Method and device for real-time prediction and fault warning of DC current based on autoregression

Abstract

The present invention provides a method and device for real-time prediction and fault warning of direct current based on autoregression, which relates to the technical field of current prediction, and aims to realize accurate and real-time prediction and fault warning of direct current with low computational cost, low economic cost, including receiving current data and then sampling to obtain a historical data set; establishing a current regression equation based on the historical data set, the current regression equation is used to describe the correlation between the historical data value and the current data value of the current; obtaining the current sampling data set, and rolling predicting the future current trend through the current regression equation according to the current sampling data set, and obtaining the current value at several moments in the future; judging whether the fault alarm triggering condition is met, and sending a corresponding alarm signal. The present invention has the advantages of high cost performance, high reliability and strong scalability.

Description

Technical Field

[0001] The present invention relates to the technical field of current prediction, and in particular to a method and device for real-time prediction of direct current and fault warning based on autoregression. Background Art

[0002] With the continuous development of the global economy and the improvement of people's quality of life, the power industry has ushered in rapid development. Whether it is a traditional power system or a new renewable energy system, current monitoring and early warning are needed to ensure the stability and safety of system operation. Current measurement, current trend prediction, and fault early warning are indispensable links in current monitoring and early warning.

[0003] As an important means of long-distance power transmission, the construction of ultra-high voltage direct current transmission projects has continued to grow in recent years, and various types of optical fiber current transformers for measuring direct current have been installed and put into operation. However, there is still a lack of monitoring methods for key state parameters (sensitivity, linearity, frequency response range, working current range, temperature drift, etc.) in the current measurement device, so it is impossible to timely detect abnormal hidden dangers, which seriously restricts the reliability of the measurement device operation. Therefore, the operation status monitoring and early warning of the direct current measurement device are of great practical significance. Among them, the prediction of current trend can provide the main data support for its fault early warning and cause analysis. The existing direct current fault early warning methods usually select individual common faults, directly perform data annotation and model construction based on the mechanism analysis of the fault, and use data-driven learning and correction of parameters related to the fault mechanism. This type of method has the following four obvious disadvantages: (1) relying on a single fault, low scalability; (2) having strict requirements on professional knowledge, which is not conducive to subsequent maintenance; (3) requiring a high amount of data and requiring a large data collection cost; (4) only being able to judge whether a fault has occurred, unable to provide intuitive current prediction values, and having weak persuasiveness. In order to observe the current value intuitively, the present invention proposes a method that combines current prediction and fault judgment, and uses the current prediction result as the direct data support for fault warning. However, similar to the current fault warning method, most previous current prediction methods also require a lot of domain knowledge and mathematical deduction ability, the design is relatively complex, and the practical application is relatively difficult. The more innovative DC current prediction method uses machine learning and big data technology to solve the above problems, but this type of method basically uses supervised methods such as random forests and neural networks, and requires manual labeling of data, such as whether it is abnormal data, corresponding fault type, etc., which greatly increases labor costs and economic costs.

[0004] Therefore, when the fault causes are complex, professional knowledge is lacking, and the amount of data is insufficient, the accuracy and sensitivity of the above methods in current prediction and fault warning will not be able to meet the growing power supply needs. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for real-time prediction and fault warning of direct current based on autoregression, which can achieve low computational cost, low economic cost and accurate real-time prediction of direct current and fault warning.

[0006] The embodiments of the present invention are implemented by the following technical solutions:

[0007] The present invention provides a method for real-time prediction and fault warning of direct current based on autoregression, comprising the following steps:

[0008] Receive current data and then perform sampling to obtain a historical acquisition data set;

[0009] Establishing a current regression equation based on the historical data set, wherein the current regression equation is used to describe the correlation between the historical data value and the current data value of the current;

[0010] Acquire a current sampling data set, and according to the current sampling data set, use the current regression equation to rollingly predict the future current trend, and obtain the current values ​​at the next m moments;

[0011] Determine whether the fault alarm triggering conditions are met. If yes, send an alarm signal; otherwise, do nothing.

[0012] Preferably, the method for obtaining the current history acquisition data set is:

[0013] Sampling the current data at unit time intervals;

[0014] The sampled current data are arranged in ascending order according to time sequence to form the historical acquisition data set.

[0015] Preferably, the unit time is 1 second.

[0016] Preferably, the method for establishing the current regression equation is:

[0017] Select the n data most recent to the current time t from the historically collected data set;

[0018] A linear equation is established based on the unknown coefficient w and the random error term b:

[0019] y=wx+b;

[0020] Get the dependent variable y:

[0021] y=I i ;

[0022] Among them, I i The current data at time i in the historical collection data set;

[0023] Get the independent variable x:

[0024] x=(I i-1 ,I i-2 ,…,I i-k+1 ,I i-k ) T ;

[0025] Among them, the elements in x are I i Current data of the previous k moments;

[0026] The above steps are performed sequentially with i=t+k-n+1,t+k-n+2,…,t-1,t to obtain multiple linear equations;

[0027] A linear equation group is formed according to the established multiple linear equations to obtain a current regression equation.

[0028] Preferably, the method for obtaining the current regression equation according to the linear equation group is:

[0029] The linear equations are fitted to obtain the unknown coefficient w, and the fitting objective function is:

[0030]

[0031] Among them, I i is the actual current at time i, The actual current at time i is predicted based on the linear equation system.

[0032] Preferably, the method for rolling prediction of the future current trend is to predict the current values ​​at the next m moments in sequence:

[0033] The independent variable x is constructed by the current sampling data set. t ,I t-1 ,…,I t-k+2 ,I t-k+1 ) T , the elements in the current sampling data set are the current data at k moments up to the current moment t;

[0034] The dependent variable is obtained by using the current regression equation according to the current sampling data set. in, is the predicted current value at time r+1;

[0035] Initialize j=0;

[0036] The operation of cutting off the updated value is performed until j=m-1:

[0037] Construct the independent variable x:

[0038]

[0039] The dependent variable is obtained by using the current regression equation according to the current sampling data set. in, is the predicted current value at time t+j+1;

[0040] Update value: j=j+1;

[0041] Get the current value at the next m moments

[0042] Preferably, the method for acquiring the current sampling data set is:

[0043] Select k data closest to the current sampling time from the historically collected data set to form the current sampling data set.

[0044] Preferably, the method for determining whether the fault alarm triggering condition is met is:

[0045] Set multiple fault alarm trigger conditions;

[0046] It is determined in sequence whether the current values ​​at the future m moments meet each of the fault alarm triggering conditions. If all of them are not met, it is determined that the fault alarm triggering conditions are not met; if any of them are met, it is determined that the fault alarm triggering conditions are met.

[0047] Preferably, the fault alarm triggering conditions include a current upper limit threshold, a current lower limit threshold, a current fluctuation amplitude and a current over-limit frequency.

[0048] The present invention also provides a direct current real-time prediction and fault warning device based on autoregression, which is applied to any of the direct current real-time prediction and fault warning methods based on autoregression described above, comprising:

[0049] A data acquisition module is used to receive current data and then perform sampling to obtain a historical acquisition data set;

[0050] An autoregressive fitting module, used to establish a current regression equation based on the historical acquisition data set, wherein the current regression equation is used to describe the correlation between the historical data value and the current data value of the current;

[0051] An autoregressive prediction module is used to obtain a current sampling data set, and according to the current sampling data set, a rolling prediction of the future current trend is made through the current regression equation to obtain the current values ​​at the next m moments;

[0052] The fault diagnosis module is used to determine whether the fault alarm triggering conditions are met. If so, an alarm signal is sent; otherwise, no operation is performed.

[0053] The technical solution of the embodiment of the present invention has at least the following advantages and beneficial effects:

[0054] The present invention can realize the real-time prediction of future current by relying on the real-time current measurement device, and combine the current prediction with the fault judgment, intuitively give a real-time early warning of the fault, timely discover and solve the potential fault of the device, and has high reliability;

[0055] The present invention adopts an unsupervised and self-supervised learning algorithm framework, which does not require prior model training and uses unlabeled data itself for online learning. The cost of model training and data labeling is reduced, which greatly improves the development and implementation efficiency of the current monitoring and early warning system.

[0056] The prediction method of the present invention has good adaptability and scalability, and can be applied to various types of optical fiber current transformers with different principles, and can be migrated to other types of measurement values ​​such as voltage, temperature, optical power, optical fiber loss, etc., thereby expanding the scope of state monitoring and fault management and comprehensively analyzing the causes of faults;

[0057] Based on the reliability and maintainability of the present invention, the research results can be used for advance testing and online monitoring of various measuring devices, and potential hidden dangers in the operation of the device can be checked through fault simulation, mechanism analysis, status monitoring and abnormal early warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic flow chart of a method for real-time prediction and fault warning of direct current based on autoregression provided in Example 1 of the present invention;

[0059] Figure 2 A schematic diagram of the system framework of the autoregressive-based DC current real-time prediction and fault warning device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0061] Example 1

[0062] This embodiment provides a method for real-time prediction and fault warning of DC current based on autoregression, see Figure 1 , including the following steps:

[0063] Step S1: receiving current data and then sampling to obtain a historical acquisition data set; the current data comes from a measuring device;

[0064] Step S2: establishing a current regression equation based on the historical data set, wherein the current regression equation is used to describe the correlation between the historical data value and the current data value of the current;

[0065] Step S3: obtaining a current sampling data set, and rolling-forecasting the future current trend through the current regression equation according to the current sampling data set to obtain the current values ​​at the next m moments;

[0066] Step S4: Determine whether the fault alarm triggering condition is met, if so, send an alarm signal, if not, do nothing.

[0067] As a preferred solution of this embodiment, the method for obtaining the current history acquisition data set is:

[0068] Sampling the current data at unit time intervals;

[0069] The sampled current data are arranged in ascending order according to time sequence to form the historical acquisition data set.

[0070] Furthermore, the unit time is 1 second.

[0071] This embodiment constructs a historical collection data set based on the collected current data, then establishes a current regression equation for current prediction based on the historical collection data set, and then performs prediction and early warning work according to the current regression equation.

[0072] As an example, in step S1, the system can send a sampling signal to the data acquisition module at intervals of 1 second to obtain a historical data set. When the historical data set has accumulated to 1,500 items, that is, the sampling lasts for 25 minutes, the current regression equation can be established to start current prediction. After that, the system can send a prediction signal to the data acquisition module at intervals of 1 minute to perform multiple predictions, that is, the prediction and early warning steps of steps S2-S3 can be executed at a frequency of 1 minute.

[0073] Based on the above, when the data acquisition module receives the sampling signal at time i, the current I at the current moment is obtained from the measurement point of the optical fiber current transformer. i , continuously accumulating historical data. When the prediction signal is received for the first time, the module has collected 1500 current measurement values ​​in the previous 25 minutes. When the data acquisition module receives the prediction signal at time t, the last n = 1500 current values ​​(I t-1499 ,I t-1498 ,…,I t-1 ,I t ) is sent to establish the current regression equation, and the last k = 500 current values ​​(It-499 ,I t-498 ,…,I t-1 ,I t ) is sent as the current sampling data set for future current prediction, and the last l = 300 current values ​​(I t-299 ,I t-298 ,…,I t-1 ,I t ) is sent for troubleshooting.

[0074] Example 2

[0075] This embodiment is based on the technical solution of Embodiment 1, and mainly further explains the establishment of the current regression equation and the prediction based on the equation.

[0076] In this embodiment, the method for establishing the current regression equation in step S2 is:

[0077] Select the n data most recent to the current time t from the historically collected data set;

[0078] A linear equation is established based on the unknown coefficient w and the random error term b:

[0079] y=wx+b;

[0080] Get the dependent variable y:

[0081] y=I i ;

[0082] Among them, I i The current data at time i in the historical collection data set;

[0083] Get the independent variable x:

[0084] x=(I i-1 ,I i-2 ,…,I i-k+1 ,I i-k ) T ;

[0085] Among them, the elements in x are I i Current data of the previous k moments;

[0086] The above steps are performed sequentially with i=t+k-n+1,t+k-n+2,…,t-1,t to obtain multiple linear equations;

[0087] A linear equation group is formed according to the established multiple linear equations to obtain a current regression equation.

[0088] Furthermore, the method for obtaining the current regression equation according to the linear equation group is:

[0089] The linear equations are fitted to obtain the unknown coefficient w, and the fitting objective function is:

[0090]

[0091] Among them, I i is the actual current at time i, The actual current at time i is predicted based on the linear equation system.

[0092] As an example, this embodiment is based on a historical acquisition data set with a length of n=1500, establishes a regression equation with k=500 unknown autoregressive coefficients, and fits the autoregressive coefficients to apply to the current prediction based on autoregression in step S3. The specific steps are:

[0093] Step S201: For each moment i, collect the time series current (I i-1 ,I i-2 ,…,I i-k+1 ,I i-k ) and the current to be predicted at the current moment The inference relationship between them is used to establish a linear regression equation with common unknown parameters. The formula is:

[0094]

[0095] Among them, the weight parameters (w1,w1,…,w k-1 ,w k ) is the common autoregressive coefficient for each time point, and the bias parameter b~N(0,1) is a white noise sequence that follows the normal distribution, which is used as a random error term to explain the impact of the measurement data error; since the number of autoregressive coefficients and the length of the input variable of the regression equation are k=500, the value range of time i is t+k-n+1≤i≤t, that is, t-999≤i≤t;

[0096] Step S202: Execute the formula in step S201 for nk=1000 sets of historical data, that is, establish an inference relationship from time t+k-n+1 to time t, corresponding to the predicted current

[0097] Step S203: fitting the unknown parameters in the regression equation by the difference between the actual current and the predicted current, and the objective function to be minimized is:

[0098]

[0099] Furthermore, in the actual fitting, to fit the autoregressive coefficients (w1, w2, …, w k-1 ,wk ), and the random error term b can be ignored, the formula is:

[0100]

[0101] On the other hand, when performing the prediction in step S3, the method of rolling prediction of the future current trend is preferably to predict the current values ​​at the next m moments in sequence:

[0102] The independent variable x is constructed by the current sampling data set. t ,I t-1 ,…,I t-k+2 ,I t-k+1 ) T , the elements in the current sampling data set are the current data at k moments up to the current moment t;

[0103] The dependent variable is obtained by using the current regression equation according to the current sampling data set. in, is the predicted current value at time t+1;

[0104] Initialize j=0;

[0105] The operation of cutting off the updated value is performed until j=m-1:

[0106] Construct the independent variable x:

[0107]

[0108] The dependent variable is obtained by using the current regression equation according to the current sampling data set. in, is the predicted current value at time t+j+1;

[0109] Update value: j=j+1;

[0110] Get the current value at the next m moments

[0111] As a preferred solution, the method for obtaining the current sampling data set is:

[0112] The k data closest to the current sampling time are selected from the historically collected data sets to form the current sampling data set.

[0113] The following is an example of executing step S3 of this embodiment:

[0114] This embodiment predicts the current values ​​at m=300 moments in the future in sequence. The specific steps of prediction are:

[0115] Step 301: Substituting the autoregressive coefficient obtained in step S203 into the regression equation established in step S201 to obtain the current regression equation;

[0116] Step S302: using the current regression equation with known coefficients and the current sampling data set with k data, the current prediction for the next time t+1 is performed, and the formula is:

[0117]

[0118] Step S303: Calculate the predicted current from time t+2 to time t+m in the future During the calculation process, if the input current at the future time t+i (i>0) cannot be obtained from the historical data, the predicted current output by the regression equation at that time is used. Instead, the formula is as follows:

[0119]

[0120] Based on the above steps, the predicted current values ​​at the next m moments can be obtained. t+1 to b t+m It is determined and distributed according to a white noise sequence that follows a normal distribution.

[0121] Example 3

[0122] This embodiment is based on the technical solution of Embodiment 1, and mainly further explains the fault alarm.

[0123] As a preferred solution of this embodiment, the method for determining whether the fault alarm triggering condition is met is:

[0124] Set multiple fault alarm trigger conditions;

[0125] It is determined in sequence whether the current values ​​at the future m moments meet each of the fault alarm triggering conditions. If all of them are not met, it is determined that the fault alarm triggering conditions are not met; if any of them are met, it is determined that the fault alarm triggering conditions are met.

[0126] In this embodiment, the fault alarm triggering conditions preferably include a current upper limit threshold, a current lower limit threshold, a current fluctuation amplitude and a current over-limit frequency.

[0127] The following is a case study of this embodiment:

[0128] In the historical data set, the length of I is l = 300 as of the current time t 历史 =(I t-299 ,I t-298 ,…,I t-1 ,I t) and future prediction data with length m = 300 When the data triggers a fault alarm condition, it is determined whether the data triggers a fault alarm condition and sends a corresponding alarm signal. The specific steps are as follows:

[0129] Step S401: setting various thresholds for current abnormality alarm, including the allowable fluctuation percentage V0 under local (next one minute) zero tolerance, the allowable fluctuation percentage V1 and the over-limit frequency L1 under local low tolerance, and the allowable fluctuation percentage V2 and the over-limit frequency L2 under global (next five minutes) low tolerance;

[0130] Step S402: Count the actual measured current I in the past 5 minutes 历史 The maximum value in max I 历史 and minimum value minI 历史 , used to calculate the upper current threshold and the lower current threshold;

[0131] Step S403: For the predicted current data within the next minute:

[0132]

[0133] If the current exceeds the zero tolerance fluctuation range at any time, the system will send a first-level alarm signal (S=1); if the number of moments exceeding the local low tolerance fluctuation range reaches the local over-limit frequency L1, a second-level alarm signal (S=2) will be sent;

[0134] If the current data I is predicted in the next five minutes 未来 If the number of moments exceeding the global low tolerance fluctuation amplitude reaches the global limit-crossing frequency L2, a third-level alarm signal (S=3) is sent; otherwise, wait for the next prediction signal.

[0135] The formula is as follows:

[0136]

[0137] Example 4

[0138] This embodiment provides a direct current real-time prediction and fault warning device based on autoregression, see Figure 2 , applied to the autoregressive DC current real-time prediction and fault warning method described in any one of the above embodiments, comprising:

[0139] The data acquisition module is used to receive current data and then perform sampling, and obtain the historical acquisition data set through the sampling signal; its main function is to obtain the acquisition signal of the detection device and the prediction signal of the system, and it can also sort out the historical acquisition data to construct the historical acquisition data set and provide it to other modules;

[0140] An autoregressive fitting module is used to establish a current regression equation based on the historical data set, wherein the current regression equation is used to describe the correlation between the historical data value and the current data value of the current; the module can obtain the autoregressive coefficient through fitting to construct a current regression equation that can be used for prediction;

[0141] An autoregressive prediction module is used to obtain a current sampling data set, and according to the current sampling data set, the future current trend is predicted by the current regression equation to obtain the current value at a certain time in the future; the future prediction data obtained by the module is sent to the fault diagnosis module;

[0142] The fault diagnosis module is used to determine whether the fault alarm triggering conditions are met. If so, an alarm signal is sent; otherwise, no operation is performed.

[0143] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for real-time prediction and fault warning of DC current based on autoregression, characterized in that: The following steps are involved: Receive current data and then perform sampling to obtain a historical acquisition data set; Establishing a current regression equation based on the historical data set, wherein the current regression equation is used to describe the correlation between the historical data value and the current data value of the current; Acquire a current sampling data set, and according to the current sampling data set, roll-forward predict the future current trend by using the current regression equation to obtain current values ​​at several moments in the future; Determine whether the fault alarm triggering conditions are met, if yes, send an alarm signal, if not, do nothing; The method for establishing the current regression equation is: Select the most recent data from the historically collected data set up to the current moment; Create a linear equation based on unknown coefficients and random error terms: y=wx+b; Get the dependent variable y: y=I i ; Among them, I i The current data at time i in the historical collection data set; Get the independent variable x: x=(I i-1 ,I i-2 ,…,I i-k+1 ,I i-k ) T ; Among them, the elements in x are I i Current data of the previous k moments; The above steps are performed sequentially with i=t+k-n+1,t+k-n+2,…,t-1,t to obtain multiple linear equations; A linear equation group is formed according to the established multiple linear equations to obtain the current regression equation; The method for obtaining the current regression equation is: The linear equations are fitted to obtain the unknown coefficients, and the fitting objective function is: Among them, I i is the actual current at time i, A predicted current is the actual current at time i obtained based on the linear equation system; The method for rolling prediction of the future current trend is to predict the current values ​​at the next m moments in sequence: The independent variable x is constructed by the current sampling data set. t ,I t-1 ,…,I t-k+2 ,I t-k+1 ) T , the elements in the current sampling data set are the current data at k moments up to the current moment t; The dependent variable is obtained by using the current regression equation according to the current sampling data set. in, is the predicted current value at time t+1; Initialize j=0; The operation of cutting off the updated value is performed until j=m-1: Construct the independent variable x: The dependent variable is obtained by using the current regression equation according to the current sampling data set. in, is the predicted current value at time t+j+1; Update value: j=j+1; Get the current value at the next m moments The method for determining whether the fault alarm triggering condition is met is: Set multiple fault alarm trigger conditions; Determine in sequence whether the current values ​​at the future m moments satisfy each of the fault alarm triggering conditions; if all of them are not satisfied, determine that the fault alarm triggering conditions are not satisfied; if any of them are satisfied, determine that the fault alarm triggering conditions are satisfied; The fault alarm triggering conditions include the current upper limit threshold, the current lower limit threshold, the current fluctuation amplitude and the current over-limit frequency; Based on the historical data collection, the length of I as of the current time is 历史 =(I t-(l-1) ,I t-(l-2) ,…,I t-1 ,I t ) and future prediction data of length m When the data triggers a fault alarm condition, it is determined whether the data triggers a fault alarm condition and sends a corresponding alarm signal. The specific steps are as follows: Step S401: setting various thresholds for current abnormality alarm, including a fluctuation percentage of 0 in a local zero tolerance situation, a fluctuation percentage of 1 and an over-limit frequency of 1 in a local low tolerance situation, and a fluctuation percentage of 2 and an over-limit frequency of 2 in a global low tolerance situation, wherein the local situation includes the next 1 minute and the global situation includes the next 5 minutes; Step S402: Count the actual measured current I in the past 5 minutes 历史 The maximum value in max I 历史 and minimum value min I 历史 , used to calculate the upper current threshold and the lower current threshold; Step S403: For the predicted current data within the next minute: If the current exceeds the zero tolerance fluctuation range at any time, the system will send a first-level alarm signal; if the number of moments exceeding the local low tolerance fluctuation range reaches the local over-limit frequency 1, a second-level alarm signal will be sent; If the current data I is predicted in the next five minutes 未来 If the number of moments exceeding the global low tolerance fluctuation range reaches the global limit-crossing frequency 2, a level 3 alarm signal is sent; otherwise, wait for the next prediction signal.

2. The method for real-time prediction and fault warning of direct current based on autoregression according to claim 1 is characterized in that: The method for obtaining the historical collection data set is: Sampling the current data at unit time intervals; The sampled current data are arranged in ascending order according to time sequence to form the historical acquisition data set.

3. The method for real-time prediction and fault warning of direct current based on autoregression according to claim 2 is characterized in that: The unit time is 1 second.

4. The method for real-time prediction and fault warning of direct current based on autoregression according to claim 1 is characterized in that: The method for obtaining the current sampling data set is: The k data closest to the current sampling time are selected from the historically collected data sets to form the current sampling data set.

5. A device for real-time prediction and fault warning of direct current based on autoregression, applied to the method for real-time prediction and fault warning of direct current based on autoregression according to any one of claims 1 to 4, characterized in that: include: A data acquisition module is used to receive current data and then perform sampling to obtain a historical acquisition data set; An autoregressive fitting module, used to establish a current regression equation based on the historical acquisition data set, wherein the current regression equation is used to describe the correlation between the historical data value and the current data value of the current; An autoregressive prediction module is used to obtain a current sampling data set, and according to the current sampling data set, a rolling prediction of the future current trend is made through the current regression equation to obtain the current values ​​at the next m moments; The fault diagnosis module is used to determine whether the fault alarm triggering conditions are met. If so, an alarm signal is sent; otherwise, no operation is performed.

Images