A VR panoramic operation and maintenance data analysis system
By introducing modules such as time series induction, dynamic pattern mining, resource allocation optimization, and performance fluctuation monitoring into the VR panoramic operation and maintenance data analysis system, the shortcomings of traditional systems in resource allocation and performance monitoring have been solved, resource utilization has been optimized, and the service quality and user experience of VR panoramic content have been improved.
Patent Information
- Application Number
- CN202410488060.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-23
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-04-23
AI Technical Summary
Traditional data analysis systems lack flexibility and efficiency when processing large-scale, high-dimensional VR panoramic operation and maintenance data, especially in resource allocation and performance monitoring. This results in an inability to respond to dynamic changes in operation and maintenance data in real time, affecting the timeliness and accuracy of operation and maintenance decisions. Furthermore, the lack of refined resource allocation strategies leads to resource waste or insufficiency, affecting the stability of VR panoramic content and user satisfaction.
The system employs a time series induction module, a dynamic pattern mining module, a resource allocation optimization module, a performance fluctuation monitoring module, a knowledge base update module, a model adaptation and adjustment module, and an incremental knowledge acquisition module. Through dynamic time warping algorithm, incremental principal component analysis, and particle swarm optimization, it achieves optimization of resource allocation and dynamic adjustment of performance monitoring.
It improved resource utilization efficiency, ensured the rational allocation of resources during operation and maintenance, significantly improved the service quality and user experience of VR panoramic content, and reduced downtime.
Smart Images

Figure CN118277793B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data analysis technology, and in particular to a VR panoramic operation and maintenance data analysis system. Background Art
[0002] The field of data analytics is a technology that focuses on extracting useful information and identifying patterns from large amounts of data through algorithms and computational methods to support decision making. The field of data analytics covers a wide range of applications, from business intelligence and market research to complex scientific research. Within this field, the development of technology enables organizations and individuals to optimize operations, improve efficiency and innovation through accurate data-driven methods.
[0003] The purpose of the VR panoramic operation and maintenance data analysis system is to provide data support for the operation and maintenance of VR panoramic content, so that the operation and maintenance team can monitor the status of VR panoramic content by analyzing the collected data, identify potential problems, and perform preventive maintenance or immediate repairs. Through in-depth analysis of the data generated during the operation and maintenance process, it aims to improve the service quality and user experience of VR panoramic content, while reducing downtime and optimizing resource allocation.
[0004] Traditional data analysis systems lack sufficient flexibility and efficiency when processing large-scale, high-dimensional VR panoramic operation and maintenance data, especially in terms of resource allocation and performance monitoring. Traditional systems use static analysis models for resource optimization and performance fluctuation monitoring, which makes it impossible to respond to dynamic changes in operation and maintenance data in real time, making it difficult to accurately identify and prevent potential problems, thus affecting the timeliness and accuracy of operation and maintenance decisions. The lack of a refined resource allocation strategy also often leads to waste or shortage of resources, which in turn affects the stability of VR panoramic content and user satisfaction.
[0005] Based on this, the present invention designs a VR panoramic operation and maintenance data analysis system to solve the above problems. Summary of the invention
[0006] The purpose of the present invention is to provide a VR panoramic operation and maintenance data analysis system to solve the problem that the traditional data analysis system proposed in the above background technology lacks sufficient flexibility and efficiency when processing large-scale, high-dimensional VR panoramic operation and maintenance data, especially in terms of resource allocation and performance monitoring. The traditional system uses static analysis models for resource optimization and performance fluctuation monitoring, which makes it impossible to respond to dynamic changes in operation and maintenance data in real time, and it is difficult to accurately identify and prevent potential problems, thereby affecting the timeliness and accuracy of operation and maintenance decisions. The lack of a refined resource allocation strategy also often leads to waste or shortage of resources, which in turn affects the stability of VR panoramic content and user satisfaction.
[0007] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a VR panoramic operation and maintenance data analysis system includes a time series induction module, a dynamic pattern mining module, a resource allocation optimization module, a performance fluctuation monitoring module, a knowledge base update module, a model adaptation adjustment module, an incremental knowledge acquisition module, and a scheduling plan formulation module;
[0008] The time series induction module performs timestamp marking based on the data collected in VR panoramic operation and maintenance, and sorts the collected data points according to the timestamps to determine the time sequence of the data and generate a time-stamped data set;
[0009] The dynamic pattern mining module uses a dynamic time warping algorithm to serialize data points based on a time-tagged data set, calculates and compares the time difference and value changes between data points, identifies common patterns and abnormal fluctuations in the data, and generates a pattern recognition set;
[0010] The resource allocation optimization module uses incremental principal component analysis based on pattern recognition sets to evaluate the efficiency of current resource allocation, discover resource usage bottlenecks, and adjust resource allocation parameters, including allocation amount and priority, to obtain an allocation optimization plan;
[0011] The performance fluctuation monitoring module continuously monitors key system performance indicators, including response time and processing capacity, based on the allocation optimization scheme, records the fluctuation of multiple performance indicators, provides data support for maintenance and optimization, and obtains performance fluctuation records;
[0012] The knowledge base update module updates and supplements the existing operation and maintenance knowledge base based on the performance fluctuation records, including knowledge items such as fault diagnosis methods and maintenance strategies, and establishes an updated knowledge base;
[0013] The model adaptation adjustment module automatically adjusts model parameters, including learning rate and weight, based on the updated knowledge base and according to the added operation and maintenance knowledge, optimizes the accuracy and adaptability of the model, and generates an adaptive adjustment model;
[0014] The incremental knowledge acquisition module analyzes the newly collected operation and maintenance data according to the adaptive adjustment model, identifies and extracts data and patterns that are not included in the existing knowledge base, and obtains an incremental knowledge set;
[0015] The scheduling plan formulation module evaluates and re-formulates the existing scheduling plan based on the incremental knowledge set, combines operation and maintenance requirements and resource availability, optimizes and adjusts the scheduling plan, and generates a resource scheduling plan.
[0016] Preferably, the time-stamped data set includes the collection time, value and data type of the data point, the pattern recognition set includes periodic fluctuations, abnormal event identification and trend change indicators, the allocation optimization plan includes adjusted resource allocation amount, resource allocation priority and resource release strategy, the performance fluctuation record includes response time record, processing capacity record and system load record, the updated knowledge base includes newly added fault handling strategies, maintenance operation steps and preventive measures, the adaptive adjustment model includes adjusted model learning rate, optimized weight allocation and updated feature selection, and the incremental knowledge set includes identified operating modes, unrecorded fault types and added maintenance proposals.
[0017] Preferably, the time series induction module includes a data time marking submodule, a data sorting submodule, and a data set construction submodule;
[0018] The data time stamp submodule traverses the data points based on the data collected in the VR panoramic operation and maintenance, assigns a determined collection timestamp to the data points, determines the collection time point reflected by the data points, and generates a time stamp list;
[0019] The data sorting submodule sorts the data points in ascending order based on the time stamp list, using the timestamp as a keyword, and verifies whether the data list is arranged according to time to obtain a sorted data list;
[0020] The data set construction submodule integrates data points based on the sorted data list, constructs a data set arranged in chronological order, provides a basis for analysis and processing, and generates a time-stamped data set.
[0021] Preferably, the dynamic pattern mining module includes a serialization processing submodule, a pattern comparison submodule, and a pattern recognition set construction submodule;
[0022] The serialization processing submodule serializes the data points in chronological order based on the time-stamped data set. If the time intervals of the data points are continuous, they are regarded as a sequence to generate a serialized data set.
[0023] The pattern comparison submodule calculates the time difference and value change of the data points in the sequence based on the serialized data set and uses the dynamic time warping algorithm. If there are similar time difference and value change patterns between the sequences, they are marked as potential similar patterns to obtain a potential pattern set;
[0024] The dynamic time warping algorithm is according to the formula:
[0025] D[i,j]=min{D[i-1,j-1]×α,D[i-1,j]×β,D[i,j-1]×γ}+(d(s i ,t j)×δ)
[0026] Calculate the distance between time series s and t and generate the value of distance matrix D;
[0027] Where D[i,j] represents the minimum cumulative distance between the first i elements of sequence s and the first j elements of sequence t, d(s i ,t j ) is the distance measure between the i-th element of sequence s and the j-th element of sequence t, α, β, γ are weight factors for adjusting the distance calculation of matching points between sequences, and δ is a parameter for sensitivity to numerical changes;
[0028] The pattern recognition set construction submodule is based on the potential pattern set and uses online stochastic gradient descent to identify common patterns and abnormal fluctuations in the data by judging the frequency of occurrence and degree of abnormality of the pattern, and generates a pattern recognition set.
[0029] Preferably, the resource allocation optimization module includes an efficiency evaluation submodule, a resource bottleneck mining submodule, and an allocation scheme construction submodule;
[0030] The efficiency evaluation submodule evaluates the allocation efficiency of current resources based on the pattern recognition set, and calculates the resource utilization rate and satisfaction rate. If the resource utilization rate is low, it is marked as a low-efficiency resource, and an efficiency evaluation record is generated;
[0031] The resource bottleneck mining submodule uses incremental principal component analysis based on efficiency evaluation records to analyze resource usage. If a resource frequently becomes a source of inefficiency, it is marked as a resource bottleneck, and a resource bottleneck list is obtained.
[0032] The incremental principal component analysis method is according to the formula:
[0033] X new =(X+ΔX)W+αS
[0034] Compute a new representation of resource usage data and obtain a list of resource bottlenecks;
[0035] Among them, X is the original resource usage data matrix, ΔX represents the change in resource usage data associated with time, W is the weight matrix, α is the adjustment coefficient, S is used for balance, and S is the sensitivity score matrix of resource usage;
[0036] The allocation scheme construction submodule uses particle swarm optimization based on the resource bottleneck list to adjust the parameters of resource allocation, including resource allocation amount and resource use priority, and obtains an allocation optimization scheme in response to the resource bottleneck problem.
[0037] Preferably, the performance fluctuation monitoring module includes a performance index monitoring submodule, a fluctuation data sorting submodule, and a fluctuation situation analysis submodule;
[0038] The performance indicator monitoring submodule monitors key system performance indicators based on the allocation optimization scheme, including response time and processing capacity, records changes in indicators, including increase or decrease amplitude and change frequency, and generates performance change records;
[0039] The fluctuation data sorting submodule screens the records whose fluctuation range exceeds a predetermined threshold based on the performance change records, records the occurrence time and fluctuation degree, and obtains a fluctuation detail record table;
[0040] The fluctuation situation analysis submodule analyzes the data in the fluctuation detail record table, identifies the fluctuation rules and performance indicators that frequently fluctuate, summarizes potential fluctuation causes, and obtains performance fluctuation records.
[0041] Preferably, the knowledge base update module includes a fluctuation cause classification submodule, an update strategy formulation submodule, and a knowledge base update execution submodule;
[0042] The fluctuation cause classification submodule classifies the fluctuation causes into known failure modes and new problems based on the performance fluctuation records, marks and classifies the new problems, and generates a classified fluctuation cause list;
[0043] The update strategy formulation submodule determines the fault handling methods and maintenance strategies that need to be added to the knowledge base based on the classified fluctuation reason list, and formulates an update strategy, including knowledge content and update priority, to obtain an update strategy solution;
[0044] The knowledge base update execution submodule executes the update strategy scheme, adds the new fault handling method and maintenance strategy to the existing operation and maintenance knowledge base, updates the knowledge base, and establishes an updated knowledge base.
[0045] Preferably, the model adaptation and adjustment module includes a parameter analysis submodule, an adaptability adjustment submodule, and a model verification submodule;
[0046] The parameter analysis submodule analyzes the compatibility of the current parameter configuration of the model based on the updated knowledge base, evaluates the impact on the existing model parameter configuration, determines the parameter items that need to be adjusted, including learning rate and weight, and generates a parameter adjustment requirement list;
[0047] The adaptive adjustment submodule adjusts the learning rate and weight of the model according to the parameter adjustment requirement list, matches the model parameters with the operation and maintenance knowledge, and performs parameter optimization to obtain the optimized model parameter configuration;
[0048] The model verification submodule uses the optimized model parameter configuration to test the model, verify the accuracy and adaptability of the adjusted model, and confirm whether the performance of the model meets expectations, thereby generating an adaptive adjustment model.
[0049] Preferably, the incremental knowledge acquisition module includes a data analysis submodule, a knowledge extraction submodule, and a knowledge integration submodule;
[0050] The data analysis submodule analyzes the newly collected operation and maintenance data based on the adaptive adjustment model, identifies patterns and anomalies in the data, marks information that the model fails to identify, and generates unidentified information records;
[0051] The knowledge extraction submodule extracts data and patterns that can be converted into operation and maintenance knowledge based on the unidentified information records, and verifies the supplementary value of the information to the existing knowledge base to form a list of convertible knowledge;
[0052] The knowledge integration submodule integrates the information and patterns in the convertible knowledge list into the existing operation and maintenance knowledge base, updates the knowledge points of the knowledge base, and obtains an incremental knowledge set.
[0053] Preferably, the scheduling scheme formulation module includes a demand analysis submodule, a resource assessment submodule, and a scheme construction submodule;
[0054] The demand analysis submodule collects and analyzes current operation and maintenance requirements based on the incremental knowledge set, determines task priorities and required resource types, identifies key operation and maintenance activities and emergency repair tasks, and generates demand analysis data;
[0055] The resource assessment submodule assesses the existing resource inventory and availability, including hardware resources, software tools, and human resources, based on the demand analysis data, determines the current status and scheduling capabilities of the resources, and obtains resource availability information;
[0056] The scheme building submodule evaluates the existing scheduling plan based on the resource availability information, re-formulates and optimizes the scheduling plan, determines whether it meets the resource utilization and operation and maintenance requirements, and generates a resource scheduling plan.
[0057] Compared with the prior art, the present invention has the following beneficial effects: the present invention adopts a dynamic time warping algorithm, incremental principal component analysis and particle swarm optimization, and in the resource allocation optimization module, by evaluating the resource allocation efficiency and exploring the resource usage bottleneck, adjusting the resource allocation parameters, and achieving the optimization of resource usage. When the efficiency evaluation record shows that a resource usage rate is low, the resource bottleneck problem is effectively solved by adjusting the resource allocation amount and priority, which not only improves the resource utilization efficiency, but also ensures the reasonable allocation and use of resources during the operation and maintenance process, thereby significantly improving the service quality and user experience of VR panoramic content and reducing the downtime. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing 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.
[0059] Figure 1 A module diagram of a VR panoramic operation and maintenance data analysis system is proposed for the present invention;
[0060] Figure 2 A system framework diagram of a VR panoramic operation and maintenance data analysis system is proposed for the present invention;
[0061] Figure 3 A flow chart of a time series induction module in a VR panoramic operation and maintenance data analysis system is proposed for the present invention;
[0062] Figure 4 The present invention proposes a flow chart of a dynamic mode mining module in a VR panoramic operation and maintenance data analysis system;
[0063] Figure 5 The present invention proposes a flow chart of resource allocation optimization module in VR panoramic operation and maintenance data analysis system;
[0064] Figure 6 A flow chart of a performance fluctuation monitoring module in a VR panoramic operation and maintenance data analysis system is proposed for the present invention;
[0065] Figure 7 The present invention proposes a flow chart of a knowledge base update module in a VR panoramic operation and maintenance data analysis system;
[0066] Figure 8 The present invention proposes a flow chart of a model adaptation adjustment module in a VR panoramic operation and maintenance data analysis system;
[0067] Fig. 9 The present invention proposes a flow chart of an incremental knowledge acquisition module in a VR panoramic operation and maintenance data analysis system;
[0068] Fig.10 The present invention proposes a flow chart of a scheduling scheme formulation module in a VR panoramic operation and maintenance data analysis system. DETAILED DESCRIPTION
[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0070] See also Figure 1 ,The present invention provides a technical solution: a VR panoramic operation and maintenance data analysis system includes a time series induction module, a dynamic pattern mining module, a resource allocation optimization module, a performance fluctuation monitoring module, a knowledge base update module, a model adaptation adjustment module, an incremental knowledge acquisition module, and a scheduling plan formulation module;
[0071] The time series induction module performs timestamp marking based on the data collected in VR panoramic operation and maintenance, and sorts the collected data points according to the timestamps to determine the time sequence of the data and generate a time-stamped data set;
[0072] The dynamic pattern mining module uses a dynamic time warping algorithm to serialize data points based on time-stamped data sets, calculates and compares the time difference and value changes between data points, identifies common patterns and abnormal fluctuations in the data, and generates a pattern recognition set;
[0073] The resource allocation optimization module uses incremental principal component analysis based on pattern recognition sets to evaluate the efficiency of current resource allocation, discover resource usage bottlenecks, and adjust resource allocation parameters, including allocation amount and priority, to obtain an allocation optimization plan;
[0074] The performance fluctuation monitoring module continuously monitors key system performance indicators, including response time and processing capacity, based on the allocation optimization scheme, records the fluctuation of multiple performance indicators, provides data support for maintenance and optimization, and obtains performance fluctuation records;
[0075] The knowledge base update module updates and supplements the existing operation and maintenance knowledge base based on the performance fluctuation records, including knowledge items such as fault diagnosis methods and maintenance strategies, and establishes an updated knowledge base;
[0076] The model adaptation module automatically adjusts model parameters, including learning rate and weight, based on the updated knowledge base and added operation and maintenance knowledge, optimizes the accuracy and adaptability of the model, and generates an adaptive adjustment model;
[0077] The incremental knowledge acquisition module analyzes the newly collected operation and maintenance data based on the adaptive adjustment model, identifies and extracts data and patterns that are not included in the existing knowledge base, and obtains an incremental knowledge set;
[0078] The scheduling plan formulation module evaluates and reformulates the existing scheduling plan based on the incremental knowledge set, combines operation and maintenance requirements and resource availability, optimizes and adjusts the scheduling plan, and generates a resource scheduling plan.
[0079] The time-stamped dataset includes the collection time, value and data type of the data points. The pattern recognition set includes periodic fluctuations, abnormal event identification and trend change indicators. The allocation optimization plan includes the adjusted resource allocation amount, resource allocation priority and resource release strategy. The performance fluctuation record includes response time record, processing capacity record and system load record. The updated knowledge base includes newly added fault handling strategies, maintenance operation steps and preventive measures. The adaptive adjustment model includes adjusted model learning rate, optimized weight allocation and updated feature selection. The incremental knowledge set includes identified operation modes, unrecorded fault types and added maintenance proposals.
[0080] See also Figure 2 and Figure 3 ,The time series induction module includes a data time marking submodule, a data sorting submodule, and a data set construction submodule;
[0081] The data time stamp submodule traverses the data points based on the data collected in VR panoramic operation and maintenance, assigns a certain collection timestamp to the data points, determines the collection time point reflected by the data points, and generates a time stamp list;
[0082] The data sorting submodule uses the timestamp as the keyword to sort the data points in ascending order based on the time stamp list, and verifies whether the data list is arranged according to time to obtain the sorted data list;
[0083] The dataset construction submodule integrates the data points based on the sorted data list, constructs a chronological dataset, provides a basis for analysis and processing, and generates a time-stamped dataset.
[0084] In the data time stamp submodule, based on the data collected by VR panoramic operation and maintenance, a traversal algorithm is used to traverse the data points through Python's for loop structure, and a certain collection timestamp is assigned to each data point. The timestamp is obtained by the system's current time function datetime.now() to ensure that the timestamp of each data point reflects its collection time point. Through the traversal operation, a time stamp list is generated.
[0085] In the data sorting submodule, the quick sort algorithm is used, based on the timestamp in the timestamp list as the sorting keyword, and the key parameter is set to the timestamp attribute of the data point through the sorted() function in Python to sort the data points in ascending order. The sorting process confirms the order of the data points by comparing the timestamp values, and rearranges the data points through the divide-and-conquer method of the quick sort algorithm to verify whether the data list is arranged in ascending time order to obtain the sorted data list.
[0086] In the dataset construction submodule, based on the sorted data list, a list merging algorithm is used to integrate the sorted data points into a new list through Python's extend() method. This process iterates the sorted data list and adds data points to the new list one by one to construct a chronologically arranged dataset. This dataset provides the basis for subsequent analysis and processing to generate a time-stamped dataset.
[0087] See also Figure 2 and Figure 4 ,The dynamic pattern mining module includes a serialization processing submodule, a pattern comparison submodule, and a pattern recognition set construction submodule;
[0088] The serialization processing submodule serializes the data points in chronological order based on the time-stamped data set. If the time intervals of the data points are continuous, they are regarded as a sequence to generate a serialized data set.
[0089] The pattern comparison submodule is based on the serialized data set and uses the dynamic time warping algorithm to calculate the time difference and value change of the data points in the sequence. If there are similar time difference and value change patterns between the sequences, they are marked as potential similar patterns to obtain the potential pattern set;
[0090] The pattern recognition set construction submodule is based on the potential pattern set and uses online stochastic gradient descent to identify common patterns and abnormal fluctuations in the data by judging the frequency of occurrence and degree of abnormality of the pattern, and generates a pattern recognition set.
[0091] In the serialization processing submodule, based on the time-tagged dataset, the time serialization method is adopted and implemented using Python's Pandas library. The time-tagged data is input through the DataFrame method, and time serialization processing is performed. The continuity of the data point time interval is detected through the resample method, and the data points of the continuous time interval are marked as a sequence to generate a serialized dataset.
[0092] In the pattern comparison submodule, based on the serialized data set, the dynamic time warping algorithm is adopted and implemented using Python's DTW library. The serialized data set is input through the dtw method, and the time difference and numerical change calculations are performed. The same time difference and numerical change patterns are identified through the calculated distance measurement, marked as potential similar patterns, and the potential pattern set is obtained.
[0093] Dynamic time warping algorithm, according to the formula:
[0094] D[i,j]=min{D[i-1,j-1]×α,D[i-1,j]×β,D[i,j-1]×γ}+(d(s i ,t j )×δ)
[0095] Calculate the distance between time series s and t and generate the value of distance matrix D;
[0096] Where D[i,j] represents the minimum cumulative distance between the first i elements of sequence s and the first j elements of sequence t, d(s i ,t j ) is the distance measure between the i-th element of sequence s and the j-th element of sequence t, α, β, γ are weight factors for adjusting the distance calculation of matching points between sequences, and δ is a parameter for sensitivity to numerical changes.
[0097] Execution process:
[0098] S1: Create a matrix whose size is based on the length of the two time series, initially filled with a large number, except for the starting point which is set to 0;
[0099] S2: For each pair of points in the time series, calculate the distance between them. This usually uses the Euclidean distance, but other distance metrics can be selected as needed;
[0100] S3: Recursively fill the entire matrix according to specific rules. The rules include taking into account the matching points between time series and the sensitivity to numerical changes, analyzing the characteristics of time series data, adjusting weight coefficients, using validation sets or cross-validation to evaluate performance, and selecting the optimal coefficient value to improve the performance of the algorithm;
[0101] S4: By comparing the distances between time series, find out the sequences with similar time change and value change patterns, and classify the identified similar time series into a group to form a set.
[0102] In the pattern recognition set construction submodule, based on the potential pattern set, an online stochastic gradient descent algorithm is used, which is implemented using Python's SciKit-Learn library. The potential pattern set is input through the SGDClassifier method, and the pattern frequency and abnormality analysis is performed. By setting the loss function to log loss, common patterns and abnormal fluctuations in the data are identified to generate a pattern recognition set.
[0103] See also Figure 2 and Figure 5 ,The resource allocation optimization module includes the efficiency evaluation submodule, the resource bottleneck mining submodule, and the allocation scheme building submodule;
[0104] The efficiency evaluation submodule evaluates the current resource allocation efficiency based on the pattern recognition set. It calculates the resource utilization rate and satisfaction rate. If the resource utilization rate is low, it is marked as an inefficient resource and an efficiency evaluation record is generated.
[0105] The resource bottleneck mining submodule uses incremental principal component analysis based on efficiency evaluation records to analyze resource usage. If a resource frequently becomes the source of inefficiency, it is marked as a resource bottleneck and a resource bottleneck list is obtained.
[0106] The allocation scheme construction submodule uses particle swarm optimization based on the resource bottleneck list to adjust the resource allocation parameters, including resource allocation amount and resource usage priority, and obtains the allocation optimization scheme in response to the resource bottleneck problem.
[0107] In the efficiency evaluation submodule, based on the pattern recognition set, the resource utilization analysis method is adopted and implemented using Python's Pandas library. The current resource usage data is input through the DataFrame method, and the resource utilization calculation is performed. Through the comparison operation, the resources below the predetermined threshold are marked as inefficient resources, and the efficiency evaluation records are generated.
[0108] In the resource bottleneck mining submodule, based on the efficiency evaluation records, the incremental principal component analysis method is adopted and implemented using Python's Scikit-learn library. The resource usage data is input through the IncrementalPCA method, and principal component analysis is performed. The resources that frequently become the source of inefficiency are identified through the explained variance ratio, marked as resource bottlenecks, and a list of resource bottlenecks is obtained.
[0109] Incremental principal component analysis method, according to the formula:
[0110] X new =(X+ΔX)W+αS
[0111] Compute a new representation of resource usage data to obtain a weighted list of resource bottlenecks;
[0112] Among them, X is the original resource usage data matrix, ΔX represents the time-related change in resource usage data, W is the weight matrix obtained by incremental principal component analysis, α is the adjustment coefficient, S is used for balance, and S is the sensitivity or importance score matrix of resource usage.
[0113] Execution process
[0114] The incremental principal component analysis method calculates a new representation of the resource usage data by multiplying the sum of the original resource usage data matrix and its time variation by the principal component weight matrix according to the formula, and then adding the adjustment coefficient multiplied by the resource usage sensitivity or importance score matrix to obtain the explained variance ratio.
[0115] In the allocation scheme construction submodule, based on the resource bottleneck list, the particle swarm optimization algorithm is adopted and implemented using Python's PySwarms library. The resource allocation parameters and resource bottleneck problems are input through the ps.single.GlobalBestPSO method, and particle swarm optimization is performed. The resource allocation amount and resource usage priority are adjusted by setting the objective function to obtain the allocation optimization plan corresponding to the resource bottleneck problem.
[0116] See also Figure 2 and Figure 6 ,The performance fluctuation monitoring module includes a performance index monitoring submodule, a fluctuation data sorting submodule, and a fluctuation situation analysis submodule;
[0117] The performance indicator monitoring submodule monitors the key system performance indicators based on the allocation optimization scheme, including response time and processing capacity, records the changes in the indicators, including the increase or decrease amplitude and change frequency, and generates performance change records;
[0118] The fluctuation data sorting submodule screens the records whose fluctuation range exceeds the predetermined threshold based on the performance change records, records the occurrence time and fluctuation degree, and obtains the fluctuation detail record table;
[0119] The fluctuation analysis submodule analyzes the data in the fluctuation detail record table, identifies the fluctuation patterns and performance indicators that frequently fluctuate, summarizes potential fluctuation causes, and obtains performance fluctuation records.
[0120] In the performance indicator monitoring submodule, based on the allocation optimization plan, the time series analysis method is adopted, and it is implemented using Python's Pandas library and Statsmodels library. The key indicator data of system performance is input through the DataFrame method, and time series analysis is performed. The change trend of response time and processing capacity is predicted through the ARIMA model, and the increase or decrease amplitude and change frequency are recorded to generate performance change records.
[0121] In the fluctuation data sorting submodule, based on the performance change records, the threshold analysis method is adopted and implemented using Python's NumPy library. The performance change records are input through the loadtxt method, and threshold filtering is performed. The fluctuation records that exceed the predetermined threshold are screened out through the logical_and method, and the occurrence time and degree of fluctuation are recorded to obtain the fluctuation details record table.
[0122] In the fluctuation analysis submodule, the data in the fluctuation details record table are analyzed. A clustering algorithm is used and implemented using Python's Scikit-learn library. The data in the fluctuation details record table is input through the KMeans method to perform cluster analysis. The inertia_ attribute is used to identify the pattern of fluctuations and the performance indicators that frequently fluctuate, summarize the potential causes of fluctuations, and obtain performance fluctuation records.
[0123] See also Figure 2 and Figure 7 ,The knowledge base update module includes a fluctuation cause classification submodule, an ,update strategy formulation submodule, and a knowledge base update execution submodule;
[0124] The fluctuation cause classification submodule divides the fluctuation causes into known failure modes and new problems based on the performance fluctuation records, marks and classifies the new problems, and generates a classified fluctuation cause list;
[0125] The update strategy formulation submodule determines the fault handling methods and maintenance strategies that need to be added to the knowledge base based on the classified fluctuation cause list, and formulates an update strategy, including knowledge content and update priority, to obtain an update strategy solution;
[0126] The knowledge base update execution submodule executes the update strategy plan, adds new fault handling methods and maintenance strategies to the existing operation and maintenance knowledge base, updates the knowledge base, and establishes an updated knowledge base.
[0127] In the fluctuation cause classification submodule, based on the performance fluctuation records, the text mining method is adopted and implemented using Python's NLTK library. The performance fluctuation records are input through the nltk.tokenize.word_tokenize method, and text segmentation is performed. The word frequency is counted through the nltk.probability.FreqDist method to distinguish known fault modes and new problems. The part-of-speech tagging is performed through the nltk.tag.pos_tag method to mark and classify new problems and generate a classified fluctuation cause list.
[0128] In the update strategy formulation submodule, according to the classified fluctuation cause list, the decision tree algorithm is adopted and implemented by Python's Scikit-learn library. The classified fluctuation cause list is input through the DecisionTreeClassifier method, and the decision tree classification is performed. The decision tree is visualized through the tree.export_graphviz method to determine the fault handling method and maintenance strategy newly added to the knowledge base. The priority queue algorithm is implemented through Python's queue.PriorityQueue class to prioritize the update strategies and obtain the update strategy plan.
[0129] In the knowledge base update execution submodule, the update strategy plan is executed, database transaction processing technology is adopted, and Python's SQLite3 library is used to implement it. The sqlite3.connect method is used to connect to the operation and maintenance knowledge base database, and transaction processing is performed. The new fault handling method and maintenance strategy are input through the cursor.execute method. The COMMIT command is used to ensure that all updates are safely applied to the database and establish an updated knowledge base.
[0130] See also Figure 2 and Figure 8 ,The model adaptation adjustment module includes a parameter analysis submodule, an adaptive adjustment submodule, and a model verification submodule;
[0131] The parameter analysis submodule analyzes the matching of the current model parameter configuration based on the updated knowledge base, evaluates the impact on the existing model parameter configuration, determines the parameters that need to be adjusted, including learning rate and weight, and generates a list of parameter adjustment requirements;
[0132] The adaptive adjustment submodule adjusts the learning rate and weight of the model according to the parameter adjustment requirement list, matches the model parameters with operation and maintenance knowledge, and optimizes the parameters to obtain the optimized model parameter configuration;
[0133] The model verification submodule uses the optimized model parameter configuration to test the model, verify the accuracy and adaptability of the adjusted model, and confirm whether the performance of the model meets expectations, generating an adaptive adjustment model.
[0134] In the parameter analysis submodule, based on the updated knowledge base, the decision tree algorithm is adopted and implemented using Python's Scikit-learn library. The current model parameters and the updated knowledge base content are input through the fit method to perform parameter analysis. The list of parameter items that need to be adjusted is output through the predict method, including specific learning rate values and weight values, to generate a list of parameter adjustment requirements.
[0135] In the adaptive adjustment submodule, according to the parameter adjustment requirement list, the gradient descent method is used and implemented using Python's TensorFlow library. The model parameter optimization process is monitored through the GradientTape method, the learning rate and weight are adjusted, and the apply_gradients method is used to apply the new learning rate and weight to the model to complete the optimization of the model parameters and obtain the optimized model parameter configuration.
[0136] In the model validation submodule, the optimized model parameter configuration is used, the cross-validation method is adopted, and the Python Scikit-learn library is used to implement it. The optimized model parameters and test data set are input through the cross_val_score method, and the model test is performed to verify the accuracy and adaptability of the adjusted model, confirm whether the model performance meets expectations, and generate an adaptive adjustment model.
[0137] See also Figure 2 and Fig. 9 ,The incremental knowledge acquisition module includes a data analysis submodule, a knowledge extraction submodule, and a knowledge integration submodule;
[0138] The data analysis submodule analyzes the newly collected operation and maintenance data based on the adaptive adjustment model, identifies patterns and anomalies in the data, marks information that the model fails to identify, and generates records of unidentified information;
[0139] The knowledge extraction submodule extracts data and patterns that can be converted into operation and maintenance knowledge based on unidentified information records, verifies the supplementary value of the information to the existing knowledge base, and forms a list of convertible knowledge;
[0140] The knowledge integration submodule integrates the information and patterns in the convertible knowledge list into the existing operation and maintenance knowledge base, updates the knowledge points of the knowledge base, and obtains an incremental knowledge set.
[0141] In the data analysis submodule, based on the adaptive adjustment model, the clustering algorithm is adopted and implemented using Python's Scikit-learn library. The newly collected operation and maintenance data is input through the KMeans method, and data clustering is performed. The clustering effect is evaluated through the inertia_ attribute, and the information that the model fails to recognize is marked, generating unrecognized information records.
[0142] In the knowledge extraction submodule, natural language processing technology is used according to the unrecognized information records, and Python's NLTK library is used to implement it. The unrecognized information is input through the nltk.tokenize.word_tokenize method, and information segmentation is performed. The word frequency is counted through the nltk.probability.FreqDist method, and the data and patterns that can be converted into operation and maintenance knowledge are extracted to form a list of convertible knowledge.
[0143] In the knowledge integration submodule, the information and patterns in the convertible knowledge list are integrated into the existing operation and maintenance knowledge base. Graph database technology is used and implemented using the Neo4j database. The convertible knowledge list is input through the CREATE statement of the Cypher query language, and knowledge point creation and relationship establishment are executed to update the knowledge points of the knowledge base and obtain an incremental knowledge set.
[0144] See also Figure 2and Fig.10 ,The scheduling scheme formulation module includes the demand analysis submodule, the resource evaluation submodule, and the scheme building submodule;
[0145] The demand analysis submodule collects and analyzes current operation and maintenance requirements based on the incremental knowledge set, determines task priorities and required resource types, identifies key operation and maintenance activities and emergency repair tasks, and generates demand analysis data;
[0146] The resource assessment submodule evaluates the existing resource inventory and availability based on the demand analysis data, including hardware resources, software tools, and human resources, determines the current status and scheduling capabilities of resources, and obtains resource availability information;
[0147] The scheme construction submodule evaluates the existing scheduling plan based on the resource availability information, re-formulates and optimizes the scheduling plan, determines whether it meets the resource utilization and operation and maintenance requirements, and generates a resource scheduling plan.
[0148] In the demand analysis submodule, based on the incremental knowledge set, the decision support system method is adopted and implemented using Python's Pandas library. The current operation and maintenance demand data is input through the DataFrame method, demand classification is performed, and task priorities are sorted through the sort_values method. Key operation and maintenance activities and emergency repair tasks are identified to generate demand analysis data.
[0149] In the resource assessment submodule, based on the demand analysis data, a resource optimization algorithm is adopted and implemented using Python's NumPy library. The existing resource inventory data is input through the array method, resource matching is performed, and the optimal resource allocation plan is determined through the linear_sum_assignment method to obtain resource availability information.
[0150] In the scheme construction submodule, a scheduling algorithm is adopted based on resource availability information and implemented using Python's SciPy library. The resource allocation plan and operation and maintenance requirements are input through the minimize method, and the scheduling plan optimization is performed. The constraints parameter is used to ensure that the scheduling plan meets the resource utilization and operation and maintenance requirements, and a resource scheduling plan is generated.
[0151] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0152] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.
Claims
1. A VR panoramic operation and maintenance data analysis system, characterized in that: The system includes a time series induction module, a dynamic pattern mining module, a resource allocation optimization module, a performance fluctuation monitoring module, a knowledge base update module, a model adaptation adjustment module, an incremental knowledge acquisition module, and a scheduling plan formulation module; The time series induction module performs timestamp marking based on the data collected in VR panoramic operation and maintenance, and sorts the collected data points according to the timestamps to determine the time sequence of the data and generate a time-stamped data set; The dynamic pattern mining module uses a dynamic time warping algorithm to serialize data points based on a time-stamped data set, and identifies common patterns and abnormal fluctuations in the data by calculating and comparing the time difference and value changes between data points, and generates a pattern recognition set, which includes periodic fluctuations, abnormal event identifiers, and trend change indicators; The resource allocation optimization module uses incremental principal component analysis based on pattern recognition sets to evaluate the efficiency of current resource allocation, discover resource usage bottlenecks, and adjust resource allocation parameters, including allocation amount and priority, to obtain an allocation optimization plan; The performance fluctuation monitoring module continuously monitors key system performance indicators, including response time and processing capacity, based on the allocation optimization scheme, records the fluctuation of multiple performance indicators, provides data support for maintenance and optimization, and obtains performance fluctuation records; The knowledge base update module updates and supplements the existing operation and maintenance knowledge base based on the performance fluctuation records, including knowledge items such as fault diagnosis methods and maintenance strategies, and establishes an updated knowledge base; The model adaptation adjustment module automatically adjusts model parameters, including learning rate and weight, based on the updated knowledge base and according to the added operation and maintenance knowledge, optimizes the accuracy and adaptability of the model, and generates an adaptive adjustment model; The incremental knowledge acquisition module analyzes the newly collected operation and maintenance data according to the adaptive adjustment model, identifies and extracts data and patterns that are not included in the existing knowledge base, and obtains an incremental knowledge set; The scheduling plan formulation module evaluates and re-formulates the existing scheduling plan based on the incremental knowledge set, combines operation and maintenance requirements and resource availability, optimizes and adjusts the scheduling plan, and generates a resource scheduling plan; The dynamic pattern mining module includes a serialization processing submodule, a pattern comparison submodule, and a pattern recognition set construction submodule; The serialization processing submodule serializes the data points in chronological order based on the time-stamped data set. If the time intervals of the data points are continuous, they are regarded as a sequence to generate a serialized data set. The pattern comparison submodule calculates the time difference and value change of the data points in the sequence based on the serialized data set and uses the dynamic time warping algorithm. If there are similar time difference and value change patterns between the sequences, they are marked as potential similar patterns to obtain a potential pattern set; The dynamic time warping algorithm is according to the formula: ; Calculate the distance between time series S and t and generate the value of distance matrix D; in, Representation sequence Before Elements and sequences Before The minimum cumulative distance between elements, Is a sequence No. Elements and sequences No. The distance measure between elements is is the weight factor for adjusting the distance calculation of matching points between sequences, is the parameter of sensitivity to numerical changes; The pattern recognition set construction submodule is based on the potential pattern set and uses online stochastic gradient descent to identify common patterns and abnormal fluctuations in the data by judging the frequency of occurrence and degree of abnormality of the pattern, thereby generating a pattern recognition set; The resource allocation optimization module includes an efficiency evaluation submodule, a resource bottleneck mining submodule, and an allocation scheme construction submodule; The efficiency evaluation submodule evaluates the allocation efficiency of current resources based on the pattern recognition set, and calculates the resource utilization rate and satisfaction rate. If the resource utilization rate is low, it is marked as a low-efficiency resource, and an efficiency evaluation record is generated; The resource bottleneck mining submodule uses incremental principal component analysis based on efficiency evaluation records to analyze resource usage. If a resource frequently becomes a source of inefficiency, it is marked as a resource bottleneck, and a resource bottleneck list is obtained. The incremental principal component analysis method is according to the formula: ; Compute a new representation of resource usage data and obtain a list of resource bottlenecks; in, Using data matrix for original resources, Indicates the change in resource usage data associated with time. is the weight matrix, is the adjustment factor, For balance, is the sensitivity scoring matrix of resource usage; The allocation scheme construction submodule uses particle swarm optimization based on the resource bottleneck list to adjust the parameters of resource allocation, including resource allocation amount and resource use priority, and obtains an allocation optimization scheme in response to the resource bottleneck problem.
2. A VR panoramic operation and maintenance data analysis system according to claim 1, characterized in that: The time-stamped data set includes the collection time, value and data type of the data point, the allocation optimization plan includes the adjusted resource allocation amount, resource allocation priority and resource release strategy, the performance fluctuation record includes response time record, processing capacity record and system load record, the updated knowledge base includes newly added fault handling strategies, maintenance operation steps and preventive measures, the adaptive adjustment model includes adjusted model learning rate, optimized weight allocation and updated feature selection, and the incremental knowledge set includes identified operation modes, unrecorded fault types and added maintenance proposals.
3. A VR panoramic operation and maintenance data analysis system according to claim 1, characterized in that: The time series induction module includes a data time marking submodule, a data sorting submodule, and a data set construction submodule; The data time stamp submodule traverses the data points based on the data collected in the VR panoramic operation and maintenance, assigns a determined collection timestamp to the data points, determines the collection time point reflected by the data points, and generates a time stamp list; The data sorting submodule sorts the data points in ascending order based on the time stamp list, using the timestamp as a keyword, and verifies whether the data list is arranged according to time to obtain a sorted data list; The data set construction submodule integrates data points based on the sorted data list, constructs a data set arranged in chronological order, provides a basis for analysis and processing, and generates a time-stamped data set.
4. A VR panoramic operation and maintenance data analysis system according to claim 1, characterized in that: The performance fluctuation monitoring module includes a performance index monitoring submodule, a fluctuation data sorting submodule, and a fluctuation situation analysis submodule; The performance indicator monitoring submodule monitors key system performance indicators based on the allocation optimization scheme, including response time and processing capacity, records changes in indicators, including increase or decrease amplitude and change frequency, and generates performance change records; The fluctuation data sorting submodule screens the records whose fluctuation range exceeds a predetermined threshold based on the performance change records, records the occurrence time and fluctuation degree, and obtains a fluctuation detail record table; The fluctuation analysis submodule analyzes the data in the fluctuation detail record table, identifies the fluctuation rules and performance indicators that frequently fluctuate, summarizes potential fluctuation causes, and obtains performance fluctuation records.
5. A VR panoramic operation and maintenance data analysis system according to claim 1, characterized in that: The knowledge base update module includes a fluctuation cause classification submodule, an update strategy formulation submodule, and a knowledge base update execution submodule; The fluctuation cause classification submodule classifies the fluctuation causes into known failure modes and new problems based on the performance fluctuation records, marks and classifies the new problems, and generates a classified fluctuation cause list; The update strategy formulation submodule determines the fault handling methods and maintenance strategies that need to be added to the knowledge base based on the classified fluctuation reason list, and formulates an update strategy, including knowledge content and update priority, to obtain an update strategy solution; The knowledge base update execution submodule executes the update strategy scheme, adds the new fault handling method and maintenance strategy to the existing operation and maintenance knowledge base, updates the knowledge base, and establishes an updated knowledge base.
6. A VR panoramic operation and maintenance data analysis system according to claim 1, characterized in that: The model adaptation and adjustment module includes a parameter analysis submodule, an adaptability adjustment submodule, and a model verification submodule; The parameter analysis submodule analyzes the matching of the current parameter configuration of the model based on the updated knowledge base, evaluates the impact on the existing model parameter configuration, determines the parameter items that need to be adjusted, including learning rate and weight, and generates a parameter adjustment requirement list; The adaptive adjustment submodule adjusts the learning rate and weight of the model according to the parameter adjustment requirement list, matches the model parameters with the operation and maintenance knowledge, and performs parameter optimization to obtain the optimized model parameter configuration; The model verification submodule uses the optimized model parameter configuration to test the model, verify the accuracy and adaptability of the adjusted model, and confirm whether the performance of the model meets expectations, thereby generating an adaptive adjustment model.
7. A VR panoramic operation and maintenance data analysis system according to claim 1, characterized in that: The incremental knowledge acquisition module includes a data analysis submodule, a knowledge extraction submodule, and a knowledge integration submodule; The data analysis submodule analyzes the newly collected operation and maintenance data based on the adaptive adjustment model, identifies patterns and anomalies in the data, marks information that the model fails to identify, and generates unidentified information records; The knowledge extraction submodule extracts data and patterns that can be converted into operation and maintenance knowledge based on the unidentified information records, and verifies the supplementary value of the information to the existing knowledge base to form a list of convertible knowledge; The knowledge integration submodule integrates the information and patterns in the convertible knowledge list into the existing operation and maintenance knowledge base, updates the knowledge points of the knowledge base, and obtains an incremental knowledge set.
8. A VR panoramic operation and maintenance data analysis system according to claim 1, characterized in that: The scheduling plan formulation module includes a demand analysis submodule, a resource evaluation submodule, and a plan construction submodule; The demand analysis submodule collects and analyzes current operation and maintenance requirements based on the incremental knowledge set, determines task priorities and required resource types, identifies key operation and maintenance activities and emergency repair tasks, and generates demand analysis data; The resource assessment submodule assesses the existing resource inventory and availability, including hardware resources, software tools, and human resources, based on the demand analysis data, determines the current status and scheduling capabilities of the resources, and obtains resource availability information; The scheme building submodule evaluates the existing scheduling plan based on the resource availability information, re-formulates and optimizes the scheduling plan, determines whether it meets the resource utilization and operation and maintenance requirements, and generates a resource scheduling plan.
