiency coefficient evaluates the preservation of original information in the dimensionality-reduced low-dimensional space and the quantitative coupling result of its correlation with the quotation error. It describes the comprehensive driving efficiency of the privacy-protected feature space on the quotation error in the cloud-based project quotation system. The larger the dimensionality reduction efficiency coefficient, the stronger the influence of the data information in the dimensionality-reduced low-dimensional space on the comprehensive evaluation of the quotation error. It can provide high information density and low redundancy information interpretation capabilities for project quotations while eliminating the direct exposure of original sensitive data, balancing the dual needs of computational efficiency and privacy protection.

[0059] Thus, the dimensionality reduction efficiency coefficient for each quotation task is obtained.

[0060] Step 3: By analyzing the correlation between the system response time and dimensionality reduction efficiency coefficient of each bidding task, and the distribution density of dimensionality reduction efficiency coefficients among different bidding tasks, the resource scheduling coefficient of each bidding task is determined in conjunction with the dimensionality reduction efficiency coefficient.

[0061] The flowchart of the method for obtaining the resource scheduling coefficient for each bidding task provided in this application embodiment is as follows: Figure 2 As shown.

[0062] Furthermore, in cloud-based engineering quotation systems, when multiple engineering projects are continuously quoted, the complex coupling relationships in their high-dimensional feature space and the uneven distribution of resource consumption lead to significant delays in the real-time response to engineering quotations. This results in highly volatile and locally intensive computational loads on cloud-based engineering quotation systems.

[0063] Therefore, clustering is performed on the dimensionality reduction efficiency coefficients to analyze the resource consumption patterns of the cloud-based engineering quotation system during the computation of different quotation tasks, thereby quantifying the centralized allocation requirements of resources. Specifically:

[0064] The system response time of each quotation task is combined with the dimensionality reduction efficiency coefficient to form a two-dimensional array;

[0065] Perform linear regression analysis on all two-dimensional arrays and obtain the regression coefficients;

[0066] It should be noted that linear regression analysis is a well-known technique and will not be elaborated upon here.

[0067] Cluster the dimensionality reduction efficiency coefficients of all bidding tasks to obtain the local density corresponding to each bidding task;

[0068] In this embodiment, the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm is used for clustering. The DBSCAN clustering algorithm is a well-known technique and will not be described in detail here. The neighborhood radius of the DBSCAN clustering algorithm is set. Minimum number of samples As another implementation method, the implementer can set it according to the actual situation.

[0069] Apply a positive mapping to the dimensionality reduction efficiency coefficients;

[0070] In this embodiment, the positive mapping process is as follows: a positive mapping is performed on the dimensionality reduction efficiency coefficients using a logarithmic function, assuming the first... The dimensionality reduction efficiency coefficient of each quotation task is denoted as . Then The result of the positive mapping is taken as the result of the positive mapping, such that the result of the positive mapping is greater than or equal to 0, where, It is a logarithmic function with base 10.

[0071] It should be noted that the logarithmic function is used to... Nonlinear smoothing is performed to suppress the data distribution skew caused by high-dimensional data discretization binning and word frequency statistics, avoid the dominance of extreme dimensionality reduction performance pricing on resource scheduling strategies, and quantify the information density and interpretability contribution of multi-source data to engineering cost prediction after dimensionality reduction.

[0072] The product of the positive mapping result, the regression coefficient, and the local density corresponding to each bidding task is used as the resource scheduling coefficient for each bidding task.

[0073] In this embodiment, the first The formula for calculating the resource scheduling coefficient for each bidding task is:

[0074]

[0075] in, For the first Resource scheduling coefficient for each quotation task. The regression coefficients are... No. Local density of each quotation task For the first The dimensionality reduction efficiency coefficient of a quotation task It is a logarithmic function with base 10.

[0076] It should be noted that the regression coefficient reflects the strength of the linear relationship between the dimensionality reduction efficiency coefficient and the system response time. The larger the regression coefficient, the stronger the global linear impact of the dimensionality reduction efficiency coefficient of the data after hash encoding and bag-of-words vectorization preprocessing on the real-time response latency. It quantifies the correlation between the privacy-preserving feature space after PCA dimensionality reduction and system resource consumption. The larger the regression coefficient, the more necessary it is to sacrifice higher computational latency to protect data security and avoid exposing the original sensitive data. The local density reflects the resource competition among bidding tasks with similar dimensionality reduction efficiency. The larger the local density, the easier it is to trigger resource competition and data access conflicts. The resource scheduling coefficient measures the comprehensive efficiency of multimodal data in the cloud engineering bidding system after privacy-preserving dimensionality reduction and resource consumption mode. The larger the resource scheduling coefficient, the more necessary it is to prioritize resource allocation to reduce real-time response latency when the bidding achieves low redundancy computation through high information density features under low sensitive data exposure.

[0077] At this point, the resource scheduling coefficient for each bidding task is obtained.

[0078] Step 4: Based on the resource scheduling coefficient, perform hierarchical scheduling processing for tasks with different quotations.

[0079] Furthermore, based on resource scheduling coefficients, a dynamic preemptive scheduling mechanism for GPU / CPU hybrid computing resources is constructed. The pricing tasks are processed in a tiered manner using these coefficients, and asynchronous pipelined parallel technology is employed to achieve load balancing for high-dimensional matrix operations, ultimately generating a real-time pricing strategy after error compensation. A modular and layered architecture is adopted to support efficient data processing and resource scheduling. At the hardware level, the cloud-based engineering pricing system integrates a GPU / CPU hybrid computing architecture, where the GPU accelerates high-dimensional matrix operations, and the CPU handles task scheduling and logic control. At the software level, the cloud-based engineering pricing system decouples feature reduction, model inference, and error correction computational stages through a dynamic preemptive scheduling mechanism and asynchronous pipelined parallel technology, achieving dynamic task allocation and load balancing. In addition, the cloud-based engineering pricing system uses Linux Cgroups container isolation technology to ensure resource isolation and data security between different tasks, preventing data leakage or contamination caused by resource contention or task overlap. The block matrix operation, mixed-precision computation, and asynchronous pipelined parallel technology are all well-known technologies in this field and will not be elaborated upon here.

[0080] In the cloud-based engineering quotation system, to optimize resource allocation and scheduling efficiency, quotation tasks are assigned to different processing queues based on their priority, and corresponding resources are allocated accordingly. By constructing a dynamic priority-driven parallel computing optimization mechanism, all quotation tasks are processed in a hierarchical manner, thereby achieving a dynamic preemptive scheduling mechanism. Specifically:

[0081] Based on the resource scheduling coefficients of all bidding tasks, a kernel density estimation curve is constructed. The resource scheduling coefficients corresponding to the cumulative distribution function (CDF) reaching the preset first value and the preset second value are calculated through the kernel density estimation curve. These are denoted as the first segmentation threshold and the second segmentation threshold, where the preset second value is greater than the preset first value.

[0082] In this embodiment, the first preset value is 20% and the second preset value is 70%. As other implementation methods, the implementer can set them according to the actual situation. Therefore, the resource scheduling coefficient corresponding to the cumulative distribution function reaching 20% ​​is recorded as the first segmentation threshold, and the resource scheduling coefficient corresponding to the cumulative distribution function reaching 70% is recorded as the second segmentation threshold.

[0083] It should be noted that the kernel density estimation curve and cumulative distribution function are well-known techniques and will not be elaborated here. The horizontal axis of the kernel density estimation curve represents the resource scheduling coefficient, and the vertical axis represents the probability density. Furthermore, since the kernel density estimation curve reflects the probability density distribution of the resource scheduling coefficient, and the CDF is the integral of the kernel density estimation curve, reflecting the probability that the resource scheduling coefficient is less than or equal to a certain value, a 20% CDF corresponds to the 20th percentile of the resource scheduling coefficient in the kernel density estimation curve. Therefore, when the CDF reaches 20%, it indicates that 20% of the bidding tasks have resource scheduling coefficients less than or equal to the first segmentation threshold, meaning that these 20% of bidding tasks with resource scheduling coefficients less than or equal to the first segmentation threshold have lower resource requirements.

[0084] Based on the first and second segmentation thresholds, the multi-level feedback queue is divided into a high-priority queue, a medium-priority queue, and a low-priority queue.

[0085] Bidding tasks with a resource scheduling coefficient less than or equal to the first segmentation threshold are assigned to a low-priority queue; bidding tasks with a resource scheduling coefficient greater than the first judgment threshold and less than or equal to the second segmentation threshold are assigned to a medium-priority queue; and bidding tasks with a resource scheduling coefficient greater than the second segmentation threshold are assigned to a high-priority queue.

[0086] It should be noted that multi-level feedback queues are a well-known technology and will not be elaborated upon here.

[0087] Therefore, bidding tasks with different resource scheduling coefficients are assigned to corresponding priority queues for resource scheduling, and resources are allocated preferentially to bidding tasks in high-priority queues.

[0088] The aforementioned dynamic preemptive scheduling mechanism effectively solves the core problem of resource overload for high-dimensional data information computation in cloud-based engineering quotation systems. It achieves precise hierarchical scheduling of computing resources, and by combining GPU TensorCore acceleration with mixed-precision block matrix computation, it reduces the time consumption of high-dimensional feature computation and decreases GPU memory usage. Asynchronous pipelined parallel technology decouples feature dimensionality reduction, model inference, and error correction stages, achieving multi-task load balancing. While ensuring privacy and security, it achieves breakthroughs in computational efficiency and accuracy for complex engineering quotation tasks, reducing real-time quotation response latency. Under the premise of ensuring enterprise data security and integrity, the cloud-based engineering quotation system can adaptively handle complex scenarios such as regional quota differences, policy changes, and multimodal engineering data coupling, reducing the quotation error rate. The block matrix computation, mixed-precision computation, and asynchronous pipelined parallel technology are all well-known technologies in this field and will not be elaborated upon here.

[0089] This application embodiment also provides a generalization processing protection device for cloud engineering quotations. The device stores a computer program, which, when executed by a processor, implements the steps of any one of the above-described methods for generalization processing protection of cloud engineering quotations.

[0090] Based on the same inventive concept as the above methods, this application also provides a generalized processing protection system for cloud engineering quotations, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described methods for generalized processing protection of cloud engineering quotations.

[0091] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.

[0092] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0093] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application. Therefore, any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of this application, without departing from the content of the technical solution of this application, shall fall within the protection scope of the technical solution of this application.

Claims

1. A generalized processing and protection method for cloud-based engineering quotations, characterized in that, The method includes the following steps: By converting the data types of the project list information and the corresponding policy document information of all projects that require batch engineering quotations, an engineering dataset is formed. Engineering quotations are then performed for each project, the quotation error of each quotation task is calculated, and the system response time of each quotation task is recorded. Dimensionality reduction analysis is performed on the engineering dataset to extract each principal component. The importance of different principal components corresponding to each bidding task and their contribution to the original data information are analyzed to determine the dimensionality reduction efficiency coefficient of each bidding task. By analyzing the correlation between the system response time and dimensionality reduction efficiency coefficient of each bidding task, as well as the distribution density of dimensionality reduction efficiency coefficients among different bidding tasks, and combining the dimensionality reduction efficiency coefficients, the resource scheduling coefficient of each bidding task is determined. Based on the resource scheduling coefficient, different bidding tasks are subject to hierarchical scheduling processing. The step of performing dimensionality reduction analysis on the engineering dataset and extracting each principal component includes: using the PCA method to reduce the dimensionality of all data in the engineering dataset of each quotation task and extracting the eigenvectors and variance contribution rates of each principal component; The further measurement process of the importance is as follows: the bidding error of each bidding task and the feature vectors of all principal components are used as inputs to the random forest model to obtain the importance score of each principal component; The dimensionality reduction efficiency coefficient is the sum of the products of the importance scores and the variance contribution rates of all principal components in each quotation task; The relevant changes are further measured using regression coefficients, specifically as follows: The system response time of each quotation task is combined with the dimensionality reduction efficiency coefficient to form a two-dimensional array; Perform linear regression analysis on the two-dimensional arrays corresponding to all quotation tasks to obtain the regression coefficients; The determination of the resource scheduling coefficient for each bidding task includes: Cluster the dimensionality reduction efficiency coefficients of all bidding tasks to obtain the local density corresponding to each bidding task; perform positive mapping on the dimensionality reduction efficiency coefficients; The resource scheduling coefficient is the product of the positive mapping result, the regression coefficient, and the local density corresponding to each bidding task; The hierarchical scheduling process for different pricing tasks includes: Based on the resource scheduling coefficients of all bidding tasks, a kernel density estimation curve is constructed. The resource scheduling coefficients corresponding to the cumulative distribution function reaching the preset first value and the preset second value are calculated through the kernel density estimation curve. These are denoted as the first segmentation threshold and the second segmentation threshold, where the preset second value is greater than the preset first value. Based on the first and second segmentation thresholds, the multi-level feedback queue is divided into a high-priority queue, a medium-priority queue, and a low-priority queue. Bidding tasks with different resource scheduling coefficients are assigned to the corresponding priority teams for resource scheduling.

2. The generalized processing and protection method for cloud engineering quotation as described in claim 1, characterized in that, The quotation error is obtained through the following process: The cloud-based engineering quotation system predicts the engineering quotation for each project and obtains the total quotation amount for each project. The formula for calculating the pricing error for each pricing task is as follows: .

3. A generalized processing and protection device for cloud-based engineering quotations, wherein the device stores a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the generalized processing protection method for cloud engineering quotations as described in any one of claims 1-2.

4. A generalized processing and protection system for cloud-based engineering quotations, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the generalized processing protection method for cloud engineering quotations as described in any one of claims 1-2.

Images