Construction Project Cost Data Disaster Recovery Method and Disaster Recovery Management System

Abstract

The present invention provides a disaster tolerance method for construction project cost data and a disaster tolerance management system. By initializing a neural network, disaster tolerance backup prediction is performed on multiple candidate data in a sample construction project cost data sequence to generate training disaster tolerance backup allocation space data. The global network learning error is determined by comparing the predicted data with the labeled data, so as to screen out effective sample data for network parameter learning, and obtain a trained disaster tolerance backup model. This method can accurately and efficiently perform disaster tolerance backup processing on construction project cost data, improve the accuracy and efficiency of data disaster tolerance, and ensure the security and reliability of construction project cost data. In practical applications, the trained disaster tolerance backup model is used to perform disaster tolerance backup processing on the construction project cost data set to be processed, effectively avoiding the risk of data loss or damage.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and more particularly, to a disaster recovery method for construction project cost data and a disaster recovery management system. Background Art

[0002] In the management process of construction projects, cost data is crucial information, which is directly related to the project's cost budget, fund allocation, and economic benefit evaluation. However, due to various force majeure factors, such as natural disasters, system failures, or human errors, the cost data faces the risk of loss or damage. Once these data are lost or damaged, it will bring great trouble to the smooth progress of the project and may even cause serious economic losses.

[0003] Traditional data disaster recovery methods often rely on simple backup strategies, such as regularly copying data to other storage devices or remote servers. However, these methods are often unable to cope when faced with large-scale and highly complex construction project cost data. On the one hand, simple backup strategies are difficult to ensure the real-time nature and integrity of data, and data may be lost or incorrect during the backup process; on the other hand, traditional disaster recovery methods lack an intelligent decision-making mechanism and cannot flexibly allocate backups according to the actual needs and importance of data, resulting in waste or insufficiency of backup resources. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide a disaster recovery method for construction project cost data, the method comprising:

[0005] Obtaining a sample construction project cost data sequence and initializing a neural network; the sample construction project cost data sequence includes a plurality of candidate sample construction project cost data; each of the candidate sample construction project cost data carries disaster recovery backup allocation space data annotation data;

[0006] Performing disaster recovery backup prediction on each candidate sample construction project cost data according to the initialized neural network, and generating training disaster recovery backup allocation space data for each candidate sample construction project cost data;

[0007] For each of the candidate sample construction project cost data, determining a global network learning error according to the training disaster recovery backup allocation space data and the disaster recovery backup allocation space data annotation data, and extracting effective sample construction project cost data from the sample construction project cost data sequence according to the global network learning errors of the candidate sample construction project cost data;

[0008] Performing network parameter learning on the initialized neural network according to the extracted effective sample construction project cost data, and outputting a trained disaster recovery backup model;

[0009] Obtain the construction project cost data set to be subject to data disaster recovery processing, and based on the trained disaster recovery backup model, and according to the predicted disaster recovery backup allocation space data, perform disaster recovery backup processing on the construction project cost data set.

[0010] On the other hand, an embodiment of the present invention further provides a disaster recovery management system, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, the embodiment of the present application predicts disaster recovery backup for multiple candidate data in the sample construction project cost data sequence through an initialized neural network, and generates training disaster recovery backup allocation space data. By comparing the predicted data with the labeled data, the global network learning error is determined, so as to screen out valid sample data for network parameter learning, and obtain a trained disaster recovery backup model. This method can accurately and efficiently perform disaster recovery backup processing on construction project cost data, improve the accuracy and efficiency of data disaster recovery, and ensure the security and reliability of construction project cost data. In practical applications, the trained disaster recovery backup model is used to perform disaster recovery backup processing on the construction project cost data set to be processed, effectively avoiding the risk of data loss or damage. Description of the Drawings

[0012] Figure 1 It is a schematic execution flowchart of the construction project cost data disaster recovery method provided by an embodiment of the present invention.

[0013] Figure 2 It is a schematic hardware architecture diagram of the disaster recovery management system provided by an embodiment of the present invention. Detailed Embodiments

[0014] The present invention will be specifically described below in conjunction with the accompanying drawings of the specification. Figure 1 It is a schematic flowchart of the construction project cost data disaster recovery method provided by an embodiment of the present invention. The construction project cost data disaster recovery method will be introduced in detail below.

[0015] Step S110, obtain a sample construction project cost data sequence and initialize a neural network. The sample construction project cost data sequence includes multiple candidate sample construction project cost data. Each of the candidate sample construction project cost data carries disaster recovery backup allocation space data labeled data.

[0016] In this embodiment, in the scenario of construction project cost management, a certain construction project database stores the cost data of numerous past construction projects. These cost data cover various aspects of information, including the procurement costs of basic building materials, the rental and purchase costs of various equipment, human resource costs, and various management costs. Part of the data in this database constitutes the sample construction project cost data sequence. For example, the cost data of one construction project includes the expenses for materials such as concrete, steel, and wood, the labor costs of the construction team, the rental costs of large-scale construction machinery, etc. And for this project, based on past experience or the evaluation after actual operation, a relatively reasonable disaster recovery backup allocation space data annotation data has been determined. For example, according to a certain ratio, the space required for disaster recovery backup has been set for various cost data respectively. For example, the concrete cost data may be allocated 1% of the total cost backup space, and the labor cost data is allocated 3% of the backup space, etc. The cost data of this construction project becomes a candidate sample construction project cost data. At the same time, the initialization of the neural network construction is based on a specific algorithm architecture. For example, a multi-layer perceptron (MLP) structure is adopted, which includes an input layer, several hidden layers, and an output layer. The number of neurons in the input layer is determined according to the number of features in the sample construction project cost data. For example, the number of neurons in the input layer is determined according to the types of materials, cost types, etc. The number of neurons and the number of layers in the hidden layer are initially set according to experience or pre-experiments. The output layer is used to output the prediction results related to the disaster recovery backup allocation space data.

[0017] Step S120: Respectively perform disaster recovery backup prediction on each candidate sample construction project cost data according to the initialized neural network, and generate the training disaster recovery backup allocation space data for each candidate sample construction project cost data.

[0018] Taking the construction project cost data mentioned above as an example, input this candidate sample construction project cost data into the initialized neural network. Inside the neural network, the data first enters the input layer, and each neuron receives the corresponding data feature value. Then, the data undergoes complex calculations and transformations in the hidden layer. Assuming that the sigmoid activation function is used in the hidden layer, when the data is transmitted between the neurons in the hidden layer, it will be linearly combined according to the connection weights between the neurons, and then undergo a non-linear transformation through the sigmoid function. For the output layer, it further calculates based on the output result of the hidden layer, so as to generate the training disaster recovery backup allocation space data for this candidate sample construction project cost data. This training disaster recovery backup allocation space data may include logical relationships such as the priority ranking of different cost components during disaster recovery backup, and the backup space allocation ratio of different cost data based on risk assessment. For example, for the material cost part, the neural network gives a backup space allocation logical structure classified by different materials according to various factors such as the price fluctuation risk of materials and the stability of material supply. For rare metal materials with large price fluctuations and unstable supply, allocate a larger proportion and higher priority backup space; for ordinary building materials with relatively stable prices and sufficient supply, allocate a smaller proportion and lower priority backup space.

[0019] Step S130, for each of the candidate sample construction project cost data, determine the global network learning error based on the training disaster recovery backup allocation space data and the disaster recovery backup allocation space data annotation data, and extract the effective sample construction project cost data from the sample construction project cost data sequence according to the global network learning error of each candidate sample construction project cost data.

[0020] Continuing with the construction project cost data as an example, for each candidate sample, first determine the first network learning error corresponding to each initialized neural network. For example, for a candidate sample, the backup space ratio of a certain material cost given in its training disaster recovery backup allocation space data is 20%, while the backup space ratio of this material cost in its disaster recovery backup allocation space data annotation is 15%. Calculate the error of this part through a specific error calculation function (such as the mean square error function), and then combine the errors of all cost components to obtain the first network learning error corresponding to this initialized neural network. If there are multiple initialized neural networks, for example, two, calculate the first network learning error corresponding to each neural network respectively. Then, determine the relative learning error between every two training disaster recovery backup allocation space data. Assume that there are differences in the backup space allocation of equipment rental costs in the training disaster recovery backup allocation space data of two initialized neural networks for the same candidate sample, and calculate the quantization value of this difference to obtain the relative learning error. Determine the sum of each first network learning error and the relative learning error as the global network learning error.

[0021] When extracting the construction project cost data of effective samples, assume there are multiple extraction methods. One method is to label the construction project cost data of the set proportion of samples with the smallest global network learning error as the construction project cost data of effective samples. For example, from 100 candidate samples, select 20% (i.e., 20) of the samples with the smallest global network learning error as the effective samples. Another method is to label the construction project cost data of samples with a global network learning error less than the set error as the construction project cost data of effective samples. For example, if the set error is 0.1 and the global network learning error of a certain candidate sample is 0.08, then this sample is labeled as an effective sample. These effective samples are data that are considered to be more valuable in the neural network learning process after comparison with the annotation data and error analysis among numerous candidate samples.

[0022] Step S140, perform network parameter learning on the initialized neural network according to the extracted construction project cost data of effective samples, and output the trained disaster recovery backup model.

[0023] Taking the construction project cost data of the previously selected valid samples as an example, network parameter learning is performed on multiple initialized neural networks based on these valid sample construction project cost data. During the network parameter learning process, parameters such as the connection weights of the neural network are adjusted through the backpropagation algorithm. For example, for the construction project cost data of a certain building project in the valid samples, according to its input features and the corresponding disaster recovery backup allocation space data annotation data, the error between the current neural network output result and the annotation data is calculated, and then this error is propagated backward from the output layer to the hidden layer and the input layer. During the propagation process, the connection weights between neurons are adjusted according to a certain learning rate. Suppose for a certain neuron connection, its current weight is 0.5, and the amount to be adjusted is calculated as -0.1 according to the error and the learning rate, then the adjusted weight becomes 0.4.

[0024] After performing such network parameter learning on multiple initialized neural networks, multiple initialized neural networks are generated. Then, the performance of these neural networks is verified. The performance verification can be carried out in various ways. For example, new sample construction project cost data that has not participated in the training is input into these neural networks, and the error between their prediction results and the actual annotation data is compared. Suppose in the test of a new sample, the prediction error of one neural network is very small for most samples, while the prediction error of another neural network is large. Then, the neural network with a small prediction error is determined from these multiple initialized neural networks as the trained disaster recovery backup model.

[0025] Step S150, obtain the construction project cost data set to be processed for data disaster recovery, and perform disaster recovery backup processing on the construction project cost data set according to the trained disaster recovery backup model and based on the predicted disaster recovery backup allocation space data.

[0026] In this embodiment, assume that there is a new large-scale construction project, and its construction project cost data set contains detailed cost information. First, use the trained disaster recovery backup model to process this construction project cost data set.

[0027] Using the disaster recovery backup model, extract the characteristic knowledge distribution data of the construction project cost data set. The specific process is as follows:

[0028] Perform data cleaning operations on the construction project cost data set to remove any possible noise data (such as abnormally high or low cost data for a certain item due to data entry errors), incorrect data (such as data errors caused by incorrect material price units), and incomplete data (such as only partial cost records for a certain cost item and missing other related cost records). Then, perform standardization processing on the construction project cost data set after data cleaning operations. For example, unify data with different units into the same standard unit. After that, perform data encoding on the standardized construction project cost data set to generate the data encoding result of the construction project cost data set. Input the data encoding result into the initial feature extraction layer of the disaster recovery backup model. Assume there are 10 neurons in the initial feature extraction layer. Through these 10 neurons, perform a linear combination operation on the data encoding result to obtain an intermediate feature extraction result. For example, add the outputs of the 10 neurons according to a certain weight to obtain an intermediate result, and then perform an activation function processing on the intermediate feature extraction result (such as using the ReLU activation function) to generate an initial feature set.

[0029] Adopt a feature cross method to perform a combination operation on the features in the initial feature set to generate a combined feature sequence. For example, cross-combine the material cost feature and the labor cost feature to obtain a new combined feature. Then, screen the combined feature sequence based on the feature importance evaluation index, which is obtained by calculating the correlation between each feature and the disaster recovery backup target. Assume that after calculating the correlation, it is found that some combined features have a very low correlation with the disaster recovery backup target, and these features will be removed from the combined feature sequence to obtain a screened combined feature sequence.

[0030] Input the screened combined feature sequence into the deep feature mining layer in the disaster recovery backup model. The deep feature mining layer consists of 3 hidden layers. The screened combined feature sequence is input into the first layer of the deep feature mining layer, which has 8 neurons. Through these 8 neurons, perform a linear combination operation on the screened combined feature sequence, and after performing an activation function processing on the result of the linear combination operation (such as using the tanh activation function), pass the output result to the next hidden layer to repeat the above linear combination and activation function processing process. The output result of the last hidden layer is used as the deep feature set.

[0031] Perform semantic analysis on the depth features in the depth feature set, for example, analyze the semantic meaning of material cost features in different disaster recovery scenarios. After grouping the depth features with similar semantics according to the semantic analysis results, perform knowledge fusion operations on the grouped depth features to generate a feature knowledge subset. For example, fuse the depth features related to the price fluctuations of different building materials into a feature knowledge subset. Combine all the feature knowledge subsets to generate an integrated feature knowledge set. Determine the dimension of the feature knowledge distribution. Suppose it is determined to be 5 dimensions, corresponding to different cost components and disaster recovery related factors respectively. Statistically analyze the integrated feature knowledge set according to the determined dimensions, and based on the statistical results, construct the structure of the feature knowledge distribution data using a matrix data structure, where the rows represent different dimensions and the columns represent the specific attributes of the feature knowledge under each dimension.

[0032] Determine the feature focusing factor of the feature knowledge distribution data, and extract the focused knowledge distribution data of the feature knowledge distribution data according to the feature focusing factor. Specifically as follows:

[0033] Determine the feature focusing factor of the feature knowledge distribution data in the disaster recovery knowledge dimension. First, analyze the knowledge content related to the classification of the disaster recovery knowledge dimension in the feature knowledge distribution data, such as knowledge content related to backup frequency, backup storage location, data recovery strategy, etc. Construct a feature association network with each feature in the knowledge content classified under the disaster recovery knowledge dimension as a node. For example, for features such as the material cost backup frequency and equipment rental cost backup frequency in the backup frequency knowledge dimension, specifically analyze the direct association relationship (such as the material cost backup frequency and the equipment rental cost backup frequency may have a direct association due to the association of the project progress) and indirect association relationship (such as an indirect association through the overall project risk assessment) between each feature, and according to the direct and indirect association relationships between the features, connect the relevant feature nodes with connection edges to generate the feature association network, which is used to reflect the mutual relationship between the features in the knowledge content classified under the disaster recovery knowledge dimension.

[0034] For each node in the feature association network, based on the degree value of the node (for example, if a certain node is connected to 3 other nodes, the degree value is 3), betweenness centrality (indicating the frequency of a node appearing on the shortest paths between other nodes, for example, if a certain node often appears on the shortest paths between other nodes, the betweenness centrality is relatively high), and closeness centrality (reflecting the average distance from a node to other nodes, for example, if the average distance from a certain node to other nodes is relatively short, the closeness centrality is relatively high), evaluate the importance of the node in the feature association network to generate the first feature importance evaluation result. Obtain a pre-set disaster tolerance strategy, such as high-frequency backup and secure storage location strategies for important cost data. According to the disaster tolerance strategy, adjust the first feature importance evaluation result to generate an adjusted feature importance evaluation result. Determine a benchmark value, for example, the average value of all feature importances. For each node, based on the relationship between the adjusted importance evaluation result of the node and the benchmark value, construct a calculation rule, and set different dimension weights for the backup frequency knowledge dimension, backup storage location knowledge dimension, and data recovery strategy knowledge dimension respectively. For example, the weight of the backup frequency knowledge dimension is 0.4, the weight of the backup storage location knowledge dimension is 0.3, and the weight of the data recovery strategy knowledge dimension is 0.3. And according to the calculation rule and dimension weights, construct a calculation framework. Invoke the calculation framework to calculate the feature focusing factor of each node in the disaster tolerance knowledge dimension based on the adjusted feature importance evaluation result and the dimension weights of the disaster tolerance knowledge dimension, and summarize the feature focusing factors of each node in the disaster tolerance knowledge dimension to determine the feature focusing factor of the feature knowledge distribution data in the disaster tolerance knowledge dimension.

[0035] Determine the focused knowledge distribution data based on the feature knowledge distribution data and the feature focusing factor of the disaster tolerance knowledge dimension. First, analyze the data structure of the feature knowledge distribution data. Based on the data structure of the feature knowledge distribution data, search for feature elements associated with the disaster tolerance knowledge dimension. For example, in data with a matrix structure, search for elements in the rows and columns related to backup frequency, backup storage location, and data recovery strategy. By searching and matching in the analyzed data structure, extract all feature elements related to the disaster tolerance knowledge dimension to generate a set of feature elements. Use the feature focusing factor of the disaster tolerance knowledge dimension to assign weights to each feature element in the set of feature elements to generate a weight sequence of the set of feature elements. The feature focusing factor is an indicator used to measure the importance of each feature element in the disaster tolerance knowledge dimension. According to a pre-set weight threshold or weight ratio, screen the set of feature elements based on the weight sequence of the set of feature elements to generate an optimized subset of feature elements. Based on the screened subset of feature elements, reconstruct the data structure to generate the focused knowledge distribution data.

[0036] Perform disaster recovery backup prediction on the focused knowledge distribution data to generate the confidence levels of the construction project cost data sets corresponding to multiple reference disaster recovery backup allocation space data. For example, for a certain cost component, the confidence levels under different backup space allocation schemes may be obtained. For instance, the confidence level for allocating 10% backup space is 0.6, and the confidence level for allocating 15% backup space is 0.8, etc. Determine the predicted disaster recovery backup allocation space data of the construction project cost data set based on the confidence levels of the construction project cost data sets corresponding to multiple reference disaster recovery backup allocation space data. This predicted disaster recovery backup allocation space data comprehensively considers the backup space allocation schemes of different cost components under different confidence levels. For example, for the material cost part, determine a backup space allocation logical structure divided by material category and risk level according to its confidence levels under different backup space allocation schemes; for the labor cost part, also determine a backup space allocation logical structure based on the importance of job positions and the risk of personnel flow, etc. Finally, perform disaster recovery backup processing on the construction project cost data set according to the predicted disaster recovery backup allocation space data, and back up the data to the corresponding disaster recovery storage locations according to the determined logical structure.

[0037] Based on the above steps, the embodiment of the present application performs disaster recovery backup prediction on multiple candidate data in the sample construction project cost data sequence through an initialized neural network to generate training disaster recovery backup allocation space data. Determine the global network learning error by comparing the predicted data with the labeled data, so as to screen out effective sample data for network parameter learning and obtain the trained disaster recovery backup model. This method can accurately and efficiently perform disaster recovery backup processing on construction project cost data, improve the accuracy and efficiency of data disaster recovery, and ensure the security and reliability of construction project cost data. In practical applications, use the trained disaster recovery backup model to perform disaster recovery backup processing on the construction project cost data set to be processed, effectively avoiding the risk of data loss or damage.

[0038] In a possible implementation manner, the initialized neural network includes two or more.

[0039] Step S120 includes: loading each of the candidate sample construction project cost data into multiple initialized neural networks to generate the training disaster recovery backup allocation space data corresponding to each of the candidate sample construction project cost data for each initialized neural network.

[0040] For each of the candidate sample construction project cost data, step S130 includes:

[0041] Step S131: Determine the first network learning error corresponding to each of the initialized neural networks based on the training disaster recovery backup allocation space data corresponding to each of the initialized neural networks and the annotation data of the disaster recovery backup allocation space data.

[0042] Step S132: Determine the global network learning error of the candidate example construction project cost data based on the first network learning error corresponding to each of the initialized neural networks.

[0043] In a possible implementation manner, step S140 includes:

[0044] Step S141: Perform network parameter learning on each of the multiple initialized neural networks based on the valid example construction project cost data to generate multiple initialized neural networks.

[0045] Step S142: Determine the trained disaster recovery backup model from the multiple initialized neural networks by performing performance verification on the multiple initialized neural networks.

[0046] In a possible implementation manner, step S132 includes:

[0047] Step S1321: Determine the relative learning error between every two of the training disaster recovery backup allocation space data.

[0048] Step S1322: Determine the sum of each of the first network learning errors and the relative learning errors as the global network learning error.

[0049] In this embodiment, taking the cost data of a large construction project as an example, assume that there are three initialized neural networks, namely neural network A, neural network B, and neural network C. For the cost data of a specific construction project, it includes detailed cost information such as the cost of building structural materials, the cost of electromechanical equipment, and the salaries of construction workers. When this candidate sample construction project cost data is loaded into neural network A, the input layer of neural network A receives the eigenvalue of this cost data, and then calculates through its internal hidden layer. The neurons in the hidden layer perform a linear combination of the input data according to the preset weights, and after the non-linear transformation of the activation function, finally obtain the training disaster recovery backup allocation space data for this candidate sample construction project cost data at the output layer. This training disaster recovery backup allocation space data is a logical structure data related to the disaster recovery backup allocation space calculated based on the internal algorithm and weights of neural network A, and may include different allocation ratios and priority relationships of different cost parts in the disaster recovery backup, etc. Similarly, when this candidate sample construction project cost data is loaded into neural network B and neural network C, the corresponding training disaster recovery backup allocation space data will be obtained respectively. And due to the differences in the initial weights and structures of each neural network, the training disaster recovery backup allocation space data obtained by these three neural networks may be different in structure and value.

[0050] For each candidate construction project cost data sample, the global network learning error is determined based on the training disaster recovery backup allocation space data and the disaster recovery backup allocation space data annotation data. For the above-mentioned candidate construction project cost data sample, the first network learning error corresponding to each initialized neural network is determined based on its corresponding training disaster recovery backup allocation space data and disaster recovery backup allocation space data annotation data. For example, for neural network A, in the training disaster recovery backup allocation space data it outputs, the disaster recovery backup allocation ratio of the building structure material cost is 10%, while in the disaster recovery backup allocation space data annotation data, the disaster recovery backup allocation ratio of the building structure material cost is 8%. The error of this part is calculated through a specific error calculation function (such as the mean square error function), and then the errors of all cost components are comprehensively calculated to obtain the first network learning error corresponding to neural network A. Similarly, similar calculations are performed on neural network B and neural network C to obtain their respective first network learning errors. Then, based on the first network learning errors corresponding to each initialized neural network, the global network learning error of the candidate construction project cost data sample is determined. Specifically, the relative learning error between every two training disaster recovery backup allocation space data is determined. Suppose the disaster recovery backup allocation ratios of neural network A and neural network B for the building structure material cost are 10% and 12% respectively. By calculating the difference between the two and quantifying this difference according to certain rules, the relative learning error of this part is obtained. Then, the sum of the first network learning errors of neural network A, neural network B, and neural network C and the calculated relative learning error is determined as the global network learning error of this candidate construction project cost data sample.

[0051] Based on the extracted effective sample construction project cost data, network parameter learning is carried out on the initialized neural network to generate a trained disaster recovery backup model. In the previous operation, effective sample construction project cost data has been extracted according to the global network learning error. Now, based on these effective sample construction project cost data, network parameter learning is carried out on multiple initialized neural networks respectively to generate multiple initialized neural networks. Taking the cost data of a construction project in these effective samples as an example, for neural network A, input this effective sample construction project cost data into neural network A. According to the error between the output result and the corresponding disaster recovery backup allocation space data annotation data, adjust the parameters such as the neuron connection weights in neural network A through the backpropagation algorithm. For example, the initial weight of a certain neuron connection in neural network A is 0.3, and the calculated weight change amount to be adjusted according to the error is -0.05, then the adjusted weight becomes 0.25. The same operation is also carried out on neural network B and neural network C, so as to generate multiple initialized neural networks after network parameter learning. Then, through performance verification of multiple initialized neural networks, a trained disaster recovery backup model is determined from multiple initialized neural networks. During the performance verification process, a new set of construction project cost data that has not participated in the previous training process is selected as the verification set. Input these verification set data into neural network A, neural network B, and neural network C after network parameter learning respectively, and calculate the error between their respective prediction results and the actual disaster recovery backup allocation space data annotation data. Suppose the average error of neural network A on the verification set data is 0.08, the average error of neural network B is 0.12, and the average error of neural network C is 0.1. Then, since the average error of neural network A is the smallest, neural network A is determined as the trained disaster recovery backup model.

[0052] When determining the global network learning error of the candidate sample construction project cost data, continue to take the previous neural networks A, B, and C as examples. For the disaster recovery backup allocation space data of the building electromechanical equipment cost, the output of neural network A is 15%, and the output of neural network B is 13%. Calculate the difference between the two as 2%. Then, according to certain quantization rules (such as factors like the proportion of this part of the cost in the entire project cost data), convert this difference into a relative learning error value. Perform the same operation for other cost components to obtain the relative learning error between every two training disaster recovery backup allocation space data. Then, determine the global network learning error as the sum of each first network learning error and these relative learning errors. This global network learning error comprehensively considers the errors of each initialized neural network itself and the relative errors caused by the output differences between them, and can more comprehensively reflect the learning effect of the candidate sample construction project cost data in multiple initialized neural networks. This method helps to accurately extract the effective sample construction project cost data in subsequent steps, thereby improving the accuracy and reliability of the trained disaster recovery backup model.

[0053] In a possible implementation manner, step S140 may further include: performing iterative network parameter learning on the initialized neural network according to the extracted effective sample construction project cost data to generate a first temporary neural network. If the first temporary neural network meets the network convergence requirement, generate the trained disaster recovery backup model according to the first temporary neural network.

[0054] If the first temporary neural network does not meet the network convergence requirement, use the first temporary neural network after network parameter learning as the iterative neural network, and iteratively execute the following steps until the generated second temporary neural network meets the network convergence requirement, and generate the trained disaster recovery backup model according to the second temporary neural network that meets the network convergence requirement:

[0055] Step A110, obtain the iterative sample construction project cost data sequence.

[0056] Step A120, perform disaster recovery backup prediction on each candidate sample construction project cost data in the iterative sample construction project cost data sequence according to the initialized neural network to generate the training disaster recovery backup allocation space data of each candidate sample construction project cost data.

[0057] Step A130, for each candidate sample construction project cost data, determine the global network learning error according to the training disaster recovery backup allocation space data and the disaster recovery backup allocation space data annotation data.

[0058] Step A140, extract the effective sample construction project cost data from the sample construction project cost data sequence according to the global network learning error of each of the candidate sample construction project cost data.

[0059] Step A150, perform iterative network parameter learning on the initialized neural network according to the extracted effective sample construction project cost data to generate a second temporary neural network. If the second temporary neural network does not meet the network convergence requirement, use the second temporary neural network as the iterative neural network.

[0060] In this embodiment, taking the construction project cost data of the previously mentioned construction project as an example, it is assumed that effective sample data has been extracted from a large number of candidate sample construction project cost data. For parameters such as the neuron connection weights in the initialized neural network, these effective sample construction project cost data are used for iterative adjustment. For example, a certain construction project cost data in the effective sample includes detailed infrastructure construction costs, decoration costs, equipment purchase costs, etc. This effective sample construction project cost data is input into the initialized neural network. The neural network processes the data according to the structure and algorithm of its input layer, hidden layer, and output layer. During the processing, the backpropagation algorithm is used to adjust the neuron connection weights according to the error between the output result and the actual data annotation data of the disaster recovery backup allocation space. For the connection weight between a certain neuron in the hidden layer and the neuron in the output layer, the initial weight may be 0.4, and the amount by which the weight needs to be adjusted calculated according to the error is -0.03. Then, after one iteration, this weight becomes 0.37. After multiple such iterations and processing multiple effective sample construction project cost data, a first temporary neural network is generated. At this time, it is necessary to determine whether the first temporary neural network meets the network convergence requirement. The network convergence requirement can be defined in various ways. For example, the network output error after several consecutive iterations is less than a certain set threshold, or the change amount of the network parameters is within a certain range, etc. If the first temporary neural network meets the network convergence requirement, generate a trained disaster recovery backup model based on the first temporary neural network. This trained disaster recovery backup model will be used for subsequent disaster recovery backup prediction and other operations on the construction project cost data set.

[0061] If the first temporary neural network does not meet the network convergence requirement, then use the first temporary neural network after network parameter learning as the iterative neural network and iteratively execute the subsequent steps. First, obtain the iterative sample construction project cost data sequence. This iterative sample construction project cost data sequence can be the part of the candidate sample construction project cost data that has not been fully utilized before, or the newly collected data related to the construction project cost. For example, some newly obtained construction project cost data of different building types (such as residential buildings, commercial buildings, etc.) or different regions constitute the iterative sample construction project cost data sequence.

[0062] Next, according to the initialized neural network, disaster recovery backup prediction is performed on each candidate sample construction project cost data in the iterative sample construction project cost data sequence to generate training disaster recovery backup allocation space data for each candidate sample construction project cost data. Taking the cost data of a commercial building project in the iterative sample construction project cost data sequence as an example, this data includes cost information such as land acquisition cost, commercial facility construction cost, and marketing cost. Input it into the initialized neural network, and the initialized neural network calculates according to its own structure and algorithm. After receiving the eigenvalue of these cost data in the input layer, through the neurons in the hidden layer for linear combination and activation function processing, finally, training disaster recovery backup allocation space data for this candidate sample construction project cost data is generated in the output layer, which includes the allocation logic of different cost parts in the disaster recovery backup. For example, the land acquisition cost determines the corresponding disaster recovery backup allocation ratio and priority according to its importance and risk factors (such as the risk of land policy changes, etc.) in the project.

[0063] For each candidate sample construction project cost data, the global network learning error is determined based on the training disaster recovery backup allocation space data and the annotation data of the disaster recovery backup allocation space data. For example, for the commercial facility construction cost part in the above commercial building project cost data, the allocation ratio in its training disaster recovery backup allocation space data is 12%, while the allocation ratio in the annotation data of the disaster recovery backup allocation space data is 10%. Calculate the error of this part through a specific error calculation function (such as the mean square error function), and then comprehensively calculate the error of all cost components to obtain the error of this candidate sample construction project cost data corresponding to the initialized neural network. Then, according to the method mentioned before, determine the relative learning error between every two training disaster recovery backup allocation space data, and determine the sum of each first network learning error and the relative learning error as the global network learning error.

[0064] Extract the effective sample construction project cost data from the sample construction project cost data sequence according to the global network learning error of each candidate sample construction project cost data. For example, the method of labeling the sample construction project cost data with the smallest set proportion of the global network learning error as the effective sample construction project cost data can be adopted. Suppose there are 100 candidate samples in the iterative sample construction project cost data sequence, and the set proportion is 20%. Then, the 20 candidate samples with the smallest global network learning error are labeled as the effective sample construction project cost data. Or the method of labeling the sample construction project cost data with the global network learning error less than the set error as the effective sample construction project cost data can be adopted. For example, if the set error is 0.1 and the global network learning error of a certain candidate sample is 0.08, then this sample is labeled as the effective sample construction project cost data.

[0065] Iteratively learn the network parameters of the initialized neural network based on the extracted valid sample construction project cost data to generate a second temporary neural network. Taking the cost data of a residential building project in these newly extracted valid sample construction project cost data as an example, input it into the initialized neural network. According to the network parameter learning method mentioned before, based on the error between the output result and the actual data annotation data of the disaster recovery backup allocation space, adjust the parameters such as the neuron connection weights of the neural network through the backpropagation algorithm. After multiple iterations, after processing multiple valid sample construction project cost data, generate a second temporary neural network. If the second temporary neural network does not meet the network convergence requirements, use the second temporary neural network as the iterative neural network and continue to iterate and execute the above steps until the generated temporary neural network meets the network convergence requirements. Generate a trained disaster recovery backup model based on the temporary neural network that meets the network convergence requirements. This process continuously optimizes the parameters of the neural network, enabling it to more accurately predict the disaster recovery backup allocation space data, thereby improving the accuracy and reliability of the entire disaster recovery backup model. Through this iterative method, different sample construction project cost data can be fully utilized, and the parameters of the neural network can be continuously adjusted to adapt to various complex construction project cost situations, ensuring the accuracy and effectiveness in disaster recovery backup prediction.

[0066] In a possible implementation manner, step A140 includes at least one of the following:

[0067] Step A141, label the sample construction project cost data with a set proportion of the minimum global network learning error as the valid sample construction project cost data.

[0068] In this embodiment, assume that there is a large sample construction project cost data sequence, which contains the cost data of numerous construction projects covering different types of buildings (such as residential buildings, commercial buildings, industrial buildings, etc.) and different construction scales. For each candidate sample construction project cost data, its global network learning error has been calculated. For example, there are 100 candidate sample construction project cost data, and the set proportion is 20%, that is, 20 samples are to be selected as valid samples. First, sort these 100 candidate samples according to the global network learning error. The global network learning error reflects the deviation degree of each candidate sample in the neural network learning process. This error is obtained by comprehensively considering factors such as the difference between the prediction result of the neural network for this sample and the actual disaster recovery backup allocation space data annotation data, and the relative difference in prediction results between different neural networks. After sorting, select the top 20 candidate sample construction project cost data with the smallest global network learning error and label them as valid sample construction project cost data. These data labeled as valid samples are of great significance in subsequent operations such as neural network parameter learning because they showed a smaller deviation from the actual annotation data in the previous learning process, can provide more accurate learning samples for the neural network, and help improve the accuracy of the neural network.

[0069] Step A142, label the sample construction project cost data with a global network learning error less than the set error as the valid sample construction project cost data.

[0070] Also in this sequence containing numerous candidate sample construction project cost data, preset an error value, for example, set the error to 0.1. Then check the global network learning error of each candidate sample construction project cost data one by one. Taking the cost data of a commercial building project as an example, the cost data of this project includes multiple parts such as building structure cost, decoration cost, and equipment installation cost. After calculating the error between the prediction result of the neural network for this sample and the actual disaster recovery backup allocation space data annotation data, the global network learning error is obtained. If the global network learning error of the cost data of this commercial building project is 0.08, since 0.08 is less than the set error of 0.1, then the cost data of this commercial building project is labeled as valid sample construction project cost data. For other candidate samples, they are judged in the same way. As long as their global network learning error is less than the set error, they are labeled as valid sample construction project cost data. The valid sample construction project cost data selected in this way meet specific requirements in terms of global network learning error, can provide a reliable data basis for the further learning of the neural network, help improve the accuracy and reliability of the disaster recovery backup model, and enable the finally trained disaster recovery backup model to more accurately perform disaster recovery backup prediction and processing on the construction project cost data set.

[0071] In a possible implementation, step S150 includes:

[0072] Step S151, using the disaster recovery and backup model, extract the characteristic knowledge distribution data of the construction project cost data set, determine the characteristic focusing factor of the characteristic knowledge distribution data, and extract the focused knowledge distribution data of the characteristic knowledge distribution data according to the characteristic focusing factor.

[0073] Step S152, perform disaster recovery and backup prediction on the focused knowledge distribution data, and generate the confidence levels of the construction project cost data set corresponding to multiple reference disaster recovery and backup allocation space data respectively.

[0074] Step S153, determine the predicted disaster recovery and backup allocation space data of the construction project cost data set according to the confidence levels of the construction project cost data set corresponding to multiple reference disaster recovery and backup allocation space data respectively.

[0075] In this embodiment, taking a construction project cost data set containing the cost information of multiple construction projects as an example, this construction project cost data set includes detailed cost data such as various building material costs, labor costs, equipment rental costs, etc. The disaster recovery backup model first processes this data set and extracts characteristic knowledge distribution data through operations. For example, for the building material cost, the model will analyze the proportion of the costs of different materials (such as steel, cement, etc.) in the total project cost, the fluctuation situation, and the correlation with other cost factors (such as construction progress, market supply situation, etc.), so as to construct the part of the characteristic knowledge distribution data of the building material cost in this construction project cost data set. For the labor cost, factors such as the salary levels of different types of work (such as bricklayers, electricians, etc.), working hours, and seasonal changes in labor demand will be considered to construct the corresponding part of the characteristic knowledge distribution data. After constructing the characteristic knowledge distribution data of the entire construction project cost data set, the characteristic focusing factors of the characteristic knowledge distribution data are then determined. Taking the building material cost as an example, analyze its key factors in the disaster recovery backup scenario. For example, the supply stability of some rare materials is more important for disaster recovery backup, so these factors will be given higher weights in the calculation of the characteristic focusing factors. Calculate the characteristic focusing factor of the building material cost part according to these weight relationships. Similar calculations are also carried out for other parts of the entire construction project cost data set (such as labor cost, equipment rental cost, etc.) to obtain their respective characteristic focusing factors. Then, based on these characteristic focusing factors, the focused knowledge distribution data of the characteristic knowledge distribution data is extracted. For example, in the building material cost part, select the knowledge data related to the factors that have a greater impact on disaster recovery backup (such as the supply stability of rare materials, the price fluctuation trend of key materials, etc.) according to the characteristic focusing factors to form the focused knowledge distribution data of the building material cost part. The same operation is also carried out for other parts of the entire construction project cost data set, and finally the complete focused knowledge distribution data is obtained.

[0076] Next, after obtaining the focused knowledge distribution data, predictions are made for different reference disaster recovery backup allocation space data. For example, for the building material cost part, assume that there are three cases for the reference disaster recovery backup allocation space data: 10%, 15%, and 20% of the total project cost as the disaster recovery backup space. The disaster recovery backup model analyzes and calculates based on the building material cost-related factors (such as material price fluctuations, supply stability, etc.) in the focused knowledge distribution data, and obtains a confidence level of 0.6 in the case of a 10% disaster recovery backup allocation space, a confidence level of 0.8 in the case of a 15% disaster recovery backup allocation space, and a confidence level of 0.9 in the case of a 20% disaster recovery backup allocation space. Similar predictions are also made for other parts of the construction project cost data set (such as labor cost, equipment rental cost, etc.) for these three reference disaster recovery backup allocation space data to obtain their respective confidence levels.

[0077] Finally, taking the entire construction project cost data set as a whole, comprehensively consider the confidence levels of various parts such as building material costs, labor costs, and equipment rental costs under different reference disaster recovery backup allocation space data. For example, the confidence level of building material costs is relatively high under a 15% disaster recovery backup allocation space, the confidence level of labor costs is relatively high under a 10% disaster recovery backup allocation space, and the confidence level of equipment rental costs is relatively high under a 20% disaster recovery backup allocation space. However, due to the need to consider the overall disaster recovery backup effect of the entire construction project cost data set, it is necessary to weigh the confidence levels of different reference disaster recovery backup allocation space data based on factors such as the proportion of each part in the total project cost and their mutual relationships. After calculation and weighing, determine the predicted disaster recovery backup allocation space data for the entire construction project cost data set. This predicted disaster recovery backup allocation space data is not a simple value, but a logical structure data that comprehensively considers the characteristics of each cost part, their mutual relationships, and the confidence levels under different disaster recovery backup allocation spaces, and can provide an accurate basis for the disaster recovery backup processing of the construction project cost data set.

[0078] In a possible implementation manner, step S151 includes:

[0079] Step S1511, perform a data cleaning operation on the construction project cost data set to remove the noise data, error data, and incomplete data existing in the construction project cost data set.

[0080] Step S1512, after performing a standardization process on the construction project cost data set that has undergone the data cleaning operation, perform data encoding on the standardized construction project cost data set to generate the data encoding result of the construction project cost data set.

[0081] Step S1513, input the data encoding result into the initial feature extraction layer of the disaster recovery backup model, perform a linear combination operation on the data encoding result through multiple neurons in the initial feature extraction layer to obtain an intermediate feature extraction result, and perform an activation function process on the intermediate feature extraction result to generate an initial feature set.

[0082] Step S1514, adopt a feature cross method to perform a combination operation on the features in the initial feature set to generate a combined feature sequence, and screen the combined feature sequence based on an importance evaluation index of the features, where the importance evaluation index is obtained by calculating the correlation between each feature and the disaster recovery backup target.

[0083] Step S1515: Input the filtered combined feature sequence into the deep feature mining layer in the disaster recovery backup model. The deep feature mining layer consists of multiple hidden layers. The filtered combined feature sequence is input into the first layer of the deep feature mining layer, and the neurons in the first layer perform a linear combination operation on the filtered combined feature sequence. After processing the result of the linear combination operation with an activation function, the output result is passed to the next hidden layer to repeat the above linear combination and activation function processing process. The output result of the last hidden layer is used as the deep feature set.

[0084] Step S1516: Perform semantic analysis on the deep features in the deep feature set, group the deep features with similar semantics according to the semantic analysis results, and then perform a knowledge fusion operation on the grouped deep features to generate a feature knowledge subset. Combine all the feature knowledge subsets to generate an integrated feature knowledge set.

[0085] Step S1517: Determine the dimension of the feature knowledge distribution, statistically analyze the integrated feature knowledge set according to the determined dimension, and construct the structure of the feature knowledge distribution data using a matrix data structure or a vector data structure according to the statistical results. Among them, if a matrix structure is adopted, the rows represent different dimensions, and the columns can represent the specific attributes of the feature knowledge under each dimension.

[0086] Step S1518: Determine the feature focusing factor of the feature knowledge distribution data in the disaster recovery knowledge dimension.

[0087] Step S1519: Determine the focused knowledge distribution data based on the feature knowledge distribution data and the feature focusing factor of the disaster recovery knowledge dimension.

[0088] In a possible implementation manner, step S1518 includes:

[0089] Step S1518-1: Analyze the knowledge content related to the classification of the disaster recovery knowledge dimension in the feature knowledge distribution data.

[0090] Step S1518-2: Construct a feature association network with each feature in the knowledge content classified by the disaster recovery knowledge dimension as a node. Specifically, analyze the direct association relationship and indirect association relationship between each feature, and connect the relevant feature nodes with connection edges according to the direct and indirect association relationships between the features to generate the feature association network. The feature association network is used to reflect the mutual relationship between the features in the knowledge content classified by the disaster recovery knowledge dimension.

[0091] Step S1518-3: For each node in the feature association network, evaluate the importance of the node in the feature association network based on the degree value, betweenness centrality, and closeness centrality of the node, and generate a first feature importance evaluation result. The degree value is the number of edges connected to the node. The betweenness centrality represents the frequency of a node appearing on the shortest path between other nodes. The closeness centrality reflects the average distance from a node to other nodes.

[0092] Step S1518-4: Obtain a pre-set disaster recovery policy, and adjust the first feature importance evaluation result according to the disaster recovery policy to generate an adjusted feature importance evaluation result.

[0093] Step S1518-5: Determine a benchmark value, where the benchmark value is the average of all feature importances or a basic value set according to a preset rule.

[0094] Step S1518-6: For each node, construct a calculation rule based on the relationship between the adjusted importance evaluation result of the node and the benchmark value, and set different dimension weights for the backup frequency knowledge dimension, backup storage location knowledge dimension, and data recovery policy knowledge dimension respectively. Then, construct a calculation framework according to the calculation rule and dimension weights.

[0095] Step S1518-7: Invoke the calculation framework to calculate the feature focusing factor of each node in the disaster recovery knowledge dimension based on the adjusted feature importance evaluation result and the dimension weights of the disaster recovery knowledge dimension, and summarize the feature focusing factors of each node in the disaster recovery knowledge dimension to determine the feature focusing factor of the feature knowledge distribution data in the disaster recovery knowledge dimension.

[0096] In this embodiment, taking a large-scale construction project cost data set as an example, this data set contains the cost information of numerous construction projects, such as the procurement costs of various building materials, the labor costs of different types of work, the rental and purchase costs of equipment, and various management costs, etc. In this data set, noise data may be manifested as abnormal fluctuations of individual data points due to accidental errors during data entry. For example, the price of a certain building material is mis-entered as a value that significantly deviates from the normal market price range; incorrect data may be due to misclassification or miscalculation of certain costs. For instance, the cost that should belong to the equipment purchase cost is wrongly included in the building material procurement cost; incomplete data may be that some projects only record part of the cost information and lack the cost data of key parts. For example, a certain construction project only records the cost of the infrastructure part and does not record the cost of the decoration part. Through data cleaning operations, specific algorithms and rules are used to identify and correct these data problems. For example, for data points that significantly deviate from the normal price range, by comparing with the market average price or the price data of similar projects, they are corrected to reasonable values; for misclassified data, they are re-classified according to the nature and use of the costs; for incomplete data, if possible, they are supplemented, otherwise the data of this project is specially marked or appropriately processed in subsequent analysis.

[0097] After standardizing the construction project cost data set that has undergone data cleaning operations, data encoding is performed on the standardized construction project cost data set to generate the data encoding result of the construction project cost data set. The purpose of standardization is to unify data of different types and units to a standard scale for subsequent calculation and analysis. For example, for the procurement cost of building materials, it may be in units of per ton or per cubic meter, while the labor cost may be in units of per person per day. Through standardization, these data with different units are converted into a unified numerical representation form. When performing data encoding, according to the pre-set encoding rules, the standardized construction project cost data set is converted into a coding form that can be recognized and processed by a computer. For example, encoding methods such as One-Hot Encoding can be used to convert different categorical variables (such as the types of building materials, types of work, etc.) into binary vector forms, and corresponding encoding conversions are also performed on numerical variables (such as cost amounts, etc.), and finally the data encoding result of the construction project cost data set is generated.

[0098] Input the data encoding result into the initial feature extraction layer of the disaster recovery and backup model. Through the linear combination operation of multiple neurons in the initial feature extraction layer on the data encoding result, obtain the intermediate feature extraction result, and perform an activation function process on the intermediate feature extraction result to generate the initial feature set. Assume that there are several neurons in the initial feature extraction layer of the disaster recovery and backup model, and each neuron has different weights for different parts of the input data encoding result. Taking the building material cost data as an example, in the data encoding result, the cost data of different building materials correspond to different encoding values respectively, and these encoding values are input into the neurons of the initial feature extraction layer. The neurons perform a linear combination operation on these input values according to their weights. For example, for the encoding value corresponding to the cost of a certain building material, the weight of neuron A is 0.3, and the weight of neuron B is 0.2, then the linear combination result is the value obtained by multiplying the encoding value by the corresponding weight and then adding them. After performing such a linear combination operation on all the encoding values, obtain the intermediate feature extraction result. Then, perform an activation function process on the intermediate feature extraction result. The activation function can adopt functions such as the ReLU (Rectified Linear Unit) function. Through the non-linear transformation of the activation function, convert the intermediate feature extraction result into the initial feature set. This initial feature set contains the feature information of the construction project cost data set after preliminary processing. These feature information have been extracted and transformed to a certain extent, providing a basis for subsequent operations.

[0099] Adopt the feature crossing method to perform a combination operation on the features in the initial feature set to generate a combined feature sequence, and screen the combined feature sequence based on the importance evaluation index of the features, where the importance evaluation index is obtained by calculating the correlation between each feature and the disaster recovery and backup target. In the initial feature set, it contains various features related to building cost, such as features of different building material costs, different types of labor costs, etc. Adopt the feature crossing method to combine these features. For example, cross-combine the building material cost feature with the labor cost feature to generate new combined features, such as "the associated feature of the cost of a certain building material and the labor cost of a specific type of work", etc., thus obtaining the combined feature sequence. Then, calculate the correlation between each feature and the disaster recovery and backup target to obtain the importance evaluation index of the features. Taking the disaster recovery and backup target as ensuring the recoverability and integrity of the construction project cost data in the event of a disaster as an example, the features with higher correlation with this target may be those features related to data that have a greater impact on the project cost and are likely to be lost or damaged in the event of a disaster, such as the impact of the supply stability and cost fluctuations of key building materials on the project cost and the recoverability of these data in the event of a disaster. According to these correlation calculation results, screen the combined feature sequence, remove those combined features with lower correlation with the disaster recovery and backup target, and obtain the screened combined feature sequence.

[0100] The filtered combined feature sequence is input into the deep feature mining layer in the disaster tolerance backup model. The deep feature mining layer consists of multiple hidden layers. The filtered combined feature sequence is input into the first layer of the deep feature mining layer, and the neurons in the first layer perform a linear combination operation on the filtered combined feature sequence. After processing the result of the linear combination operation with an activation function, the output result is passed to the next hidden layer to repeat the above linear combination and activation function processing process. The output result of the last hidden layer is used as the deep feature set. Suppose the deep feature mining layer has three hidden layers. The filtered combined feature sequence is input into the first hidden layer, and there are several neurons in this layer. For example, for a certain combined feature in the combined feature sequence, the neurons in the first hidden layer perform a linear combination operation on this combined feature according to its weight, and then process it through an activation function (such as the tanh function) to obtain the output result and pass it to the second hidden layer. The neurons in the second hidden layer also perform a linear combination and activation function processing on the input result, and then pass the result to the third hidden layer. After the same operation in the third hidden layer, its output result is used as the deep feature set. This deep feature set contains the feature information of the construction project cost data set after deep mining and transformation. These feature information are more abstract and advanced, and can better reflect the internal structure of the data and the information related to disaster tolerance backup.

[0101] Semantic analysis is performed on the deep features in the deep feature set, and according to the results of the semantic analysis, the deep features with similar semantics are grouped. After that, a knowledge fusion operation is performed on the grouped deep features to generate feature knowledge subsets, and all the feature knowledge subsets are combined to generate an integrated feature knowledge set. In the deep feature set, different deep features have different semantic meanings. For example, some deep features may be semantically related to the cost fluctuations of building materials at different time periods, and some other deep features may be semantically related to the labor costs of different types of work at different construction stages. Through semantic analysis, the deep features with similar semantics are identified. For example, the deep features related to the cost fluctuations of building materials are grouped together. Then a knowledge fusion operation is performed on the grouped deep features to integrate the deep features in the same group. For example, through weighted average or logical operations, etc., they are fused into a feature knowledge subset. After performing such operations on all groups, all the obtained feature knowledge subsets are combined to generate an integrated feature knowledge set. This integrated feature knowledge set synthesizes various feature information in the deep feature set, and through semantic analysis and knowledge fusion, it is more refined and meaningful.

[0102] Determine the dimensions of the characteristic knowledge distribution, count the integrated characteristic knowledge set according to the determined dimensions, and based on the statistical results, construct the structure of the characteristic knowledge distribution data using a matrix data structure or a vector data structure. Among them, if a matrix structure is adopted, the rows represent different dimensions, and the columns can represent the specific attributes of the characteristic knowledge under each dimension. For example, the determined dimensions can include the dimension of construction material costs, the dimension of labor costs, the dimension of equipment costs, etc. For the integrated characteristic knowledge set, count according to these dimensions, and count information such as the quantity and proportion of the characteristic knowledge under each dimension. If a matrix data structure is used to construct the structure of the characteristic knowledge distribution data, with the dimensions of construction material costs, labor costs, and equipment costs as rows, for each row, the columns can represent different attributes, such as the fluctuation range of costs, the composition ratio of expenses, etc. Through such statistics and structure construction, generate the characteristic knowledge distribution data, which can clearly show the distribution of the characteristic knowledge of the construction project cost data set under different dimensions.

[0103] When determining the characteristic focus factor of the characteristic knowledge distribution data and extracting the focused knowledge distribution data of the characteristic knowledge distribution data, first determine the characteristic focus factor of the characteristic knowledge distribution data in the disaster tolerance knowledge dimension. Analyze the knowledge content related to the classification of the disaster tolerance knowledge dimension in the characteristic knowledge distribution data. For example, in the construction project cost data, the content related to the disaster tolerance knowledge dimension may include knowledge content such as the frequency of data backup, the location of backup storage, and the data recovery strategy. Taking each characteristic in the knowledge content classified under the disaster tolerance knowledge dimension as a node, construct a characteristic association network. For example, for characteristics such as the backup frequency of construction material costs and the backup frequency of labor costs under the backup frequency knowledge dimension, analyze their direct association relationships (such as the backup frequency of construction material costs and the backup frequency of labor costs may have a direct association due to the overall project budget control) and indirect association relationships (such as having an indirect association through the project's risk assessment and management strategy), and according to these relationships, connect the relevant characteristic nodes with connecting edges to generate a characteristic association network. This characteristic association network can reflect the mutual relationships among the characteristics in the knowledge content classified under the disaster tolerance knowledge dimension.

[0104] For each node in the feature association network, based on the degree value, betweenness centrality, and closeness centrality of the node, evaluate the importance of the node in the feature association network to generate the first feature importance evaluation result. For example, for a node related to the backup frequency of building material costs, its degree value represents the number of edges connected to this node. If there are 3 edges connected to it, then the degree value is 3; the betweenness centrality represents the frequency of this node appearing on the shortest path between other nodes. If this node appears on the shortest paths between many other nodes, then the betweenness centrality is relatively high; the closeness centrality reflects the average distance from this node to other nodes. If the average distance from this node to other nodes is relatively short, then the closeness centrality is relatively high. Based on these metrics, evaluate the importance of the node in the feature association network to obtain the first feature importance evaluation result. Obtain a pre-set disaster recovery strategy, such as adopting a high-frequency backup and a secure storage location for important cost data. According to this disaster recovery strategy, adjust the first feature importance evaluation result to generate an adjusted feature importance evaluation result.

[0105] Determine a benchmark value, which can be the average value of all feature importances or a base value set according to preset rules. For example, if the average value of all feature importances is 0.5, then this 0.5 is used as the benchmark value. For each node, based on the relationship between the adjusted importance evaluation result of the node and the benchmark value, construct a calculation rule, and set different dimension weights for the backup frequency knowledge dimension, backup storage location knowledge dimension, and data recovery strategy knowledge dimension respectively. For example, the weight of the backup frequency knowledge dimension is 0.4, the weight of the backup storage location knowledge dimension is 0.3, and the weight of the data recovery strategy knowledge dimension is 0.3. And according to the calculation rule and dimension weights, construct a calculation framework. Invoke this calculation framework based on the adjusted feature importance evaluation result and the dimension weights of the disaster recovery knowledge dimension to calculate the feature focusing factor of each node under the disaster recovery knowledge dimension, and summarize the feature focusing factors of each node under the disaster recovery knowledge dimension to determine the feature focusing factor of the feature knowledge distribution data in the disaster recovery knowledge dimension.

[0106] Determine the focused knowledge distribution data based on the feature knowledge distribution data and the feature focusing factor of the disaster recovery knowledge dimension. First, analyze the data structure of the feature knowledge distribution data. Based on the data structure of the feature knowledge distribution data, search for feature elements associated with the disaster recovery knowledge dimension. For example, in data with a matrix structure, search for elements in the rows and columns related to backup frequency, backup storage location, and data recovery strategy. By searching and matching in the parsed data structure, extract all feature elements related to the disaster recovery knowledge dimension to generate a set of feature elements. Use the feature focusing factor of the disaster recovery knowledge dimension to assign weights to each feature element in the set of feature elements to generate a weight sequence of the set of feature elements. The feature focusing factor is an indicator used to measure the importance of each feature element under the disaster recovery knowledge dimension. According to a preset weight threshold or weight ratio, and based on the weight sequence of the set of feature elements, screen the set of feature elements to generate an optimized subset of feature elements. Rebuild the data structure based on the screened subset of feature elements to generate the focused knowledge distribution data. This focused knowledge distribution data is more focused on content related to the disaster recovery knowledge dimension and can provide a more targeted data basis for subsequent operations such as disaster recovery backup prediction.

[0107] In one possible implementation manner, step S1519 includes:

[0108] Step S1519-1, analyze the data structure of the feature knowledge distribution data. Based on the data structure of the feature knowledge distribution data, search for feature elements associated with the disaster recovery knowledge dimension. By searching and matching in the parsed data structure, extract all feature elements related to the disaster recovery knowledge dimension to generate a set of feature elements. The disaster recovery knowledge dimension includes a backup frequency knowledge dimension, a backup storage location knowledge dimension, and a data recovery strategy knowledge dimension.

[0109] Step S1519-2, use the feature focusing factor of the disaster recovery knowledge dimension to assign weights to each feature element in the set of feature elements to generate a weight sequence of the set of feature elements. The feature focusing factor is an indicator used to measure the importance of each feature element under the disaster recovery knowledge dimension.

[0110] Step S1519-3, according to a preset weight threshold or weight ratio, and based on the weight sequence of the set of feature elements, screen the set of feature elements to generate an optimized subset of feature elements.

[0111] Step S1519-4, rebuild the data structure based on the screened subset of feature elements to generate the focused knowledge distribution data.

[0112] In this embodiment, taking the characteristic knowledge distribution data of a construction project cost dataset as an example, assuming that the data structure is represented in matrix form, the rows represent different cost components, such as building materials, labor costs, equipment leasing, etc., and the columns represent various attributes under each cost component, such as cost fluctuation range, seasonal impact, etc. For the disaster tolerance knowledge dimension, the backup frequency knowledge dimension may be related to the update frequency of the cost data at different time periods, the backup storage location knowledge dimension may be related to the server location or storage medium where the cost data is stored, and the data recovery strategy knowledge dimension may be related to the recovery process and cost in case of data loss or damage.

[0113] When parsing the characteristic knowledge distribution data of this matrix structure, from the perspective of rows, for the row of building materials, it may be found that some attributes are related to the backup frequency knowledge dimension. For example, the supply stability of a certain special building material is low, resulting in the need for high-frequency backup of its cost data. Then this attribute related to the backup frequency is a characteristic element associated with the disaster tolerance knowledge dimension. From the perspective of columns, in the column of cost fluctuation range, it may be found that the cost fluctuation of labor costs is related to the data recovery strategy knowledge dimension, because the fluctuation of labor costs may affect the budget adjustment during data recovery. This is also a characteristic element related to the disaster tolerance knowledge dimension. By carefully searching and matching the entire matrix structure, all characteristic elements related to the backup frequency knowledge dimension, backup storage location knowledge dimension, and data recovery strategy knowledge dimension are extracted to form a set of characteristic elements.

[0114] Next, use the characteristic focusing factor of the disaster tolerance knowledge dimension to assign weights to each characteristic element in the set of characteristic elements, generating a weight sequence of the set of characteristic elements. The characteristic focusing factor is an index used to measure the importance of each characteristic element under the disaster tolerance knowledge dimension. Continuing with the above construction project cost dataset as an example, assume that the characteristic focusing factors of each characteristic element under the disaster tolerance knowledge dimension have been obtained in the previous calculation. For a characteristic element related to the backup frequency of building materials in the set of characteristic elements, its characteristic focusing factor may be relatively high, indicating that this characteristic element is more important under the disaster tolerance knowledge dimension. According to this characteristic focusing factor, assign a relatively high weight to this characteristic element, such as 0.8. For another characteristic element related to the data recovery strategy of labor costs, if its characteristic focusing factor is relatively low, a weight of 0.3 may be assigned to it. In this way, weights are assigned to each characteristic element in the set of characteristic elements according to its corresponding characteristic focusing factor, thereby generating a weight sequence of the set of characteristic elements.

[0115] Then, according to the pre-set weight threshold or weight ratio, and based on the weight sequence of the feature element set, the feature element set is screened to generate an optimized subset of feature elements. The pre-set weight threshold or weight ratio is determined according to the actual disaster recovery requirements and experience. For example, if the weight threshold is set to 0.5, for each feature element in the feature element set, if its weight is greater than or equal to 0.5, it is retained in the optimized subset of feature elements. Feature elements with weights less than 0.5 are discarded. Or in the way of weight ratio, assuming that the top 60% of the feature elements with higher weights are selected, then after sorting the feature element set by weight, the top 60% of the feature elements are selected to form an optimized subset of feature elements. Taking the feature elements related to the backup frequency of building materials as an example, if its weight is 0.8, which is greater than the weight threshold of 0.5, then it will be retained in the optimized subset of feature elements; while if the weight of a feature element related to the storage location of equipment rental is 0.4, which is less than the weight threshold of 0.5, it will be discarded.

[0116] Finally, based on the screened subset of feature elements, a data structure is reconstructed to generate focused knowledge distribution data. Since the screened subset of feature elements only contains feature elements that are highly relevant to the disaster recovery knowledge dimension and have high importance, reconstructing the data structure based on these feature elements can more accurately reflect the knowledge distribution related to disaster recovery. If the previous feature knowledge distribution data adopted a matrix structure, then when reconstructing the focused knowledge distribution data, only the rows and columns related to the screened subset of feature elements may be retained. For example, in the previous matrix, if the feature element related to the backup frequency of building materials is retained, then in the new focused knowledge distribution data structure, the columns related to the backup frequency in the row of building materials will be retained, as well as the rows and columns related to other retained feature elements. Through such reconstruction, the generated focused knowledge distribution data is more focused on the content related to the disaster recovery knowledge dimension, can provide a more targeted data basis for subsequent operations such as disaster recovery backup prediction, and helps to improve the accuracy and efficiency in the disaster recovery backup processing of construction project cost data.

[0117] Figure 2 The hardware structure diagram of the disaster recovery management system 100 for implementing the above-mentioned disaster recovery method for construction project cost data provided by the embodiment of the present invention is shown, as Figure 2 shown, the disaster recovery management system 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

[0118] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data obtained from an external terminal. In some embodiments, the machine-readable storage medium 120 may store the data and / or instructions that the disaster recovery management system 100 uses to execute or complete the exemplary methods described in the present invention.

[0119] In a specific implementation process, one or more processors 110 execute the computer-executable instructions stored in the machine-readable storage medium 120, so that the processors 110 can execute the construction project cost data disaster recovery method in the above method embodiments. The processors 110, the machine-readable storage medium 120, and the communication unit 140 are connected through a bus 130. The processors 110 can be used to control the sending and receiving actions of the communication unit 140.

[0120] For the specific implementation process of the processors 110, reference may be made to the respective method embodiments executed by the disaster recovery management system 100 above. Their implementation principles and technical effects are similar, and will not be elaborated here in this embodiment.

[0121] In addition, an embodiment of the present invention also provides a readable storage medium, in which computer-executable instructions are preset. When a processor executes the computer-executable instructions, the above construction project cost data disaster recovery method is implemented.

[0122] It should be noted that, in order to simplify the description of the present invention disclosure and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, sometimes multiple features are incorporated into one embodiment, drawing, or description thereof.

Claims

1. A construction project cost data disaster recovery method, characterized in that The method includes: Obtaining a sample construction project cost data sequence and initializing a neural network; the sample construction project cost data sequence includes a plurality of candidate sample construction project cost data; each of the candidate sample construction project cost data carries disaster recovery backup allocation space data annotation data; Performing disaster recovery backup prediction on each candidate sample construction project cost data respectively according to the initialized neural network, and generating training disaster recovery backup allocation space data for each candidate sample construction project cost data; For each of the candidate sample construction project cost data, determining a global network learning error according to the training disaster recovery backup allocation space data and the disaster recovery backup allocation space data annotation data, and extracting effective sample construction project cost data from the sample construction project cost data sequence according to the global network learning errors of the candidate sample construction project cost data; Performing network parameter learning on the initialized neural network according to the extracted effective sample construction project cost data, and outputting a trained disaster recovery backup model; Obtaining a construction project cost data set to be subjected to data disaster recovery processing, and performing disaster recovery backup processing on the construction project cost data set according to the trained disaster recovery backup model and according to the predicted disaster recovery backup allocation space data; There are two or more of the initialized neural networks; The performing disaster recovery backup prediction on each candidate sample construction project cost data respectively according to the initialized neural network, and generating training disaster recovery backup allocation space data for each candidate sample construction project cost data includes: Loading each of the candidate sample construction project cost data into a plurality of initialized neural networks respectively, and generating training disaster recovery backup allocation space data corresponding to each of the initialized neural networks for each of the candidate sample construction project cost data; For each of the candidate sample construction project cost data, the determining a global network learning error according to the training disaster recovery backup allocation space data and the disaster recovery backup allocation space data annotation data includes: Determining a first network learning error corresponding to each of the initialized neural networks according to the training disaster recovery backup allocation space data and the disaster recovery backup allocation space data annotation data corresponding to each of the initialized neural networks; Determining the global network learning error of the candidate sample construction project cost data according to the first network learning errors corresponding to the initialized neural networks; 2. The construction project cost data disaster recovery method according to claim 1, characterized in that The performing network parameter learning on the initialized neural network according to the extracted effective sample construction project cost data, and outputting a trained disaster recovery backup model includes: Performing network parameter learning on the plurality of initialized neural networks respectively according to the effective sample construction project cost data, and generating a plurality of initialized neural networks; Determining the trained disaster recovery backup model from the plurality of initialized neural networks through performance verification of the plurality of initialized neural networks; 3. The construction project cost data disaster recovery method according to claim 1, characterized in that The determining the global network learning error of the candidate sample construction project cost data according to the first network learning errors corresponding to the initialized neural networks includes: Determine the relative learning error between every two of the training disaster recovery backup allocation space data; Determine the sum of each of the first network learning errors and the relative learning error as the global network learning error.

4. The construction project cost data disaster recovery method according to claim 1, wherein The network parameter learning of the initialized neural network based on the extracted valid example construction project cost data, and output the trained disaster recovery backup model, including: Perform iterative network parameter learning on the initialized neural network based on the extracted valid example construction project cost data to generate a first temporary neural network. If the first temporary neural network meets the network convergence requirement, generate the trained disaster recovery backup model based on the first temporary neural network; If the first temporary neural network does not meet the network convergence requirement, the method further includes: Use the first temporary neural network after network parameter learning as an iterative neural network, and iteratively execute the following steps until the generated second temporary neural network meets the network convergence requirement, and generate the trained disaster recovery backup model based on the second temporary neural network that meets the network convergence requirement: Obtain an iterative example construction project cost data sequence; Perform disaster recovery backup prediction on each candidate example construction project cost data in the iterative example construction project cost data sequence based on the initialized neural network to generate training disaster recovery backup allocation space data for each candidate example construction project cost data; For each of the candidate example construction project cost data, determine the global network learning error based on the training disaster recovery backup allocation space data and the data labeled with the disaster recovery backup allocation space data; Extract valid example construction project cost data from the example construction project cost data sequence based on the global network learning error of each of the candidate example construction project cost data; Perform iterative network parameter learning on the initialized neural network based on the extracted valid example construction project cost data to generate a second temporary neural network. If the second temporary neural network does not meet the network convergence requirement, use the second temporary neural network as an iterative neural network.

5. The construction project cost data disaster recovery method according to claim 1, wherein, The extracting valid example construction project cost data from the example construction project cost data sequence based on the global network learning error of each of the candidate example construction project cost data includes at least one of the following: Label a set proportion of the example construction project cost data with the smallest global network learning error as the valid example construction project cost data; Label the example construction project cost data with a global network learning error less than a set error as the valid example construction project cost data.

6. The construction project cost data disaster recovery method according to claim 1, characterized in that The disaster recovery backup processing of the construction project cost data set based on the trained disaster recovery backup model and according to the predicted disaster recovery backup allocation space data includes: Using the disaster recovery backup model, extract the characteristic knowledge distribution data of the construction project cost data set, determine the characteristic focusing factor of the characteristic knowledge distribution data, and extract the focused knowledge distribution data of the characteristic knowledge distribution data based on the characteristic focusing factor; Perform disaster recovery backup prediction on the focused knowledge distribution data to generate the confidence levels of the construction project cost data sets corresponding to multiple reference disaster recovery backup allocation space data respectively. Determine the predicted disaster recovery backup allocation space data of the construction project cost data set according to the confidence levels of the construction project cost data sets corresponding to multiple reference disaster recovery backup allocation space data respectively.

7. The construction project cost data disaster recovery method according to claim 6, characterized in that, The step of using the disaster recovery backup model to extract the characteristic knowledge distribution data of the construction project cost data set includes: Perform data cleaning operations on the construction project cost data set to remove the noise data, error data and incomplete data existing in the construction project cost data set. After performing standardization processing on the construction project cost data set that has undergone data cleaning operations, perform data encoding on the standardized construction project cost data set to generate the data encoding result of the construction project cost data set. Input the data encoding result into the initial feature extraction layer of the disaster recovery backup model, perform linear combination operations on the data encoding result through multiple neurons in the initial feature extraction layer to obtain an intermediate feature extraction result, and perform activation function processing on the intermediate feature extraction result to generate an initial feature set. Adopt a feature cross method to perform combination operations on the features in the initial feature set to generate a combined feature sequence, and screen the combined feature sequence based on the importance evaluation index of the features, where the importance evaluation index is obtained by calculating the correlation between each feature and the disaster recovery backup target. Input the screened combined feature sequence into the deep feature mining layer in the disaster recovery backup model. The deep feature mining layer consists of multiple hidden layers. The screened combined feature sequence is input into the first layer of the deep feature mining layer, and linear combination operations are performed on the screened combined feature sequence through the neurons in the first layer. After performing activation function processing on the result of the linear combination operation, the output result is passed to the next hidden layer to repeat the above linear combination and activation function processing process, and the output result of the last hidden layer is used as the deep feature set. Perform semantic analysis on the deep features in the deep feature set, group the deep features with similar semantics according to the semantic analysis results, and perform knowledge fusion operations on the grouped deep features to generate feature knowledge subsets, and combine all the feature knowledge subsets to generate an integrated feature knowledge set. Determine the dimension of the feature knowledge distribution, count the integrated feature knowledge set according to the determined dimension, and construct the structure of the feature knowledge distribution data using a matrix data structure or a vector data structure according to the statistical results. Among them, if a matrix structure is adopted, the rows represent different dimensions, and the columns can represent the specific attributes of the feature knowledge under each dimension. The step of determining the feature focusing factor of the feature knowledge distribution data and extracting the focused knowledge distribution data of the feature knowledge distribution data according to the feature focusing factor includes: Determine the feature focusing factor of the feature knowledge distribution data in the disaster recovery knowledge dimension. Determine the focused knowledge distribution data based on the feature knowledge distribution data and the feature focusing factor of the disaster tolerance knowledge dimension; Among them, the step of determining the feature focusing factor of the feature knowledge distribution data in the disaster tolerance knowledge dimension includes: Analyze the knowledge content related to the classification of the disaster tolerance knowledge dimension in the feature knowledge distribution data; Taking each feature in the knowledge content under the classification of the disaster tolerance knowledge dimension as a node, construct a feature association network. Specifically, analyze the direct association relationship and indirect association relationship between each feature, and according to the direct and indirect association relationships between the features, connect the relevant feature nodes with connection edges to generate the feature association network. The feature association network is used to reflect the mutual relationship between each feature in the knowledge content under the classification of the disaster tolerance knowledge dimension; For each node in the feature association network, evaluate the importance of the node in the feature association network based on the degree value, betweenness centrality, and closeness centrality of the node, and generate the first feature importance evaluation result. The degree value is the number of edges connected to the node, the betweenness centrality represents the frequency of a node appearing on the shortest path between other nodes, and the closeness centrality reflects the average distance from a node to other nodes; Obtain the pre-set disaster tolerance strategy, and adjust the first feature importance evaluation result according to the disaster tolerance strategy to generate an adjusted feature importance evaluation result; Determine a benchmark value, which is the average value of all feature importances or a basic value set according to preset rules; For each node, construct a calculation rule according to the relationship between the adjusted importance evaluation result of the node and the benchmark value, and set different dimension weights for the backup frequency knowledge dimension, backup storage location knowledge dimension, and data recovery strategy knowledge dimension respectively. According to the calculation rule and dimension weights, construct a calculation framework; Call the calculation framework to calculate the feature focusing factor of each node in the disaster tolerance knowledge dimension based on the adjusted feature importance evaluation result and the dimension weights of the disaster tolerance knowledge dimension, and summarize the feature focusing factors of each node in the disaster tolerance knowledge dimension to determine the feature focusing factor of the feature knowledge distribution data in the disaster tolerance knowledge dimension.

8. The construction project cost data disaster recovery method according to claim 7, characterized in that The step of determining the focused knowledge distribution data based on the feature knowledge distribution data and the feature focusing factor of the disaster tolerance knowledge dimension includes: Analyze the data structure of the feature knowledge distribution data. Based on the data structure of the feature knowledge distribution data, search for feature elements associated with the disaster tolerance knowledge dimension. By searching and matching in the analyzed data structure, extract all feature elements related to the disaster tolerance knowledge dimension to generate a set of feature elements. The disaster tolerance knowledge dimension includes the backup frequency knowledge dimension, backup storage location knowledge dimension, and data recovery strategy knowledge dimension; Use the feature focusing factor of the disaster recovery knowledge dimension to assign weights to each feature element in the feature element set, and generate a weight sequence of the feature element set. The feature focusing factor is an index used to measure the importance of each feature element under the disaster recovery knowledge dimension; According to a preset weight threshold or weight ratio, screen the feature element set according to the weight sequence of the feature element set, and generate an optimized feature element subset; Based on the screened feature element subset, reconstruct the data structure to generate the focused knowledge distribution data.

9. A disaster recovery management system, characterized in that, The disaster recovery management system includes a processor and a memory. The memory is connected to the processor. The memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the construction project cost data disaster recovery method according to any one of claims 1-8 above.

Images