An industrial multi-project multi-level indicator calculation method, system, device and storage medium

Abstract

The present invention relates to the technical field of business data processing, and discloses a method, system, device and storage medium for calculating industrial multi-project and multi-level indicators, including: collecting multi-dimensional data from different industrial projects and levels, and preprocessing; using the preprocessed data to establish an analysis model, and analyzing the dependencies between various data indicators; based on the analyzed dependencies, forming an initial hierarchical structure, and automatically adjusting the hierarchical structure according to data and business requirements; optimizing the calculation path selection and resource allocation. By dynamically analyzing the dependencies between data, forming and automatically adjusting the hierarchical structure, the system can flexibly respond to complex industrial data processing needs. The calculation path selection and resource allocation are optimized, and the efficiency of data processing and the accuracy of decision-making are improved. The intelligence and practicality of industrial multi-project and multi-level indicator calculation are significantly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of business data processing, and specifically to an industrial multi-project and multi-level indicator calculation method, system, equipment and storage medium. Background Art

[0002] In modern industrial environments, efficient data processing and decision support systems play a vital role in improving production efficiency and optimizing resource management. With the improvement of industrial automation and intelligence, more and more companies are beginning to rely on advanced data analysis methods to process and analyze complex data sets from different projects and levels. In this context, industrial multi-project and multi-level indicator calculation technologies have developed rapidly, which enable companies to extract valuable information from large-scale data sets to support more accurate decision making. Traditional data processing methods usually rely on batch processing mode and adopt static data analysis frameworks, which often show the shortcomings of slow processing speed and poor adaptability when processing dynamically changing industrial data.

[0003] The main shortcoming of existing technologies is that their static data processing mode is difficult to adapt to rapidly changing production conditions and market demands. Especially in a multi-project and multi-level environment, traditional methods are often unable to effectively identify and process complex dependencies between data, resulting in data analysis results that cannot fully reflect the actual business situation. In addition, existing technologies lack flexibility and automation in data preprocessing, indicator calculation, and resource allocation optimization, making the data processing process time-consuming and inefficient, and making it difficult to achieve data-driven dynamic decision support. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are the problems of insufficient data dependency analysis, inability to dynamically adjust the hierarchical structure, and difficulty in achieving intelligent optimization of calculation path selection and resource allocation in the process of industrial multi-project and multi-level data processing.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for calculating industrial multi-project and multi-level indicators, comprising:

[0007] Collect multi-dimensional data from different industrial projects and levels and perform pre-processing.

[0008] Using the preprocessed data, an analysis model is established to analyze the dependencies between various data indicators.

[0009] Based on the analyzed dependencies, the dependencies between the indicators are extracted from the analysis model.

[0010] Dependency includes the strength of dependency and the direction of dependency.

[0011] The strength of the dependency relationship includes the correlation coefficient and mutual information.

[0012] The directionality of dependency includes the direction of causality.

[0013] The dependency data is structured to generate a dependency matrix, which is expressed as follows:

[0014]

[0015] Among them, R represents the dependency matrix, R ij Indicates the degree of dependence of indicator i on indicator j. Each R ij It is a composite value that combines the dependence strength and direction, which can be further expressed as:

[0016] R ij =[ρ ij ,I ij ,D ij ]

[0017] Among them, ρ ij Represents the correlation coefficient between index i and index j, I ij represents the mutual information between index i and index j, D ij It represents the causal directionality of indicator i to indicator j. Based on the dependency matrix, the indicators are clustered using agglomerative hierarchical clustering, and the indicators with strong dependencies are aggregated in the same level to form an initial hierarchical structure.

[0018] The dependency matrix is ​​set as the adjacency matrix of a weighted graph, where nodes represent indicators and edge weights represent the dependency relationship between indicators, and the degree matrix of the weighted graph is constructed.

[0019] The formula of the degree matrix is ​​expressed as:

[0020]

[0021] in, represents the element of the i-th row and i-th column of the degree matrix of node i, n represents the total number of indicators, R ij represents the degree of dependence of indicator i on indicator j, α represents the adjustment parameter, 1-R ij Represents the inverse of the degree of dependence.

[0022] The Laplace matrix of the weighted graph is defined as the difference between the degree matrix and the dependency matrix. The Laplace matrix is ​​normalized to obtain a symmetric normalized Laplace matrix, which is expressed as follows:

[0023]

[0024] in, represents the symmetric normalized Laplace matrix; D * represents the degree matrix, (D * ) -1 / 2 Denotes the optimized degree matrix D * The inverse square root matrix of * represents the optimized Laplacian matrix.

[0025] Calculate the eigenvalues ​​and eigenvectors of the normalized Laplace matrix, select the eigenvectors of the smallest k eigenvalues, and construct the characteristic matrix. The formula is expressed as:

[0026]

[0027] Among them, U * represents the final feature matrix, argmin represents finding the value that minimizes the objective function, Tr(·) represents the trace operation of the matrix, and U T represents the transposed matrix of matrix U, represents the symmetric normalized Laplace matrix, U represents the process characteristic matrix, U T U = I represents the orthogonal constraint, δ represents the regularization parameter, and k represents the number of eigenvectors in the feature matrix; u i represents the i-th eigenvector of the process characteristic matrix, Represents vector u i The gradient of Represents vector u i Squared norm of the gradient.

[0028] Each row of the feature matrix is ​​regarded as a new data point, and the K-means clustering algorithm is used to divide these data points into k clusters.

[0029] The result of K-means clustering is hierarchical division, where each cluster represents a level, generating an initial hierarchical structure.

[0030] Each level contains a group of indicators assigned by the clustering algorithm, ensuring strong dependencies between these indicators and forming an initial hierarchical structure.

[0031] The hierarchical structure is represented as a hierarchical matrix, and the formula is expressed as:

[0032]

[0033] in, Indicates the membership degree of the i-th indicator in the optimized hierarchical matrix to the k-th level; Represents the characteristic matrix U of the i-th index after optimization * The vector representation in ck represents the center vector of the kth cluster, Represents the feature vector With cluster center c k The square of the Euclidean distance between 2 represents the square of the bandwidth parameter of the Gaussian kernel function, represents the Gaussian kernel function; c l Represents the lth cluster center vector.

[0034] According to data and business requirements, a dynamic threshold τ of dependency strength is set. Only indicator pairs with dependency strength exceeding τ will be considered at the same level.

[0035] The formula for the intensity-dependent dynamic threshold is:

[0036]

[0037] Among them, Θ(τ) represents the dynamic threshold adjustment function of the dependent strength, τ represents the threshold of the dependent strength, R ij represents the dependency matrix; represents the smoothed logistic regression function, η represents the smoothing parameter, Represents the inverse operation of the logistic regression function.

[0038] The value of Θ(τ) is thresholded from 0 to 1.

[0039] Cross-validation and A / B testing are used to verify the hierarchical structure to ensure its rationality and robustness. The formula is expressed as:

[0040]

[0041] in, represents the cross validation loss function, Indicates averaging the results of cross-validation, and n indicates the number of cross-validation folds; represents trace operation, U i represents the feature matrix in the i-th fold, represents the transpose of the feature matrix in the i-th fold, represents the symmetric normalized Laplace matrix, δ represents the regularization parameter, Denotes the feature vector u j The square of the gradient norm of , Θ(τ) represents the intensity-dependent dynamic threshold adjustment function.

[0042] The value threshold of is from 0 to positive infinity. The smaller the value, the better the fitting effect of the hierarchical structure.

[0043] Categorize the verification results at each level. When the values ​​of are all lower than the preset global threshold, the hierarchical structure is determined to have passed the verification and proceed to the next step of adjustment.

[0044] When the verification result of a certain layer exceeds the global threshold, the hierarchical structure is determined to be an item that needs to be adjusted and is readjusted.

[0045] when When the values ​​of are globally close to the threshold, but the dependence strength is greater than the preset upper limit, it is determined that the layer is too dependent on the overall structure, resulting in a single point failure of the system.

[0046] After classifying the verification results of each level and readjusting the hierarchical structure that exceeds the global threshold, the entire hierarchical structure is globally optimized, and the secondary verification mechanism is triggered. The hierarchical structure is re-verified using cross-validation and A / B testing to achieve automatic adjustment of the hierarchical structure.

[0047] Based on the adjusted hierarchical structure, the computing path selection and resource allocation are optimized.

[0048] As a preferred solution of the industrial multi-project and multi-level indicator calculation method described in the present invention, the multi-dimensional data includes production efficiency data, quality control data, energy and resource consumption data, equipment health status data, maintenance prediction and health management data, supply chain optimization data, environmental compliance and sustainability data, safety monitoring data, cost-benefit analysis data, and real-time operation data.

[0049] The preprocessing includes: cleaning the multidimensional data, processing missing values, outliers and deleting duplicate data; converting the multidimensional data and integrating multi-source data; labeling and classifying the multidimensional data and performing data consistency checks.

[0050] As a preferred solution of the industrial multi-project and multi-level indicator calculation method described in the present invention, an analysis model is established, including feature extraction of pre-processed multidimensional data, using a random forest model to learn known dependencies, and combining an unsupervised learning method of association rule mining to discover potential dependencies.

[0051] As a preferred solution of the industrial multi-project and multi-level indicator calculation method described in the present invention, the optimization of calculation path selection and resource allocation includes constructing an optimization model using a multi-objective optimization algorithm to optimize the calculation path selection and resource allocation.

[0052] The formula of the optimization model is expressed as:

[0053]

[0054] Among them, Y represents the result of optimizing the calculation path selection and resource allocation, x represents the decision variable of the calculation path selection, and w i represents the weight of the i-th target, min x represents minimizing the decision variable x, f i (x,θ) represents the function related to the i-th target, c j represents the cost of the jth resource, x j represents the allocation of the jth resource, e -∈h(x) represents an exponential decay term, ∈ represents an optimization parameter, h(x) represents the constraint function, and Γ(α+1) represents the gamma function.

[0055] The value threshold of Y is a positive real number. The larger the value, the lower the efficiency in resource allocation and path selection.

[0056] As a preferred solution of the industrial multi-project and multi-level indicator calculation system described in the present invention, it includes: a data acquisition and processing module, a model establishment and adjustment level module, and a solution optimization module.

[0057] The data acquisition and processing module is used to collect multi-dimensional data from different industrial projects and levels and perform pre-processing.

[0058] The model building and hierarchical adjustment module is used to use preprocessed data to build analysis models, analyze the dependencies between various data indicators, form an initial hierarchical structure based on the analyzed dependencies, and automatically adjust the hierarchical structure according to data and business needs.

[0059] The solution optimization module is used to optimize the calculation path selection and resource allocation based on the adjusted hierarchical structure.

[0060] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a step of an industrial multi-project and multi-level indicator calculation method.

[0061] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of an industrial multi-project and multi-level indicator calculation method.

[0062] Beneficial effects of the present invention: The industrial multi-project multi-level indicator calculation method provided by the present invention forms and automatically adjusts the hierarchical structure by dynamically analyzing the dependencies between data, so that the system can flexibly respond to complex industrial data processing needs. Combined with the multi-objective optimization algorithm, the calculation path selection and resource allocation are optimized, and the efficiency of data processing and the accuracy of decision-making are improved. In addition, the present invention effectively solves the problems of rigid hierarchical structure and inefficient resource allocation in traditional methods, and significantly improves the intelligence and practicality of industrial multi-project multi-level indicator calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0064] Figure 1 An overall flow chart of an industrial multi-project and multi-level indicator calculation method provided for the first embodiment of the present invention. DETAILED DESCRIPTION

[0065] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0066] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides an industrial multi-project multi-level indicator calculation method, comprising:

[0067] S1: Collect multi-dimensional data from different industrial projects and levels and perform pre-processing.

[0068] Multi-dimensional data includes production efficiency data, quality control data, energy and resource consumption data, equipment health status data, maintenance prediction and health management data, supply chain optimization data, environmental compliance and sustainability data, safety monitoring data, cost-benefit analysis data, and real-time operation data.

[0069] The preprocessing includes: cleaning the multidimensional data, processing missing values, outliers and deleting duplicate data; converting the multidimensional data and integrating multi-source data; labeling and classifying the multidimensional data and performing data consistency checks.

[0070] Furthermore, by collecting and preprocessing multi-dimensional data from different industrial projects and levels, the integrity and consistency of the data are ensured, avoiding the interference of data noise, missing values ​​and outliers on subsequent analysis. The preprocessed data is of higher quality, laying a solid foundation for subsequent analysis and model building.

[0071] Furthermore, through a systematic data processing process, the reliability and accuracy of data analysis can be improved, thereby ensuring the reliability of basic data in the entire industrial indicator calculation process, which is crucial for achieving accurate decision-making and resource optimization.

[0072] S2: Use the preprocessed data to establish an analysis model and analyze the dependencies between various data indicators.

[0073] Building an analysis model includes extracting features from preprocessed multidimensional data, using a random forest model to learn known dependencies, and combining unsupervised learning methods for association rule mining to discover potential dependencies.

[0074] Furthermore, advanced machine learning technology can be used to achieve deep mining of industrial data, ensure comprehensive analysis of multi-dimensional data, and dynamically track its changes, providing a scientific basis for forming and adjusting hierarchical structures.

[0075] S3: Based on the analyzed dependencies, an initial hierarchical structure is formed, and the hierarchical structure is automatically adjusted according to data and business requirements.

[0076] Forming the initial hierarchical structure includes extracting the dependencies between indicators from the analysis model.

[0077] Dependency includes the strength of dependency and the direction of dependency.

[0078] The strength of the dependency relationship includes the correlation coefficient and mutual information.

[0079] The directionality of dependency includes the direction of causality.

[0080] The dependency data is structured to generate a dependency matrix, which is expressed as follows:

[0081]

[0082] Among them, R represents the dependency matrix, R ij Indicates the degree of dependence of indicator i on indicator j. Each R ij It is a composite value that combines the dependence strength and direction, which can be further expressed as:

[0083] R ij =[ρ ij ,I ij,D ij ]

[0084] Among them, ρ ij Represents the correlation coefficient between index i and index j, I ij represents the mutual information between index i and index j, D ij Indicates the direction of causal relationship between indicator i and indicator j.

[0085] Based on the dependency matrix, the indicators are clustered using agglomerative hierarchical clustering, and the indicators with strong dependencies are aggregated in the same level to form an initial hierarchical structure.

[0086] The dependency matrix is ​​set as an adjacency matrix of a weighted graph, in which nodes represent indicators and edge weights represent dependency relationships between indicators, and a degree matrix of the weighted graph is constructed.

[0087] The formula of the degree matrix is ​​expressed as:

[0088]

[0089] in, represents the element of the i-th row and i-th column of the degree matrix of node i, n represents the total number of indicators, R ij represents the degree of dependence of indicator i on indicator j, α represents the adjustment parameter, 1-R ij Represents the inverse of the degree of dependence.

[0090] The Laplace matrix of the weighted graph is defined as the difference between the degree matrix and the dependency matrix. The Laplace matrix is ​​normalized to obtain a symmetric normalized Laplace matrix, which is expressed as follows:

[0091]

[0092] in, represents the symmetric normalized Laplace matrix; D * represents the degree matrix, (D * ) -1 / 2 Denotes the optimized degree matrix D * The inverse square root matrix of * represents the optimized Laplacian matrix.

[0093] Calculate the eigenvalues ​​and eigenvectors of the normalized Laplace matrix, select the eigenvectors of the smallest k eigenvalues, and construct the characteristic matrix. The formula is expressed as:

[0094]

[0095] Among them, U *represents the final feature matrix, argmin represents finding the value that minimizes the objective function, Tr(·) represents the trace operation of the matrix, and U T represents the transposed matrix of matrix U, represents the symmetric normalized Laplace matrix, U represents the process characteristic matrix, U T U = I represents the orthogonal constraint, δ represents the regularization parameter, and k represents the number of eigenvectors in the feature matrix; u i represents the i-th eigenvector of the process characteristic matrix, Represents vector u i The gradient of Represents vector u i Squared norm of the gradient.

[0096] Each row of the feature matrix is ​​regarded as a new data point, and the K-means clustering algorithm is used to divide these data points into k clusters.

[0097] The result of K-means clustering is hierarchical division, where each cluster represents a level, generating an initial hierarchical structure.

[0098] Each level contains groups of indicators assigned by the clustering algorithm, ensuring strong dependencies between these indicators.

[0099] The hierarchical structure is represented as a hierarchical matrix, and the formula is expressed as:

[0100]

[0101] in, Indicates the membership degree of the i-th indicator in the optimized hierarchical matrix to the k-th level; Represents the characteristic matrix U of the i-th index after optimization * The vector representation in c k represents the center vector of the kth cluster, Represents the feature vector With cluster center c k The square of the Euclidean distance between 2 represents the square of the bandwidth parameter of the Gaussian kernel function, represents the Gaussian kernel function. c l Represents the lth cluster center vector.

[0102] Automatically adjusting the hierarchical structure includes setting a dynamic threshold τ of the dependency strength, and only pairs of indicators with a dependency strength exceeding τ will be considered as the same level.

[0103] The formula for the intensity-dependent dynamic threshold is:

[0104]

[0105] Among them, Θ(τ) represents the dynamic threshold adjustment function of the dependent strength, τ represents the threshold of the dependent strength, R ij represents the dependency matrix; represents the smoothed logistic regression function, η represents the smoothing parameter, Represents the inverse operation of the logistic regression function.

[0106] The value of Θ(τ) is thresholded from 0 to 1.

[0107] Cross-validation and A / B testing are used to verify the hierarchical structure to ensure its rationality and robustness. The formula is expressed as:

[0108]

[0109] in, represents the cross validation loss function, Indicates averaging the results of cross-validation, and n indicates the number of cross-validation folds; represents trace operation, U i represents the feature matrix in the i-th fold, represents the transpose of the feature matrix in the i-th fold, represents the symmetric normalized Laplace matrix, δ represents the regularization parameter, Denotes the feature vector u j The square of the gradient norm of , Θ(τ) represents the intensity-dependent dynamic threshold adjustment function.

[0110] The value threshold of is from 0 to positive infinity. The smaller the value, the better the fitting effect of the hierarchical structure.

[0111] Categorize the verification results at each level. When the values ​​of are all lower than the preset global threshold, the hierarchical structure is determined to have passed the verification and proceed to the next step of adjustment.

[0112] When the verification result of a certain layer exceeds the global threshold, the hierarchical structure is determined to be an item that needs to be adjusted and is readjusted.

[0113] when When the values ​​of are globally close to the threshold, but the dependence strength is greater than the preset upper limit, it is determined that the layer is too dependent on the overall structure, resulting in a single point failure of the system.

[0114] After classifying the verification results of each level and readjusting the hierarchical structure that exceeds the global threshold, the entire hierarchical structure is globally optimized, and the secondary verification mechanism is triggered to re-verify the hierarchical structure using cross-validation and A / B testing.

[0115] Furthermore, by forming an initial hierarchical structure and automatically adjusting the hierarchical structure according to business needs, the flexibility and adaptability of the system are guaranteed. This process can dynamically respond to changes in data, optimize the hierarchical structure, and make decisions more accurate and efficient.

[0116] Furthermore, through a dynamic adjustment mechanism, the limitations of the traditional static hierarchical structure can be broken, the system's adaptability in complex industrial environments can be enhanced, and it can be ensured that the hierarchical structure can reflect changes in actual business needs and data relationships in real time.

[0117] S4: Based on the adjusted hierarchical structure, the computing path selection and resource allocation are optimized.

[0118] Optimizing the calculation path selection and resource allocation includes building an optimization model using a multi-objective optimization algorithm to optimize the calculation path selection and resource allocation.

[0119] The formula of the optimization model is expressed as:

[0120]

[0121] Among them, Y represents the result of optimizing the calculation path selection and resource allocation, x represents the decision variable of the calculation path selection, and w i represents the weight of the i-th target, min x represents minimizing the decision variable x, f i (x,θ) represents the function related to the i-th target, c j represents the cost of the jth resource, x j represents the allocation of the jth resource, e -∈h(x) represents an exponential decay term, ∈ represents an optimization parameter, h(x) represents the constraint function, and Γ(α+1) represents the gamma function.

[0122] The value threshold of Y is a positive real number. The larger the value, the lower the efficiency in resource allocation and path selection.

[0123] Furthermore, by optimizing the calculation path selection and resource allocation, the overall efficiency of the system is improved, ensuring maximum utilization under limited resources. The multi-objective optimization algorithm can balance multiple objectives to achieve the global optimal effect.

[0124] Furthermore, improving the intelligence of resource management and path selection can avoid resource waste and irrational path selection, thereby maximizing the efficiency of the industrial system and improving the competitiveness and production efficiency of enterprises.

[0125] On the other hand, this embodiment also provides an industrial multi-project multi-level indicator calculation system, which includes:

[0126] The data collection and preprocessing module collects multi-dimensional data from different industrial projects and levels and performs preprocessing operations.

[0127] The analysis model building module uses the preprocessed multidimensional data to analyze the dependencies between various data indicators.

[0128] The hierarchical structure formation and adjustment module forms an initial hierarchical structure based on the dependencies in the analysis model and automatically adjusts the hierarchical structure according to data and business requirements.

[0129] The multi-objective optimization and resource allocation module uses a multi-objective optimization algorithm to optimize calculation path selection and resource allocation.

[0130] The hierarchical verification and optimization module verifies the formed hierarchical structure and uses cross-validation and A / B testing to ensure its rationality and robustness.

[0131] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0132] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0133] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0134] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0135] Example 2, the experiment selected three different projects (Project A, Project B, Project C) of an industrial enterprise, which covered different production links and management levels. In order to fully verify the advantages of the present invention, multi-dimensional data were collected from each project, including production efficiency data (unit: %), equipment health status data (unit: %), resource consumption data (unit: tons), quality control data (unit: times / 1000 pieces), maintenance cost data (unit: yuan), and energy consumption data (unit: kWh). The data collection period is 30 days, and preliminary preprocessing is carried out, including data cleaning, missing value filling, outlier processing and data standardization steps to ensure the quality and consistency of the data.

[0136] After data preprocessing, the processed data was used to build an analysis model. The model used the random forest algorithm and association rule mining method to analyze the dependencies between various data indicators. After the model was built, the initial hierarchical structure was formed through the dependency matrix. Due to the large differences in production environments and requirements of different projects, the hierarchical structure was automatically adjusted according to actual business needs after it was formed to ensure the best match between the data dependencies and business logic of each project.

[0137] Subsequently, based on the adjusted hierarchical structure, the calculation path selection and resource allocation were optimized. A multi-objective optimization algorithm was used to set multiple optimization goals including resource utilization, production efficiency, and maintenance cost. Through the dynamic adjustment of the algorithm, the optimal path selection and resource allocation under different hierarchical structures were achieved. During the experiment, three rounds of adjustments were made for each project. After each adjustment, the rationality of the hierarchical structure was re-verified, and finally an optimized resource allocation plan was generated. The experimental data are shown in Table 1. The experimental data are shown in Table 1.

[0138] Table 1 Experimental data table

[0139]

[0140] Through the comparative analysis of the above table data, it can be clearly seen that the implementation of the present invention has a significant optimization effect on the data of each dimension of Project A, Project B, and Project C. First, in terms of production efficiency, the efficiency of each project has been improved, among which the production efficiency of Project A has increased from 85% to 92%, Project B has increased from 78% to 88%, and Project C has increased from 82% to 90%. This significant improvement is mainly due to the fact that the dynamically adjusted hierarchical structure can better reflect the actual business needs of each project, thereby optimizing the production path and resource allocation.

[0141] Secondly, in terms of equipment health, the health indicators of equipment in each project have also been significantly improved, indicating that through the optimized resource allocation, equipment maintenance has been more effectively managed and the occurrence of equipment failures has been reduced. In particular, in Project B, the equipment health status increased from 65% to 78%. This change shows that the adjusted hierarchical structure can better balance the allocation of maintenance resources and improve the operational reliability of equipment.

[0142] In terms of resource consumption and energy consumption, the resource consumption and energy consumption of each project have decreased. The resource consumption of Project A dropped from 154 tons to 132 tons, and the energy consumption dropped from 20310 kWh to 18056 kWh; Project B and Project C also showed similar optimization effects. This shows that the multi-objective optimization algorithm plays a significant role in resource allocation, and the optimized calculation path can minimize resource waste and improve resource utilization.

[0143] In addition, quality control and maintenance costs have also improved significantly. For example, the quality control issues of Project A have been reduced from 15 times / 1,000 pieces to 10 times / 1,000 pieces, and the maintenance cost has been reduced from 50,000 yuan to 45,000 yuan. This further proves that by adjusting the hierarchical structure, the system can more accurately identify and solve the weak links in quality control and reduce maintenance costs through more efficient resource allocation.

[0144] The present invention shows significant advantages in processing complex industrial multi-project multi-level data by adopting a multi-objective optimization algorithm, especially in terms of improving production efficiency, optimizing resource utilization and enhancing system robustness. Compared with the traditional single-objective optimization algorithm, the multi-objective optimization algorithm can more effectively coordinate the balance between different objectives, so that the system can achieve overall performance improvement in multiple dimensions. In specific applications, the algorithm can respond to changes in business data in real time, achieve optimal resource allocation by dynamically adjusting the hierarchical structure and optimizing path selection, and maintain efficient operation of the system when dealing with emergencies. In addition, the system has the ability of continuous learning and self-optimization, and can continuously adjust the optimization strategy according to historical data to ensure continuous efficiency after long-term operation. Through visualization tools, users can view the system status in real time and adjust the optimization target according to actual needs, further enhancing the user-friendliness and practicality of the system.

[0145] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for calculating industrial multi-project and multi-level indicators, characterized in that: include: Collect multi-dimensional data from different industrial projects and levels and perform pre-processing; Using the preprocessed data, an analysis model is established to analyze the dependencies between various data indicators; Based on the analyzed dependencies, the dependencies between the indicators are extracted from the analysis model; Dependency relationships include the strength of dependency relationships and the directionality of dependency relationships; The strength of the dependency includes,correlation coefficient and mutual information; The directionality of dependency relationships includes,the direction of causality; The dependency data is structured to generate a dependency matrix, which is expressed as follows: Among them, R represents the dependency matrix, R ij Indicates the degree of dependence of indicator i on indicator j. Each R ij It is a composite value that combines the dependence strength and direction, which can be further expressed as: R ij =[ρ ij ,I ij ,D ij ] Among them, ρ ij Represents the correlation coefficient between index i and index j, I ij represents the mutual information between index i and index j, D ij Indicates the causal directionality of indicator i to indicator j. Based on the dependency matrix, the indicators are clustered using agglomerative hierarchical clustering, and indicators with strong dependencies are aggregated in the same level to form an initial hierarchical structure. Set the dependency matrix to an adjacency matrix of a weighted graph, where nodes represent indicators and edge weights represent the dependency relationship between indicators, and construct the degree matrix of the weighted graph; The formula of the degree matrix is ​​expressed as: in, represents the element of the i-th row and i-th column of the degree matrix of node i, n represents the total number of indicators, R ij represents the degree of dependence of indicator i on indicator j, α represents the adjustment parameter, 1-R ij represents the inverse of the degree of dependence; The Laplace matrix of the weighted graph is defined as the difference between the degree matrix and the dependency matrix. The Laplace matrix is ​​normalized to obtain a symmetric normalized Laplace matrix, which is expressed as follows: in, represents the symmetric normalized Laplace matrix; D * represents the degree matrix, (D * ) -1 / 2 Denotes the optimized degree matrix D * The inverse square root matrix of * represents the optimized Laplacian matrix; Calculate the eigenvalues ​​and eigenvectors of the normalized Laplace matrix, select the eigenvectors of the smallest k eigenvalues, and construct the characteristic matrix. The formula is expressed as: Among them, U * represents the final feature matrix, argmin represents finding the value that minimizes the objective function, Tr(·) represents the trace operation of the matrix, and U T represents the transposed matrix of matrix U, represents the symmetric normalized Laplace matrix, U represents the process characteristic matrix, U T U = I represents the orthogonal constraint, δ represents the regularization parameter, and k represents the number of eigenvectors in the feature matrix; u i represents the i-th eigenvector of the process characteristic matrix, Represents vector u i The gradient of Represents vector u i The squared norm of the gradient; Treat each row of the feature matrix as a new data point and use the K-means clustering algorithm to divide these data points into k clusters; The result of K-means clustering is hierarchical division, where each cluster represents a level, generating an initial hierarchical structure; Each level contains a group of indicators assigned by the clustering algorithm, ensuring strong dependencies between these indicators to form an initial hierarchical structure; The hierarchical structure is represented as a hierarchical matrix, and the formula is expressed as: in, Indicates the membership degree of the i-th indicator in the optimized hierarchical matrix to the k-th level; Represents the characteristic matrix U of the i-th index after optimization * The vector representation in c k represents the center vector of the kth cluster, Represents the feature vector With cluster center c k The square of the Euclidean distance between 2 represents the square of the bandwidth parameter of the Gaussian kernel function, represents the Gaussian kernel function; c l represents the lth cluster center vector; According to data and business requirements, a dynamic threshold τ of dependency strength is set. Only indicator pairs with dependency strength exceeding τ will be considered at the same level. The formula for the intensity-dependent dynamic threshold is: Among them, Θ(τ) represents the dynamic threshold adjustment function of the dependent strength, τ represents the threshold of the dependent strength, R ij represents the dependency matrix; represents the smoothed logistic regression function, η represents the smoothing parameter, Represents the inverse operation of the logistic regression function; The value threshold of Θ(τ) is between 0 and 1; Cross-validation and A / B testing are used to verify the hierarchical structure to ensure its rationality and robustness. The formula is expressed as: in, represents the cross validation loss function, Indicates averaging the results of cross-validation, and n indicates the number of cross-validation folds; represents trace operation, U i represents the feature matrix in the i-th fold, represents the transpose of the feature matrix in the i-th fold, represents the symmetric normalized Laplace matrix, δ represents the regularization parameter, Denotes the feature vector u j The square of the gradient norm of , Θ(τ) represents the intensity-dependent dynamic threshold adjustment function; The value threshold of is from 0 to positive infinity. The smaller the value, the better the fitting effect of the hierarchical structure. Categorize the verification results at each level. When the values ​​of are all lower than the preset global threshold, the hierarchical structure is determined to have passed the verification and enters the next step of adjustment; When the verification result of a certain layer exceeds the global threshold, the hierarchical structure is determined to be an item that needs to be adjusted and is readjusted; when When the values ​​of are all globally close to the threshold, but the dependence strength is greater than the preset upper limit, it is determined that the layer is too dependent on the overall structure, resulting in a single point failure of the system; After classifying the verification results of each level and readjusting the hierarchical structure that exceeds the global threshold, the entire hierarchical structure is globally optimized, and the secondary verification mechanism is triggered. The hierarchical structure is re-verified using cross-validation and A / B testing to achieve automatic adjustment of the hierarchical structure. Based on the adjusted hierarchical structure, the computing path selection and resource allocation are optimized.

2. The method for calculating industrial multi-project and multi-level indicators as claimed in claim 1, characterized in that: The multi-dimensional data includes production efficiency data, quality control data, energy and resource consumption data, equipment health status data, maintenance prediction and health management data, supply chain optimization data, environmental compliance and sustainability data, safety monitoring data, cost-benefit analysis data, and real-time operation data; The preprocessing includes: cleaning the multidimensional data, processing missing values, outliers and deleting duplicate data; converting the multidimensional data and integrating multi-source data; labeling and classifying the multidimensional data and performing data consistency checks.

3. The method for calculating industrial multi-project and multi-level indicators as claimed in claim 1, characterized in that: Establishing an analysis model includes extracting features from preprocessed multidimensional data, using a random forest model to learn known dependencies, and combining unsupervised learning methods of association rule mining to discover potential dependencies.

4. The method for calculating industrial multi-project and multi-level indicators as claimed in claim 1, characterized in that: The optimization of computing path selection and resource allocation includes: using a multi-objective optimization algorithm to build an optimization model to optimize computing path selection and resource allocation; The formula of the optimization model is expressed as: Among them, Y represents the result of optimizing the calculation path selection and resource allocation, x represents the decision variable of the calculation path selection, and w i represents the weight of the i-th target, min x represents minimizing the decision variable x, f i (x,θ) represents the function related to the i-th target, c j represents the cost of the jth resource, x j represents the allocation of the jth resource, e -∈h(x) represents an exponential decay term, ∈ represents an optimization parameter, h(x) represents the constraint function, and Γ(α+1) represents the gamma function; The value threshold of Y is a positive real number. The larger the value, the lower the efficiency in resource allocation and path selection.

5. An industrial multi-project multi-level indicator calculation system according to any one of claims 1 to 4, characterized in that: It includes data collection and processing module, model building and adjustment level module, and solution optimization module; The data acquisition and processing module is used to collect multi-dimensional data from different industrial projects and levels and perform pre-processing; The model building and hierarchical adjustment module is used to use the preprocessed data to build an analysis model, analyze the dependencies between various data indicators, form an initial hierarchical structure based on the analyzed dependencies, and automatically adjust the hierarchical structure according to data and business needs; The solution optimization module is used to optimize the calculation path selection and resource allocation based on the adjusted hierarchical structure.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the processor implements the steps of an industrial multi-project and multi-level indicator calculation method described in any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of an industrial multi-project and multi-level indicator calculation method described in any one of claims 1 to 4 are implemented.

Images