A remote maintenance method for a converged communication system with monitoring function

Abstract

The present invention discloses a remote maintenance method for a fusion communication system with a monitoring function, including S1, a monitoring and collection stage, S2, an analysis and processing stage, S3, an alarm sending stage, S4, a fault prediction stage, and S5, an operation and maintenance guidance stage. The remote maintenance method for a fusion communication system with a monitoring function collects data related to the equipment and data such as the current, voltage, and temperature of the equipment when it is working through various sensors in the S1 stage, and compares the collected data with the data of the equipment in the sample library when it is working normally in the S2 stage to determine whether there is a fault. When the data is abnormal, the S3 stage sends the alarm information to the management personnel using the communication system, and then the S4 stage starts to predict the results of the hidden faults of the equipment through the prediction algorithm, and then the S5 stage outputs the maintenance plan according to the predicted results, thereby solving the problem that the existing equipment maintenance is prone to cause large economic losses.

Description

Technical Field

[0001] The present invention relates to the technical field of remote maintenance, and in particular to a remote maintenance method for a fusion communication system with a monitoring function. Background Art

[0002] Relay gateway plays an important role in the communication system, which is reflected in the connection of different networks: Relay gateway is the key equipment connecting PSTN (Public Switched Telephone Network) and NGN (Next Generation Network). It is located in the edge access layer of the NGN network, realizing the conversion between IP packets and TDM (Time Division Multiplexing) signals, so that the voice in the IP domain and the circuit domain can be connected and interoperable1; Voice processing function: The relay gateway has voice processing functions, including voice signal encoding and decoding, echo control and elimination, etc., to ensure the quality of voice communication; Call processing and control: The relay gateway can control calls according to the commands of the soft switch, such as connection, interruption, etc., and can also detect and report the user status on the PSTN side, such as busy, long-term ringing without answer, etc.; Maintenance and management: The relay gateway can establish and maintain the association with the soft switch After detecting a connection interruption, it can restore the connection after fault recovery or congestion elimination, and report the configuration status and abnormal faults of the underlying connection; support multiple signaling and protocols: the relay gateway supports multiple signaling and protocols, such as No.7, NO.1, Q.931, Q.sig, MGCP, H.323 and SIP, etc., to ensure interoperability between different networks, and the relay gateway is a key device for realizing network interconnection and voice communication in the communication system. It enables different networks to communicate and collaborate effectively by converting signals and processing calls. In order to ensure the normal operation of the communication system, it needs to be maintained.

[0003] The existing maintenance method is mostly to notify staff to perform on-site maintenance after equipment failure, which affects related work and causes great economic losses. In view of the above problems, it is necessary to improve the existing equipment. Summary of the invention

[0004] The purpose of the present invention is to provide a remote maintenance method for a converged communication system with a monitoring function to solve the problem that the existing maintenance methods proposed in the above background technology are mostly to notify staff to perform on-site maintenance after equipment failure, thereby affecting related work and causing greater economic losses.

[0005] To achieve the above object, the present invention provides the following technical solution: a remote maintenance method of a converged communication system with a monitoring function, comprising:

[0006] S1, monitoring and collection stage: collect relevant temperature data, current data, voltage data and other relevant information through various sensors and instruments;

[0007] S2, analysis and processing stage: compare the collected accumulated data with the set normal data range value stored in the sample library as the sample, and analyze its operation status based on this;

[0008] S3, warning sending stage: This stage is triggered when there is a difference between the collected data and the normal data, and the warning information is sent to the relevant management personnel through the communication system;

[0009] S4, fault prediction stage: predict the upcoming fault information based on the existing data information through the prediction algorithm;

[0010] S5, operation and maintenance oriented stage: the maintenance plan is automatically output according to the prediction results, and the export of the maintenance plan is based on the big data repository. At the same time, the big data repository and the sample library are both connected to the Internet.

[0011] Preferably, the S1 stage data collection instrument includes various sensors, monitoring probes and GPS positioning equipment, and the GPS positioning equipment attaches a time stamp to the electromechanical data.

[0012] Preferably, the S2 stage compares the data collected in the S1 stage with the algorithm in the sample library through a comparison algorithm.

[0013] Preferably, the communication system in the S3 stage sends warning information in the form of text messages, emails and faxes, and the number of text messages, emails and faxes sent is selected according to the setting requirements and one or more are sent simultaneously.

[0014] Preferably, the S4 stage is completed through the difference autoregressive moving average model, abbreviated as ARIMA, and the process of ARIMA is to first transform the non-stationary time series into a stationary time series by selecting a difference operation of an appropriate order, then establish an autoregressive moving average model for the stationary series, and then predict the original data.

[0015] Preferably, the autoregressive moving average formula is:

[0016]

[0017] Among them, y 1 is the original time series; φ d The day of yi-1 is a stationary time series after difference operation; ε i is a white noise random error sequence with zero mean; δ i and θ i are the parameters of the model; P and q are the orders of the autoregressive and moving average processes, respectively.

[0018] Preferably, the prediction process of the autoregressive moving average model includes the following steps:

[0019] Step 1: Obtain the time series data of the observed system;

[0020] Step 2: Plot the data to see if it is a stationary time series. If it is a non-stationary time series, perform a d-order difference operation to convert it into a stationary time series. The d here is the d in the ARIMA (p, d, q) model. If it is a stationary series, use the ARMA (p, q) model. The difference between the ARIMA (p, d, q) model and the ARMA (p, q) model is that the characteristic polynomial of the autoregressive part of the former contains d unit roots.

[0021] Preferably, the prediction process of the autoregressive moving average model further includes:

[0022] Step 3: Obtain the autocorrelation coefficient ACF and partial autocorrelation coefficient PACF of the obtained stationary time series respectively, and obtain the optimal layer p and order q by analyzing the autocorrelation graph and partial autocorrelation graph. The ARIMA model is obtained by the above-obtained d, q, and p;

[0023] Step 4: Model diagnosis, perform diagnostic analysis to confirm that the obtained model is indeed consistent with the observed data characteristics; if not, return to step 3.

[0024] Preferably, when the ARIMA model is used, the original data is generally divided into linear and nonlinear parts for separate prediction, and the SVM and GRNN methods are used in the nonlinear part to improve the accuracy of the prediction effect.

[0025] Preferably, the SVM is a support vector machine, which is characterized in that it is difficult to classify and solve many original data in low-dimensional space, but a nonlinear transformation of such data can be performed through a suitable kernel function in order to achieve the purpose of linear separability. The kernel function is usually required to be constructed in the following format:

[0026] k(x i +y i )=ω(x i )·ω(x j ).

[0027] Compared with the prior art, the invention has the following beneficial effects: the remote maintenance method of the fusion communication system with monitoring function,

[0028] The present invention collects equipment-related data and data such as current, voltage, temperature, etc. when the equipment is working through various sensors in the S1 stage, and compares the collected data with various data when the equipment is working normally in the sample library in the S2 stage to determine whether there is a fault in the device. When there is an abnormality in the data, the S3 stage sends the alarm information to the management personnel through the communication system, and then the S4 stage starts to predict the results of the hidden faults of the equipment through the prediction algorithm. Then the S5 stage automatically outputs the maintenance plan according to the prediction results, and thereby solves the problem of existing equipment maintenance, which mostly notifies the staff to perform on-site maintenance after the equipment fails, thereby affecting related work and causing large economic losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the overall workflow of the present invention;

[0030] Figure 2 It is the overall flow chart of the ARIMA model of the present invention;

[0031] Figure 3 This is a flowchart of the parallel random forest based on spark in the present invention. DETAILED DESCRIPTION

[0032] 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.

[0033] See also Figure 1-3 The present invention provides a technical solution: a remote maintenance method of a converged communication system with a monitoring function, comprising:

[0034] S1, monitoring and collection stage: collect relevant temperature data, current data, voltage data and other relevant information through various sensors and instruments;

[0035] S2, analysis and processing stage: compare the collected accumulated data with the set normal data range value stored in the sample library as the sample, and analyze its operation status based on this;

[0036] S3, warning sending stage: This stage is triggered when there is a difference between the collected data and the normal data, and the warning information is sent to the relevant management personnel through the communication system;

[0037] S4, fault prediction stage: predict the upcoming fault information based on the existing data information through the prediction algorithm;

[0038] S5, operation and maintenance oriented stage: the maintenance plan is automatically output according to the prediction results, and the export of the maintenance plan is based on the big data repository. At the same time, the big data repository and the sample library are both connected to the Internet.

[0039] The acquisition instruments in the S1 stage include various sensors, monitoring probes and GPS positioning devices, and the GPS positioning device attaches a corresponding time stamp to the electromechanical data.

[0040] In the S2 stage, the data collected in the S1 stage are compared with the algorithm in the sample library through the comparison algorithm.

[0041] The communication system in the S3 stage sends warning information in the form of SMS, email and fax, and the number of SMS, email and fax sent is selected according to the setting needs, and one or more of them are sent simultaneously.

[0042] The S4 stage is completed through the difference autoregressive moving average model, abbreviated as ARIMA. The process of ARIMA is to first transform the non-stationary time series into a stationary time series by selecting a difference operation of appropriate order, then establish an autoregressive moving average model for the stationary series, and then predict the original data.

[0043] The autoregressive moving average formula is:

[0044]

[0045] Among them, y 1 is the original time series; φ d The day of yi-1 is a stationary time series after difference operation; ε i is a white noise random error sequence with zero mean; δ i and θ i are the parameters of the model; P and q are the orders of the autoregressive and moving average processes, respectively.

[0046] The forecasting process of the autoregressive moving average model includes the following steps:

[0047] Step 1: Obtain the time series data of the observed system;

[0048] Step 2: Plot the data to see if it is a stationary time series. If it is a non-stationary time series, perform a d-order difference operation to convert it into a stationary time series. The d here is the d in the ARIMA (p, d, q) model. If it is a stationary series, use the ARMA (p, q) model. The difference between the ARIMA (p, d, q) model and the ARMA (p, q) model is that the characteristic polynomial of the autoregressive part of the former contains d unit roots.

[0049] The forecasting process of the autoregressive moving average model also includes:

[0050] Step 3: Obtain the autocorrelation coefficient ACF and partial autocorrelation coefficient PACF of the obtained stationary time series respectively, and obtain the optimal layer p and order q by analyzing the autocorrelation graph and partial autocorrelation graph. The ARIMA model is obtained by the above-obtained d, q, and p;

[0051] Step 4: Model diagnosis, perform diagnostic analysis to confirm that the obtained model is indeed consistent with the observed data characteristics; if not, return to step 3.

[0052] When using the ARIMA model, the original data is usually divided into linear and nonlinear parts for separate prediction. In the nonlinear part, SVM and GRNN methods are used to improve the accuracy of the prediction effect.

[0053] SVM is support vector machine. It is difficult to classify and solve many low-dimensional space raw data. However, it can transform such data into a nonlinear form through a suitable kernel function in order to achieve linear separability. The kernel function is usually constructed in the following format: k(x i +y i )=ω(x i )·ω(x j ).

[0054] Specifically, in optimizing SVM, the penalty parameters and kernel function parameters are usually adjusted to obtain a higher prediction and classification accuracy. When adjusting, the common method is to try to take values ​​within a certain range. In addition, the parameters can also be optimized through methods such as genetic algorithms and particle swarm networks.

[0055] Embodiment 2

[0056] This embodiment is a further description of the above embodiment. It should be understood that this embodiment includes all the aforementioned technical features and is further described in detail.

[0057] The prediction of the S4 stage also includes the gray theory method. The principle of gray theory for data prediction is to model and process some data that have a greater impact on the prediction results, find relevant information in it, and then predict the data to get the corresponding results. The model building steps are: first determine a set of data sequences, then perform a cumulative calculation on the sequence and generate a new sequence. Construct a matrix for the two sequences, solve its matrix coefficients, and then use the established GM (1,1) model to calculate the data prediction value and its error value. Among them, since the CM (1,1) model is an exponential model, before training the data sequence, it is necessary to pre-process it with methods such as exponential weighting and sliding average to improve the accuracy of the prediction results. Because the original gray theory has problems such as unsatisfactory prediction results for data with large volatility, gray theory is often combined with Markov model, BP neural network and wavelet analysis to test data in order to improve the data prediction effect.

[0058] Embodiment 3

[0059] This embodiment is a further description of the above embodiment. It should be understood that this embodiment includes all the aforementioned technical features and is further described in detail.

[0060] The prediction in the S4 stage also includes the random forest method. Random forest is an integrated learning method that uses decision trees as the basic learning unit. It contains multiple decision trees trained by bagging integrated learning theory and random subspace method. The sample to be classified is input, and each decision tree generates each classification result. The final classification result is decided by voting on the results of each decision tree. Random forest is an integrated learning method of multiple decision trees. It can not only overcome some shortcomings of decision trees, but also has good scalability and parallelism. It can effectively solve the problem of fast processing of big data and has good application prospects for prediction in big data environment. The random forest algorithm can not only realize the parallelization of data processing, but also realize the parallelization of training units. As shown in the figure, the steps for implementing the random forest algorithm in parallel on the Spark platform are as follows:

[0061] Step 1: Parallel sampling of the original data set at each node: Repeatedly use the Bootstrap sampling method to sample the original training set D to obtain k sub-training sets. At the same time, the data that are not sampled constitute the out-of-bag (0OB) data set. Finally, we can obtain k 0OB data sets, which can be used to test the prediction performance of the random forest model;

[0062] Step 2: Decision tree modeling at each node: Use the above k sub-training sets to build k decision models. It is worth noting that not all features are used in the process of tree building. Each tree randomly selects m features from all features for modeling;

[0063] Step 3: Obtain random forest model: collect k decision tree models to construct a random forest model;

[0064] Step 4: Vote for the predicted samples, and the one with the most votes is the final category.

[0065] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A remote maintenance method for a converged communication system with monitoring function, characterized in that: include: S1, monitoring and collection stage: collect relevant temperature data, current data, voltage data and other relevant information through various sensors and instruments; S2, analysis and processing stage: compare the collected accumulated data with the set normal data range value stored in the sample library as the sample, and analyze its operation status based on this; S3, warning sending stage: This stage is triggered when there is a difference between the collected data and the normal data, and the warning information is sent to the relevant management personnel through the communication system; S4, fault prediction stage: predict the upcoming fault information based on the existing data information through the prediction algorithm; S5, operation and maintenance oriented stage: the maintenance plan is automatically output according to the prediction results, and the export of the maintenance plan is based on the big data repository. At the same time, the big data repository and the sample library are both connected to the Internet.

2. The remote maintenance method of a converged communication system with monitoring function according to claim 1, characterized in that: The S1 stage acquisition instruments include various sensors, monitoring probes and GPS positioning devices, and the GPS positioning device attaches a time stamp to the electromechanical data.

3. The remote maintenance method of a converged communication system with monitoring function according to claim 1, characterized in that: The S2 stage compares the data collected in the S1 stage with the algorithm in the sample library through a comparison algorithm.

4. The remote maintenance method of a converged communication system with monitoring function according to claim 1, characterized in that: The communication system in the S3 stage sends warning information in the form of text messages, emails and faxes, and the number of text messages, emails and faxes sent is selected according to the setting requirements, and one or more of them are sent simultaneously.

5. The remote maintenance method of a converged communication system with monitoring function according to claim 1, characterized in that: The S4 stage is completed through the difference autoregressive moving average model, abbreviated as ARIMA, and the process of ARIMA is to first transform the non-stationary time series into a stationary time series by selecting a difference operation of appropriate order, then establish an autoregressive moving average model for the stationary series, and then predict the original data.

6. The remote maintenance method of a converged communication system with monitoring function according to claim 5, characterized in that: The autoregressive moving average formula is: Among them, y1 is the original time series; φ d The day of yi-1 is a stationary time series after difference operation; ε i is a white noise random error sequence with zero mean; δ i and θ i are the parameters of the model; P and q are the orders of the autoregressive and moving average processes, respectively.

7. The remote maintenance method of a converged communication system with monitoring function according to claim 5, characterized in that: The prediction process of the autoregressive moving average model includes the following steps: Step 1: Obtain the time series data of the observed system; Step 2: Plot the data to see if it is a stationary time series. If it is a non-stationary time series, perform a d-order difference operation to convert it into a stationary time series. The d here is the d in the ARIMA (p, d, q) model. If it is a stationary series, use the ARMA (p, q) model. The difference between the ARIMA (p, d, q) model and the ARMA (p, q) model is that the characteristic polynomial of the autoregressive part of the former contains d unit roots.

8. The remote maintenance method of a converged communication system with monitoring function according to claim 5, characterized in that: The prediction process of the autoregressive moving average model also includes: Step 3: Obtain the autocorrelation coefficient ACF and partial autocorrelation coefficient PACF of the obtained stationary time series respectively, and obtain the optimal layer p and order q by analyzing the autocorrelation graph and partial autocorrelation graph. The ARIMA model is obtained by the above-obtained d, q, and p; Step 4: Model diagnosis, perform diagnostic analysis to confirm that the obtained model is indeed consistent with the observed data characteristics; if not, return to step 3.

9. The remote maintenance method of a converged communication system with monitoring function according to claim 1, characterized in that: When the ARIMA model is used, the original data is generally divided into linear and nonlinear parts for separate prediction, and the SVM and GRNN methods are used in the nonlinear part to improve the accuracy of the prediction effect.

10. The remote maintenance method of a converged communication system with monitoring function according to claim 9, characterized in that: The SVM is a support vector machine. It is difficult to classify and solve many raw data in low-dimensional space. However, a suitable kernel function can be used to perform a nonlinear transformation on such data in order to achieve linear separability. The kernel function is usually constructed in the following format: k(x i +y i )=ω(x i )·ω(x j )。

Images