Large area dynamic detection method for remote communication fault of power consumption information acquisition terminal

With the rapid development of power system technology, the requirements for the stability of power equipment operation are higher and higher. In order to ensure the stability of power equipment operation, this paper puts forward the dynamic detection method of long-distance communication fault of power information acquisition terminal. Through the collection of remote communication fault dynamic information, data preprocessing. And the fault area is accurately located to determine the fault location. Through the power information acquisition terminal to determine the abnormal value, the detection effect of communication fault can be realized. According to the experimental comparison, it can be seen that compared with the traditional fault detection methods, the large interval dynamic detection method of power information acquisition terminal remote communication fault has higher detection accuracy and significant advantages.

1 Introduction data and detailed information of fault maintenance personnel [6] . Fault sample library refers to the process of remote communication fault detection, which adopts data mining technology to mine effective rules from massive production data and real-time status data list. The signal preprocessor is responsible for preprocessing the signals collected by the collector. In the process of fault detection, a series of digital signals can be obtained after the signal is converted by high-speed ADC. The signal is input into buffer RAM after frame processing, and the high-efficiency processing is realized by TMS320C5402 processor. In the high-precision DAC converter, the processed signal is input, the analog signal is restored, and the power amplifier is used to amplify the output, which ensures that the preprocessing result is more accurate and can effectively dig out the hidden law of the fault [7] . Compared with the host log data, the type and quantity of network data are more complex and huge, which makes the remote communication fault detection of network attacks more difficult. In order to integrate and analyze network packets effectively, we must understand a series of TCP / IP structures. The model of TCP / P protocol is simplified, and the combination of protocol stack is given. TPC and PI packet format, respectively, carry the key messages needed for analysis and processing. The original data packet collected by the data acquisition module is decrypted through the Ethernet frame to determine whether the protocol type belongs to the same IP address. If there is, the PI data layer parsing needs to be handed over to the IP parsing packet; otherwise, it will be discarded immediately. This paper deals with the electronic circuit fault in communication, and the processing flow is shown in Figure 1.

Communication fault location algorithm
In the design of the overall work flow of multi-channel communication signal fault detection, this paper adopts a master-slave handshake protocol mode and flow of control and terminal. The terminal workstation will upload the collected network node information to the information acquisition module . The acquisition module  completes the data format conversion, noise reduction  and other preprocessing, carries out broadcast  communication, collects the fault information of multichannel communication network, and transmits the  instructions of the upper computer chip down. Centralized analysis and processing of the collected fault information, and determine the location, area and type of communication network fault. With the master-slave handshake protocol flow of communication fault detection, on the one hand, the location of communication network fault can be verified, that is, whether the communication fault occurs on the node or on the link [8] . On the other hand, even if there are counting errors and network delay faults in the communication network, they can be completely and accurately identified. The segmentation feature extraction of irregular fault data can extend the processing time of the whole extraction operation by stabilizing the data redundancy dimension. On this basis, the sample space of the fault data to be extracted is continuously expanded, so as to fundamentally reduce the irregular fault data The purpose of the overall dimension. Let f represent a segmented set of n fault control data, then the irregularity suppression factor coefficient of each set can be expressed as: Where, d represents n different irregular inhibition factor coefficients, representative coefficient acquisition parameters,  represents the weighted vector of factor coefficients,  is a fixed weighted period. The segmentation feature extraction results of fault data can be expressed as follows: Where, ( ) d  represents the definition function related to the average number of n irregular suppression factor coefficients, and l represents the standard feature extraction operator. The network structure includes a database server and a network terminal server. The database server is mainly installed in the Oracle database, and the servers are all installed in Apache server and deployed on the corresponding network page for the business logic in the foreground page [9] . Axis2 server is deployed in the network terminal server. Both database server and network terminal server are public servers. The required database server is public server, which is connected by router, and the servers connected by different routers are located in the subnet. According to the structure diagram, the communication fault is quickly The purpose of fault location algorithm is to find all possible faults and ensure the normal operation of remote communication of acquisition terminal [10] . According to the process shown in Figure 2, the specific analysis of its steps is as follows: obtain two warning information in any warning set; find the node that can explain the two warning information at the same time; use these nodes to replace the original warning information; repeat the acquisition Warning information, until there is no intersection description between the two warning information [11] . The signal model is designed by taking x and y as the signal acquisition module to collect the transmitted signal and the received signal.
Where s(t) is the original signal, k is the sampling times, E is the coefficient of the original signal of the multi-channel communication network, and n(t) is the function of the environmental noise of the fault detection.
Cross correlation operation is performed between the transmitted signal x(t) and the received signal y(t), then: Considering the environmental noise, the correlation between the two signals is rewritten as: The correlation between the transmitted signal and the received signal in the multi-channel communication network is identified. Based on the quadratic correlation algorithm, the accuracy of network delay estimation can be improved under the condition of low signal-to-noise ratio, and the detailed information contained in the multi-channel communication fault signal can be accurately extracted. Therefore, the block diagram of dynamic fault detection is shown in Figure 3. Due to the frequency difference between multichannel communication signal and noise signal, it can be assumed that the useful signal is not correlated with noise signal in fault signal feature recognition: signal in multi-channel communication network can be accurately extracted, so as to realize the location and identification of fault node or link, and eliminate the communication fault of multi-channel network in time [12] . After the normal communication module sends the message to the optimal communication module, the other sub modules recursively forward the message to all other sub modules, and finally get most of the same messages as the sub modules. WiFi is used for communication between communication modules, and fault detection mechanism can prevent error information from entering, so as to avoid wrong operation and congestion of communication protocol [13] . In optical communication, fault detection algorithm needs to complete the following three tasks. The detection scheme must ensure that all communication subsystems (communication module, optical signal) complete the normal message processing. The normal sub can retrieve the correct signal according to the detection scheme. If a sub works normally, the rest of the communication network can receive the correct optical signal. In order to improve the accuracy and efficiency of pattern recognition, neural network algorithm is selected to detect the error information in the communication process. Using the labeled communication network training set elements as reference, the difference between the network information and the real port label information can be detected to generate the prediction area information. Then, back propagation algorithm is used to adjust the weight and offset of the network. After enough training, the neural network can successfully detect the error signal in the optical communication network.

The realization of remote communication fault interval detection
Remote communication fault detection based on data mining relies on the implicit assumption, that is, the task of switching from a failed server to another running server. Therefore, the remote communication transmission fault data mining is divided into K groups and e groups, each group contains at least one object. Therefore, based on the situation of K failure and e taking over its task, the data mining of transmission failure data in remote communication is divided into K group and E group, and this exchange mechanism is described in detail. In the case of no fault, the K and e servers receive the sensor and all the packets sent by the sensor [14] . Only K responds to these packets, computes the necessary control packets and sends them to the specified actuator node. When e detects the missing data packet (indicating that K is faulty), it enters a cycle to replace the inactive K and send the control data packet to the designated actuator. The control process running on E and used to back up K in case of fault must be designed to adapt to the situation of missing data packet. Similarly, control cannot easily lose control packets, which is to overcome the possibility of losing at most one packet when switching between K and E. In this case, a simple solution is the "keep previous samples" technique until a new control packet is received. The possibility of telecommunication failure can be represented by three possible events: contention conflict, bandwidth unavailability and channel error. Let P denote the failure probability caused by the above three possible events, then: Where P e is the error probability caused by the channel, P 2 is the collision probability caused by contention, q is the probability that the base station accepts the bandwidth request, and T is the response time or waiting time. By rearranging, we can get the following results: In the process of fault detection, due to the characteristics of the equipment itself, the detection results contain a certain amount of noise. In order to improve the reliability of fault detection results, sub GHz technology should be used for the second noise reduction processing. Let p represent the basic physical period coefficient of real-time detection 2 2 ( Where y represents the standard detection vector of irregular control fault data, S  、 S   are represent the first-order detection operator and the second-order detection operator respectively,  、   represent the operation coefficients of the irregular control fault data when the fixed-point operator is obtained. In order to reduce the noise figure as much as possible, the parasitic parameters generated by the parasitic capacitance in the structure are calculated. The 50 ohm impedance is input to estimate the noise performance, and the maximum transmission power gain is achieved when the matching network is input, so that it can obtain better anti noise performance at lower working frequency. According to the noise reduction, the fault label can be obtained, and then the fault detection results of unknown samples can be obtained.

Analysis of experimental results
The maximum number of bytes allowed to be selected by the local host through TCP is 1024bytes 3.1.2. The local host protocol transmission control protocol (TCP) is the host environment through which frames are transmitted from the sender to the receiver. In order to ensure the accuracy of the experiment, the two remote communication fault detection methods are placed in the same test environment, using the same experimental data.
The experimental data information is shown in the table 1. Using the network security packet tracking mode produced by a domestic company, the cycle is guaranteed to be 4 days, and the total data flow is 62GB. In order to facilitate the subsequent training and testing, this paper synthesizes the common network feature data stream and extracts the features. Table 3 and 4 shows the sample data for training and testing. The specific parameters are set as follows: for each mobile phone, in the normal call mode, 10 segments of zero input, single frequency input and frequency mixing input are collected respectively, and 500 ms target data is selected from each segment; during frame processing, the frame length is set as 20 ms, the acquisition frequency is 44.0 kHz, and there are about 526 data points in each frame; the data type is marked as-T. The others are labeled t. Based on the above steps, the fault detection effects of the traditional method and the method in this paper are compared and analyzed. The specific detection results are shown in the following figure. According to the comparison chart of the experimental results, it is obvious that compared with the traditional remote communication fault detection method, with the increase of the number of tests, the accuracy of the detection results of this method increases, and has been higher than the traditional method.