Intelligent operation and maintenance management system and method based on Internet of Things

Abstract

The invention discloses an intelligent operation and maintenance management system and method based on the Internet of Things, and relates to the technical field of intelligent operation and maintenance management. Comprising an operation and maintenance acquisition module, a storage management analysis module, a delay management analysis module and an execution module. In an existing system, redundant data processing efficiency is low, and redundant data cannot be effectively identified and optimally managed, so that resource waste and low data analysis efficiency are caused; in the prior art, delay processing is not intelligent enough, a traditional system does not comprehensively consider multiple delay factors, adjustment in the network congestion or data transmission process is not flexible enough, delay in the transmission process is high, and real-time performance and user experience are affected, in the prior art, through optimization of a storage management analysis module and a delay management analysis module, real-time performance and user experience are improved. The key problems of redundant data processing efficiency and data transmission delay in the intelligent operation and maintenance management system of the Internet of Things are solved, and the overall performance and response speed of the system are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent operation and maintenance management technology, and specifically to an intelligent operation and maintenance management system and method based on the Internet of Things. Background Art

[0002] Existing IoT-based intelligent operation and maintenance management systems use sensors, wireless communication technology, and big data analysis to monitor and manage equipment operating status in real time, and implement remote equipment diagnosis and fault warning. These systems have been widely used in many industries, such as manufacturing, energy, and transportation.

[0003] However, the current system still has some problems. When the current IoT system processes data from multiple devices, there is often a large amount of redundant data, which not only increases storage pressure, but also affects data analysis efficiency. The existing system lacks an effective redundant data identification and processing mechanism, resulting in repeated calculation and storage of data, wasting resources. The networked system faces problems such as network delays, processing delays and queuing delays during data transmission, and the delays will directly affect the real-time performance and user experience of the system. The existing system often fails to comprehensively consider various network delays, resulting in unstable data transmission and slow response, affecting system performance and timely data processing. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned problems and to propose an intelligent operation and maintenance management system and method based on the Internet of Things.

[0005] The purpose of the present invention can be achieved by the following technical solutions: a smart operation and maintenance management system based on the Internet of Things, including a storage management and analysis module and a delay management and analysis module;

[0006] The storage management and analysis module is used to perform redundancy analysis on real-time data of multiple devices, specifically:

[0007] Preprocess the original data and then calculate the similarity between the real-time data of each device:

[0008] Calculate the Euclidean distance to get the reciprocal of the Euclidean distance R ij ;

[0009] The Pearson correlation coefficient method was used to calculate the correlation r ij ;

[0010] The cosine similarity is calculated to obtain the cosine similarity cos'(i,j);

[0011] Then call the reciprocal of the distance R ij , correlation ij and cosine similarity cos'(i,j), and then perform comprehensive analysis to obtain the redundancy coefficient;

[0012] The delay management analysis module is used to analyze the time required for data to be transmitted from the source end to the target end, specifically:

[0013] Perform edge processing on the data to be transmitted to obtain edge-processed data;

[0014] Perform propagation delay optimization on the edge processed data to obtain the target to be selected;

[0015] The data queuing delay after edge processing is optimized to predict changes in network traffic and obtain predicted values ​​for future traffic.

[0016] As a preferred embodiment of the present invention, it also includes an operation and maintenance collection module and an execution module;

[0017] The operation and maintenance collection module obtains real-time data of various devices through the Internet of Things, including: environmental monitoring data, equipment status data, fault monitoring and early warning, operation efficiency and performance data, and safety monitoring data, and transmits it to the storage management and analysis module in real time for processing to predict the status of the monitoring equipment;

[0018] The execution module receives the data merging signal and performs the data merging operation according to this signal; it is responsible for performing specific operations in the process of optimizing data transmission delay; it is also responsible for intelligently adjusting the network traffic according to the output of the traffic prediction model; it is also responsible for intelligently adjusting the network traffic according to the output of the traffic prediction model.

[0019] As a preferred embodiment of the present invention, the reciprocal of the Euclidean distance R is obtained. ij The specific process is:

[0020] Select the data source for comparison. If the data of device i and device j in time T are respectively: x i = [ x i1 ,x i2 ,…,x in ],x j =[x j1 ,x j2 ,…,x jn ]; Calculate the straight-line distance between two data points in n-dimensional space, using the formula: Output the Euclidean distance D between device i and device j in the entire time window T ij ;x ik and x jk is the data value of device i and device j at the kth moment; the inverse of the distance is used to measure the redundancy coefficient: Where R ij is the inverse of the Euclidean distance.

[0021] As a preferred embodiment of the present invention, the correlation r ij The specific process is:

[0022] x ik and x jk Substitute into the Pearson correlation coefficient calculation formula: Output the correlation r between the data of device i and device j ij ,in and are the means of the data of device i and device j respectively.

[0023] As a preferred embodiment of the present invention, the specific process of obtaining the cosine similarity cos'(i,j) is as follows: ik and x jk Substitute into the cosine similarity formula: Output the cosine similarity cos'(i,j) between the data of device i and device j.

[0024] As a preferred implementation of the present invention, the specific process of obtaining the redundancy coefficient is:

[0025] The reciprocal of the distance R ij , correlation ij Substitute the cosine similarity cos'(i,j) into the XYZ coordinate system and convert the correlation r ij Substitute the X-axis and convert the reciprocal of the distance R ij Substitute into the Y axis, substitute the cosine similarity cos'(i,j) into the Z axis, and then substitute the correlation r ij Substituting into the X axis, we get the reciprocal R of the distance between the endpoint and the ij The endpoint obtained by substituting into the Y axis, the endpoint obtained by substituting the cosine similarity cos'(i,j) into the Z axis, and the starting point after substituting the cosine similarity cos'(i,j) into the Z axis are connected to obtain a redundant coefficient calculation diagram; and the model volume in the redundant coefficient calculation diagram is calculated, and the model volume value is used as the redundant coefficient; then the minimum volume and the maximum volume are obtained from the model in the redundant coefficient calculation diagram. Similarly, the minimum volume value and the maximum volume value of the model are obtained as the value range of the redundant coefficient; then a merging threshold is set in the value range of the redundant coefficient. If the redundant coefficient is greater than the merging threshold, a data merging signal is generated.

[0026] As a preferred implementation of the present invention, the specific process of performing edge processing on the data to be transmitted to obtain edge-processed data is as follows:

[0027] First, the noisy data is processed and the noise is removed using a filter, using the formula: Get f(t), f(t) represents the value of the Gaussian function, which is the response or weight at a certain time t; σ is the standard deviation; the value of each point is calculated by the weighted average of the points around it; each data point is weighted by a smoothing function so that smaller fluctuations or noise in the data are weakened; then use the principal component analysis technique, by projecting the original data into a new orthogonal coordinate system, specifically: standardizing the data is the first step of principal component analysis, and principal component analysis is sensitive to the scale of the data; standardization ensures that the mean of each feature is 0 and the standard deviation is 1, avoiding the unfair impact of differences in the range of feature values ​​on the results; through the formula: Get the standardized data x', where x1 is the original data, μ1 is the mean of the feature, and σ1 is the standard deviation of the feature; then proceed to the second step to calculate the covariance matrix for the standardized data x'; if you have n samples and m features, the standardized data matrix X is an n×m matrix, representing the values ​​of n samples and m features; the covariance matrix C calculation formula is: Output covariance matrix C, X T It is the transposed matrix of X, with a size of m×n, that is, the dimensions of features and samples are exchanged.

[0028] As a preferred implementation mode of the present invention, the specific process of performing propagation delay optimization processing on the edge processed data to obtain the target to be selected is as follows:

[0029] The physical transmission delay is obtained by dividing the distance between the data source and the target by the signal propagation speed, which is recorded as Tb1. If the data source has data with a similarity of more than 90% with the target, the distance between the data source and the data is obtained. If the distance between the data with a similarity of more than 90% with the target and the data source is less than the distance between the data source and the target, it is recorded as the candidate data. The missing part between the candidate data and the target is selected from the data source and combined with the corresponding candidate data to obtain the candidate target. The propagation delay of the candidate target is calculated by the formula: The propagation delay Tb2 of the target to be selected is calculated; d2 is the physical distance between the target to be selected and the target, v is the data transmission speed, T CL is the data processing delay, that is, the time it takes for data to be processed at the source or target end; data processing delay T CL Through calculation, we can get: perform parallel calculation on the data, obtain the number of processing cores of the corresponding processing CPU, record it as h, and divide the processing task into h subtasks for parallel execution; through the formula: Calculated, where D o is the size of the data to be processed corresponding to subtask o, and P is the data processing rate; Tb1 is compared with Tb2, and if Tb2 is smaller than Tb1, the target to be selected corresponding to Tb2 is selected as the original target.

[0030] As a preferred implementation of the present invention, the specific process of obtaining the predicted value of future flow is as follows:

[0031] Obtain historical traffic data, network status data and external environment data, and construct a multi-dimensional feature input matrix X with historical traffic data, network status data and external environment data t ; Then normalize the historical traffic data, network status data and external environment data; convert the time series data into supervised learning form. If the data of the past u time is used to predict the data y of the next time t+1 , and y t+1 =YC(X t ); then X t Substitute into the model training and calculate the minimized loss function through the formula: Output loss function value λ, Z is the number of samples, y t+w ' is the predicted value of the model, that is, the value at the future time t+w predicted based on the input features; y t+w is the actual value, that is, the real data at the future time t+w; when the model training is completed, X t Input historical traffic and network status characteristics, and predict the traffic at time u in the future through the formula: f(X t )=y t+u 'Get, f represents the mapping function of the LSTM model to the input features, y t+u ' is the predicted value of future traffic.

[0032] The intelligent operation and maintenance management method based on the Internet of Things includes:

[0033] Step 1: Obtain real-time data from multiple devices through the Internet of Things;

[0034] Step 2: Preprocess and standardize the real-time data of multiple devices and calculate the redundancy between the data; use three methods, namely, Euclidean distance, Pearson correlation coefficient and cosine similarity, to measure the similarity between devices, and substitute the results into the XYZ coordinate system to generate a redundancy coefficient calculation diagram; set the merging threshold according to the model volume and the redundancy coefficient range, and if the redundancy exceeds the threshold, trigger the data merging signal to achieve data optimization and merging;

[0035] Step 3: Optimize data transmission delay, including propagation delay, processing delay and queuing delay. Optimize propagation path selection and reduce delay through edge processing denoising and dimensionality reduction. Use models to predict traffic changes and intelligently adjust routing, caching and priority queues to cope with traffic fluctuations.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. The present invention improves data management efficiency by performing redundant analysis on real-time data of multiple devices through a storage management and analysis module; first, the module pre-processes the original data, including time alignment, standardization (normalization or z-score), outlier and missing value processing; then, the Euclidean distance, Pearson correlation coefficient, cosine similarity and other methods are used to calculate the data similarity between devices, and then quantify the redundancy. The Euclidean distance is used to measure the straight-line distance of data points in multidimensional space, the Pearson correlation coefficient reflects the linear relationship of the data, and the cosine similarity evaluates the angular similarity of the data vectors; by substituting these indicators into the XYZ coordinate system, the module comprehensively calculates the redundancy coefficient and measures the degree of redundancy by the model volume; finally, if the redundancy coefficient exceeds the set merging threshold, the system will generate a data merging signal and perform a merging operation to optimize the redundant management and processing efficiency of device data.

[0038] 2. The present invention optimizes various types of delays in the data transmission process through a delay management analysis module to improve network performance and user experience; the present invention comprehensively considers network transmission, processing and queuing delays, and uses edge computing to pre-process, reduce dimensions and process noise on data to ensure the real-time nature of data transmission. By intelligently optimizing the physical network transmission path and processing delays, the optimal data target is selected, and routing, caching and traffic shaping are adjusted according to predicted traffic when the network is congested to reduce queuing delays; using historical traffic data, network status and external environment information, the module can predict future traffic changes and make adjustments in advance to avoid network bottlenecks and overloads, improve data transmission efficiency and response speed, and thus significantly improve overall network performance and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0040] Figure 1 It is a principle block diagram of the present invention;

[0041] Figure 2 It is a diagram of the method steps of the present invention;

[0042] Figure 3 This is the redundancy coefficient calculation diagram of the present invention. DETAILED DESCRIPTION

[0043] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. 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.

[0044] It should be understood that the terms “include” and “comprising” used in the specification and claims of the present disclosure indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0045] It should also be understood that the terms used in this disclosure are only for the purpose of describing specific embodiments and are not intended to limit the disclosure. As used in this disclosure and claims, the singular forms of "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise. It should also be further understood that the term "and / or" used in this disclosure and claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations.

[0046] See also Figure 1 As shown, the intelligent operation and maintenance management system based on the Internet of Things includes: an operation and maintenance collection module, a storage management and analysis module, a delay management and analysis module, and an execution module;

[0047] The operation and maintenance collection module obtains real-time data of various devices through the Internet of Things, including: environmental monitoring data, equipment status data, fault monitoring and early warning, operating efficiency and performance data, and safety monitoring data, and transmits it to the storage management and analysis module in real time for processing, predicts the status of the monitoring equipment, optimizes the operation of the operation and maintenance management system, and improves the efficiency of energy management.

[0048] The storage management and analysis module is used to perform redundancy analysis on real-time data of multiple devices, specifically:

[0049] First, preprocess the raw data, align the data from different devices according to the timestamp, standardize the data of each device, use normalization or z-score standardization, and remove outliers and missing values;

[0050] Then calculate the similarity (consistency) between the real-time data of each device:

[0051] Calculate the Euclidean distance: First, select the data source for comparison, usually data within the same time window; for example, select a time period T for comparison, which may be an hour, a day, or a time segment of a real-time data stream; suppose the data of device i and device j within time T are: x i = [ x i1 ,x i2 ,…,x in ],x j =[x j1 ,xj2 ,…,x jn ]; then calculate the straight-line distance between two data points in n-dimensional space, using the formula: Output the Euclidean distance D between device i and device j in the entire time window T ij ;x ik and x jk is the data value of device i and device j at the kth moment; the inverse of the distance is used to measure the redundancy coefficient: Where R ij is the reciprocal of the Euclidean distance, and R ij The closer it is to 1, the higher the redundancy;

[0052] The Pearson correlation coefficient method is used to calculate the linear relationship and correlation between device i and device j. ik and x jk Substitute into the Pearson correlation coefficient calculation formula: Output the correlation r between the data of device i and device j ij ,in and are the means of the data of device i and device j respectively. If the correlation r ij The closer it is to 1, the higher the redundancy between the two data sources;

[0053] Use cosine similarity to calculate the angle between two data vectors to measure their similarity; ik and x jk Substitute into the cosine similarity formula: Output the cosine similarity cos'(i,j) between the data of device i and device j; the value of cosine similarity ranges from -1 to 1, and the closer the value is to 1, the higher the redundancy.

[0054] See also Figure 3 As shown, finally, call the reciprocal of the distance R ij , correlation ij And cosine similarity cos'(i,j) for comprehensive analysis, the reciprocal of the distance R ij , correlation ij Substitute the cosine similarity cos'(i,j) into the XYZ coordinate system and convert the correlation r ij Substitute the X-axis and convert the reciprocal of the distance R ij Substitute into the Y axis, substitute the cosine similarity cos'(i,j) into the Z axis, and then substitute the correlation r ij Substituting into the X axis, we get the reciprocal R of the distance between the endpoint and the ijThe endpoint obtained by substituting into the Y axis, the endpoint obtained by substituting the cosine similarity cos'(i,j) into the Z axis, and the starting point after substituting the cosine similarity cos'(i,j) into the Z axis are connected to obtain a redundant coefficient calculation diagram; and the model volume in the redundant coefficient calculation diagram is calculated, and the model volume value is used as the redundant coefficient; then the minimum volume and the maximum volume are obtained from the model in the redundant coefficient calculation diagram. Similarly, the minimum volume value and the maximum volume value of the model are obtained as the value range of the redundant coefficient; then a merging threshold is set in the value range of the redundant coefficient. If the redundant coefficient is greater than the merging threshold, a data merging signal is generated, and the execution module executes the data merging.

[0055] The delay management analysis module is used to analyze the time (delay) required for data to be transmitted from the source to the target, specifically:

[0056] Data delay is mainly composed of network transmission delay, processing delay and queuing delay; network delay and bandwidth limitation may affect the real-time performance of data transmission; in the process of data transmission, the model for calculating the total delay is: Tz = Ts + Tb + Tp + Tc to calculate the total delay Tz, where Ts is the transmission delay, which depends on the data size and network bandwidth; Tb is the propagation delay, which depends on the physical distance and the speed of light; Tp is the queuing delay, which depends on the buffer of the network router and network congestion; Tc is the processing delay, which involves the encoding, decoding, and processing of data;

[0057] First, edge processing is performed on the data to be transmitted: First, the data source (such as sensors, devices, etc.) collects raw data, which may be high-dimensional, complex, or contain noise; the noisy data is processed and filters (such as mean filters, Gaussian filters, etc.) are used to remove noise, using the formula: Get f(t), f(t) represents the value of the Gaussian function, which is the response or weight at a certain time t; σ is the standard deviation; then the value of each point is calculated by the weighted average of the points around it; and follow the Gaussian distribution, the points close to the center have a larger weight, and the points far from the center have a smaller weight; use a smoothing function to weight each data point, so that the smaller fluctuations or noise in the data are weakened, while the overall trend of the signal is retained; then use the principal component analysis technique (principal component analysis is a commonly used dimensionality reduction method), and then project the original data into a new orthogonal coordinate system (principal component space) to reduce the dimension of the data while retaining the main information of the data as much as possible; specifically: standardizing the data is the first step of principal component analysis, and principal component analysis is sensitive to the scale of the data; standardization ensures that the mean of each feature is 0 and the standard deviation is 1, avoiding the unfair impact of the range of eigenvalues ​​on the results; through the formula: Get the standardized data x', where x1 is the original data, μ1 is the mean of the feature, and σ1 is the standard deviation of the feature; then proceed to the second step, calculate the covariance matrix (the covariance matrix represents the correlation between the features in the data) for the standardized data x'; if you have n samples and m features, the standardized data matrix X is an n×m matrix, representing the values ​​of n samples and m features; the covariance matrix C calculation formula is: Output covariance matrix C, X T is the transposed matrix of X, with a size of m×n, that is, the dimensions of features and samples are exchanged; suppose there is a data set containing 100 samples and 20 features; through principal component analysis, the data is reduced from 20 dimensions to 2 or 3 dimensions, while trying to maintain the variance of the original data; this not only reduces the storage and computing requirements of the data, but also improves the efficiency of subsequent processing.

[0058] Optimize the data propagation delay after edge processing: the physical network distance usually refers to the actual physical distance between the data source and the target node; the longer the distance, the longer the signal transmission time; the physical transmission delay is obtained by dividing the distance between the data source and the target by the signal propagation speed, denoted as Tb1; if the data source has data with a similarity of more than 90% with the target, obtain the distance between the data source and the data; if the distance between the data with a target similarity of more than 90% and the data source is less than the distance between the data source and the target, it is recorded as the candidate data; and the missing part between the candidate data and the target is selected from the data source and combined with the corresponding candidate data to obtain the candidate target, and the propagation delay of the candidate target is calculated: by the formula: The propagation delay Tb2 of the target to be selected is calculated; where d2 is the physical distance between the target to be selected and the target, v is the data transmission speed (such as network bandwidth, which determines the data transmission rate), T CL is the data processing delay, that is, the time it takes for data to be processed at the source or target end; data processing delay T CL By calculation, we can get: by parallel computing the data, we can get the number of processing cores of the corresponding processing CPU, which is recorded as h. The processing task can be divided into h subtasks for parallel execution; by formula: Calculated, where D o is the size of the data to be processed corresponding to subtask o, and P is the data processing rate; Tb1 is compared with Tb2, and if Tb2 is smaller than Tb1, the target to be selected corresponding to Tb2 is selected as the original target.

[0059] Optimize the queuing delay of data after edge processing: predict changes in network (IoT) traffic, and intelligently adjust routing, caching, and priority queues to reduce queuing delays. Specifically:

[0060] By obtaining historical traffic data (past network traffic, such as the number of packets and bytes per second), network status data and external environment data; network status data includes: bandwidth utilization, current load, packet loss rate, delay, etc.; external environment data includes: such as user behavior, seasonal fluctuations, weather changes, holidays, etc.; the data is used to construct a multi-dimensional feature input matrix X t , where each column represents a feature and each row represents the feature value at a time point; user behavior represents the user's demand and behavior for network services within a specific time, and features are generated based on the number of user visits, page views, etc.

[0061] Seasonal fluctuations mean that seasonal fluctuations affect network traffic, especially specific seasons (such as summer and winter vacations, shopping seasons, etc.) will produce obvious fluctuations; capture seasonal fluctuations and construct the following features: Seasonal indicator coefficient: create virtual variables for different seasons of the year (spring, summer, autumn, and winter); for example, set spring to 1 and other seasons to 0; similarly, seasonal features can be created based on months, weeks, etc. to capture periodic changes; similarly, weather changes are feature constructed, and virtual variables are created based on weather conditions (sunny, cloudy, rainy, etc.) to identify different types of weather; feature holidays are constructed, and features are established based on holiday types (public holidays, company holidays, etc.), and continuous feature values ​​can also be created based on the start and end dates of holidays.

[0062] Then normalize the historical traffic data to the range of [0,1]; convert the time series data into supervised learning form. If the data of the past u (u>0) time is used to predict the data of the next time step; for example, if the data of the past 5 minutes is used to predict the traffic of the next 1 minute, it is expressed as: X t =[y t-1 ,y t-2 ,y t-3 ,y t-4 ,y t-5 ], where y t represents the flow value at time t, where y t+1 represents the flow value at time t+1 (i.e. the predicted value at future time t+1), and y t+1 =YC(X t ); then X t Substitute into LSTM model training and calculate the minimized loss function through the formula: Output loss function value λ, Z is the number of samples, y t+w 'The predicted value of the model, that is, the value at the future time t+w predicted based on the input features (such as past traffic data); y t+w is the actual value, that is, the real data at the future time t+w; when the model training is completed, X tInput historical traffic and network status characteristics, and predict the traffic at time u in the future through the formula: f(X t )=y t+u 'Get, where f represents the mapping function of the LSTM model to the input features, y t+u ' is the predicted value of future traffic.

[0063] If a router or network device predicts that it will face a sudden traffic peak, causing network congestion, the execution module will adjust the traffic distribution in advance based on the predicted traffic and select a path with less load for traffic transmission to avoid network bottlenecks; if it is predicted that traffic will increase significantly in the future, the execution module will expand cache resources in advance and pre-store popular content to reduce latency and improve user experience; if a traffic peak is predicted, the execution module will perform traffic shaping to prioritize the transmission of real-time traffic by limiting the rate of certain non-real-time data streams to avoid network overload.

[0064] See also Figure 2 As shown, the present invention also provides an intelligent operation and maintenance management method based on the Internet of Things, and the specific steps are:

[0065] Step 1: Obtain real-time data from multiple devices through the Internet of Things;

[0066] Step 2: Preprocess and standardize the real-time data of multiple devices and calculate the redundancy between the data. Use three methods, namely, Euclidean distance, Pearson correlation coefficient and cosine similarity, to measure the similarity between devices, and substitute the results into the XYZ coordinate system to generate a redundancy coefficient calculation diagram; set the merging threshold according to the model volume and the redundancy coefficient range, and if the redundancy exceeds the threshold, trigger the data merging signal to achieve data optimization and merging;

[0067] Step 3: Optimize data transmission delay, including propagation delay, processing delay, and queuing delay. Through edge processing denoising and dimensionality reduction, optimize the propagation path selection and reduce delay; use the LSTM model to predict traffic changes, intelligently adjust routing, cache, and priority queues to cope with traffic fluctuations, avoid network congestion, and improve data transmission real-time performance and user experience.

[0068] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. 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. An intelligent operation and maintenance management system based on the Internet of Things, including a storage management and analysis module and a delay management and analysis module; the characteristics are: The storage management and analysis module is used to perform redundancy analysis on real-time data of multiple devices, specifically: Preprocess the original data and then calculate the similarity between the real-time data of each device: Calculate the Euclidean distance to get the reciprocal of the Euclidean distance R ij ; The Pearson correlation coefficient method was used to calculate the correlation r ij ; The cosine similarity is calculated to obtain the cosine similarity cos'(i,j); Then call the reciprocal of the distance R ij , correlation ij and cosine similarity cos'(i,j), and then perform comprehensive analysis to obtain the redundancy coefficient; The delay management analysis module is used to analyze the time required for data to be transmitted from the source end to the target end, specifically: Perform edge processing on the data to be transmitted to obtain edge-processed data; Perform propagation delay optimization on the edge processed data to obtain the target to be selected; The data queuing delay after edge processing is optimized to predict changes in network traffic and obtain predicted values ​​for future traffic.

2. The intelligent operation and maintenance management system based on the Internet of Things according to claim 1 is characterized in that: It also includes operation and maintenance collection module and execution module; The operation and maintenance collection module obtains real-time data of various devices through the Internet of Things, including: environmental monitoring data, equipment status data, fault monitoring and early warning, operation efficiency and performance data, and safety monitoring data, and transmits it to the storage management and analysis module in real time for processing to predict the status of the monitoring equipment; The execution module receives the data merging signal and performs the data merging operation according to this signal; it is responsible for performing specific operations in the process of optimizing data transmission delay; it is also responsible for intelligently adjusting the network traffic according to the output of the traffic prediction model; it is also responsible for intelligently adjusting the network traffic according to the output of the traffic prediction model.

3. The intelligent operation and maintenance management system based on the Internet of Things according to claim 1 is characterized in that: The reciprocal of the Euclidean distance R is obtained ij The specific process is: Select the data source for comparison. If the data of device i and device j in time T are respectively: x i =[x i1 ,x i2 ,…,x in ], Calculate the straight-line distance between two data points in n-dimensional space using the formula: Output the Euclidean distance D between device i and device j in the entire time window T ij ;x ik and x jk is the data value of device i and device j at time k; Then use the inverse of the distance to measure the redundancy coefficient: Where R ij is the inverse of the Euclidean distance.

4. The intelligent operation and maintenance management system based on the Internet of Things according to claim 3 is characterized in that: The obtained correlation r ij The specific process is: x ik and x jk Substitute into the Pearson correlation coefficient calculation formula: Output the correlation r between the data of device i and device j ij ,in and are the means of the data of device i and device j respectively.

5. The smart operation and maintenance management system and method based on the Internet of Things according to claim 4 is characterized in that: The specific process of obtaining the cosine similarity cos'(i,j) is as follows: ik and x jk Substitute into the cosine similarity formula: Output the cosine similarity cos'(i,j) between the data of device i and device j.

6. The intelligent operation and maintenance management system based on the Internet of Things according to claim 5 is characterized in that: The specific process of obtaining the redundancy coefficient is as follows: The reciprocal of the distance R ij , correlation ij Substitute the cosine similarity cos'(i,j) into the XYZ coordinate system and convert the correlation r ij Substitute the X-axis and convert the reciprocal of the distance R ij Substitute into the Y axis, substitute the cosine similarity cos'(i,j) into the Z axis, and then substitute the correlation r ij Substituting into the X axis, we get the reciprocal R of the distance between the endpoint and the ij Substitute the endpoint obtained by substituting into the Y axis, the endpoint obtained by substituting the cosine similarity cos'(i, j) into the Z axis, and the starting point after substituting the cosine similarity cos'(i, j) into the Z axis to obtain the redundancy coefficient calculation graph; And calculate the volume of the model in the redundant coefficient calculation diagram, and use the model volume value as the redundant coefficient; then obtain the minimum volume and maximum volume of the model in the redundant coefficient calculation diagram, and similarly, obtain the minimum volume value and the maximum volume value of the model as the value range of the redundant coefficient; Then, a merging threshold is set in the value range of the redundancy coefficient, and if the redundancy coefficient is greater than the merging threshold, a data merging signal is generated.

7. The intelligent operation and maintenance management system based on the Internet of Things according to claim 1 is characterized in that: The specific process of performing edge processing on the data to be transmitted to obtain edge processed data is as follows: First, the noisy data is processed and the noise is removed using a filter, using the formula: Get f(t), f(t) represents the value of the Gaussian function, which is the response or weight at a certain time t; σ is the standard deviation; the value of each point is calculated by the weighted average of the points around it; each data point is weighted by a smoothing function so that smaller fluctuations or noise in the data are weakened; then use the principal component analysis technique, by projecting the original data into a new orthogonal coordinate system, specifically: standardizing the data is the first step of principal component analysis, and principal component analysis is sensitive to the scale of the data; standardization ensures that the mean of each feature is 0 and the standard deviation is 1, avoiding the unfair impact of differences in the range of feature values ​​on the results; through the formula: Get the standardized data x', where x1 is the original data, μ1 is the mean of the feature, and σ1 is the standard deviation of the feature; then proceed to the second step to calculate the covariance matrix for the standardized data x'; if you have n samples and m features, the standardized data matrix X is an n×m matrix, representing the values ​​of n samples and m features; the covariance matrix C calculation formula is: Output covariance matrix C, X T It is the transposed matrix of X, with a size of m×n, that is, the dimensions of features and samples are exchanged.

8. The intelligent operation and maintenance management system based on the Internet of Things according to claim 1 is characterized in that: The specific process of performing propagation delay optimization processing on the edge processed data to obtain the target to be selected is as follows: The physical transmission delay is obtained by dividing the distance between the data source and the target by the signal propagation speed, denoted as Tb1; If there is data in the data source with a similarity of more than 90% with the target, the distance between the data source and the data is obtained. If the distance between the data with a similarity of more than 90% with the target and the data source is smaller than the distance between the data source and the target, it is recorded as the candidate data; and the missing part between the candidate data and the target is selected from the data source and combined with the corresponding candidate data to obtain the candidate target, and the propagation delay of the candidate target is calculated: by formula: The propagation delay Tb2 of the target to be selected is calculated; d2 is the physical distance between the target to be selected and the target, v is the data transmission speed, T CL is the data processing delay, that is, the time it takes for data to be processed at the source or target end; data processing delay T CL Through calculation, we can get: perform parallel calculation on the data, obtain the number of processing cores of the corresponding processing CPU, record it as h, and divide the processing task into h subtasks for parallel execution; through the formula: Calculated, where D o is the size of the data to be processed corresponding to subtask o, and P is the data processing rate; Tb1 is compared with Tb2, and if Tb2 is smaller than Tb1, the target to be selected corresponding to Tb2 is selected as the original target.

9. The intelligent operation and maintenance management system based on the Internet of Things according to claim 1 is characterized in that: The specific process of obtaining the predicted value of future traffic is as follows: Obtain historical traffic data, network status data and external environment data, and construct a multi-dimensional feature input matrix X with historical traffic data, network status data and external environment data t ; Then normalize the historical traffic data, network status data and external environment data; convert the time series data into supervised learning form. If the data of the past u time is used to predict the data y of the next time t+1 , and y t+1 =YC(X t ); then X t Substitute into the model training and calculate the minimized loss function through the formula: Output loss function value λ, Z is the number of samples, y t+w ' is the predicted value of the model, that is, the value at the future time t+w predicted based on the input features; y t+w is the actual value, that is, the real data at the future time t+w; when the model training is completed, X t Input historical traffic and network status characteristics, and predict the traffic at time u in the future through the formula: f(X t )=y t+u 'Get, f represents the mapping function of the LSTM model to the input features, y t+u ' is the predicted value of future traffic.

10. The intelligent operation and maintenance management method based on the Internet of Things is characterized by: Applied to implement the smart operation and maintenance management system based on the Internet of Things as described in any one of claims 1 to 9, the method comprises: Step 1: Obtain real-time data from multiple devices through the Internet of Things; Step 2: Preprocess and standardize the real-time data of multiple devices and calculate the redundancy between the data; use three methods, namely, Euclidean distance, Pearson correlation coefficient and cosine similarity, to measure the similarity between devices, and substitute the results into the XYZ coordinate system to generate a redundancy coefficient calculation diagram; set the merging threshold according to the model volume and the redundancy coefficient range, and if the redundancy exceeds the threshold, trigger the data merging signal to achieve data optimization and merging; Step 3: Optimize data transmission delay, including propagation delay, processing delay and queuing delay; Through edge processing denoising and dimensionality reduction, the propagation path selection is optimized and latency is reduced. The model is used to predict traffic changes and intelligently adjust routing, caching and priority queues to cope with traffic fluctuations.

Images