Big data-based communication engineering informatization system and method

Abstract

The application discloses a communication engineering informatization system and method based on big data, relates to the technical field of engineering informatization, and realizes real-time access to an engineering management system, a sensor network and a user operation log, constructs a dynamic knowledge graph containing a sub-project topological relationship, data generation space-time distribution and a cross-project calling chain, generates a classification scheme through a sliding window clustering algorithm based on historical sub-project quantity distribution characteristics, dynamically calibrates a critical threshold of data storage priority integration in combination with an engineering complexity factor, analyzes sub-project dependency relationship for a new engineering, activates an optimization process when a joint index of the sub-project quantity and topological complexity exceeds the threshold, analyzes sub-project correlation scores based on space-time correlation and business semantic characteristics of historical calling data, and generates a storage integration strategy, thereby improving clustering efficiency, reducing manual configuration requirements and improving engineering deployment efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of engineering informatization technology, in particular to a communication engineering informatization system and method based on big data. BACKGROUND

[0002] With the continuous development and popularization of information technology, the status and role of communication engineering become more and more important. As the foundation and pillar of information society construction, communication engineering faces increasing pressure and challenges with the continuous development of economy and the acceleration of informationization process.

[0003] The prior art determines the complexity of the project based on a fixed threshold or a single index, which is difficult to dynamically adapt to different scales and types of communication engineering. For example, when dividing the high complexity sub-project cluster, the traditional method often uses a preset global threshold, which cannot distinguish the difference between large core network engineering and small base station engineering, and ignores the actual interaction intensity and topological characteristics between sub-projects. The resource optimization strategy lacks pertinence, simple engineering is over-optimized and resources are wasted, and the real complex engineering faces performance bottlenecks due to missed judgment.

[0004] Therefore, the present application discloses a communication engineering informatization system and method based on big data to solve the above problems. SUMMARY

[0005] The purpose of the present application is to provide a communication engineering informatization system and method based on big data to solve the problems in the prior art.

[0006] To achieve the above purpose, the present application provides the following technical scheme: a communication engineering informatization method based on big data, the method comprising the following steps:

[0007] S1: Real-time access to the project management system, sensor network and user operation log, and construct a dynamic knowledge graph containing sub-project topological relationship, data generation spatio-temporal distribution and cross-project call chain;

[0008] S2: Based on the historical sub-project quantity distribution characteristics, generate a classification scheme through a sliding window clustering algorithm, and dynamically calibrate the critical threshold value of data storage priority integration combined with the engineering complexity factor;

[0009] S3: Analyze the sub-project dependency relationship of the new project, and activate the optimization process when the joint index of sub-project quantity and topological complexity exceeds the threshold value;

[0010] S4: Based on the spatio-temporal correlation and business semantic characteristics of historical call data, analyze the sub-project association score, and generate a storage integration strategy.

[0011] According to the above content, in S1, the following content is included:

[0012] S101: Collect historical communication engineering project decomposition into a set of sub-project quantities, denoted as P={P1, P2,..., P i ..., I}, wherein P i represents the number of sub-projects decomposed from the i-th historical communication engineering project; i∈[1, I]; I represents the total number of historical communication engineering projects; simultaneously obtain the sub-project name information, engineering type label, data volume information, calling information and construction duration of the historical communication engineering project decomposition; the calling information of the sub-project data includes the number of calls, the called data volume information and the standard deviation of the calling delay; normalize the number of calls, the called data volume information and the standard deviation of the calling delay; construct a sub-project calling space-time matrix based on the calling information of the processed sub-project data; denote the sub-project calling space-time matrix of the i-th historical communication engineering project as T i ;

[0013] ;

[0014] wherein M is equal to the number of sub-projects decomposed from the i-th historical communication engineering project; the sub-project calling space-time matrix T i contains elements (C （m，n） , q （m，n） ,△t （m，n） ), m, n∈[1, M], m and n are positive integers; C （m，n） represents the number of times that project m is called when project n initiates a calling instruction; q （m，n） represents the average data volume of a single call of project m when project n initiates a calling instruction,△t （m，n） represents the standard deviation of the calling delay of project m when project n initiates a calling instruction;

[0015] S102: Analyze the data similarity based on the engineering space-time matrix of each historical communication engineering project; extract the calling data of each historical communication engineering project containing the same-named sub-project pair, and analyze the data similarity according to the calling data combined with the data similarity function; in the data similarity function, if q （j，ab） is greater than or equal to q thr , the output is 1, otherwise the output is 0; q （j，ab） represents the average data volume of a single call of project a when project b initiates a calling instruction in the j-th historical communication engineering project containing the same-named sub-projects a and b, j∈[1, J]; J represents the total number of historical communication engineering projects containing the same-named sub-projects a and b, J<I; q thr represents the called data volume threshold, which is preset by the system;

[0016] According to the historical communication engineering project text report set, the text similarity is analyzed; the text similarity average analysis includes: using BERT to generate vectors for the text reports of sub-project a and sub-project b, calculating the cosine similarity between the vectors based on the cosine function, and calculating the average value;

[0017] The cross-engineering knowledge graph is constructed by combining the data similarity, the text similarity and the time decay coefficient, denoted as G= (V, E); wherein V represents a node set, and the node set includes a sub-project name and a corresponding engineering type label; E represents an edge set, and the weight of the connection edge between sub-project a and sub-project b is denoted as W （a，b） :

[0018] ;

[0019] wherein η j represents the time decay coefficient of the jth historical communication engineering project, η j = exp (-t j ); wherein t j is equal to the quotient of the difference between the completion time of the historical communication engineering project and the current time and a preset constant; λ is a weight coefficient, and the weight coefficient is a system preset constant; S sem (a, b) represents the text similarity average value of sub-project a and sub-project b.

[0020] The normalized processing of the calling information (calling times, data volume, time delay standard deviation) eliminates the dimensional differences of the data, so that the subsequent analysis is more fair and reliable. The sub-project calling space-time matrix (Ti) provides a structured representation and can accurately capture the space-time dependency relationship (such as calling frequency and data transmission mode) between sub-projects, avoiding the ambiguity of dependency modeling in traditional methods.

[0021] The combination of data similarity analysis (based on threshold function) and text similarity analysis (BERT vector and cosine similarity) realizes multi-modal data fusion. This not only can identify behavior-based similarity (such as high-frequency calling sub-projects), but also can capture semantic similarity (such as engineering terms in report text), so as to generate a more comprehensive knowledge graph. The time decay coefficient (ηj) gives higher weight to newer projects, ensuring dynamic updating of the knowledge graph and improving the timeliness of recommendation or prediction (reducing the interference of outdated data).

[0022] The edge weight formula (W(a, b) = λ * S_sem(a, b) + (1-λ) * (1 / J) * Σ(ηj * data similarity output)) of the knowledge graph (G) balances the text, data, and timing factors. The parameter λ (system preset weight coefficient) allows flexible adjustment of the contribution of text and data similarity to adapt to different engineering scenarios (such as setting λ higher for projects with rich text reports). This enables the graph to adaptively integrate heterogeneous data, supports efficient query of sub-project associations in subsequent modules (such as storage optimization of S4), and reduces the need for manual intervention.

[0023] According to the above content, the following content is included in S2:

[0024] S201: Grouping each historical communication engineering project according to the engineering type label; generating a historical communication engineering project sequence in ascending order of the number of sub-projects in each engineering type, calculating the sliding window size of each engineering type, and the sliding window size is equal to the arithmetic square root of the number of historical communication engineering projects in the corresponding engineering type rounded up;

[0025] The historical communication engineering project sequence is split using a sliding window. If there are more than a number threshold of historical communication engineering projects in a single sliding window, and the difference in the number of sub-projects is not more than a preset constant, the corresponding sliding window is marked as a high-density window, and the number threshold is equal to the product of the sliding window size and a preset coefficient rounded up. The high-density windows are merged according to the interval overlap relationship to generate clusters, the clustering center of each cluster is analyzed, the clustering center is equal to the average number of sub-project call times, the sub-project call times include the number of call initiation and the number of calls, and a clustering center set is generated;

[0026] S202: Obtain the number of adjacent edges of each knowledge graph node, and record the number of adjacent edges of the e-th knowledge graph node v e as A e , extract the elements related to the knowledge graph node v e from the sub-project call spatio-temporal matrix to which the knowledge graph node v e belongs, analyze the data initiated by the knowledge graph node v e and the called data to obtain the total number of call initiation, the total number of calls, the average data volume of single call initiation, the average data volume of single call, the average standard deviation of call initiation delay, and the average standard deviation of call delay, generate a feature vector V e based on the total number of call initiation, the total number of calls, the average data volume of single call initiation, the average data volume of single call, the average standard deviation of call initiation delay, and the average standard deviation of call delay; perform PageRank calculation on the knowledge graph node v e , and record the PageRank value as PageRank(ve ), the computational complexity factor of each cluster; the complexity factor of the fth cluster I f is denoted as θ f :

[0027] ;

[0028] where |I f | represents the number of sub-projects in the cluster I f , t e represents the construction time of the knowledge graph node v e , and ||V e || represents the length of the feature vector V e .

[0029] S203: Obtain the number of adjacent edges corresponding to all sub-projects under the same engineering type, and calculate the median of the number of adjacent edges, denoted as median(A); based on the median of the number of adjacent edges, the cluster center set and the complexity factor set, analyze the node dynamic threshold Z = a1 x median(A) + a2 x max(θ f x C f ); where C f represents the cluster center of the fth cluster.

[0030] The sliding window mechanism and the high-density window marking can automatically identify similar patterns in historical projects. This avoids the defect that traditional clustering methods (such as K-means) need to pre-set the number of clusters, and improves the clustering efficiency. Merging overlapping windows to generate clusters and calculating cluster centers enables the system to focus on high-frequency interaction sub-projects and provides a basis for resource allocation. The complexity factor integrates cluster size, node importance, feature vector length and construction time. Highlighting complex clusters with tight construction periods helps to prioritize high-risk projects.

[0031] The node dynamic threshold combines the statistical median and the maximum complexity-cluster center product. This makes the threshold adaptive to different engineering types, avoiding false positives caused by fixed thresholds. After marking high-complexity clusters, the system can automatically trigger alarms or optimizations to prevent resource bottlenecks (such as storage or computing overload) and potentially reduce the risk of project delays.

[0032] According to the above content, in S3, the following content is included:

[0033] Obtaining a sub-project set of a current communication engineering, constructing a directed acyclic graph based on the sub-project set of the current communication engineering, a node set in the directed acyclic graph being equal to the sub-project set of the current communication engineering, and a directed edge representing that there is a dependency relationship between the sub-projects; dividing the nodes into several modules according to a community discovery algorithm and analyzing a module degree; analyzing a topology complexity, the topology complexity being equal to a critical path length multiplied by the module degree divided by a total number of the sub-projects; if both the number of the sub-projects of the current communication engineering and the topology complexity are greater than corresponding threshold values, activating a storage optimization process, and the number of the sub-projects of the current communication engineering corresponding to the threshold values being equal to a node dynamic threshold value.

[0034] The DAG construction intuitively represents the sub-project dependencies (such as data flow or task order), the community discovery algorithm (such as the Louvain method) divides the modules and calculates the module degree (measures the module cohesion), and can efficiently identify the functional units in the engineering. The topology complexity formula comprehensively considers the structural characteristics: the critical path length reflects the longest dependency chain (affects the project duration), the module degree reflects the modularization degree (a high value indicates easy management), and the total number of the sub-projects normalizes the complexity. This provides an objective complexity index to replace subjective evaluation and improves the management accuracy.

[0035] According to the above content, in S4, the following content is included:

[0036] S401: Extracting a sub-project set of a current communication engineering, in a cross-engineering knowledge graph, retrieving a corresponding node for each sub-project name, extracting all associated edges connected to the corresponding node and their weights to form a sub-graph edge set;

[0037] S402: Analyzing the association score for each pair of sub-projects and their weights, and recording the association score of the current communication engineering sub-projects u and v as O （u，v） :

[0038] ;

[0039] Wherein, X （u，v） represents the number of historical call records of the sub-projects u and v in the current communication engineering, b x represents the data volume of the xth call record, △t x represents the standard deviation of the call delay of the xth call record, ξ() represents the Sigmoid function; Sim sem (u, v) represents the average text similarity of the sub-projects u and v;

[0040] If the association score is greater than the corresponding threshold value, mark the corresponding sub-project pair as a physical storage collocation candidate; according to the descending order of the association scores of the candidate pairs, deploy the data and services of the corresponding sub-projects on the same storage partition in turn.

[0041] The association score formula combines historical calling behavior (data volume bx and time delay standard deviation Δtx) and semantic similarity. The sigmoid function normalizes the calling data to enhance robustness (avoid extreme value interference). High-scoring item pairs (such as frequently called and large data volume) are preferentially co-located in the same storage partition, reducing data access latency (especially for real-time communication engineering), and potentially improving I / O performance. Deploy the candidate pairs in descending order of score to enable the system to prioritize high-impact sub-items (such as core modules) and maximize storage resource utilization. Combined with the activation mechanism of S3, a closed-loop optimization is formed to reduce the need for manual configuration and improve engineering deployment efficiency

[0042] In another aspect of the present application, a big data-based communication engineering informatization system is provided, which is applied to the big data-based communication engineering informatization method described above and includes a calling data analysis module, a clustering analysis module, a real-time project screening module, and an association storage module.

[0043] The calling data analysis module is used to access the engineering management system, sensor network, and user operation log in real time, and construct a dynamic knowledge graph containing sub-item topological relationship, data generation spatio-temporal distribution, and cross-project calling chain.

[0044] The clustering analysis module is used to generate a classification scheme based on historical sub-item quantity distribution characteristics through a sliding window clustering algorithm, and dynamically calibrate the critical threshold of data storage priority integration in combination with the engineering complexity factor.

[0045] The real-time project screening module is used to analyze the sub-item dependency relationship of a new project, and activate the optimization process when the joint index of sub-item quantity and topological complexity exceeds the threshold.

[0046] The association storage module is used to analyze sub-item association scores based on the spatio-temporal correlation and business semantic characteristics of historical calling data, and generate a storage integration strategy.

[0047] According to the above content, the calling data analysis module includes a spatio-temporal matrix analysis unit and a cross-project knowledge graph analysis unit.

[0048] The spatio-temporal matrix analysis unit is used to collect a set of historical communication engineering project decomposition sub-items, synchronously acquire sub-item name information, engineering type label, data volume information, calling information, and construction duration of historical communication engineering project decomposition, and normalize the calling frequency, called data volume information, and calling time delay standard deviation of the sub-item data. A sub-item calling spatio-temporal matrix is constructed based on the calling information of the processed sub-item data.

[0049] The cross-engineering knowledge graph analysis unit is configured to analyze data similarity based on engineering space-time matrix of each historical communication engineering project; analyze text similarity based on a set of text reports of the historical communication engineering project; and construct the cross-engineering knowledge graph in combination with the data similarity, the text similarity, and a time decay coefficient.

[0050] According to the above, the clustering analysis module includes a clustering unit and a dynamic threshold analysis unit.

[0051] The clustering unit is configured to group each historical communication engineering project according to an engineering type label; perform sliding window density clustering; and output a clustering center set.

[0052] The dynamic threshold analysis unit is configured to obtain a number of adjacent edges of each knowledge graph node, analyze calling instructions and called data initiated by the knowledge graph node, and calculate a complexity factor in each cluster; and analyze a node dynamic threshold based on a median of the number of adjacent edges, the clustering center set, and the complexity factor set.

[0053] According to the above, the real-time project screening module includes a topological complexity analysis unit and an optimization activation unit.

[0054] The topological complexity analysis unit is configured to obtain a sub-project set of a current communication engineering project, construct a directed acyclic graph based on the sub-project set of the current communication engineering project, set a node set in the directed acyclic graph equal to the sub-project set of the current communication engineering project, and set a directed edge to represent an existing dependency relationship between sub-projects; divide the nodes into a plurality of modules according to a community discovery algorithm and analyze a module degree; and analyze a topological complexity, which is equal to a key path length multiplied by the module degree and divided by a total number of sub-projects.

[0055] The optimization activation unit is configured to activate a storage optimization process if both a sub-project quantity of the current communication engineering project and the topological complexity are greater than corresponding threshold values, and the sub-project quantity of the current communication engineering project corresponds to the node dynamic threshold.

[0056] According to the above, the correlation storage module includes a sub-graph edge set analysis unit and an optimization deployment unit.

[0057] The sub-graph edge set analysis unit is configured to extract a sub-project set of a current communication engineering project, search for a corresponding node in a cross-engineering knowledge graph for each sub-project name, extract all correlation edges and weights connected to the corresponding node to form a sub-graph edge set.

[0058] The optimization deployment unit is configured to analyze a correlation score for each pair of sub-projects and weights, mark a corresponding pair of sub-projects as a physical storage collocation candidate if the correlation score is greater than a corresponding threshold value, and deploy data and services of the corresponding sub-projects on the same storage partition in a descending order of the correlation scores of the candidate pairs.

[0059] Compared with the prior art, the beneficial effects of the present application are: by normalizing the call information, the dimensional difference of data is eliminated, making the subsequent analysis more fair and reliable. The sub-project call space-time matrix provides a structured representation, which can accurately capture the space-time dependency between sub-projects, avoiding the ambiguity of dependency modeling in traditional methods; the combination of data similarity analysis and text similarity analysis realizes multi-modal data fusion. This not only can identify behavior-based similarity, but also can capture semantic similarity, so as to generate a more comprehensive knowledge graph. The time decay coefficient gives higher weight to newer projects, ensuring dynamic updating of the knowledge graph and improving the timeliness of recommendation or prediction. The edge weight formula of the knowledge graph balances the text, data and time sequence factors. The parameters allow flexible adjustment of the contribution of text and data similarity, adapting to different engineering scenarios. This makes the graph adaptively integrate heterogeneous data, supporting efficient query of sub-project associations in subsequent modules, reducing the need for manual intervention. The sliding window mechanism and high-density window marking can automatically identify similar patterns in historical projects. This avoids the need to pre-set the number of clusters in traditional clustering methods, improving clustering efficiency. Merging overlapping windows generates clusters and calculates cluster centers, enabling the system to focus on high-frequency interaction sub-projects and providing a basis for resource allocation. The complexity factor integrates cluster size, node importance, feature vector length and construction duration. It highlights complex clusters with tight construction periods, helping to prioritize high-risk projects. The node dynamic threshold combines the statistical median and the maximum complexity-cluster center product. This makes the threshold adaptive to different engineering types, avoiding false positives caused by fixed thresholds. After marking high-complexity clusters, the system can automatically trigger alerts or optimizations to prevent resource bottlenecks and potentially reduce the risk of project delays. The DAG construction intuitively represents sub-project dependencies, and the community discovery algorithm divides modules and calculates modularity, enabling efficient identification of functional units in the project. The topological complexity formula integrates structural features: key path length reflects the longest dependency chain, modularity reflects the degree of modularity, and the total number of sub-projects normalizes complexity. This provides an objective complexity indicator, replacing subjective evaluation and improving management accuracy. The association score formula combines historical call behavior and semantic similarity. The sigmoid function normalizes call data, enhancing robustness. High-scoring sub-projects are preferentially co-located in the same storage partition, reducing data access latency and potentially improving I / O performance. Deploying candidate pairs in descending order of score enables the system to prioritize high-impact sub-projects, maximizing storage resource utilization. Combined with the activation mechanism of S3, a closed-loop optimization is formed, reducing the need for manual configuration and improving engineering deployment efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0060] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate the present application and are used to explain the present application together with the embodiments of the present application, and do not constitute a limitation of the present application. In the drawings:

[0061] Figure 1This is a flowchart illustrating the information-based communication engineering method based on big data according to the present invention.

[0062] Figure 2 This is a schematic diagram of the structure of the communication engineering information system based on big data according to the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Example 1: Please refer to Figure 1 This invention provides a technical solution: a communication engineering informatization method based on big data, which includes the following steps:

[0065] S1: Real-time access to the engineering management system, sensor network and user operation logs to build a dynamic knowledge graph that includes sub-project topology, spatiotemporal distribution of data generation and cross-project call chain;

[0066] S1 includes the following:

[0067] S101: Collect the set of the number of sub-projects into which historical communication engineering projects are decomposed, denoted as P = {P1, P2, ..., P...} i ..., P I}, where P i Let represent the number of sub-projects into which the i-th historical communication project is decomposed; i∈[1,I]; I represents the total number of historical communication projects; synchronously acquire the sub-project name information, project type label, data volume information, call information, and construction duration of the historical communication project decomposition; the call information of the sub-project data includes the number of calls, the data volume information of the calls, and the standard deviation of the call delay; normalize the number of calls, the data volume information of the calls, and the standard deviation of the call delay; construct the sub-project call spatiotemporal matrix based on the call information of the processed sub-project data, and denote the sub-project call spatiotemporal matrix of the i-th historical communication project as T. i ;

[0068] ;

[0069] Where M equals the number of sub-projects into which the i-th historical communication project is decomposed; the sub-projects call the spatiotemporal matrix T i Containing elements (C （m，n） q （m，n） , △t （m，n）), m, n ∈ [1, M], m and n are positive integers; C （m，n） represents the number of times that project m is called when project n initiates the call instruction; q （m，n） represents the average amount of data of a single call of project m when project n initiates the call instruction, △t （m，n） represents the standard deviation of the call delay of project m when project n initiates the call instruction.

[0070] S102: Analyze the data similarity based on the engineering space-time matrix of each historical communication engineering project; extract the call data of each historical communication engineering project containing the same-name sub-project pair, and analyze the data similarity according to the call data combined with the data similarity function; in the data similarity function, if q （j，ab） is greater than or equal to q thr , the output is 1, otherwise the output is 0; q （j，ab） represents the average amount of data of a single call of project a when project b initiates the call instruction in the jth historical communication engineering project containing the same-name sub-projects a and b, j ∈ [1, J]; J represents the total number of historical communication engineering projects containing the same-name sub-projects a and b, J < I; q thr represents the call data threshold, which is preset by the system.

[0071] Analyze the text similarity according to the text report set of the historical communication engineering project; the average text similarity analysis includes: generating vectors for the text reports of sub-projects a and b using BERT, calculating the cosine similarity between the vectors based on the cosine function, and calculating the average value.

[0072] Construct a cross-engineering knowledge graph combining data similarity, text similarity and time decay coefficient, denoted as G= (V, E); where V represents the node set, which contains the sub-project name and the corresponding engineering type label; E represents the edge set, and the weight of the connection edge between sub-projects a and b is denoted as W （a，b） :

[0073] ;

[0074] where η j represents the time decay coefficient of the jth historical communication engineering project, η j = exp (-t j ) ; where t j is equal to the difference between the completion time of the historical communication engineering project and the current time and the quotient of the preset constant; λ is the weight coefficient, which is a system preset constant; S sem (a, b) represents the average text similarity of sub-projects a and b.

[0075] S2: Based on the historical sub-project quantity distribution characteristics, a classification scheme is generated by a sliding window clustering algorithm, and a critical threshold value for dynamically calibrating data storage priority integration is combined with the engineering complexity factor;

[0076] In S2, the following contents are included:

[0077] S201: Grouping each historical communication engineering project according to the engineering type label; generating a historical communication engineering project sequence in ascending order according to the number of sub-projects in each engineering type, calculating the sliding window size of each engineering type, and the sliding window size is equal to the arithmetic square root of the number of historical communication engineering projects in the engineering type, rounded up;

[0078] The historical communication engineering project sequence is split by a sliding window; if there are more than a number threshold of historical communication engineering projects in a single sliding window, and the difference in the number of sub-projects is not more than a preset constant, then the sliding window is marked as a high-density window, and the number threshold is equal to the product of the sliding window size and a preset coefficient, rounded up; the high-density windows are merged to generate clusters according to the interval overlap relationship, the clustering center of each cluster is analyzed, the clustering center is equal to the average number of sub-project call times, the sub-project call times include the number of call times and the number of called times, and the clustering center set is generated;

[0079] S202: Obtain the number of adjacent edges of each knowledge graph node, and the number of adjacent edges of the e-th knowledge graph node v e is denoted as A e , the elements related to the knowledge graph node v e are extracted from the sub-project call space-time matrix to which the knowledge graph node v e belongs, the knowledge graph node v e initiation call instruction and called data are analyzed to obtain the total number of initiation calls, the total number of calls, the average data volume of single initiation call, the average data volume of single call, the average standard deviation of initiation call time delay, and the average standard deviation of call time delay, and a feature vector V e is generated based on the total number of initiation calls, the total number of calls, the average data volume of single initiation call, the average data volume of single call, the average standard deviation of initiation call time delay, and the average standard deviation of call time delay; the PageRank value of the knowledge graph node v e is obtained by performing PageRank calculation, denoted as PageRank(v e ), and the complexity factor of each cluster is calculated; the complexity factor of the f-th cluster I f is denoted as θ f :

[0080] ;

[0081] Where, |If | represents cluster I f the number of neutron items, t e represents a knowledge graph node v e the construction length of the node, ||v e || represents the length of the feature vector V e ;

[0082] S203: Obtain the number of adjacent edges corresponding to all sub-projects under the same engineering type, and calculate the median of the number of adjacent edges, denoted as median(A); based on the median of the number of adjacent edges, the cluster center set and the complexity factor set, analyze the node dynamic threshold Z = a1 x median(A) + a2 x max(θ f x C f ); wherein C f represents the cluster center of the fth cluster.

[0083] S3: Perform sub-project dependency analysis on the new project, and activate the optimization process when the joint index of the number of sub-projects and the topology complexity exceeds the threshold;

[0084] In S3, the following contents are included:

[0085] Obtain the sub-project set of the current communication engineering, and construct a directed acyclic graph based on the sub-project set of the current communication engineering. The node set in the directed acyclic graph is equal to the sub-project set of the current communication engineering, and the directed edge represents that there is a dependency relationship between the sub-projects. Divide the nodes into several modules according to the community discovery algorithm and analyze the module degree. Analyze the topology complexity, which is equal to the length of the critical path multiplied by the module degree divided by the total number of sub-projects. If the number of sub-projects of the current communication engineering and the topology complexity are both greater than the corresponding threshold, activate the storage optimization process, and the number of sub-projects of the current communication engineering corresponds to the threshold value equal to the node dynamic threshold.

[0086] S4: Based on the spatio-temporal correlation and business semantic features of historical call data, analyze the sub-project association score, and generate a storage integration strategy.

[0087] In S4, the following contents are included:

[0088] S401: Extract the sub-project set of the current communication engineering, and in the cross-engineering knowledge graph, retrieve the corresponding node for each sub-project name, extract all associated edges connected to the corresponding node and their weights, and form a sub-graph edge set;

[0089] S402: Analyze the association score for each pair of sub-projects and their weights, and denote the association score of the current communication engineering sub-projects u and v as O （u，v） :

[0090] ;

[0091] wherein, X （u，v） represents the historical call record number of the sub-project u and the sub-project v in the current communication engineering, b x represents the data volume of the xth call record, △t x represents the call delay standard deviation of the xth call record, ξ() represents a Sigmoid function; Sim sem (u, v) represents the text similarity average of the sub-project u and the sub-project v;

[0092] If the association score is greater than the corresponding threshold value, the corresponding sub-project pair is marked as a physical storage collocation candidate; according to the descending order of the association score of the candidate pair, the data and the service of the corresponding sub-project are deployed on the same storage partition in turn.

[0093] Embodiment 2: Please refer to Figure 2 The application provides a technical solution: a communication engineering informatization system based on big data, which comprises a call data analysis module, a clustering analysis module, a real-time project screening module and an associated storage module.

[0094] The call data analysis module is used for real-time access to an engineering management system, a sensor network and user operation logs, and for constructing a dynamic knowledge graph comprising sub-project topological relationships, data generation spatio-temporal distribution and cross-project call chains.

[0095] The clustering analysis module is used for generating a classification scheme through a sliding window clustering algorithm based on historical sub-project quantity distribution characteristics, and dynamically calibrating a critical threshold value of data storage priority integration in combination with an engineering complexity factor.

[0096] The real-time project screening module is used for sub-project dependency relationship analysis of a new project, and activates an optimization process when a joint index of the sub-project quantity and topological complexity exceeds a threshold value.

[0097] The associated storage module is used for analyzing sub-project association scores based on spatio-temporal correlation and business semantic characteristics of historical call data, and generating a storage integration strategy.

[0098] The call data analysis module comprises a spatio-temporal matrix analysis unit and a cross-engineering knowledge graph analysis unit.

[0099] The spatio-temporal matrix analysis unit is used for collecting a historical communication engineering project decomposition into a sub-project quantity set, synchronously acquiring sub-project name information, engineering type labels, data volume information, call information and construction duration of the historical communication engineering project decomposition, and the call information of the sub-project data comprises call frequency, called data volume information and call delay standard deviation, and the call frequency, the called data volume information and the call delay standard deviation are normalized; a sub-project call spatio-temporal matrix is constructed based on the call information of the processed sub-project data.

[0100] The cross-project knowledge graph analysis unit is configured to analyze data similarity based on engineering space-time matrix of each historical communication engineering project; analyze text similarity based on a set of text reports of the historical communication engineering project; and construct a cross-project knowledge graph in combination with the data similarity, the text similarity, and a time decay coefficient.

[0101] The clustering analysis module comprises a clustering unit and a dynamic threshold analysis unit.

[0102] The clustering unit is configured to group each historical communication engineering project according to an engineering type label; perform sliding window density clustering, and output a clustering center set.

[0103] The dynamic threshold analysis unit is configured to obtain a number of adjacent edges of each knowledge graph node, analyze a calling instruction initiated by the knowledge graph node and called data, and calculate a complexity factor in each cluster; and analyze a node dynamic threshold based on a median of the number of adjacent edges, the clustering center set, and the complexity factor set.

[0104] The real-time project screening module comprises a topological complexity analysis unit and an optimization activation unit.

[0105] The topological complexity analysis unit is configured to obtain a sub-project set of a current communication engineering project, construct a directed acyclic graph based on the sub-project set of the current communication engineering project, a node set in the directed acyclic graph being equal to the sub-project set of the current communication engineering project, and a directed edge indicating that there is a dependency relationship between the sub-projects; divide the nodes into a plurality of modules according to a community discovery algorithm and analyze a module degree; and analyze a topological complexity, the topological complexity being equal to a key path length multiplied by the module degree and divided by a total number of the sub-projects.

[0106] The optimization activation unit is configured to activate a storage optimization process if both a number of sub-projects of the current communication engineering project and the topological complexity are greater than corresponding thresholds, the number of sub-projects of the current communication engineering project corresponding to a threshold being equal to the node dynamic threshold.

[0107] The correlation storage module comprises a sub-graph edge set analysis unit and an optimization deployment unit.

[0108] The sub-graph edge set analysis unit is configured to extract a sub-project set of the current communication engineering project, search for a corresponding node in the cross-project knowledge graph for each sub-project name, extract all correlation edges and weights connected to the corresponding node to form a sub-graph edge set.

[0109] The optimization deployment unit is configured to analyze a correlation score for each pair of sub-projects and the weight thereof, mark a corresponding pair of sub-projects as a physical storage collocation candidate if the correlation score is greater than a corresponding threshold, and deploy data and services of the corresponding sub-projects on the same storage partition in a descending order of the correlation scores of the candidate pairs.

[0110] It is to be understood that the phrases such as first and second, and the like, can refer to different entities or operations without necessarily implying any actual relationship or order between such entities or operations. Also, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0111] It will be apparent to those skilled in the art that the application is not limited to the details of the foregoing exemplary embodiments and that the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. No feature of the application is considered critical unless expressly stated in the claims.

Claims

1. A communication engineering information method based on big data, characterized in that, The method includes the following steps: S1: Real-time access to the engineering management system, sensor network and user operation logs to build a dynamic knowledge graph that includes sub-project topology, spatiotemporal distribution of data generation and cross-project call chain; S2: Based on the distribution characteristics of the number of historical sub-projects, a classification scheme is generated through a sliding window clustering algorithm, and the critical threshold for priority integration of data storage is dynamically calibrated by combining the engineering complexity factor. S2 includes the following: S201: Group each historical communication project according to the project type label; generate a sequence of historical communication projects in each project type by sorting them in ascending order according to the number of sub-projects decomposed into them; calculate the sliding window size of each project type, where the sliding window size is equal to the arithmetic square root of the number of historical communication projects in the project type, rounded up. A sliding window is used to split the historical communication engineering project sequence; If the number of sub-projects in a single sliding window that exceed a certain threshold does not exceed a preset constant, then the sliding window to which it belongs is marked as a high-density window. The number threshold is equal to the floor function of the product of the sliding window size and a preset coefficient. The high-density windows are merged according to the interval overlap relationship to generate clusters. The cluster center of each cluster is analyzed. The cluster center is equal to the average number of calls to sub-projects within the cluster. The number of calls to sub-projects includes the number of calls initiated and the number of calls received. A set of cluster centers is generated. S202: Obtain the number of neighboring edges for each knowledge graph node, and set the e-th knowledge graph node v... e The number of adjacent edges is denoted as A. e From knowledge graph node v e The sub-project calls the spatiotemporal matrix to extract the knowledge graph node v. e The relevant elements, for knowledge graph node v e The call instructions and the called data are analyzed to obtain the total number of calls initiated, the total number of calls made, the average data volume per call initiated, the average data volume per call made, the average standard deviation of call initiation latency, and the average standard deviation of call made latency. Based on these parameters, a feature vector V is generated. e For knowledge graph node v e Perform PageRank calculation and obtain the PageRank value, denoted as PageRank(v). e ), calculate the complexity factor for each cluster; for the f-th cluster I f The complexity factor is denoted as θ f ; S203: Obtain the number of adjacent edges for all sub-projects under the same project type, and calculate the median of the number of adjacent edges, denoted as median(A); based on the median of the number of adjacent edges, the cluster center set, and the complexity factor set, analyze the dynamic threshold of the nodes Z=α1×median(A)+α2×max(θ) f ×C f ); where C f This represents the cluster center of the f-th cluster; S3: Perform sub-project dependency analysis on the new project. When the joint index of the number of sub-projects and topology complexity exceeds the threshold, the optimization process is activated. S4: Based on the spatiotemporal correlation and business semantic features of historical call data, analyze the sub-project association score and generate storage integration strategy.

2. The information-based communication engineering method based on big data according to claim 1, characterized in that: S1 includes the following: S101: Collect the set of the number of sub-projects into which historical communication engineering projects are decomposed, denoted as P = {P1, P2, ..., P...} i ..., P I }, where P i Let represent the number of sub-projects into which the i-th historical communication project is decomposed; i∈[1,I]; I represents the total number of historical communication projects; synchronously acquire the sub-project name information, project type label, data volume information, call information, and construction duration of the historical communication project decomposition; the call information of the sub-project data includes the number of calls, the data volume information of the calls, and the standard deviation of the call delay; normalize the number of calls, the data volume information of the calls, and the standard deviation of the call delay; construct the sub-project call spatiotemporal matrix based on the call information of the processed sub-project data, and denote the sub-project call spatiotemporal matrix of the i-th historical communication project as T. i ; S102: Analyze data similarity based on the spatiotemporal matrix of each historical communication engineering project; extract call data containing pairs of sub-projects with the same name from each historical communication engineering project, and analyze data similarity based on the call data and a data similarity function; in the data similarity function, if q （j，ab） Greater than or equal to q thr The output is 1 if the output is q, otherwise the output is 0. （j，ab） This represents the average amount of data called by project a in a single instance when project b initiates a call command in the j-th historical communication project that simultaneously contains sub-projects a and b with the same name, j∈[1,J]; J represents the total number of historical communication projects that simultaneously contain sub-projects a and b with the same name, J<1; q thr This indicates the threshold for the amount of data being called, which is a system preset. Text similarity analysis is performed based on a collection of text reports from historical communication engineering projects. The mean text similarity analysis includes: generating vectors from the text reports of sub-project a and sub-project b using BERT, calculating the cosine similarity between the vectors based on the cosine function, and calculating the mean. A cross-engineering knowledge graph, denoted as G=(V,E), is constructed by combining data similarity, text similarity, and time decay coefficient; where V represents the set of nodes, which includes sub-project names and corresponding engineering type labels; and E represents the set of edges, with the weight of the edge connecting sub-project a and sub-project b denoted as W. （a，b） .

3. The information-based communication engineering method based on big data according to claim 2, characterized in that: S3 includes the following: Obtain the set of sub-projects of the current communication project, construct a directed acyclic graph (DAG) based on the set of sub-projects of the current communication project, where the set of nodes in the DAG is equal to the set of sub-projects of the current communication project, and directed edges represent dependencies between sub-projects if and only if they exist; divide the nodes into several modules according to the community detection algorithm and analyze the module degree; analyze the topology complexity, which is equal to the critical path length multiplied by the module degree divided by the total number of sub-projects; if the number of sub-projects and the topology complexity of the current communication project are both greater than the corresponding thresholds, activate the storage optimization process, where the threshold corresponding to the number of sub-projects of the current communication project is equal to the node dynamic threshold.

4. The information-based communication engineering method based on big data according to claim 3, characterized in that: S4 includes the following: S401: Extract the set of sub-projects of the current communication project. In the cross-project knowledge graph, for each sub-project name, retrieve the corresponding node, extract all the associated edges connected to the corresponding node and their weights, and form a subgraph edge set. S402: For each pair of sub-items and their weights, analyze the correlation score and record the correlation score between the current communication engineering sub-item u and sub-item v as O. （u，v） ; If the association score is greater than the corresponding threshold, the corresponding sub-item pair is marked as a candidate for physical storage co-location; according to the descending order of the association scores of the candidate pairs, the data and services of the corresponding sub-items are deployed on the same storage partition in turn.

5. A communication engineering information system based on big data, wherein the system is applied to the communication engineering information method based on big data as described in any one of claims 1-4, characterized in that, The system includes a data analysis module, a cluster analysis module, a real-time project filtering module, and an associated storage module; The data analysis module is used to access the engineering management system, sensor network and user operation logs in real time to build a dynamic knowledge graph that includes sub-project topology, spatiotemporal distribution of data generation and cross-project call chain. The clustering analysis module is used to generate a classification scheme based on the distribution characteristics of the number of historical sub-projects, using a sliding window clustering algorithm, and dynamically calibrating the critical threshold for priority integration of data storage by combining the engineering complexity factor. The real-time project filtering module is used to parse the sub-project dependencies of new projects. When the joint index of the number of sub-projects and topology complexity exceeds the threshold, the optimization process is activated. The associated storage module is used to analyze the sub-project association score and generate storage integration strategy based on the spatiotemporal correlation and business semantic characteristics of historical call data.

6. The communication engineering information system based on big data according to claim 5, characterized in that: The data analysis module includes a spatiotemporal matrix analysis unit and a cross-engineering knowledge graph analysis unit. The spatiotemporal matrix analysis unit is used to collect the set of the number of sub-projects into which historical communication engineering projects are decomposed, and simultaneously acquire the sub-project name information, project type label, data volume information, call information and construction duration of the historical communication engineering projects. The call information of the sub-project data includes the number of calls, the amount of data called, and the standard deviation of the call delay. The number of calls, the amount of data called, and the standard deviation of the call delay are normalized. Based on the call information of the processed sub-project data, a spatiotemporal matrix of sub-project calls is constructed. The cross-engineering knowledge graph analysis unit is used to analyze data similarity based on the spatiotemporal matrix of each historical communication engineering project; analyze text similarity based on the text report set of historical communication engineering projects; and construct a cross-engineering knowledge graph by combining data similarity, text similarity, and time decay coefficient.

7. The communication engineering information system based on big data according to claim 5, characterized in that: The clustering analysis module includes clustering units and dynamic threshold analysis units; The clustering unit is used to group each historical communication project according to the project type label; it performs sliding window density clustering and outputs the cluster center set; The dynamic threshold analysis unit is used to obtain the number of neighboring edges of each knowledge graph node, analyze the calling instructions initiated by the knowledge graph node and the data called, calculate the complexity factor in each cluster, and analyze the dynamic threshold of the node based on the median of the number of neighboring edges, the cluster center set, and the complexity factor set.

8. The communication engineering information system based on big data according to claim 5, characterized in that: The real-time project filtering module includes a topology complexity analysis unit and an optimization activation unit; The topology complexity analysis unit is used to obtain the set of sub-projects of the current communication project, construct a directed acyclic graph based on the set of sub-projects of the current communication project, the set of nodes in the directed acyclic graph is equal to the set of sub-projects of the current communication project, and a directed edge indicates that there is a dependency relationship between sub-projects if and only if there is one; divide the nodes into several modules according to the community detection algorithm and analyze the module degree; analyze the topology complexity, which is equal to the critical path length multiplied by the module degree divided by the total number of sub-projects; The optimization activation unit is used to activate the storage optimization process if the number of sub-projects and the topology complexity of the current communication project are both greater than the corresponding thresholds. The threshold corresponding to the number of sub-projects in the current communication project is equal to the node dynamic threshold.

9. The communication engineering information system based on big data according to claim 5, characterized in that: The associated storage module includes a subgraph edge set analysis unit and an optimized deployment unit; The subgraph edge set analysis unit is used to extract the set of sub-projects of the current communication project. In the cross-project knowledge graph, for each sub-project name, the corresponding node is retrieved, and all associated edges connected to the corresponding node and their weights are extracted to form a subgraph edge set. The optimized deployment unit is used to analyze the correlation score of each pair of sub-projects and their weights. If the correlation score is greater than the corresponding threshold, the corresponding sub-project pair is marked as a candidate for physical storage co-location. The data and services of the corresponding sub-projects are deployed on the same storage partition in descending order of the correlation scores of the candidate pairs.

Images