Data processing method and system of integrated digital intelligent construction platform

Abstract

The invention provides a data processing method and system for an integrated digital smart construction platform, and the method comprises the steps: obtaining a to-be-processed engineering construction data set, detecting a target cost consumption event in the to-be-processed engineering construction data set, and obtaining a data range of the target cost consumption event; and extracting a current data range representation vector corresponding to the target cost consumption event data range, and obtaining a reference contrast data range representation vector corresponding to each specification grade. And determining a data range specification level of the target cost consumption event data range according to a common property measurement coefficient of the current data range representation vector and a contrast data range representation vector corresponding to each specification level, and determining a cost consumption specification level corresponding to the to-be-processed engineering construction data set according to the data range specification level of the target cost consumption event data range. According to the invention, the reliability of cost consumption specification grade determination can be improved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a data processing method and system for an integrated digital intelligent construction platform. Background Art

[0002] In the field of modern engineering construction, with the continuous expansion of the scale of engineering projects and the increasing complexity, cost management has become a crucial link. Engineering construction involves many links, including raw material procurement, personnel deployment, equipment use, etc., and each link will incur cost consumption. These cost consumptions are interrelated and influence each other, forming a complex cost system. Traditional cost management methods often rely on experience and rough estimates, and it is difficult to accurately grasp the degree of standardization of cost consumption. For example, in construction projects, the cost of concrete is estimated only based on the approximate amount and price range of previous projects, and it is impossible to fully consider the impact of special requirements, quality standards, and construction environment factors in different projects on costs.

[0003] Existing cost management technologies have obvious limitations in determining the standard level of cost consumption. At present, most methods lack the ability to deeply mine and accurately analyze engineering construction data. On the one hand, the data processing is not sophisticated enough. For the massive data generated during the construction process, there is no effective means to accurately extract key information related to the standard level of cost consumption. For example, when faced with engineering construction data containing various building material costs, labor costs, equipment rental costs and other data, it is difficult to determine which data features truly reflect whether the cost consumption meets the standard requirements, which leads to a lack of accurate basis for determining the cost standard level. On the other hand, there is a lack of unified and effective comparison standards. Since the cost consumption in engineering construction is affected by many factors and there are large differences between different projects, the existing cost management technology has not established a systematic cost consumption standard level determination system that can adapt to different situations. This makes it difficult to reliably compare and determine the cost consumption standard level between different projects or different engineering stages. Summary of the invention

[0004] In view of this, the present application provides a data processing method and system for an integrated digital intelligent construction platform.

[0005] The technical solution of the present application is implemented as follows: In the first aspect, the present application provides a data processing method for an integrated digital intelligent construction platform, the method comprising: obtaining a data set of engineering construction to be processed, detecting a target cost consumption event in the data set of engineering construction to be processed, and obtaining a target cost consumption event data range; extracting a current data range representation vector corresponding to the target cost consumption event data range, and obtaining a benchmark control data range representation vector corresponding to each specification level, wherein the benchmark control data range representation vector is obtained by extracting the representation vectors of each control cost consumption event data range of the corresponding specification level, obtaining each control cost consumption event data range representation vector, and integrating each control cost consumption event data range representation vector; determining a target control data range representation vector corresponding to the current data range representation vector in the control data range representation vector corresponding to each specification level according to the commonality measurement coefficients of the current data range representation vector and the control data range representation vectors corresponding to each specification level; determining the specification level corresponding to the target control data range representation vector as the data range specification level corresponding to the target cost consumption event data range, and determining the cost consumption specification level corresponding to the engineering construction data set to be processed according to the data range specification level of the target cost consumption event data range.

[0006] On the other hand, the present application provides a computer system, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and when the processor executes the program, the steps in the method described in the first aspect above are implemented.

[0007] The beneficial effects of the present application include: the present application is based on detecting the target cost consumption event in the engineering construction data set to be processed, obtaining the target cost consumption event data range, extracting the current data range representation vector corresponding to the target cost consumption event data range, and obtaining the benchmark control data range representation vector corresponding to each specification level, the benchmark control data range representation vector is to extract the representation vector of each control cost consumption event data range of the corresponding specification level respectively, obtain each control cost consumption event data range representation vector, and integrate each control cost consumption event data range representation vector to obtain; according to the commonality measurement coefficient between the current data range representation vector and the control data range representation vector corresponding to each specification level, determine the target control data range representation vector corresponding to the current data range representation vector in the control data range representation vector corresponding to each specification level, determine the specification level corresponding to the target control data range representation vector as the data range specification level corresponding to the target cost consumption event data range, and determine the cost consumption specification level corresponding to the engineering construction data set to be processed according to the data range specification level of the target cost consumption event data range. That is, the data range specification level of the target cost consumption event data range is determined based on the commonality measurement coefficients of the current data range representation vector and the control data range representation vectors corresponding to each specification level. This can improve the accuracy of determining the cost consumption specification level of the target cost consumption event data range. Then, the cost consumption specification level corresponding to the engineering construction data set to be processed is determined based on the data range specification level of the target cost consumption event data range, so as to increase the reliability of the obtained cost consumption specification level.

[0008] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and are used together with the specification to illustrate the technical solution of the present application.

[0010] Figure 1 A schematic diagram of the implementation flow of a data processing method for an integrated digital intelligent construction platform provided in an embodiment of the present application; Figure 2 A hardware entity diagram of a computer system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application are further elaborated in detail below in conjunction with the drawings and embodiments. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0012] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. The terms "first / second / third" involved are merely used to distinguish similar objects and do not represent a specific order for the objects. It is understood that "first / second / third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0013] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing this application and are not intended to limit this application.

[0014] The embodiment of the present application provides a data processing method for an integrated digital intelligent construction platform, which can be executed by a processor of a computer system. The computer system can refer to a server, a laptop, a tablet computer, a desktop computer, a mobile device (such as a mobile phone, a portable video player, a personal digital assistant, a dedicated messaging device, a portable gaming device), and other devices with data processing capabilities.

[0015] Figure 1 A schematic diagram of the implementation flow of a data processing method for an integrated digital intelligent construction platform provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes the following steps: Step S100: Acquire a data set of engineering construction data to be processed, detect target cost consumption events in the data set of engineering construction data to be processed, and obtain a data range of the target cost consumption events.

[0016] In an embodiment of the present application, the engineering construction data set to be processed is a data set related to engineering construction, and these data, for example, cover multiple aspects of the engineering construction process, such as procurement data of building materials, working hours records of construction personnel, and use time of mechanical equipment. The computer system obtains these data from a storage device or a related data source through a pre-set interface or data acquisition module. Among these data, for numerical data, it can be directly recorded as the corresponding numerical value, or recorded after normalization. For discrete data, it can be recorded after unique hot encoding so that the computer system can use it for calculation. Of course, for different discrete non-numerical data, different methods can be used for conversion, such as encoding (such as unique hot encoding, sequential variable encoding), word bag model, TF-IDF, logical mapping, etc., without specific limitation.

[0017] Target cost consumption events refer to specific events that will have a consumption impact on costs during the construction process. For example, when building a building, the pouring of concrete will consume a lot of costs, including the cost of concrete raw materials, transportation costs, the rental cost of pouring equipment, and labor costs, etc. This can be regarded as a target cost consumption event. The computer system detects these events through a pre-set algorithm or model. The algorithm or model here can be based on data mining technology, such as an association rule mining algorithm. Assuming that there is a data table in the engineering construction data set, the table records the relationship between different construction processes and cost consumption. The association rule mining algorithm can analyze the correlation between these data to find those construction processes associated with high cost consumption and identify them as target cost consumption events.

[0018] After detecting the target cost consumption event, the computer system determines the target cost consumption event data range. This data range refers to the set of data associated with the target cost consumption event. Taking concrete pouring as an example, the relevant data includes, for example, the supplier information of concrete, the price of different batches of concrete, the model and transportation distance of the transport vehicle, the power and working hours of the equipment used during pouring, the number of construction workers involved in pouring and their salary levels, etc. These data together constitute the data range of the target cost consumption event of concrete pouring. The computer system can filter out these data related to the target cost consumption event from the entire engineering construction data set to be processed by data tagging or indexing to form the target cost consumption event data range.

[0019] The computer system can use a variety of technical means to detect target cost consumption events and obtain their data range. For example, a rule-based system can be used to pre-set a series of rules to identify target cost consumption events. For example, the rule can be set as "when the cost of a certain engineering activity exceeds a certain threshold, it will be marked as a target cost consumption event." The formula for calculating the cost can be: cost = unit price × quantity + other expenses. Among them, the unit price can be the unit price of the material, the rental unit price of the equipment, or the daily salary of the labor, etc. The quantity is the corresponding material usage, equipment usage time, or labor working days. Other expenses include miscellaneous expenses such as transportation fees and management fees. Through such calculations and rule judgments, the computer system can accurately obtain the target cost consumption events in the engineering construction data set to be processed and determine its data range, laying the foundation for subsequent data processing steps.

[0020] Step S200: Extract the current data range representation vector corresponding to the target cost consumption event data range, and obtain the benchmark control data range representation vector corresponding to each specification level. The benchmark control data range representation vector is obtained by extracting the representation vectors of each control cost consumption event data range of the corresponding specification level, obtaining each control cost consumption event data range representation vector, and integrating each control cost consumption event data range representation vector.

[0021] In the embodiment of the present application, step S200 extracts the current data range representation vector corresponding to the target cost consumption event data range, and obtains the benchmark control data range representation vector corresponding to each specification level.

[0022] For extracting the current data range characterization vector corresponding to the target cost consumption event data range, the computer system processes the target cost consumption event data range obtained in step S100. The target cost consumption event data range includes various data related to a specific cost consumption event. For example, in a construction project, if the target cost consumption event is concrete pouring, its data range includes various factors such as the price of raw materials for concrete, transportation distance, and the length of time the pouring equipment is used. The characterization vector is a vector form that can represent data characteristics. The computer system converts these data into vectors through a specific algorithm. A feasible technical means is the principal component analysis (PCA) algorithm, which converts multiple related variables into a few unrelated comprehensive variables, and these comprehensive variables can constitute a characterization vector. For example, assuming that there are n variables in the target cost consumption event data range, the PCA algorithm can construct a characterization vector through the following steps: calculating the covariance matrix of the variables, finding the eigenvalues ​​and eigenvectors of the covariance matrix, selecting the first k eigenvectors (k<n) according to the size of the eigenvalues, and projecting the original data into the space formed by these k eigenvectors. The projection result is the characterization vector. Next, the computer system obtains the characterization vector of the benchmark control data range corresponding to each specification level. The specification level is a measure of the cost consumption and other aspects of the construction process. For example, a high specification level means stricter control standards and higher quality requirements in terms of cost consumption, while a low specification level is relatively loose. Different specification levels have corresponding control cost consumption event data ranges, which are also data sets related to various cost consumption events. Taking construction engineering as an example, the data on the quality standards, procurement channels, and usage control of steel under high specification levels are different from those under low specification levels. The computer system also needs to convert these control cost consumption event data ranges into representation vectors. This process is similar to the previous extraction of the representation vector corresponding to the target cost consumption event data range. However, here the operation is performed separately for each specification level. For each specification level, it contains multiple control cost consumption event data ranges. The computer system converts these different control cost consumption event data ranges into representation vectors and then integrates them. The purpose of integration is to obtain a benchmark control data range representation vector that can comprehensively represent the characteristics of the specification level. For example, for m control cost consumption event data ranges under a specification level, the computer system first converts each data range into a representation vector, and then integrates them by weighted average. Assuming that the i-th characterization vector is Vi and its corresponding weight is wi (the weight can be determined based on the importance of the data or other pre-set rules), then the integrated benchmark control data range characterization vector V=∑(wi * Vi) (i ranges from 1 to m).

[0023] By executing step S200, the computer system provides the necessary data basis for the subsequent determination of the data range standard level of the target cost consumption event data range. It converts the target cost consumption event data range and the benchmark control data range corresponding to each standard level into a comparable representation vector form, so that in the subsequent steps, the standard level of the target cost consumption event data range can be determined by comparing the relationship between these representation vectors, thereby providing a basis for determining the cost consumption standard level corresponding to the entire engineering construction data set to be processed. This processing method based on representation vectors helps to improve the accuracy and efficiency of data processing, so that the standardization of cost consumption can be better evaluated in a complex engineering construction data environment.

[0024] Step S300: Determine a target control data range representation vector corresponding to the current data range representation vector from the control data range representation vectors corresponding to each standard level based on the commonality measurement coefficients between the current data range representation vector and the control data range representation vectors corresponding to each standard level.

[0025] In an embodiment of the present application, step S300 is executed by a computer system. The key to this step is to determine the target control data range representation vector corresponding to the current data range representation vector in the control data range representation vectors corresponding to each standard level based on the commonality measurement coefficients of the current data range representation vector and the control data range representation vectors corresponding to each standard level.

[0026] The current data range characterization vector is a vector extracted from the target cost consumption event data range in step S200, which comprehensively represents the characteristics of the target cost consumption event related data. For example, in a construction project, if the target cost consumption event is an exterior wall decoration project, then the current data range characterization vector, for example, includes the vector characteristics formed after processing factors such as the cost of exterior wall materials, the difficulty of construction technology, and the decoration area. The control data range characterization vectors corresponding to the specification level were previously constructed for each specification level, and they represent the cost consumption event data characteristics under different specification requirements. For example, the control data range characterization vector under a high specification level reflects the data characteristics under higher standard requirements for the quality of exterior wall decoration materials, the qualifications of construction personnel, and construction safety measures, while the vector characteristics under a low specification level are relatively loose. The commonality measurement coefficient is a numerical value used to measure the similarity or commonality between two vectors.

[0027] The computer system can use a variety of technical means to determine this commonality metric coefficient. One feasible method is to calculate the distance between vectors, such as the Euclidean distance. Suppose the current data range representation vector is , the reference data range representation vector corresponding to a certain standard level is , the Euclidean distance formula is This distance value can be used as a representation of the commonality measurement coefficient. The smaller the distance, the higher the degree of commonality between the two vectors.

[0028] The computer system performs such calculations on the control data range representation vector corresponding to each standard level, and obtains the commonality measurement coefficients between the current data range representation vector and the control data range representation vectors corresponding to each standard level. Then, the target control data range representation vector is determined in the control data range representation vectors corresponding to each standard level based on these commonality measurement coefficients. For example, assuming that there are three standard levels: high standard level, medium standard level and low standard level, the computer system calculates the commonality measurement coefficients between the current data range representation vector and the control data range representation vectors corresponding to these three standard levels respectively. If the calculation result shows that the commonality measurement coefficient between the current data range representation vector and the control data range representation vector corresponding to the medium standard level is the smallest, then the computer system determines the control data range representation vector corresponding to the medium standard level as the target control data range representation vector corresponding to the current data range representation vector.

[0029] This method of determining the target control data range characterization vector based on the commonality measurement coefficient helps to establish an association between the target cost consumption event data range and each specification level, thereby laying the foundation for the subsequent accurate determination of the data range specification level of the target cost consumption event data range, and further provides a key basis for determining the cost consumption specification level corresponding to the engineering construction data set to be processed, which is of great significance in the cost control and specification management of the entire engineering construction. It enables the computer system to scientifically and reasonably judge the specification level of the target cost consumption event data based on the relationship between data features in a complex engineering construction data environment.

[0030] Step S400: Determine the specification level corresponding to the target control data range characterization vector as the data range specification level corresponding to the target cost consumption event data range, and determine the cost consumption specification level corresponding to the engineering construction data set to be processed based on the data range specification level of the target cost consumption event data range.

[0031] In the embodiment of the present application, step S400 is executed by a computer system, and this step is mainly to determine the cost consumption standard level corresponding to the engineering construction data set to be processed based on the results of the previous steps.

[0032] In step S300, the computer system has determined the target control data range representation vector, and this target control data range representation vector corresponds to a specific specification level. The specification level here is a measurement standard classification of cost consumption and other conditions in the construction process of the project. Different specification levels have different cost consumption standards, quality requirements and management specifications. For example, in a construction project, a high specification level requires the use of higher quality building materials, more stringent quality inspections during the construction process, and the corresponding cost consumption has a strict control range in all aspects; a low specification level has relatively loose requirements in terms of material quality, inspection frequency, etc., and the range of cost consumption is also relatively wide.

[0033] The computer system determines the specification level corresponding to the target control data range characterization vector as the data range specification level corresponding to the target cost consumption event data range. This means that by comparing and analyzing the target cost consumption event data range with the control data range characterization vectors corresponding to each specification level, the specification level that best matches it is found, and this specification level becomes the specification level of the target cost consumption event data range itself. Next, the cost consumption specification level corresponding to the engineering construction data set to be processed is determined based on the data range specification level of the target cost consumption event data range. The computer system can use weighted averaging technology to implement this determination process. Assume that there are multiple target cost consumption events in the engineering construction data set to be processed, and each target cost consumption event data range has its determined data range specification level. Let the data range specification level of the i-th target cost consumption event data range be , and its corresponding weight is (The weight can be determined based on the importance of the target cost consumption event in the entire project, the cost ratio, and other factors). Then the cost consumption standard level L corresponding to the construction data set to be processed can be obtained by the formula To calculate (where n is the number of target cost consumption events).

[0034] For example, in a construction project data set, there are three target cost consumption events: concrete pouring, steel bar erection, and wall masonry. The data range specification level of concrete pouring has been determined to be a medium specification level with a weight of 0.3; the data range specification level of steel bar erection is a high specification level with a weight of 0.4; the data range specification level of wall masonry is a low specification level with a weight of 0.3. According to the above formula, the cost consumption specification level corresponding to the project construction data set to be processed can be calculated. This method comprehensively considers the specification level and importance of the data range of each target cost consumption event, thereby deriving the cost consumption specification level corresponding to the entire project construction data set to be processed, which helps to comprehensively and accurately evaluate the cost consumption standardization level of the entire project construction, so as to facilitate the subsequent cost management, quality control, and project schedule scheduling.

[0035] As an implementation manner, step S100, detecting the target cost consumption event in the engineering construction data set to be processed and obtaining the target cost consumption event data range may include: Step S110: extracting a data set filter characterization vector of the engineering construction data set to be processed, performing normalization processing according to the data set filter characterization vector, and obtaining a normalized characterization vector; Step S120: activating the normalized representation vector to obtain an activated representation vector, extracting a data range based on the activated representation vector, and obtaining a target cost consumption event data range.

[0036] In the embodiment of the present application, steps S110-S120 in the specific implementation of step S100 are executed by a computer system. These two steps are the specific operation process of detecting the target cost consumption event in the engineering construction data set to be processed and obtaining the target cost consumption event data range.

[0037] In step S110 , the computer system extracts a data set filter characterization vector of the engineering construction data set to be processed, and performs normalization processing on the data set filter characterization vector to obtain a normalized characterization vector.

[0038] The construction data set to be processed is a data set containing various types of information in the construction process, such as the types, quantities, prices of construction materials, working hours, wages of construction workers, use of construction equipment and many other aspects. The data set filter representation vector is a vector representation of the original data set after feature extraction and screening. For example, in a data set of a construction project, for example, it contains a large number of material procurement records. The computer system can extract key feature information through a specific algorithm, such as the main suppliers of different materials, the average purchase price fluctuation range and other information, and combine this information into a vector, which is the data set filter representation vector.

[0039] Normalization is an operation that maps data to a specific interval. Its purpose is to eliminate the impact of different dimensions between data and help improve the accuracy and stability of data processing. For example, in an engineering construction data set, the amount of building materials is, for example, hundreds of tons, while the number of construction workers is, for example, only dozens of people. The magnitude of these two data is very different. If normalization is not performed, in the subsequent data analysis and processing, large amounts of data will have a dominant effect on the results, thereby masking the role of other data. Computer systems can use a variety of normalization techniques, one of which is the minimum-maximum normalization method. Let the original data be x, and its minimum value is , the maximum value is , the normalized data The formula can be In this way, each data in the original data set is converted to the interval [0, 1] to obtain a normalized representation vector.

[0040] In step S120, the computer system activates the normalized representation vector to obtain an activated representation vector, extracts a data range based on the activated representation vector, and obtains a target cost consumption event data range.

[0041] The activation operation is the process of performing nonlinear transformation on the normalized representation vector. The purpose is to introduce nonlinear factors so that the data can better adapt to complex actual situations. For example, in neural network-related algorithms, the activation function can be the Sigmoid function, the ReLU function, etc. Taking the Sigmoid function as an example, let the element in the normalized representation vector be x, and the element after activation by the Sigmoid function is Through this activation operation, each element in the normalized representation vector is transformed to obtain the activated representation vector.

[0042] Data range extraction based on activation characterization vectors is an operation to determine the data range of the target cost consumption event based on the characteristic information in the activation characterization vectors. For example, in a construction engineering data set, after the activation characterization vectors are obtained through the previous processing, the computer system can identify the data related to the target cost consumption event based on the characteristic values ​​in the vectors. Assuming that a certain characteristic value in the activation characterization vector is associated with the high cost consumption of building materials, the computer system can filter out all data related to the cost consumption of the building materials from the entire data set based on this characteristic value and its related algorithm logic, such as the purchase records, transportation costs, storage losses and other data of the materials, which together constitute the data range of the target cost consumption event.

[0043] In summary, steps S110-S120 accurately detect and determine the target cost consumption event data range from the engineering construction data set to be processed through a series of data processing operations, laying the foundation for the subsequent data processing flow of the integrated digital intelligent construction platform. These operations help improve the analysis and processing capabilities of engineering construction data, enabling more accurate decision-making and control in aspects such as cost consumption management.

[0044] As an implementation manner, in step S100, before obtaining the engineering construction data set to be processed, detecting the target cost consumption event in the engineering construction data set to be processed, and obtaining the target cost consumption event data range, the method may further include: Step S101: obtaining the data range of each comparison cost consumption event corresponding to the preset standard level; Step S102: extracting characterization vectors from each comparison cost consumption event data range to obtain characterization vectors from each comparison cost consumption event data range; Step S103: Integrate the data range representation vectors of each comparison cost consumption event to obtain a reference comparison data range representation vector corresponding to a preset standard level.

[0045] In step S101, the computer system obtains the data range of each comparison cost consumption event corresponding to the preset specification level. The preset specification level is a different standard level pre-set in the field of engineering construction. These levels have different degrees of specification requirements for various aspects of the engineering construction process, including cost consumption, quality control, construction technology, etc. For example, in construction engineering, three preset specification levels of high, medium and low can be set. The data range of the comparison cost consumption event is a data set under a specific cost consumption event associated with each preset specification level. Taking the cost consumption event of concrete pouring as an example, at a high specification level, the data range of the comparison cost consumption event includes, for example, data on the use of specific high-quality concrete raw materials (such as strength grade, durability index and other related data), precise batching ratio, advanced pouring equipment (performance parameters of equipment, maintenance cost and other data) and highly qualified construction personnel (human cost and other data corresponding to the qualification level); at a low specification level, the quality requirements of concrete raw materials are relatively low, and the equipment performance and personnel qualification requirements will be relaxed accordingly. These different data sets are the data ranges of the comparison cost consumption event of concrete pouring under each preset specification level. The computer system may obtain these data ranges from a pre-established engineering construction specification database or related standard documents.

[0046] In step S102, the computer system extracts the characterization vectors of each comparison cost consumption event data range to obtain the characterization vectors of each comparison cost consumption event data range. Characterization vector extraction is the process of converting a complex data range into a vector form that can represent its characteristics. For a data set such as the comparison cost consumption event data range that contains multiple types of data, the computer system uses effective technical means to extract features and convert them into vectors. For example, for the comparison cost consumption event data range of the above concrete pouring, the computer system can use the principal component analysis (PCA) algorithm to extract the characterization vector. Assuming that the comparison cost consumption event data range of concrete pouring at a high specification level contains n variables (such as different quality indicators of concrete raw materials, multiple performance parameters of equipment, and several dimensions related to personnel qualifications, etc.), the PCA algorithm first calculates the covariance matrix of these n variables, and then finds the eigenvalues ​​and eigenvectors of the covariance matrix. According to the size of the eigenvalue, the first k eigenvectors (k<n) are selected, and the original data is projected into the space formed by these k eigenvectors. The projection result is the characterization vector of the comparison cost consumption event data range of concrete pouring at a high specification level. The same method is applied to each comparison cost consumption event data range under other preset standard levels to obtain the corresponding representation vector.

[0047] In step S103, the computer system integrates the data range characterization vectors of each comparison cost consumption event to obtain a benchmark comparison data range characterization vector corresponding to the preset specification level. The purpose of the integration is to obtain a vector that can comprehensively represent the characteristics of all comparison cost consumption events under a preset specification level. Taking the high specification level in the construction project just mentioned as an example, assume that there are multiple comparison cost consumption events such as concrete pouring, steel bar binding, and formwork installation under this specification level, and their respective data range characterization vectors have been obtained. The computer system can use a weighted average method for integration. Let the i-th comparison cost consumption event data range characterization vector be , and its corresponding weight is (The weight can be determined based on the importance of the reference cost consumption event in the entire project construction, the cost ratio or other relevant factors), then the benchmark reference data range representation vector corresponding to the integrated preset specification level is (where m is the number of control cost consumption events under the preset specification level). In this way, the computer system integrates the data range representation vectors of each control cost consumption event into a benchmark control data range representation vector, which will play an important role in subsequent operations such as comparison with the target cost consumption event data range representation vector, and help determine the specification level of the target cost consumption event data range, thereby providing a basis for the cost consumption specification level evaluation of the entire engineering construction data set.

[0048] As an implementation manner, step S103, integrating the data range representation vectors of each comparison cost consumption event to obtain a reference comparison data range representation vector corresponding to a preset standard level, may include: Step S1031: combining the characterization vectors of the ranges of the comparison cost consumption event data to obtain a combined characterization vector; Step S1032: Perform feature extraction based on the combined characterization vector to obtain a reference comparison data range characterization vector corresponding to a preset standard level.

[0049] In step S1031, the computer system combines the characterization vectors of the data ranges of the control cost consumption events to obtain a combined characterization vector. The characterization vectors of the data ranges of the control cost consumption events are vectors obtained after feature extraction of each data range of the control cost consumption events in the previous step. These vectors represent the feature information of different control cost consumption events. For example, in a construction project, it is assumed that there are three control cost consumption events: concrete pouring, steel bar binding, and formwork installation. For the data range of the control cost consumption event of concrete pouring, the computer system obtains the characterization vector by a specific algorithm (such as a principal component analysis algorithm). ,in Represents the various components of the vector, which correspond to the raw material characteristics of concrete, pouring equipment parameters, construction personnel operations and other related features; for the comparison cost consumption event data range of steel bar binding, the characterization vector is obtained , whose components are related to the specifications of the steel bars, the binding process, the required tools and other characteristics; for the comparison cost consumption event data range of the template installation, the characterization vector is obtained , whose components are related to the material, size, installation method and other characteristics of the template. The computer system combines these vectors together to form a new vector, which is the combined representation vector. The combination method can be simply to arrange the vectors in order, such as the combined representation vector .

[0050] Next is step S1032, where the computer system performs feature extraction based on the combined characterization vector to obtain a baseline control data range characterization vector corresponding to the preset standard level. After obtaining the combined characterization vector, although this vector contains the characteristic information of each control cost consumption event, there are problems such as information redundancy or high dimensionality, and further feature extraction is required to obtain a vector that can more comprehensively represent the characteristics of all control cost consumption events under the preset standard level, that is, the baseline control data range characterization vector. The computer system can continue to use the principal component analysis (PCA) algorithm to extract features. Assume that the combined characterization vector (Here m = 3n, which is the dimension of the vector after the previous combination). First, the computer system calculates the covariance matrix C of the vector. The elements of the covariance matrix C are represents the covariance of the i-th component and the j-th component in the vector V. The computer system then finds the eigenvalues ​​of the covariance matrix C and the corresponding eigenvector . Sort by the size of the eigenvalues, select the eigenvectors corresponding to the first k larger eigenvalues ​​(k<m), and set the selected eigenvector to be Finally, the combined representation vector V is projected into the space formed by these k feature vectors to obtain the reference comparison data range representation vector corresponding to the preset standard level. ,in Representation vector V and eigenvector Through such a feature extraction process, the computer system obtains a reference comparison data range representation vector that can more concisely and effectively represent the characteristics of all comparison cost consumption events under the preset specification level. This vector will play an important role in subsequent operations such as comparison with the target cost consumption event data range representation vector, and will help to accurately determine the specification level of the target cost consumption event data range, thereby providing a basis for determining the cost consumption specification level of the entire engineering construction data set.

[0051] As an implementation method, the method may further include: Step S1021: loading each comparison cost consumption event data range into the reference representation vector extraction network; Step S1022: extracting representation vectors from each comparison cost consumption event data range based on the representation vector extraction component in the reference representation vector extraction network to obtain representation vectors for each comparison cost consumption event data range; Step S1023: Based on the representation vector integration component in the benchmark representation vector extraction network, each comparison cost consumption event data range representation vector is integrated to obtain a benchmark comparison data range representation vector corresponding to a preset standard level.

[0052] In step S1021 of the embodiment of the present application, the computer system loads each comparison cost consumption event data range into the benchmark characterization vector extraction network. The comparison cost consumption event data range is a data set related to specific cost consumption events under different preset specification levels, and these data contain various information in the construction process. For example, in a construction project, for the comparison cost consumption event data range of concrete pouring, for example, it includes the raw material characteristics of concrete (such as the grade of cement, the particle size and gradation of aggregates, etc.), the mix ratio of concrete, the model and performance parameters of pouring equipment (such as the pumping pressure of concrete pump trucks, the distribution radius, etc.), the skill level related indicators of construction personnel (such as the training time of workers, the years of work experience, etc.) and the data related to the construction environment (such as the temperature and humidity during pouring, etc.). The computer system loads these complex data ranges into the benchmark characterization vector extraction network, which is a model structure specially designed to process such data and extract characterization vectors.

[0053] In step S1022, the representation vector extraction component in the benchmark representation vector extraction network extracts the representation vectors of each comparison cost consumption event data range to obtain the representation vectors of each comparison cost consumption event data range. The representation vector extraction component is an important component in the benchmark representation vector extraction network. Its function is to extract features from the input data and convert it into a representation vector. For example, assume that the component adopts an autoencoder structure based on deep learning. The autoencoder consists of two parts: an encoder and a decoder. For the input comparison cost consumption event data range, such as the concrete pouring related data mentioned above, the encoder part will gradually compress the data and map the high-dimensional input data to a low-dimensional latent space representation. This process can be regarded as a nonlinear transformation, which is implemented by the neuron activation function of a multi-layer neural network (such as the ReLU function, for the input x, ReLU(x)=max(0,x)). In this process, each layer of the encoder will learn abstract features of the data at different levels. Assume that the input data passes through a multi-layer encoder to obtain a low-dimensional vector , this vector is a form of representation vector for the data range of the control cost consumption event of concrete pouring. For other control cost consumption event data ranges, such as steel bar binding and formwork installation, they will also be processed in the representation vector extraction component in the same way to obtain their corresponding representation vectors.

[0054] Finally, step S1023 is performed, based on the characterization vector integration component in the benchmark characterization vector extraction network, to integrate the characterization vectors of the data range of each comparison cost consumption event to obtain the benchmark comparison data range characterization vector corresponding to the preset specification level. The task of the characterization vector integration component is to combine the individual comparison cost consumption event data range characterization vectors obtained previously into a vector that can represent the characteristics of the entire preset specification level. For example, assuming that for a preset specification level, there are three comparison cost consumption events: concrete pouring, steel bar binding, and formwork installation, and their corresponding characterization vectors are and The representation vector integration components can be integrated by weighted average. Let the weights be and (These weights can be determined based on the importance of each reference cost consumption event in the entire project, the cost ratio or other relevant factors), then the benchmark reference data range representation vector corresponding to the integrated preset specification level is Through such integration operations, the computer system uses the representation vector integration component in the benchmark representation vector extraction network to integrate the representation vectors of each comparison cost consumption event data range into a benchmark comparison data range representation vector. This vector can comprehensively reflect the characteristics of various cost consumption events under the preset standard level, and provides an important basis for subsequent comparison with the target cost consumption event data range representation vector and determination of the standard level of the target cost consumption event data range.

[0055] This method of processing the data range of cost consumption events by extracting different components in the network through benchmark characterization vectors helps to more efficiently and accurately obtain benchmark data range characterization vectors that can represent the characteristics of preset specification levels in a complex engineering construction data environment, thereby improving the data processing accuracy and efficiency of the entire integrated digital intelligent construction platform, and providing strong support for the standardized management of cost consumption in the engineering construction process.

[0056] As an implementation method, the baseline representation vector extraction network is calibrated by the following process: Step S10: obtaining samples of each data set and corresponding sample level prior information, and loading samples of each data set into the data set detection network to be trained; Step S20: extracting representation vectors from samples of each data set based on the representation vector extraction component to be trained in the detection network of the data set to be trained, and obtaining representation vectors of samples of each data set; Step S30: integrating the representation vectors of samples of each data set based on the representation vector integration component to be trained in the detection network of the data set to be trained, to obtain a sample integrated representation vector; Step S40: performing level mapping on the sample integration representation vector based on the level mapping component to be trained in the detection network of the data set to be trained, and obtaining the sample cost consumption standard level; Step S50: optimizing the detection network of the training data set according to the sample level prior information and the sample cost consumption standard level to obtain an optimized level mapping network; Step S60: using the optimized level mapping network as the to-be-trained data set detection network, and jumping to the steps of obtaining each data set sample and the corresponding sample level prior information, and loading each data set sample into the to-be-trained data set detection network for loop operation until the preset loop stop condition is met, and the target level mapping network is obtained; Step S70: Obtain a reference representation vector extraction network according to the target representation vector extraction component and the target representation vector integration component in the target level mapping network.

[0057] In step S10 of the embodiment of the present application, the computer system obtains each data set sample and the corresponding sample level prior information, and loads each data set sample into the data set detection network to be trained. The data set sample is a representative data set selected from the engineering construction data, which covers various information in the engineering construction process, such as material cost, construction time, equipment usage, etc. For example, in the data set sample of the construction project, for example, it contains cost consumption data of different stages of a certain construction project, such as the amount of concrete and steel used and the corresponding cost in the infrastructure construction stage, as well as information such as the working hours and wages of construction personnel in this stage. The sample level prior information is the level information known in advance about the level of these data set samples, for example, it can be divided into high, medium and low levels according to the standard degree of cost consumption. The computer system loads these data set samples into the data set detection network to be trained, which is an initial neural network model that has not been trained and optimized. Its structure can include an input layer, a hidden layer and an output layer, the input layer is used to receive the data set sample, the hidden layer is used to extract and process the data, and the output layer is used to output the result corresponding to the sample level prior information.

[0058] In step S20, the representation vector extraction component to be trained in the training data set detection network extracts the representation vectors of each data set sample to obtain the representation vectors of each data set sample. The representation vector extraction component to be trained is a part of the training data set detection network, and its function is to extract features of the input data set samples and convert them into representation vectors. For example, it is assumed that a method based on a convolutional neural network (CNN) is used to extract representation vectors. For the cost consumption data matrix in the input construction engineering data set sample (which can be regarded as a two-dimensional data structure, where rows represent different cost items and columns represent different times or stages), the convolution kernel in the convolution layer slides on the data matrix to perform convolution operations. The convolution kernel is like a filter, which performs weighted summation with the local area in the data matrix through specific weight parameters to extract local features. Assume that the size of the convolution kernel is 3×3, the step size is 1, and it slides on a 5×5 cost consumption data matrix, and each time the weighted sum of the 3×3 area and the convolution kernel is calculated to obtain a new feature map. After multiple convolutional layers, downsampling is performed through a pooling layer (such as maximum pooling, taking the maximum value of a local area) to reduce the dimension of the data. Finally, the result is mapped to a vector of fixed dimension through a fully connected layer. This vector is the data set sample representation vector. For each data set sample, the corresponding data set sample representation vector is obtained through this process.

[0059] In step S30, the training vector integration component in the training dataset detection network integrates the sample representation vectors of each dataset to obtain a sample integration representation vector. The function of the training vector integration component is to combine the sample representation vectors of each dataset obtained previously into a vector that can represent the sample characteristics of the entire dataset. For example, suppose there are three sample representation vectors of the dataset: and The integration can be performed by simple addition and averaging, that is, adding the elements in the same position and dividing by the number of samples to obtain the sample integration representation vector Of course, you can also use weighted average to integrate according to factors such as the importance of the data. This is just a simple example.

[0060] In step S40, the sample integrated representation vector is graded based on the to-be-trained grade mapping component in the to-be-trained data set detection network to obtain the sample cost consumption standard grade. The to-be-trained grade mapping component is a part that maps the integrated sample integrated representation vector to the corresponding cost consumption standard grade. For example, this component can be a classifier based on a multi-layer perceptron (MLP). Assume that the sample integrated representation vector Input to the level mapping component to be trained, the multilayer perceptron processes this vector through multiple neuron layers. The calculation method of each neuron can be ,in are the elements of the input vector, is the connection weight, b is the bias, and f is the activation function (such as Sigmoid function or ReLU function). After multiple layers of such neuron calculations, a result representing the standard level of cost consumption is obtained at the output layer, for example, output 1 represents a high level, 2 represents a medium level, 3 represents a low level, etc.

[0061] In step S50, the training data set detection network is optimized according to the sample level prior information and the sample cost consumption standard level to obtain an optimized level mapping network. In this step, the computer system adjusts the parameters of the network according to the difference between the sample level prior information known in advance and the sample cost consumption standard level obtained through the network. For example, the back propagation algorithm is used for optimization. Assuming that the sample level prior information is a high level, and the sample cost consumption standard level output by the network is a medium level, there is a difference. The back propagation algorithm starts from the output layer and according to the error function (such as the mean square error function ,in is the sample level prior information, is the level of the network output) to calculate the gradient of the error to the parameters of each layer in the network (such as weights and biases). Then, according to this gradient, a certain learning rate (such as =0.01) to update the network parameters so that the next output of the network is closer to the sample level prior information. Through multiple iterations of this optimization process, the optimized level mapping network is obtained.

[0062] In step S60, the optimized level mapping network is used as the data set detection network to be trained, and the steps of obtaining each data set sample and the corresponding sample level prior information and loading each data set sample into the data set detection network to be trained are looped until the preset loop stop condition is met to obtain the target level mapping network. The preset loop stop condition can be to reach a certain number of iterations, such as 1000 times, or to stop the loop when the performance index of the network (such as the accuracy rate reaches more than 90%) meets the requirements. Each loop will use a new data set sample or resample the original data set sample, and continuously adjust the parameters of the network so that the performance of the network is continuously improved. After multiple loops, the target level mapping network obtained has a high accuracy when performing level mapping on the data set samples.

[0063] In step S70, a benchmark representation vector extraction network is obtained based on the target representation vector extraction component and the target representation vector integration component in the target level mapping network. The target representation vector extraction component and the target representation vector integration component are components that are obtained after optimization training in the target level mapping network and are capable of accurately extracting and integrating representation vectors. The computer system extracts these two components and combines them into a benchmark representation vector extraction network. This benchmark representation vector extraction network can be used in subsequent data processing on the integrated digital smart construction platform to perform representation vector extraction and integration operations on each comparison cost consumption event data range, providing a basis for determining the standard level of the target cost consumption event data range.

[0064] Through training, the computer system obtains an accurate and effective benchmark representation vector extraction network, which helps to improve the accuracy and reliability of cost consumption specification level judgment in the entire engineering construction data processing process, thereby better managing cost consumption in engineering construction.

[0065] As an implementation manner, step S300, based on the commonality measurement coefficients of the current data range representation vector and the reference data range representation vectors corresponding to each specification level, determines the target reference data range representation vector corresponding to the current data range representation vector from the reference data range representation vectors corresponding to each specification level, which may include: Step S310: determining the vector distances between the current data range characterization vector and the reference data range characterization vectors corresponding to each standard level, and obtaining each commonality measurement coefficient; Step S320: Determine a target commonality measurement coefficient among the commonality measurement coefficients, and determine a comparison data range representation vector corresponding to the target commonality measurement coefficient as a target comparison data range representation vector corresponding to the current data range representation vector.

[0066] In step S310 of the embodiment of the present application, the computer system determines the vector distance between the current data range characterization vector and the control data range characterization vector corresponding to each specification level, and obtains each commonality measurement coefficient. The current data range characterization vector is a vector obtained after feature extraction of the target cost consumption event data range, which contains the comprehensive feature information of the target cost consumption event related data. For example, in a construction project, if the target cost consumption event is concrete pouring, then the current data range characterization vector, for example, contains the vector representation formed by the feature extraction of factors such as the price of concrete raw materials, transportation costs, and the use cost of pouring equipment. The control data range characterization vector corresponding to the specification level is the vector extracted from the control cost consumption event data range under different specification levels, and these vectors reflect the characteristics of the cost consumption event data under different specification levels. For example, the control data range characterization vector under a high specification level, for example, contains the vector representation of information such as high-quality raw material requirements and cost factors corresponding to strict construction process control after processing; the vector representation under a low specification level is, for example, relatively loose in terms of raw material quality and construction process requirements, which is reflected in the different vector features.

[0067] Vector distance is an indicator to measure the difference between two vectors. Computer systems can use a variety of technical means to calculate vector distance, thereby obtaining a commonality measurement coefficient. One feasible method is Euclidean distance. Suppose the current data range representation vector is , the reference data range representation vector corresponding to a certain standard level is , the Euclidean distance formula is , this distance value can be used as a representation of the commonality measurement coefficient. The smaller the distance, the higher the degree of commonality between the two vectors, that is, the more similar the characteristics of the current data range representation vector and the reference data range representation vector corresponding to the specification level are. The computer system calculates the vector distance for the reference data range representation vector corresponding to each specification level in this way, thereby obtaining the commonality measurement coefficient corresponding to each specification level.

[0068] In step S320, the computer system determines the target commonality measurement coefficient among the various commonality measurement coefficients, and determines the control data range representation vector corresponding to the target commonality measurement coefficient as the target control data range representation vector corresponding to the current data range representation vector. After obtaining the commonality measurement coefficients corresponding to each specification level, the computer system finds the target commonality measurement coefficient therefrom. For example, assuming there are three specification levels: high specification level, medium specification level, and low specification level, the computer system calculates the commonality measurement coefficients of the current data range representation vector and the control data range representation vectors corresponding to these three specification levels as d_1 (corresponding to the high specification level), d_2 (corresponding to the medium specification level), and d_3 (corresponding to the low specification level). If d_2 is the minimum value of the three coefficients, then d_2 is the target commonality measurement coefficient.

[0069] The computer system determines the control data range characterization vector corresponding to the target commonality measurement coefficient as the target control data range characterization vector corresponding to the current data range characterization vector. Continuing with the above example, if d_2 is the target commonality measurement coefficient, then the control data range characterization vector corresponding to the medium specification level is determined as the target control data range characterization vector corresponding to the current data range characterization vector. This means that among the control data range characterization vectors corresponding to all specification levels, the current data range characterization vector and the control data range characterization vector corresponding to the medium specification level are most similar in characteristics, thereby establishing a corresponding relationship between the two. This corresponding relationship is of great significance for the subsequent determination of the data range specification level of the target cost consumption event data range, and is one of the key steps in the entire data processing process to accurately evaluate the cost consumption specification level. In this way, the computer system can accurately find the control data range characterization vector corresponding to the specification level that best matches the target cost consumption event data range in a complex engineering construction data environment based on the relationship between vectors, providing a basis for further cost consumption specification level determination.

[0070] As an implementation method, the standard level includes a cost consumption standard level and a cost consumption non-standard level. Based on this, step S300, according to the commonality measurement coefficients of the current data range representation vector and the reference data range representation vectors corresponding to each standard level, determines the target reference data range representation vector corresponding to the current data range representation vector in the reference data range representation vectors corresponding to each standard level, which may include: Step S300A: determining the vector distance between the current data range characterization vector and the reference data range characterization vector corresponding to the cost consumption specification level, and obtaining the cost consumption specification commonality measurement coefficient; Step S300B: determining the vector distance between the current data range characterization vector and the comparison data range characterization vector corresponding to the cost consumption irregularity level, and obtaining the cost consumption irregularity commonality measurement coefficient; Step S300C: Based on the loss between the common measurement coefficient of cost consumption standardization and the common measurement coefficient of cost consumption non-standardization, determine the target control data range representation vector corresponding to the current data range representation vector from the control data range representation vector corresponding to the cost consumption standardization level and the control data range representation vector corresponding to the cost consumption non-standardization level.

[0071] In the embodiment S300A of the present application, the computer system determines the vector distance between the current data range characterization vector and the control data range characterization vector corresponding to the cost consumption specification level, and obtains the cost consumption specification commonality measurement coefficient. The current data range characterization vector is extracted from the target cost consumption event data range, and it comprehensively reflects the characteristics of the target cost consumption event related data. For example, in a construction project, if the target cost consumption event is the use of steel bars in a certain structure, the current data range characterization vector, for example, contains the purchase price range of steel bars, the usage ratio of steel bars of different specifications, and the steel bar processing loss rate. The control data range characterization vector corresponding to the cost consumption specification level is a vector formed after feature extraction based on various data related to the use of steel bars (such as the price range, standard usage, standard processing loss, etc. of purchasing steel bars according to high quality standards) under the set cost consumption specification standard.

[0072] In order to determine the vector distance, the computer system can use a variety of technical means, such as the Euclidean distance calculation method. Suppose the current data range representation vector is , the corresponding control data range representation vector of the cost consumption standard level is , the Euclidean distance formula is , this distance value is a reflection of the commonality measurement coefficient of cost consumption specification. The smaller the distance, the more similar the characteristics of the current data range representation vector and the control data range representation vector corresponding to the cost consumption specification level are, that is, the higher the commonality measurement coefficient of cost consumption specification is.

[0073] In step S300B, the computer system determines the vector distance between the current data range characterization vector and the reference data range characterization vector corresponding to the cost consumption irregularity level, and obtains the cost consumption irregularity commonality measurement coefficient. The reference data range characterization vector corresponding to the cost consumption irregularity level is a vector extracted according to the relatively loose or non-compliant cost consumption situation (such as the use of lower quality but cheaper steel bars, the use of overspending or large processing losses, etc.). The above-mentioned Euclidean distance formula is also used to calculate the vector distance. Assume that the current data range characterization vector is still , the control data range representation vector corresponding to the non-standard level of cost consumption is , the calculated distance , this distance value is the commonality measurement coefficient of cost consumption non-standardization.

[0074] In step S300C, the computer system determines the target control data range representation vector corresponding to the current data range representation vector from the control data range representation vector corresponding to the cost consumption standard level and the control data range representation vector corresponding to the cost consumption non-standard level based on the loss between the common measurement coefficient of the cost consumption standard and the common measurement coefficient of the cost consumption non-standard level.

[0075] The loss here is an indicator to measure the difference between the cost consumption standard commonality measurement coefficient and the cost consumption non-standard commonality measurement coefficient. For example, the loss can be simply expressed by the absolute value of the difference between the two. Let the loss be L, then L=|cost consumption standard commonality measurement coefficient-cost consumption non-standard commonality measurement coefficient|.

[0076] Assuming that in the case of steel bar use in construction projects, if the loss L is greater than the critical loss value (this critical loss value is a pre-set threshold value used to distinguish between two situations), then the computer system will compare the cost consumption standard commonality measurement coefficient and the cost consumption non-standard commonality measurement coefficient. If the cost consumption standard commonality measurement coefficient is greater than the cost consumption non-standard commonality measurement coefficient, this means that the current data range representation vector and the control data range representation vector corresponding to the cost consumption standard level are closer in characteristics, so the computer system determines the control data range representation vector corresponding to the cost consumption standard level as the target control data range representation vector corresponding to the current data range representation vector. Conversely, if the cost consumption non-standard commonality measurement coefficient is greater than the cost consumption standard commonality measurement coefficient, the control data range representation vector corresponding to the cost consumption non-standard level is determined as the target control data range representation vector corresponding to the current data range representation vector.

[0077] However, if the loss L is not greater than the critical loss value, it means that the current data range characterization vector and the control data range characterization vector corresponding to the cost consumption standard level and the cost consumption non-standard level are not distinguished. In this case, the computer system uses the preset candidate level as the data range standard level of the target cost consumption event data range. This preset candidate level can be a default level pre-set according to the actual engineering situation, such as a medium-level standard level. When such a difficult-to-distinguish situation occurs, this default level is assigned to the data range standard level of the target cost consumption event data range.

[0078] Through the operation of steps S300A-S300C, the computer system can accurately determine the target control data range characterization vector corresponding to the current data range characterization vector under the consideration of the two situations of cost consumption standard level and cost consumption non-standard level, thereby laying the foundation for the subsequent determination of the data range standard level of the target cost consumption event data range, further improving the data processing flow of the entire integrated digital intelligent construction platform, and improving the accuracy of the cost consumption standard level determination in engineering construction. This method based on vector distance and loss judgment adapts to the complexity of engineering construction data and can reasonably determine the target control data range characterization vector in different cost consumption situations.

[0079] As an implementation manner, step S300C, based on the loss between the commonality metric coefficient of the cost consumption standard and the commonality metric coefficient of the cost consumption non-standardization, determines the target comparison data range representation vector corresponding to the current data range representation vector from the comparison data range representation vector corresponding to the cost consumption standard level and the comparison data range representation vector corresponding to the cost consumption non-standardization level, which may include: Step S300C1: determining the loss between the standard commonality measurement coefficient of cost consumption and the non-standard commonality measurement coefficient of cost consumption, and if the loss is greater than the critical loss value, comparing the standard commonality measurement coefficient of cost consumption and the non-standard commonality measurement coefficient of cost consumption; Step S300C2: if the cost consumption standard commonality measurement coefficient is greater than the cost consumption non-standard commonality measurement coefficient, then the control data range representation vector corresponding to the cost consumption standard level is determined as the target control data range representation vector corresponding to the current data range representation vector; Step S300C3: if the commonality metric coefficient of cost consumption irregularity is greater than the commonality metric coefficient of cost consumption norm, the control data range representation vector corresponding to the cost consumption irregularity level is determined as the target control data range representation vector corresponding to the current data range representation vector; Step S300C4: If the loss is not greater than the critical loss value, the preset candidate level is used as the data range specification level of the target cost consumption event data range.

[0080] In the embodiment S300C1 of the present application, the computer system determines the loss between the standard commonality measurement coefficient of cost consumption and the non-standard commonality measurement coefficient of cost consumption. If the loss is greater than the critical loss value, the standard commonality measurement coefficient of cost consumption and the non-standard commonality measurement coefficient of cost consumption are compared.

[0081] The cost consumption standard commonality measurement coefficient and the cost consumption non-standard commonality measurement coefficient are indicators that measure the similarity between the current data range representation vector and the control data range representation vector corresponding to the cost consumption standard level and the cost consumption non-standard level, respectively. For example, in a construction project, it is assumed that the current data range representation vector is a vector representation of the cost of concrete pouring, which contains processed feature information of factors such as price fluctuations of concrete raw materials, the relationship between transportation distance and cost, and the cost of using pouring equipment. The control data range representation vector corresponding to the cost consumption standard level, for example, contains the cost information related to concrete pouring under high standard specifications, such as the price range of high-quality raw materials, the cost corresponding to short-distance transportation (to ensure quality), and the cost of efficient and accurate pouring equipment. The control data range representation vector corresponding to the cost consumption non-standard level, for example, contains the cost information corresponding to the use of lower quality raw materials, longer transportation distance (for example, which will affect the quality of concrete but the cost is low), and low-performance pouring equipment.

[0082] The computer system calculates the loss between these two commonality metric coefficients. One way to calculate it can be the absolute value of the difference. Suppose the commonality metric coefficient of the cost consumption specification is , the commonality coefficient of cost consumption non-standardization is ,loss The critical loss value is a pre-set threshold value used to determine whether the difference between two commonality measurement coefficients is significant enough. For example, this critical loss value can be derived based on a large amount of historical engineering data and empirical analysis. Assume that the critical loss value is set to 0.5. If the loss L calculated by the computer system is greater than 0.5, it means that there is a significant difference between the cost consumption standard commonality measurement coefficient and the cost consumption non-standard commonality measurement coefficient. At this time, the computer system further compares the two coefficients.

[0083] In step S300C2, if the cost consumption standard commonality measurement coefficient is greater than the cost consumption non-standard commonality measurement coefficient, the control data range representation vector corresponding to the cost consumption standard level is determined as the target control data range representation vector corresponding to the current data range representation vector.

[0084] Continuing with the example of concrete pouring in the construction project, if the computer system calculates the cost consumption standard commonality metric coefficient Greater than the cost consumption irregular commonality measurement coefficient , which indicates that the current data range characterization vector and the control data range characterization vector corresponding to the cost consumption specification level are more similar in characteristics. In other words, from the perspective of the various cost components and their relationships, the current concrete pouring cost-related data is closer to the situation of being implemented in accordance with the specification standards. For example, the price fluctuations of concrete raw materials are within the price range of high-quality materials allowed by the specifications, and the transportation distance and the cost of pouring equipment are also more in line with the specifications. Therefore, the computer system determines the control data range characterization vector corresponding to the cost consumption specification level as the target control data range characterization vector corresponding to the current data range characterization vector, which helps to determine the data range specification level of the target cost consumption event data range in the subsequent determination, and is more inclined to judge it as meeting the specification level.

[0085] In step S300C3, if the commonality metric coefficient of the cost consumption irregularity is greater than the commonality metric coefficient of the cost consumption standardization, the control data range representation vector corresponding to the cost consumption irregularity level is determined as the target control data range representation vector corresponding to the current data range representation vector.

[0086] Taking concrete pouring as an example, if the computer system finds that the cost consumption is not standardized, the common measurement coefficient Greater than the cost consumption standard commonality measurement coefficient , which means that the current data range characterization vector is closer in characteristics to the control data range characterization vector corresponding to the non-standard level of cost consumption. For example, although the price of raw materials is low, the quality is not up to standard, the transportation distance is long and the protective measures during transportation are insufficient (resulting in, for example, affecting the quality of concrete but low cost), the performance of the pouring equipment is low and the maintenance cost is unreasonable, etc., which makes the overall cost situation more consistent with the characteristics of the non-standard level. Therefore, the computer system determines the control data range characterization vector corresponding to the non-standard level of cost consumption as the target control data range characterization vector corresponding to the current data range characterization vector, which will tend to be determined as a situation related to the non-standard level when subsequently determining the data range standard level of the target cost consumption event data range.

[0087] In step S300C4, if the loss is not greater than the critical loss value, the preset candidate level is used as the data range specification level of the target cost consumption event data range.

[0088] The preset candidate level is a default level pre-set in engineering construction. When the difference between the cost consumption standard commonality measurement coefficient and the cost consumption non-standard commonality measurement coefficient is not obvious (that is, the loss is not greater than the critical loss value), it is difficult for the computer system to accurately determine whether the current data range representation vector is more inclined to the cost consumption standard level or the control data range representation vector corresponding to the non-standard level. For example, in a construction project, there is such a situation that the price of concrete raw materials is between the price range corresponding to the standard level and the non-standard level, and the transportation distance and the cost of pouring equipment do not obviously tend to standard or non-standard characteristics. At this time, in order to avoid wrong judgments, the computer system adopts a preset candidate level. This preset candidate level can be set according to the overall situation of the project, the experience of previous similar projects, or the general default standard in the industry, for example, it can be set to a medium standard level. In this way, when the data characteristics are unclear and the tendency to standard or non-standard levels cannot be accurately determined, the preset candidate level is used as the data range standard level of the target cost consumption event data range, which ensures that in a complex data environment, a relatively reasonable judgment can still be made on the standard level of the target cost consumption event data range, and a compromise processing method is provided for determining the cost consumption standard level corresponding to the entire engineering construction data set.

[0089] Through the operations of steps S300C1-S300C4, the computer system can accurately determine the target comparison data range representation vector corresponding to the current data range representation vector in the comparison data range representation vectors corresponding to the cost consumption standard level and the cost consumption non-standard level based on the relationship between the common measurement coefficients of the cost consumption standards and the common measurement coefficients of the cost consumption non-standard level, thereby further improving the data processing flow of the integrated digital intelligent construction platform and improving the accuracy and rationality of the cost consumption standard level determination.

[0090] As an implementation method, the method of the present application may further include: Step S1: loading the comparison data range representation vectors corresponding to the target cost consumption event data range and each specification level into the cost consumption specification level determination network; Step S2: Based on the cost consumption specification level, the representation vector extraction component in the network is determined to extract the current data range representation vector corresponding to the target cost consumption event data range, and based on the commonality measurement coefficients of the current data range representation vector and the reference data range representation vectors corresponding to each specification level, the target reference data range representation vector corresponding to the current data range representation vector is determined in the reference data range representation vectors corresponding to each specification level, and the specification level corresponding to the target reference data range representation vector is determined as the data range specification level corresponding to the target cost consumption event data range.

[0091] In an embodiment of the present application, steps S1-S2 are executed by a computer system. These two steps mainly focus on comparing the target cost consumption event data range with the data range characterization vectors corresponding to each specification level to determine the data range specification level of the target cost consumption event data range.

[0092] In step S1, the computer system loads the target cost consumption event data range and the reference data range representation vector corresponding to each specification level into the cost consumption specification level determination network. The target cost consumption event data range is a data set obtained through a series of previous operations (such as detecting the target cost consumption event from the construction data set of the project to be processed), which contains various types of information related to the specific cost consumption event. For example, in a construction project, if the target cost consumption event is an external wall insulation project, this data range includes, for example, the purchase cost of insulation materials, the working hours of construction personnel, the rental cost of construction equipment, etc. The reference data range representation vector is a vector obtained by extracting and integrating the representation vectors of the reference cost consumption event data range under each specification level, and these vectors reflect the characteristics of cost consumption event data under different specification levels. For example, the reference data range representation vector under a high specification level reflects the high quality requirements for external wall insulation materials (such as high insulation performance indicators and high fire protection levels), high qualification requirements for construction personnel (thus high labor costs), advanced construction equipment (high equipment rental costs) and other related data characteristics; the vector under a low specification level reflects relatively low requirements and cost characteristics. The computer system loads these data and vectors into the cost consumption specification level determination network. This network is a model structure specifically used to determine the cost consumption specification level based on input data. It can be built based on a neural network and has the ability to analyze and process input data and output relevant results.

[0093] In step S2, based on the cost consumption specification level, the representation vector extraction component in the network is determined to extract the current data range representation vector corresponding to the target cost consumption event data range, and based on the commonality measurement coefficients of the current data range representation vector and the reference data range representation vectors corresponding to each specification level, the target reference data range representation vector corresponding to the current data range representation vector is determined in the reference data range representation vectors corresponding to each specification level, and the specification level corresponding to the target reference data range representation vector is determined as the data range specification level corresponding to the target cost consumption event data range.

[0094] The representation vector extraction component in the cost consumption specification level determination network is responsible for feature extraction of the target cost consumption event data range to obtain the current data range representation vector. For example, this component can be implemented using the principal component analysis (PCA) algorithm. Assuming that the data in the target cost consumption event data range can form a matrix X (rows represent different data records, and columns represent different variables, such as the various components of the cost), the PCA algorithm first calculates the covariance matrix C of X, and then solves the eigenvalues ​​and eigenvectors of C. Select the first k eigenvectors (k is less than the number of original variables) according to the size of the eigenvalue, and project the original data into the space formed by these k eigenvectors. The projection result is the current data range representation vector.

[0095] After obtaining the current data range representation vector, the computer system determines the commonality metric coefficient between it and the reference data range representation vector corresponding to each standard level. This can be achieved by calculating the vector distance, such as the Euclidean distance. Suppose the current data range representation vector is , the reference data range representation vector corresponding to a certain standard level is , the Euclidean distance formula is , this distance value can be used as a representation of the commonality measurement coefficient. The computer system calculates such a distance for the control data range representation vector corresponding to each specification level, and then finds the control data range representation vector with the smallest distance (that is, the commonality measurement coefficient best meets a certain standard). This vector is the target control data range representation vector corresponding to the current data range representation vector. Finally, the computer system determines the specification level corresponding to the target control data range representation vector as the data range specification level of the target cost consumption event data range. For example, if the target control data range representation vector corresponds to a high specification level, then the data range specification level of the target cost consumption event data range of the exterior wall insulation project is determined to be a high specification level. This helps to accurately judge the degree of standardization of the target cost consumption event in terms of cost consumption, and provides an important basis for determining the cost consumption specification level corresponding to the entire engineering construction data set.

[0096] As an implementation method, the cost consumption standard level determination network is adjusted through the following process: Step S01: obtaining target cost consumption event data range samples and corresponding data range sample level prior information; Step S02: Load the target cost consumption event data range sample and the control data range representation vector corresponding to each standard level into the cost consumption standard level determination network to be trained to perform data range standard level detection, and obtain the sample data range standard level corresponding to the target cost consumption event data range sample; Step S03: optimizing the cost consumption standard level determination network to be trained according to the sample data range standard level and the data range sample level prior information to obtain an optimized cost consumption standard level determination network; Step S04: Use the optimized cost consumption specification level determination network as the cost consumption specification level determination network to be trained, and jump to the step of obtaining the target cost consumption event data range sample and the corresponding data range sample level prior information to perform a loop operation until the preset loop stop condition is met to obtain the cost consumption specification level determination network.

[0097] In step S01 of the embodiment of the present application, the computer system obtains the target cost consumption event data range sample and the corresponding data range sample level prior information. The target cost consumption event data range sample is a representative target cost consumption event-related data set selected from the engineering construction data. For example, in a construction project, if concrete pouring is used as a target cost consumption event, then the target cost consumption event data range sample, for example, includes data on price fluctuations of concrete raw materials, labor costs under different pouring methods, pouring equipment used and its maintenance costs, and the cost of transporting concrete. These data reflect the relevant situation of the cost consumption event of concrete pouring in a specific engineering scenario. The data range sample level prior information is the specification level information known in advance about the data range samples. For example, according to the design requirements, construction standards, and industry specifications of the project, it can be predetermined that the target cost consumption event data range sample of a certain concrete pouring belongs to a high specification level (for example, the cost consumption under the conditions of using high-quality raw materials, high-precision construction equipment, and highly skilled construction personnel corresponds to a high specification level), or a medium specification level, a low specification level, etc.

[0098] In step S02, the computer system loads the target cost consumption event data range samples and the control data range characterization vectors corresponding to each specification level into the cost consumption specification level determination network to be trained for data range specification level detection, and obtains the sample data range specification level corresponding to the target cost consumption event data range samples. The cost consumption specification level determination network to be trained is an initial network model that has not been fully trained. Its structure can be constructed based on a neural network, including an input layer, a hidden layer, and an output layer. The input layer receives the target cost consumption event data range samples and the control data range characterization vectors corresponding to each specification level. For example, for the target cost consumption event data range samples of concrete pouring, and the control data range characterization vectors corresponding to different specification levels such as high, medium, and low (these control data range characterization vectors are extracted and integrated according to the cost consumption characteristics related to concrete pouring under different specification levels, such as the control data range characterization vector under the high specification level contains the feature representation of relevant cost factors such as high-quality raw materials and efficient equipment), they are input together into the cost consumption specification level determination network to be trained.

[0099] Inside the network, the input data is processed through the hidden layer. The neurons in the hidden layer can use different activation functions, such as the ReLU function (for input x, ReLU(x)=max(0,x)) to introduce nonlinear factors, so as to better handle complex data relationships. After being processed by the hidden layer, the sample data range specification level corresponding to the target cost consumption event data range sample is obtained in the output layer. For example, the output result is a numerical value or code representing the specification level. If 1 is set to represent a high specification level, 2 to represent a medium specification level, and 3 to represent a low specification level, then the network output 1 means that the computer system determines that the sample data range specification level corresponding to the target cost consumption event data range sample is a high specification level.

[0100] In step S03, the computer system optimizes the cost consumption specification level determination network to be trained according to the sample data range specification level and the data range sample level prior information to obtain the optimized cost consumption specification level determination network. In this step, the computer system adjusts the parameters of the network according to the difference between the previously known correct (data range sample level prior information) and the (sample data range specification level) predicted by the network. For example, if the data range sample level prior information indicates that the target cost consumption event data range sample of concrete pouring belongs to a high specification level, and the sample data range specification level output by the network is a medium specification level, there is an error.

[0101] Computer systems can be optimized using the backpropagation algorithm. Assuming that the output of the network is y (sample data range specification level), the correct result is t (data range sample level prior information), a loss function can be defined, such as the mean square error (MSE) function (Here n represents the number of samples, and in the case of a single sample n=1). By calculating the gradient of the loss function with respect to the parameters of each layer in the network (such as weight w and bias b) and , and then according to a certain learning rate (For example = 0.01) to update the parameters, that is Through multiple such iterative update processes, the output result of the network is closer to the correct level, thereby obtaining an optimized cost consumption specification level determination network.

[0102] In step S04, the computer system uses the optimized cost consumption specification level determination network as the cost consumption specification level determination network to be trained, and jumps to the step of obtaining the target cost consumption event data range sample and the corresponding data range sample level prior information to perform a loop operation until the preset loop stop condition is met to obtain the cost consumption specification level determination network. The preset loop stop condition can be in various forms, such as reaching a certain number of iterations (such as 1000 times), or when the accuracy of the network on the verification data set reaches a certain threshold (such as 95%), etc.

[0103] Each cycle will use a new target cost consumption event data range sample or resample the original sample, and continuously adjust the network parameters to continuously improve the network performance. For example, in each cycle, new concrete pouring target cost consumption event data range samples and their corresponding correct level information can be obtained from different construction engineering cases to continue training the network. After multiple such loop operations, the final cost consumption specification level determination network can accurately output the correct data range specification level based on the input target cost consumption event data range and the control data range representation vector corresponding to each specification level. This calibrated network can effectively determine the data range specification level of the target cost consumption event data range in the subsequent engineering construction data processing, and then provide a reliable basis for determining the cost consumption specification level corresponding to the entire engineering construction data set, which is helpful to better manage cost consumption, ensure engineering quality and comply with relevant specifications and standards during the engineering construction process.

[0104] An embodiment of the present application provides a computer system, including a memory and a processor, wherein the memory stores a computer program that can be executed on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0105] The embodiment of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium can be transient or non-transient.

[0106] An embodiment of the present application provides a computer program, including a computer-readable code. When the computer-readable code is run in a computer device, a processor in the computer device executes some or all of the steps for implementing the above method.

[0107] The embodiment of the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and when the computer program is read and executed by a computer, some or all of the steps in the above method are implemented. The computer program product can be implemented specifically by hardware, software or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium, and in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK) and the like.

[0108] It should be noted here that the description of the various embodiments above tends to emphasize the differences between the various embodiments, and the same or similar aspects can be referenced to each other. The description of the above device, storage medium, computer program and computer program product embodiments is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the embodiments of the device, storage medium, computer program and computer program product of this application, please refer to the description of the method embodiment of this application for understanding.

[0109] Figure 2 A hardware entity diagram of a computer system provided in an embodiment of the present application is shown in FIG. Figure 2 As shown, the hardware entity of the computer system 1000 includes: a processor 1001 and a memory 1002, wherein the memory 1002 stores a computer program that can be run on the processor 1001, and the processor 1001 implements the steps in the method of any of the above embodiments when executing the program.

[0110] The memory 1002 stores computer programs that can be run on the processor. The memory 1002 is configured to store instructions and applications executable by the processor 1001. It can also cache data to be processed or processed by the processor 1001 and various modules in the computer system 1000 (for example, image data, audio data, voice communication data, and video communication data). This can be achieved through flash memory (FLASH) or random access memory (Random Access Memory, RAM).

[0111] When the processor 1001 executes the program, the steps of the data processing method of any one of the above-mentioned integrated digital intelligent construction platforms are implemented. The processor 1001 generally controls the overall operation of the computer system 1000.

[0112] An embodiment of the present application provides a computer storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the steps of the data processing method of the integrated digital intelligent construction platform of any of the above embodiments.

[0113] It should be noted here that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For the technical details not disclosed in the storage medium and device embodiments of the present application, please refer to the description of the method embodiments of the present application for understanding. The above processor can be at least one of a target application integrated circuit (Application Specific Integrated Circuit, ASIC), a digital signal processor (Digital Signal Processor, DSP), a digital signal processing device (Digital Signal Processing Device, DSPD), a programmable logic device (Programmable Logic Device, PLD), a field programmable gate array (Field Programmable Gate Array, FPGA), a central processing unit (Central Processing Unit, CPU), a controller, a microcontroller, and a microprocessor. It can be understood that the electronic device that realizes the above processor function can also be other, and the embodiments of the present application are not specifically limited.

[0114] The above-mentioned computer storage medium / memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory (Flash Memory), a magnetic surface memory, an optical disk, or a compact disc read-only memory (CD-ROM) and the like; it can also be various terminals including one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0115] It should be understood that the "one embodiment" or "one embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in one embodiment" appearing in various places throughout the specification may not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in various embodiments of the present application, the size of the sequence number of each step / process mentioned above does not mean the order of execution, and the execution order of each step / process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present application. The above-mentioned sequence number of the embodiment of the present application is only for description and does not represent the advantages and disadvantages of the embodiment. It should be noted that, in this article, the term "include", "comprise" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or device. Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0116] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0117] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0118] In addition, all functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may be a separate unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0119] A person skilled in the art can understand that all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: mobile storage devices, read-only memories (ROM), magnetic disks or optical disks, etc., various media that can store program codes.

[0120] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0121] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A data processing method for an integrated digital intelligent construction platform, characterized in that: The method includes: obtaining a data set of engineering construction to be processed, detecting a target cost consumption event in the data set of engineering construction to be processed, and obtaining a data range of the target cost consumption event; extracting a current data range representation vector corresponding to the target cost consumption event data range, and obtaining a benchmark control data range representation vector corresponding to each specification level, wherein the benchmark control data range representation vector is obtained by extracting the representation vectors of each control cost consumption event data range of the corresponding specification level, respectively, and obtaining each control cost consumption event data range representation vector, and integrating each control cost consumption event data range representation vector; determining a target control data range representation vector corresponding to the current data range representation vector in the control data range representation vectors corresponding to each specification level according to the commonality measurement coefficients of the current data range representation vector and the control data range representation vectors corresponding to each specification level; determining the specification level corresponding to the target control data range representation vector as the data range specification level corresponding to the target cost consumption event data range, and determining the cost consumption specification level corresponding to the engineering construction data set to be processed according to the data range specification level of the target cost consumption event data range.

2. The method according to claim 1, characterized in that: Before obtaining the engineering construction data set to be processed, detecting the target cost consumption events in the engineering construction data set to be processed, and obtaining the target cost consumption event data range, the method also includes: obtaining each comparison cost consumption event data range corresponding to a preset specification level; extracting a representation vector for each comparison cost consumption event data range to obtain a representation vector for each comparison cost consumption event data range; integrating the representation vectors for each comparison cost consumption event data range to obtain a benchmark comparison data range representation vector corresponding to the preset specification level.

3. The method according to claim 2, characterized in that The integrating the data range representation vectors of the various control cost consumption events to obtain the benchmark control data range representation vector corresponding to the preset specification level includes: combining the data range representation vectors of the various control cost consumption events to obtain a combined representation vector; performing feature extraction based on the combined representation vector to obtain the benchmark control data range representation vector corresponding to the preset specification level.

4. The method according to claim 2, characterized in that: The method also includes: loading the data ranges of each control cost consumption event into a benchmark representation vector extraction network; extracting representation vectors of each control cost consumption event data range based on a representation vector extraction component in the benchmark representation vector extraction network to obtain representation vectors of each control cost consumption event data range; integrating the representation vectors of each control cost consumption event data range based on a representation vector integration component in the benchmark representation vector extraction network to obtain a benchmark control data range representation vector corresponding to the preset specification level.

5. The method according to claim 4, characterized in that The benchmark representation vector extraction network is calibrated by the following process: obtaining samples of each data set and corresponding sample level prior information, and loading the samples of each data set into the data set detection network to be trained; Extracting representation vectors from each of the data set samples based on the representation vector extraction component to be trained in the data set detection network to be trained, to obtain representation vectors for each of the data set samples; Based on the to-be-trained representation vector integration component in the to-be-trained data set detection network, the representation vectors of the samples of each data set are integrated to obtain a sample integrated representation vector; based on the to-be-trained level mapping component in the to-be-trained data set detection network, the level mapping of the sample integrated representation vector is performed to obtain a sample cost consumption standard level; according to the sample level prior information and the sample cost consumption standard level, the to-be-trained data set detection network is optimized to obtain an optimized level mapping network; the optimized level mapping network is used as the to-be-trained data set detection network, and the steps of jumping to obtaining each data set sample and the corresponding sample level prior information, and loading each data set sample into the to-be-trained data set detection network are looped until a preset loop stop condition is met to obtain a target level mapping network; The reference representation vector extraction network is obtained according to the target representation vector extraction component and the target representation vector integration component in the target level mapping network.

6. The method according to claim 1, characterized in that The detecting of the target cost consumption event in the engineering construction data set to be processed and obtaining the target cost consumption event data range includes: extracting the data set filter representation vector of the engineering construction data set to be processed, performing normalization processing based on the data set filter representation vector, and obtaining a normalized representation vector; activating the normalized representation vector to obtain an activation representation vector, performing data range extraction based on the activation representation vector, and obtaining the target cost consumption event data range; determining the target control data range representation vector corresponding to the current data range representation vector in the control data range representation vectors corresponding to the respective specification levels based on the commonality measurement coefficients of the current data range representation vector and the control data range representation vectors corresponding to the respective specification levels, includes: determining the vector distance between the current data range representation vector and the control data range representation vectors corresponding to the respective specification levels, and obtaining each commonality measurement coefficient; determining the target commonality measurement coefficient among the respective commonality measurement coefficients, and determining the control data range representation vector corresponding to the target commonality measurement coefficient as the target control data range representation vector corresponding to the current data range representation vector.

7. The method according to claim 1, characterized in that The standard level includes a cost consumption standard level and a cost consumption non-standard level; the determining, based on the commonality measurement coefficients of the current data range representation vector and the control data range representation vectors corresponding to each standard level, a target control data range representation vector corresponding to the current data range representation vector in the control data range representation vectors corresponding to each standard level includes: determining the vector distance between the current data range representation vector and the control data range representation vector corresponding to the cost consumption standard level to obtain the cost consumption standard commonality measurement coefficient; determining the vector distance between the current data range representation vector and the control data range representation vector corresponding to the cost consumption non-standard level to obtain the cost consumption non-standard commonality measurement coefficient; determining, based on the loss between the cost consumption standard commonality measurement coefficient and the cost consumption non-standard commonality measurement coefficient, a target control data range representation vector corresponding to the current data range representation vector in the control data range representation vector corresponding to the cost consumption standard level and the control data range representation vector corresponding to the cost consumption non-standard level.

8. The method according to claim 7, characterized in that The method of determining the target control data range representation vector corresponding to the current data range representation vector in the control data range representation vector corresponding to the cost consumption standard level and the control data range representation vector corresponding to the cost consumption non-standard level based on the loss between the cost consumption standard commonality measurement coefficient and the cost consumption non-standardization commonality measurement coefficient, includes: determining the loss between the cost consumption standard commonality measurement coefficient and the cost consumption non-standardization commonality measurement coefficient, and if the loss is greater than a critical loss value, comparing the cost consumption standard commonality measurement coefficient and the cost consumption non-standardization commonality measurement coefficient; if the cost consumption standard commonality measurement coefficient is greater than the cost consumption non-standardization commonality measurement coefficient, determining the control data range representation vector corresponding to the cost consumption standard level as the target control data range representation vector corresponding to the current data range representation vector; if the cost consumption non-standardization commonality measurement coefficient is greater than the cost consumption standard commonality measurement coefficient, determining the control data range representation vector corresponding to the cost consumption non-standardization level as the target control data range representation vector corresponding to the current data range representation vector; if the loss is not greater than the critical loss value, using a preset candidate level as the data range specification level of the target cost consumption event data range.

9. The method according to claim 1, characterized in that: The method also includes: loading the comparison data range representation vectors corresponding to the target cost consumption event data range and the various specification levels into the cost consumption specification level determination network; extracting the current data range representation vector corresponding to the target cost consumption event data range based on the representation vector extraction component in the cost consumption specification level determination network, and determining the target comparison data range representation vector corresponding to the current data range representation vector in the comparison data range representation vectors corresponding to the various specification levels according to the commonality measurement coefficients of the current data range representation vector and the comparison data range representation vectors corresponding to the various specification levels, and determining the specification level corresponding to the target comparison data range representation vector as the data range specification level corresponding to the target cost consumption event data range; wherein, the cost consumption specification level determination network is calibrated through the following process: obtaining the target cost consumption Event data range samples and corresponding data range sample level prior information; loading the target cost consumption event data range samples and the control data range representation vectors corresponding to each specification level into the cost consumption specification level determination network to be trained for data range specification level detection, and obtaining the sample data range specification level corresponding to the target cost consumption event data range samples; optimizing the cost consumption specification level determination network to be trained according to the sample data range specification level and the data range sample level prior information, and obtaining the optimized cost consumption specification level determination network; using the optimized cost consumption specification level determination network as the cost consumption specification level determination network to be trained, and jumping to the step of obtaining the target cost consumption event data range samples and the corresponding data range sample level prior information to perform a loop operation until the preset loop stop condition is met, and obtaining the cost consumption specification level determination network.

10. A computer system comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, wherein: When the processor executes the program, the steps in the method according to any one of claims 1 to 9 are implemented.

Images