Remote power monitoring system based on Internet of Things

Abstract

The invention belongs to the technical field of power system monitoring, and particularly relates to a remote power monitoring system based on the Internet of Things, which comprises a data acquisition module, a communication module, a data processing and analysis module, an alarm module and a remote control module, the data acquisition module is used for acquiring real-time data during operation of equipment in a power system; an alarm prompt can be correspondingly output through the alarm module, the alarm prompt is displayed on a display screen of a dispatching room, a dispatcher can carry out dispatching management according to the newly-appearing alarm prompt, and in addition, the remote control module can carry out remote control on equipment according to the alarm prompt, and the remote control comprises operations such as shutdown and frequency reduction. And the alarm module is also used for filtering part of alarm prompts, and can automatically filter alarm prompts which appear frequently but do not cause substantial faults to the equipment, so that the alarm prompts of key equipment are more striking, and a dispatcher can identify and make emergency disposal decisions conveniently.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system monitoring, and in particular to a remote power monitoring system based on the Internet of Things. Background Art

[0002] As the complexity and scale of power systems increase, traditional local monitoring methods can no longer meet the needs of modern power systems. Therefore, existing technologies use remote monitoring with sensors to meet the monitoring needs of modern power systems.

[0003] Remote monitoring of power systems refers to the use of communication technology to monitor and control remote operating equipment; its main task is to collect relevant real-time information on the operating status of the power system and each power plant and substation to the dispatching master station, and send commands from the dispatching master station to power plants and substations to remotely control and adjust the equipment.

[0004] However, in the practical application stage of the above-mentioned prior art, by comparing the characteristic parameters of the equipment during normal operation with the real-time data of the equipment during operation obtained by the sensor, it is judged whether there is a fault in the operation of the equipment or whether the equipment is about to fail. In essence, the above process has no defects, but in the actual operation of the power system, there are many factors that will cause the real-time data of the equipment to jump, but the jump does not indicate that the equipment is about to fail. Therefore, the characteristic parameters of the equipment during normal operation are compared with the real-time data, and the judgment result of data abnormality can be obtained from the data level, and then an alarm prompt is output. Based on this, in the long-term remote monitoring process of the power system, there are a large number of irrelevant alarm prompts, which may affect the dispatcher's identification of the failure prediction of key equipment;

[0005] To this end, the present invention provides a remote power monitoring system based on the Internet of Things. Summary of the invention

[0006] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve the technical problem is: a remote power monitoring system based on the Internet of Things described in the present invention comprises:

[0008] Data acquisition module, used to collect real-time data of equipment operation in the power system;

[0009] A communication module, used for transmitting real-time data to a data processing module through a network;

[0010] The data processing and analysis module is used to receive real-time data, process and analyze it, and generate analysis results;

[0011] An alarm module is used to generate alarm prompts based on the analysis results;

[0012] The remote control module is used to send control instructions to the device according to the alarm prompt.

[0013] Preferably, the data acquisition module includes a sensor array, and collects real-time data of the device based on the sensor array, and the real-time data includes voltage, current, temperature and power;

[0014] The sensor array and the communication module use wired or wireless communication methods, the wired communication method includes Ethernet, and the wireless communication method includes WiFi, 4G / 5G.

[0015] Preferably, the method by which the data processing and analysis module processes real-time data is:

[0016] Receiving real-time data and preprocessing the real-time data, wherein the preprocessing includes cleaning, denoising, filtering and standardization;

[0017] Perform feature extraction on preprocessed real-time data to obtain key parameters;

[0018] Set the monitoring time window and obtain the actual values ​​of key parameters changing over time within the monitoring time window;

[0019] Under the condition of equal time intervals within the monitoring time window, the actual values ​​of multiple key parameters changing over time are extracted to construct a key parameter set.

[0020] Preferably, the method by which the data processing and analysis module analyzes real-time data is:

[0021] Get key parameter set;

[0022] Inputting key parameter sets into the fault prediction model;

[0023] Output a predicted change curve representing key parameters based on the fault prediction model;

[0024] Based on the predicted change curve, determine whether the key parameters of the equipment will fail:

[0025] If there is a node on the prediction change curve where the key parameter exceeds the threshold, a fault prediction signal is output; otherwise, no fault prediction signal is output;

[0026] The fault diagnosis results of all key parameters of the equipment are counted and combined with the predicted fault time as the analysis result output; the predicted fault time is the time corresponding to the node exceeding the threshold on the predicted change curve, denoted as T X .

[0027] Preferably, the method for constructing the fault prediction model is:

[0028] Obtain historical operating data of equipment in the power system;

[0029] Build a fault prediction model based on machine learning algorithms;

[0030] Preprocess the historical operation data and divide the historical operation data into training set and test set;

[0031] Use the training set to train the fault prediction model and adjust the model parameters to improve the prediction accuracy;

[0032] Use the test set to verify the prediction accuracy of the fault prediction model and evaluate the performance of the model through methods such as cross-validation;

[0033] Adjust and optimize the model based on the verification results.

[0034] Preferably, the method of outputting a predicted change curve representing key parameters based on a fault prediction model further includes:

[0035] Setting a prediction time window, wherein the prediction time window is used to monitor actual values ​​of key parameters changing over time;

[0036] Output the predicted change curve of key parameters based on the fault prediction model before the prediction time window begins;

[0037] Compare the difference between the actual change curve and the predicted change curve of the key parameters at the end of the prediction time window;

[0038] According to the formula:

[0039]

[0040] Among them, RE is the relative error, f(t i ) is the key parameter at time node t i The actual value of g(t i ) is the key parameter at time node t i The predicted value is taken from the predicted change curve; n is the number of time nodes involved in the calculation, t i is the time node, the i Taken from the forecast time window;

[0041] When RE is less than the preset value, it means that the predicted change curve is consistent with the actual change curve of the key parameter, otherwise it means that the predicted change curve is inconsistent with the actual change curve of the key parameter.

[0042] Preferably, the method for generating an alarm prompt according to the analysis result is:

[0043] Establish important equipment sets and non-important equipment sets based on all equipment in the power system;

[0044] Obtain analysis results, analyze fault diagnosis results, predicted fault time and corresponding equipment in the analysis results;

[0045] Determine whether the device belongs to the important device set:

[0046] When the device belongs to the important device set and the fault diagnosis result is that a fault will occur, a first alarm signal is output;

[0047] When the device belongs to the important device set and the fault diagnosis result is that no fault will occur, the first alarm signal is not output;

[0048] When the device belongs to the non-important device set and the fault diagnosis result is that a fault will occur, a second alarm signal is output;

[0049] When the device belongs to the non-important device set and the fault diagnosis result is that no fault will occur, the second alarm signal is not output;

[0050] The alarm prompt includes a first alarm signal and a second alarm signal; the first alarm signal and the second alarm signal are displayed on the main control display screen in different colors respectively.

[0051] Preferably, the method for generating an alarm prompt according to the analysis result also includes:

[0052] Extract tag parameters;

[0053] comparing the marked parameters with the key parameters in the analysis results;

[0054] If the marked parameter matches the key parameter in the analysis result, the alarm prompt corresponding to the key parameter is automatically filtered, otherwise the alarm prompt is not filtered;

[0055] The method for determining the marking parameters is:

[0056] Acquire historical alarm data, wherein the historical alarm data consists of alarm prompts and analysis results;

[0057] Extract alarm prompts from historical alarm data;

[0058] Determine the equipment failure probability corresponding to the alarm prompt in the historical alarm data;

[0059] The key parameters corresponding to the alarm prompts below the probability threshold are defined as marking parameters.

[0060] Preferably, it also includes a data storage module for storing real-time data of the power system collected by the data acquisition module, and constructing historical operation data based on the real-time data; the data storage module is also used to store the analysis results output by the data processing and analysis module, and generate logs in combination with alarm prompts.

[0061] Preferably, the machine learning algorithm includes one or more combinations of a random forest algorithm, a support vector machine algorithm or a neural network algorithm.

[0062] The beneficial effects of the present invention are as follows:

[0063] The remote power monitoring system based on the Internet of Things described in the present invention analyzes real-time data, including analyzing whether the real-time data indicates that a fault has occurred in the equipment at the current moment, or whether the equipment will fail in the future, and outputs the analysis results according to the content of the analysis; according to the content of the analysis results, the alarm module can output an alarm prompt accordingly, and the alarm prompt is displayed on the display screen in the dispatching room. The dispatcher can perform dispatch management according to the newly appeared alarm prompt. In addition, the remote control module can also remotely control the equipment according to the alarm prompt, including shutting down, reducing the frequency and other operations; the alarm module is also used to filter some alarm prompts, and can automatically filter alarm prompts that appear frequently but do not cause actual faults to the equipment, thereby making the alarm prompts of key equipment more eye-catching, which is convenient for the dispatcher to identify and make emergency disposal decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] The present invention will be further described below in conjunction with the accompanying drawings.

[0065] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0066] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0067] like Figure 1 As shown, a remote power monitoring system based on the Internet of Things described in an embodiment of the present invention includes a data acquisition module, a communication module, a data processing and analysis module, an alarm module and a remote control module; the data acquisition module is used to collect real-time data of equipment in the power system during operation; the communication module is used to transmit real-time data to the data processing module through the network; the data processing and analysis module is used to receive real-time data, process and analyze it, and generate analysis results; the alarm module is used to generate alarm prompts according to the analysis results; the remote control module is used to send control instructions to the device according to the alarm prompts.

[0068] In one embodiment of the present invention, for remote monitoring of the power system, it is first necessary to build a data acquisition module for monitoring the operation data of the equipment, collect the real-time data of the equipment during operation based on the data acquisition module, and compare the real-time data with the normal data of the equipment during operation to determine or predict whether the equipment has a fault; in addition, it is also necessary to set up a communication module, upload the real-time data of different equipment during operation collected by the data acquisition module to the data processing and analysis module through the network based on the communication module, and use the data processing and analysis module to pre-process the real-time data. The real-time data after pre-processing can be used for analysis. The analysis of the real-time data includes analyzing whether the real-time data indicates that the equipment has a fault at the current moment, or whether the equipment will fail in the future, and outputting the analysis results according to the content of the analysis; according to the content of the analysis result, the alarm module can output an alarm prompt correspondingly, and the alarm prompt is displayed on the display screen of the dispatching room. The dispatcher can perform dispatch management according to the newly appeared alarm prompt. In addition, the remote control module can also remotely control the equipment according to the alarm prompt, including shutting down, reducing the frequency and other operations;

[0069] It is worth noting that the above-mentioned alarm module is also used to filter some alarm prompts. In the process of monitoring the operation of the power system, as the power system runs, its different equipment may generate a large number of alarm prompts due to environmental factors, and most of the alarm prompts may not cause equipment failure problems. Therefore, if a large number of alarm prompts are displayed on the display screen of the dispatching room, it will affect the dispatcher's identification of the alarm prompts of key equipment. For example, it will cause the dispatcher to ignore the alarm prompts of key equipment, thereby missing the best time for emergency response and incurring greater losses. Therefore, the alarm module is also used to filter some alarm prompts. It can automatically filter out alarm prompts that appear frequently but do not cause actual failures to the equipment, thereby making the alarm prompts of key equipment more eye-catching and convenient for dispatchers to identify and make emergency response decisions.

[0070] In one embodiment, the data acquisition module includes a sensor array, and collects real-time data of the device based on the sensor array, and the real-time data includes voltage, current, temperature and power;

[0071] The sensor array and the communication module use wired or wireless communication methods, the wired communication method includes Ethernet, and the wireless communication method includes WiFi, 4G / 5G.

[0072] In one embodiment, the method by which the data processing and analysis module processes real-time data is:

[0073] Receiving real-time data and preprocessing the real-time data, wherein the preprocessing includes cleaning, denoising, filtering and standardization;

[0074] Perform feature extraction on preprocessed real-time data to obtain key parameters;

[0075] Set the monitoring time window and obtain the actual values ​​of key parameters changing over time within the monitoring time window;

[0076] Under the condition of equal time intervals within the monitoring time window, the actual values ​​of multiple key parameters changing over time are extracted to construct a key parameter set.

[0077] Based on the above, after obtaining the real-time data of the equipment during operation through the sensor array, the real-time data must be preprocessed to make the real-time data more valuable for calculation and analysis. In addition, in order to facilitate the analysis of real-time data, in one embodiment of the present invention, feature extraction is performed on the preprocessed real-time data to extract key parameters reflecting the operation of the equipment, and redundant and irrelevant features, such as environmental parameter data, are eliminated to simplify the data model and improve the analysis efficiency. The key parameters extracted from different devices are different. For example, the oil temperature of the transformer is easily affected by the ambient temperature. In the high temperature season in summer, the heat dissipation condition of the transformer is not good, and its oil temperature rises significantly, which is a normal phenomenon; secondly, in order to reflect the changes of the key parameters of the equipment during operation over time, a monitoring time window is set to obtain the actual parameters of the equipment during operation for a period of time, which are recorded as actual values ​​here, and then extract multiple actual values ​​of the corresponding key parameters at different time nodes by equal time intervals. A key parameter set is constructed based on multiple actual values. The above key parameter set can be used to train a fault prediction model. After the fault prediction model is trained, the key parameters can be predicted, so as to determine whether the key parameters will fail, which is conducive to realizing the state prediction and fault prediction of the equipment operation in the power system, and facilitating the dispatching personnel to make emergency disposal decisions in advance to reduce losses;

[0078] Exemplarily, the operation data of the transformer equipment is monitored, with the monitoring current as the key parameter. In addition, the key parameters of the transformer also include temperature, gas, humidity, etc. By monitoring the current of the transformer equipment, the load condition of the transformer is understood, and it is determined whether the transformer has abnormal conditions such as overload and underload. The monitoring time window can be set to 6 hours. From the beginning to the end of the monitoring time window, 12 actual values ​​of the temperature are monitored and extracted at equal time intervals, such as 0.5 hours. The key parameter set of the temperature is constructed with these 12 actual values ​​for the subsequent output temperature prediction change curve; in addition, in order to distinguish the same key parameters of different equipment, the key parameters can be named. Exemplarily, taking the transformer temperature and oil temperature as examples, the key parameter of the transformer temperature is recorded as B RT , and the key parameter of transformer oil temperature is recorded as B YT ; Based on this, a distinction can be made; B can be understood as the abbreviation of transformer, while RT and YT represent transformer temperature and oil temperature respectively.

[0079] In one embodiment, the method by which the data processing and analysis module analyzes real-time data is:

[0080] Get key parameter set;

[0081] Inputting key parameter sets into the fault prediction model;

[0082] Output a predicted change curve representing key parameters based on the fault prediction model;

[0083] Based on the predicted change curve, determine whether the key parameters of the equipment will fail:

[0084] If there is a node on the prediction change curve where the key parameter exceeds the threshold, a fault prediction signal is output; otherwise, no fault prediction signal is output;

[0085] The fault diagnosis results of all key parameters of the equipment are counted and combined with the predicted fault time as the analysis result output; the predicted fault time is the time corresponding to the node exceeding the threshold on the predicted change curve, denoted as T X .

[0086] Based on the above, after the key parameter set of the equipment is obtained based on the monitoring time window, by inputting the key parameter set into the fault prediction model, the fault prediction model can be used to output a predicted change curve representing the key parameters, that is, the key parameter set obtained through the monitoring time window, through data analysis, predicts and outputs the curve of the key parameter changing with time, which is recorded as the predicted change curve of the key parameter. By analyzing the predicted change curve of the key parameter, it can be determined whether there is a node exceeding the threshold on the predicted change curve of the key parameter, and the threshold is the normal range of the key parameter. When there is a node exceeding the threshold on the predicted change curve of the key parameter, an early warning signal is output, and it can be determined that the key parameter is predicted to fail, and according to the time corresponding to the node on the predicted change curve where the key parameter exceeds the threshold, the time when the equipment fails can be predicted. By outputting the fault diagnosis result and the predicted fault time as the analysis result, it is convenient for the dispatcher to remotely monitor the operation of each device in the power system and predict the operation status and fault diagnosis of each device, so that emergency disposal decisions can be made effectively in advance to reduce losses; exemplary, the key parameter set of transformer oil temperature can be recorded as:

[0087] In one embodiment, the method for constructing the fault prediction model is:

[0088] Obtain historical operating data of equipment in the power system;

[0089] Build a fault prediction model based on machine learning algorithms;

[0090] Preprocess the historical operation data and divide the historical operation data into training set and test set;

[0091] Use the training set to train the fault prediction model and adjust the model parameters to improve the prediction accuracy;

[0092] Use the test set to verify the prediction accuracy of the fault prediction model and evaluate the performance of the model through methods such as cross-validation;

[0093] Adjust and optimize the model based on the verification results.

[0094] In one embodiment, the method of outputting a predicted change curve representing a key parameter based on a fault prediction model further includes:

[0095] Setting a prediction time window, wherein the prediction time window is used to monitor actual values ​​of key parameters changing over time;

[0096] Output the predicted change curve of key parameters based on the fault prediction model before the prediction time window begins;

[0097] Compare the difference between the actual change curve and the predicted change curve of the key parameters at the end of the prediction time window;

[0098] According to the formula:

[0099]

[0100] Among them, RE is the relative error, f(t i ) is the key parameter at time node t i The actual value of g(t i ) is the key parameter at time node t i The predicted value is taken from the predicted change curve; n is the number of time nodes involved in the calculation, t i is the time node, the i Taken from the forecast time window;

[0101] When RE is less than the preset value, it means that the predicted change curve is consistent with the actual change curve of the key parameter, otherwise it means that the predicted change curve is inconsistent with the actual change curve of the key parameter.

[0102] Based on the above, the key parameter set characterizing the operating status of the equipment is obtained by monitoring the time window, and the key parameter set is input into the fault prediction model. Then, the predicted change curve characterizing the key parameters is output based on the fault prediction model. It is judged whether the key parameters have a fault or are about to fail by the predicted change curve. Based on this, there is a prerequisite, that is, the accuracy of the fault prediction model must be high enough. Otherwise, due to the large accuracy error of the fault prediction model, the predicted change curve will not match the actual change curve of the key parameters, so as to generate a false alarm, thereby affecting and interfering with the dispatching personnel in identifying the operating status of the key equipment. In one embodiment of the present invention, before the key parameter set is input into the fault prediction model, it is also necessary to verify the accuracy of the fault prediction model. By setting the prediction time window The prediction time window can be set after the monitoring time window, and the prediction change curve output at the beginning of the prediction time window is used to perform error analysis in combination with the actual change curve of the key parameters at the end of the prediction time window. According to the above formula, the relative error between the actual change curve and the prediction change curve of the key parameters can be calculated. The calculation of the relative error can be used to measure whether the accuracy of the fault prediction model meets the requirements; it is worth noting that the prediction time window and the monitoring time window can be understood as a preset time period, the monitoring time window is used to obtain the actual value of the key parameter changing over time, and the prediction time window is used to judge the difference between the predicted change curve and the actual change curve of the key parameter; secondly, the preset value for comparison with RE is artificially set, and 10% is taken here;

[0103] Exemplary,

[0104] Taking the temperature of the transformer as the key parameter, the predicted change curve of the key parameter within the prediction time window is output. The prediction time window here is [t1, t3]. Based on the predicted change curve, the predicted values ​​g(t i ),as follows:

[0105] f(t1)=10 g(t1)=9

[0106] f(t2)=15 g(t2)=14

[0107] f(t3)=20 g(t3)=21

[0108] Among them, t1, t2, t3 and other time intervals are set; f(t1) represents the actual value of the temperature at time t1, and g(t1) represents the predicted value of the temperature at time t1;

[0109] Based on the above, calculate the absolute error at each time node:

[0110] |f(t1)-g(t1)|=|10-9|=1

[0111] |f(t2)-g(t2)|=|15-14|=1

[0112] |f(t3)-g(t3)|=|20-21|=1

[0113] Calculate the relative error:

[0114]

[0115] Based on the above, when the preset value is 10%, the relative error RE = 7.14% < 10%; it can be judged that the accuracy of the fault prediction model is within the requirements, so the key parameter set is input into the fault prediction model, and then the predicted change curve output by the fault prediction model can be used to characterize the actual value of the key parameter changing over time. On the contrary, if the relative error RE> 10%, the predicted change curve output by the fault prediction model cannot be used to characterize the actual value of the key parameter changing over time, and the fault prediction model needs to be optimized and the parameters adjusted to achieve improved accuracy.

[0116] In one embodiment, the method for generating an alarm prompt according to the analysis result is:

[0117] Establish important equipment sets and non-important equipment sets based on all equipment in the power system;

[0118] Obtain analysis results, analyze fault diagnosis results, predicted fault time and corresponding equipment in the analysis results;

[0119] Determine whether the device belongs to the important device set:

[0120] When the device belongs to the important device set and the fault diagnosis result is that a fault will occur, a first alarm signal is output;

[0121] When the device belongs to the important device set and the fault diagnosis result is that no fault will occur, the first alarm signal is not output;

[0122] When the device belongs to the non-important device set and the fault diagnosis result is that a fault will occur, a second alarm signal is output;

[0123] When the device belongs to the non-important device set and the fault diagnosis result is that no fault will occur, the second alarm signal is not output;

[0124] The alarm prompt includes a first alarm signal and a second alarm signal; the first alarm signal and the second alarm signal are displayed on the main control display screen in different colors respectively.

[0125] Based on the above, after the data processing and analysis module outputs the analysis results, the alarm module can output an alarm prompt based on the analysis results. In one embodiment of the present invention, the analysis results are obtained and parsed to obtain the fault diagnosis results, predicted fault time and corresponding equipment in the analysis results. It should be noted that in the entire power system, there are important equipment and auxiliary equipment, etc. Therefore, for the dispatcher, if too much attention is paid to the operation alarm prompts of the auxiliary equipment, the operation alarm prompts of the important equipment may be ignored, thereby interfering with the dispatcher. For example, in the CDT protocol, the position change telesignal and substation status change information are usually given a higher priority because they require a quick response, and are required to be delivered to the dispatcher within 1 second. Based on this, all equipment in the power system is managed in zones, including establishing important equipment sets and non-important equipment sets. For equipment predicted to fail as a result of the fault diagnosis in the analysis results, the equipment is identified to determine whether the equipment exists in the important equipment set. If the equipment exists in the important equipment set, a first alarm signal is output; if the equipment does not exist in the important equipment set, a second alarm signal is output; if the fault diagnosis result predicts that no failure will occur, neither the first alarm signal nor the second alarm signal is output. It is worth noting that in order to distinguish the first alarm signal from the second alarm signal, different colors are used to distinguish the first alarm signal from the second alarm signal on the display screen in the dispatching room to avoid interference with dispatching.

[0126] In one embodiment, the method for generating an alarm prompt according to the analysis result further includes:

[0127] Extract tag parameters;

[0128] comparing the marked parameters with the key parameters in the analysis results;

[0129] If the marked parameter matches the key parameter in the analysis result, the alarm prompt corresponding to the key parameter is automatically filtered, otherwise the alarm prompt is not filtered;

[0130] The method for determining the marking parameters is:

[0131] Acquire historical alarm data, wherein the historical alarm data consists of alarm prompts and analysis results;

[0132] Extract alarm prompts from historical alarm data;

[0133] Determine the equipment failure probability corresponding to the alarm prompt in the historical alarm data;

[0134] The key parameters corresponding to the alarm prompts below the probability threshold are defined as marking parameters.

[0135] Since there are a large number of alarm prompts in the process of remote monitoring of the power system, it will interfere with the dispatch. In one embodiment of the present invention, by obtaining historical alarm data, extracting the marking parameters in the historical alarm data, comparing the marking parameters with the key parameters in the analysis results, and judging whether the two represent an operating parameter, the overhead transmission line is exposed to the atmospheric environment for a long time and is easily affected by extreme weather. For example, meteorological disasters such as heavy rain, sandstorms, and ice cover may cause damage to the hardware, resulting in line failure. In addition, strong winds, lightning strikes, etc. may also cause the line to be temporarily grounded or short-circuited, thereby triggering an alarm. Although these alarm messages are frequent, they do not necessarily mean that there is a serious fault in the line. Sometimes it is just a transient phenomenon. For example, the oil temperature of the oil-immersed transformer is greatly affected by the ambient temperature. In the hot summer season, if the heat dissipation condition of the transformer is not good, the oil temperature may rise, triggering an alarm of excessive oil temperature; but as long as the oil temperature does not exceed the allowable value of the transformer and the heat dissipation system is normal, this alarm generally does not need to be processed immediately. Based on the above, through analysis, if the marked parameter and the key parameter in the analysis result represent the same operating parameter, the priority of the alarm prompt of the equipment that frequently triggers alarms due to environmental factors or slight fluctuations will be dynamically reduced, and the frequency of display or appearance will be reduced to avoid affecting the dispatcher from viewing the alarm prompts with higher priority; the above priority indicates whether the key parameter is important to the operating status of the equipment, and can also be understood as whether the key parameter can directly or indirectly cause equipment failure. If the key parameter can directly or indirectly cause equipment failure, the priority is higher, otherwise the alarm prompt corresponding to the key parameter can be filtered out; in addition, through the above filtering of the alarm prompts corresponding to the marked parameters, the number of dispatches can be reduced and accurate monitoring and prediction of faults can be achieved under the premise of remote monitoring, which is conducive to the optimization of dispatch management.

[0136] It is worth noting that the marking parameters are obtained through training. Based on the operation of the remote power system, when an alarm prompt appears, the key parameters associated with the alarm prompt can be analyzed based on big data to calculate the probability of equipment failure caused by the alarm prompt of the key parameter. If the probability of equipment failure is small, it means that the alarm prompt of the key parameter is not important for the remote monitoring of the power system, that is, it will not affect the operation of the power system. On the contrary, if there are a large number of such alarm prompts, it may interfere with the dispatcher's identification of important key parameters of the equipment; for example, taking the transformer oil temperature as an example, during the remote monitoring of the power system, due to the high temperature in summer, the oil temperature rises and an alarm prompt is generated, but the alarm prompt has no significant impact on the normal operation of the power system, so even if an alarm prompt is generated, it can be ignored. Based on this, the oil temperature is defined as a marking parameter through big data, denoted as B YT , if the alarm received at the current moment contains B YT , the alarm prompt will be automatically filtered.

[0137] In one embodiment, it also includes a data storage module for storing real-time data of the power system collected by the data acquisition module, and constructing historical operation data based on the real-time data; the data storage module is also used to store the analysis results output by the data processing and analysis module, and generate logs in combination with alarm prompts.

[0138] In one embodiment, the machine learning algorithm includes one or more combinations of a random forest algorithm, a support vector machine algorithm, or a neural network algorithm.

[0139] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A remote power monitoring system based on the Internet of Things, characterized in that: include: Data acquisition module, used to collect real-time data of equipment operation in the power system; A communication module, used for transmitting real-time data to a data processing module through a network; The data processing and analysis module is used to receive real-time data, process and analyze it, and generate analysis results; An alarm module is used to generate alarm prompts based on the analysis results; The remote control module is used to send control instructions to the device according to the alarm prompt.

2. The remote power monitoring system based on the Internet of Things according to claim 1 is characterized in that: The data acquisition module includes a sensor array, and collects real-time data of the device based on the sensor array, wherein the real-time data includes voltage, current, temperature and power; The sensor array and the communication module use wired or wireless communication methods, the wired communication method includes Ethernet, and the wireless communication method includes WiFi, 4G / 5G.

3. The remote power monitoring system based on the Internet of Things according to claim 2 is characterized in that: The method by which the data processing and analysis module processes real-time data is as follows: Receiving real-time data and preprocessing the real-time data, wherein the preprocessing includes cleaning, denoising, filtering and standardization; Perform feature extraction on preprocessed real-time data to obtain key parameters; Set the monitoring time window and obtain the actual values ​​of key parameters changing over time within the monitoring time window; Under the condition of equal time intervals within the monitoring time window, the actual values ​​of multiple key parameters changing over time are extracted to construct a key parameter set.

4. The remote power monitoring system based on the Internet of Things according to claim 3 is characterized in that: The method by which the data processing and analysis module analyzes real-time data is as follows: Get key parameter set; Inputting key parameter sets into the fault prediction model; Output a predicted change curve representing key parameters based on the fault prediction model; Based on the predicted change curve, determine whether the key parameters of the equipment will fail: If there is a node on the prediction change curve where the key parameter exceeds the threshold, a fault prediction signal is output; otherwise, no fault prediction signal is output; The fault diagnosis results of all key parameters of the equipment are counted and combined with the predicted fault time as the analysis result output; the predicted fault time is the time corresponding to the node exceeding the threshold on the predicted change curve, denoted as T X .

5. The remote power monitoring system based on the Internet of Things according to claim 4 is characterized in that: The method for constructing the fault prediction model is: Obtain historical operating data of equipment in the power system; Build a fault prediction model based on machine learning algorithms; Preprocess the historical operation data and divide the historical operation data into training set and test set; Use the training set to train the fault prediction model and adjust the model parameters to improve the prediction accuracy; Use the test set to verify the prediction accuracy of the fault prediction model and evaluate the performance of the model through methods such as cross-validation; Adjust and optimize the model based on the verification results.

6. The remote power monitoring system based on the Internet of Things according to claim 5 is characterized in that: The method of outputting a predicted change curve representing a key parameter based on a fault prediction model also includes: Setting a prediction time window, wherein the prediction time window is used to monitor actual values ​​of key parameters changing over time; Output the predicted change curve of key parameters based on the fault prediction model before the prediction time window begins; Compare the difference between the actual change curve and the predicted change curve of the key parameters at the end of the prediction time window; According to the formula: Among them, RE is the relative error, f(t i ) is the key parameter at time node t i The actual value of g(t i ) is the key parameter at time node t i The predicted value is taken from the predicted change curve; n is the number of time nodes involved in the calculation, t i is the time node, the i Taken from the forecast time window; When RE is less than the preset value, it means that the predicted change curve is consistent with the actual change curve of the key parameter, otherwise it means that the predicted change curve is inconsistent with the actual change curve of the key parameter.

7. The remote power monitoring system based on the Internet of Things according to claim 6 is characterized in that: The method for generating an alarm prompt according to the analysis result is: Establish important equipment sets and non-important equipment sets based on all equipment in the power system; Obtain analysis results, analyze fault diagnosis results, predicted fault time and corresponding equipment in the analysis results; Determine whether the device belongs to the important device set: When the device belongs to the important device set and the fault diagnosis result is that a fault will occur, a first alarm signal is output; When the device belongs to the important device set and the fault diagnosis result is that no fault will occur, the first alarm signal is not output; When the device belongs to the non-important device set and the fault diagnosis result is that a fault will occur, a second alarm signal is output; When the device belongs to the non-important device set and the fault diagnosis result is that no fault will occur, the second alarm signal is not output; The alarm prompt includes a first alarm signal and a second alarm signal; the first alarm signal and the second alarm signal are displayed on the main control display screen in different colors respectively.

8. The remote power monitoring system based on the Internet of Things according to claim 7 is characterized in that: The method for generating an alarm prompt according to the analysis result also includes: Extract tag parameters; comparing the marked parameters with the key parameters in the analysis results; If the marked parameter matches the key parameter in the analysis result, the alarm prompt corresponding to the key parameter is automatically filtered, otherwise the alarm prompt is not filtered; The method for determining the marking parameters is: Acquire historical alarm data, wherein the historical alarm data consists of alarm prompts and analysis results; Extract alarm prompts from historical alarm data; Determine the equipment failure probability corresponding to the alarm prompt in the historical alarm data; The key parameters corresponding to the alarm prompts below the probability threshold are defined as marking parameters.

9. The remote power monitoring system based on the Internet of Things according to claim 8, characterized in that: It also includes a data storage module for storing real-time data of the power system collected by the data acquisition module, and constructing historical operation data based on the real-time data; the data storage module is also used to store the analysis results output by the data processing and analysis module, and generate logs in combination with alarm prompts.

10. The remote power monitoring system based on the Internet of Things according to claim 9, characterized in that: The machine learning algorithm includes one or more combinations of a random forest algorithm, a support vector machine algorithm or a neural network algorithm.

Images