A big data-based fault monitoring and early warning system and method

Abstract

The application discloses a kind of based on big data's fault monitoring early warning system and method, it is related to fault monitoring early warning technical field, including state monitoring data acquisition module, effective state parameter analysis module, linear regression equation construction module, fault monitoring set division module and monitoring parameter implementation module;The state monitoring data acquisition module is used to extract the state monitoring data of the equipment to be monitored in effective monitoring period;The effective state parameter analysis module is used to screen different state monitoring data and output effective state parameter;The linear regression equation construction module is used to construct the time-state parameter deviation linear regression equation of the equipment to be monitored;The fault monitoring set division module is used to divide the fault monitoring set of the equipment to be monitored based on regression equation;The monitoring parameter implementation module is used to implement the monitoring priority and monitoring period of different fault monitoring.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault monitoring and early warning, in particular to a fault monitoring and early warning system and method based on big data. BACKGROUND

[0002] At present, the market of intelligent monitoring products mainly concentrates on the transformation link of smart grid, and the state intelligent monitoring system construction of transformer substation has been taken as an important content of smart transformer substation in the smart grid planning report of State Grid Corporation of China. The smart transformer substation realizes the online monitoring of the main equipment and important parameters such as the temperature rise of the key point of the switch cabinet, the oil chromatography of the transformer, the partial discharge of the combined electric appliance, and the full current of the lightning arrester through state monitoring units, and provides basic data support for the management of power grid equipment. After the real-time state information is analyzed and processed by an expert system, preliminary decisions can be made to realize the self-diagnosis function of the intelligent equipment in the station. With the transformation project of China's medium and low voltage power grid entering the scale stage, the intelligent monitoring device of power equipment is a foundation for the transformation, and the market potential is great.

[0003] However, the existing intelligent fault monitoring system often judges whether different monitoring devices exist in a fault state for early warning according to the set reasonable threshold interval of each monitoring device, lacks the analysis of the correlation between different monitoring devices, and the problems of the life stage of the monitoring device and the influence of the life stage on the monitoring priority and cycle adjustment. SUMMARY

[0004] The purpose of the present application is to provide a fault monitoring and early warning system and method based on big data to solve the problems raised in the background.

[0005] In order to solve the above technical problems, the present application provides the following technical scheme: a fault monitoring and early warning method based on big data, comprising the following analysis steps:

[0006] Step S1: obtaining the state monitoring data of the to-be-monitored device in the effective monitoring period, the effective monitoring period being the continuous monitoring period recorded by the sensor when the state monitoring parameter of the to-be-monitored device belongs to the corresponding parameter threshold interval; screening the effective state parameters output by different state monitoring data in the effective monitoring period;

[0007] Step S2: based on the effective state parameters, constructing a time-state parameter deviation linear regression equation of the to-be-monitored device;

[0008] Step S3: based on the time-state parameter deviation linear regression equation, predicting the state parameter deviation prediction value in the adjacent effective monitoring period, and obtaining the actual state parameter deviation value in the adjacent effective monitoring period, to divide the fault monitoring set of the to-be-monitored device;

[0009] Step S4: based on different fault monitoring sets, implementing monitoring priority and monitoring period of different fault monitoring.

[0010] Further, step S1 includes the following analysis steps:

[0011] Step S11: obtaining state monitoring parameter a of the ith to-be-monitored device at any same monitoring time as the monitoring starting point i , and marking the ith to-be-monitored device corresponding state monitoring parameter a i belongs to parameter threshold interval a i0 , and the continuous monitoring time length h i , comparing the continuous monitoring time length h of the m to-be-monitored devices i , output the continuous monitoring time length corresponding to the minimum value min[h i ] as the effective monitoring period, wherein m represents the total number of to-be-monitored devices; the minimum value is selected in consideration of the need to meet the parameters of all to-be-monitored devices within the monitoring period in the monitoring system; the effective monitoring period without limiting the monitoring starting point can adapt to various state moments of the to-be-monitored devices, and control different devices in the same monitoring interval to make the parameter analysis more accurate;

[0012] Step S12: obtaining the use time length H of the ith to-be-monitored device before the effective monitoring period i , the use time length is the time length corresponding to the monitoring starting point of the effective monitoring period from the first recording of the state monitoring parameter of the to-be-monitored device; extracting the minimum value min[H i ] and the maximum value max[H i ] of the use time length of the to-be-monitored device; if max[H i ]-min[H i ] is greater than or equal to the difference threshold value, then mark the to-be-monitored device corresponding to the use time length less than or equal to the average use time length H0 as an abnormal monitoring device, H0=(1 / m)∑H i ;

[0013] The abnormal monitoring device corresponding to the use time length is analyzed in order to exclude the devices with large differences in use time length from other to-be-monitored devices, because the state parameter fluctuation amplitudes corresponding to the initial use and the use after a long time are different as the device is used, and the same dimension analysis cannot be effectively performed;

[0014] Step S13: marking the device corresponding to the device maintenance record existing in the average use time length of the to-be-monitored device before the effective monitoring period as an abnormal monitoring device; the device maintenance record refers to the record of repairing or replacing the to-be-monitored device; removing the abnormal monitoring device from the to-be-monitored device and outputting the effective monitoring device and the effective state parameter within the effective monitoring period. Analyzing the maintenance record because the maintenance may change the state parameter of the device.

[0015] Further, the constructing of the time-state parameter deviation linear regression equation of the to-be-monitored equipment comprises the following analysis steps:

[0016] Step S21: obtaining the difference A between the effective state parameter corresponding to the jth effective monitoring equipment and the parameter threshold value j generating a sequence P according to the time development order in the effective monitoring period j , j≤n, n represents the number of effective monitoring equipment, n

[0017] generating a sequence T according to the time sequence based on the time length value of the monitoring unit in the effective monitoring period j ; based on the time length value is to take the time length from each monitoring unit of the effective monitoring period to the first record of the state monitoring parameter of the to-be-monitored equipment as the basis;

[0018] Step S22: constructing the time-state parameter deviation linear regression equation as y=ax+b, wherein the dependent variable y represents the state parameter deviation value, the independent variable x represents the time, a represents the slope of the regression equation, and b represents the intercept of the regression equation;

[0019] Step S23: recording the loss function , wherein k≥1 and k is an ordered positive integer set, k represents the number of columns in the time value xk and the state parameter deviation value yk, and w represents the total number of columns; according to the formula:

[0020]

[0021] solving to obtain , ;

[0022] Step S24: substituting the sequence P j and the sequence T j , solving to obtain the slope a and the intercept b of the regression equation; and substituting into the regression equation y=ax+b, to obtain the regression equation determined by the slope and the intercept.

[0023] Further, step S3 comprises the following analysis steps:

[0024] Step S31: obtaining the time length of the pth target period after the effective monitoring period of the effective monitoring equipment, inputting the time length into the time-state parameter deviation linear regression equation, obtaining the state parameter deviation prediction value corresponding to each monitoring unit in the pth target period and constructing a prediction set Q pThe target period refers to a period length less than the effective monitoring period, and the target period contains at least two monitoring units;

[0025] Step S32: Calculate the abnormal value C in the pth prediction set Qp p , C p =U p / V p , wherein U p represents the number of state parameter deviation prediction values in the pth prediction set that are less than the actual state parameter deviation value and the actual state parameter deviation value does not belong to the parameter threshold interval; V p represents the number of state parameter deviation prediction values in the pth prediction set;

[0026] Step S33: Traverse and analyze the number z1 of target periods corresponding to the same abnormal value of any two effective monitoring devices in z target periods, calculate the correlation index g, g=z0*(z1 / z), p≤z; mark the effective monitoring device corresponding to g>g0 as the first correlation device, g0 represents the correlation index threshold, and z0 represents the abnormal average value;

[0027] Step S34: Extract all effective monitoring devices marked as the first correlation device, and store the effective monitoring devices constructing the closed-loop correlation as the first fault monitoring set; the closed-loop correlation refers to the first first correlation device E1, the second first correlation device E2,..., the Nth first correlation device E N Any two effective monitoring devices are first correlation devices; when there is an effective monitoring device whose number of target periods corresponding to the same abnormal value in z target periods is 0, the output is an independent monitoring device, which is stored in the fourth fault monitoring set;

[0028] Step S35: When there is only one marking record for the two effective monitoring devices marked as the first correlation device, output the effective monitoring device as a linear correlation device, and store it in the third fault monitoring set; when the two effective monitoring devices marked as the first correlation device are marked several times and cannot form a closed-loop correlation, output as an open correlation device, and store it in the second fault monitoring set; and store the data of the effective monitoring devices marked as the first correlation device according to the analysis order corresponding to the first fault monitoring set to the fourth fault monitoring set;

[0029] Step S36: When the effective monitoring device is stored in any fault monitoring set and meets the storage condition of another fault monitoring set; calculate the influence index R of the two effective monitoring devices corresponding to the fault monitoring set,

[0030] R=s1*g+s2*(a0 / a r )+s3*(b0 / b r )

[0031] wherein g represents the correlation index of two effective monitoring devices meeting the fault monitoring set storage condition, a r represents the absolute value of the slope difference of the linear regression equation corresponding to the two effective monitoring devices meeting the fault monitoring set storage condition, a0 represents the average value of the slope of the linear regression equation corresponding to the two effective monitoring devices, b r represents the absolute value of the intercept difference of the linear regression equation corresponding to the two effective monitoring devices meeting the fault monitoring set storage condition, b0 represents the average value of the intercept of the linear regression equation corresponding to the two effective monitoring devices; s1, s2 and s3 respectively represent the corresponding reference coefficients and the values are greater than 0 and less than 1;

[0032] The stored fault monitoring set is marked as a parent set, and the stored fault monitoring set is marked as a child set.

[0033] The influence index R1 corresponding to the parent set is compared with the influence index R2 corresponding to the child set.

[0034] If R1≥R2, the same effective monitoring device is stored in the parent set, and the different effective monitoring device is stored in the child set.

[0035] If R1<R2, all the effective monitoring devices involved in the above analysis are stored in the parent set.

[0036] The monitoring devices are divided into different fault monitoring sets in order to effectively monitor a large number of devices, and different monitoring sets are set to distinguish the devices and quickly limit and investigate the influence range of the fault device when any device fails, so as to realize the effect of early warning and avoid greater loss.

[0037] Further, step S4 includes the following analysis steps:

[0038] When the device to be monitored is maintained or the monitoring system warns of abnormal parameters of the non-monitoring device, the processing priority of the first fault monitoring set is greater than that of the second fault monitoring set, greater than that of the third fault monitoring set, and greater than that of the fourth fault monitoring set.

[0039] And the maintenance period corresponding to the first fault monitoring set is smaller than that of the second fault monitoring set, smaller than that of the third fault monitoring set, and smaller than that of the fourth fault monitoring set.

[0040] Because the devices stored in the first fault monitoring set have strong correlation and large influence range, when corresponding to abnormal fault investigation, the first fault monitoring set is preferred.

[0041] The device corresponding to the device stored in the first fault monitoring set has similar device life cycle and possibly large use history, and the limited time corresponding to the maintenance period is the shortest maintenance period of the first period, because in the device with a larger life cycle, the closer to the end of life, the greater the possibility of abnormality; therefore, different maintenance periods are set to improve the safety and stability of the monitoring device in the first fault monitoring set to the overall system.

[0042] The fault monitoring and early warning system comprises a state monitoring data acquisition module, an effective state parameter analysis module, a linear regression equation construction module, a fault monitoring set division module and a monitoring parameter implementation module.

[0043] The state monitoring data acquisition module is used to extract the state monitoring data of the device to be monitored within the effective monitoring period.

[0044] The effective state parameter analysis module is used to screen different state monitoring data and output effective state parameters.

[0045] The linear regression equation construction module is used to construct a time-state parameter deviation linear regression equation of the device to be monitored.

[0046] The fault monitoring set division module is used to divide the fault monitoring set of the device to be monitored based on the regression equation.

[0047] Further, the effective state parameter analysis module comprises an effective monitoring period output unit, an abnormal monitoring device marking unit and an effective state parameter output unit.

[0048] The effective monitoring period output unit is used to compare the continuous monitoring time corresponding to the device to be monitored, and output the continuous monitoring time corresponding to the minimum value as the effective monitoring period.

[0049] The abnormal monitoring device marking unit is used to extract the minimum value and the maximum value of the use time corresponding to the device to be monitored; if the difference between the maximum value and the minimum value is greater than or equal to the difference threshold value, mark the device to be monitored corresponding to the use time less than or equal to the average use time as an abnormal monitoring device; and mark the device corresponding to the device maintenance record existing in the average use time of the device to be monitored before the effective monitoring period as an abnormal monitoring device.

[0050] The effective state parameter output unit is used to eliminate the abnormal monitoring device in the device to be monitored and output the effective monitoring device, and the effective state parameter within the effective monitoring period.

[0051] Further, the linear regression equation construction module comprises a sequence generation unit, a linear regression equation setting unit and a parameter calculation unit.

[0052] The number sequence generating unit is configured to generate a number sequence of the difference Aj between the jth effective monitoring device corresponding effective state parameter and the parameter threshold value in the time sequence of the effective monitoring period, and generate a number sequence of the monitoring units in the effective monitoring period in the time sequence based on the time length value;

[0053] The linear regression equation setting unit is configured to set a time-state parameter deviation linear regression equation;

[0054] The parameter calculating unit is configured to substitute the number sequence and solve the loss function to calculate the slope and intercept of the linear regression equation, and output the linear regression equation with the determined slope and intercept.

[0055] Further, the fault monitoring set division module includes a prediction set constructing unit, an abnormal value calculating unit, a correlation index calculating unit, a fault monitoring set generating unit, and a storage abnormality analyzing unit.

[0056] The prediction set constructing unit is configured to input the time length into the time-state parameter deviation linear regression equation to obtain the state parameter deviation prediction value corresponding to each monitoring unit in the target period and form a prediction set of the target period.

[0057] The abnormal value calculating unit is configured to calculate the abnormal value in the prediction set.

[0058] The correlation index calculating unit is configured to analyze the number of target periods in which any two effective monitoring devices have the same abnormal value in the target period, and calculate the correlation index.

[0059] The fault monitoring set generating unit is configured to generate fault monitoring sets meeting different storage conditions.

[0060] The storage abnormality analyzing unit is configured to analyze the repetition abnormality when the effective monitoring devices store the fault monitoring sets.

[0061] Further, the storage abnormality analyzing unit includes an influence index calculating unit and a storage path executing unit.

[0062] The influence index calculating unit is configured to calculate the influence index of the two effective monitoring devices corresponding to the fault monitoring sets when the effective monitoring devices meet the storage condition of another fault monitoring set after being stored in any fault monitoring set.

[0063] The storage path executing unit is configured to retain the same effective monitoring devices in the parent set and the different effective monitoring devices in the child set when the influence index of the parent set is greater than the influence index of the child set, and store all the effective monitoring devices involved in the above analysis in the parent set when the influence index of the parent set is less than the influence index of the child set. Mark the first stored fault monitoring set as the parent set and the second stored fault monitoring set as the child set.

[0064] Compared with the prior art, the present application has the beneficial effects that: the present application predicts the actual state parameter deviation value by constructing the time-state parameter deviation linear regression equation of different to-be-monitored devices, and combines the life stage of the to-be-monitored device with the similarity degree of other devices to perform set division; the monitoring device is divided into different fault monitoring sets in order to effectively monitor a large number of devices reasonably, and setting different monitoring sets can distinguish the devices and quickly limit and troubleshoot the influence range that can be affected by the fault device when any device fails, thereby realizing the prerequisite early warning effect and avoiding greater losses; the analysis of the fault early warning system is more intelligent, not limited to the fluctuation of numerical values, and the correlation between different devices is associated, and the priority and monitoring period are dynamically adjusted. BRIEF DESCRIPTION OF DRAWINGS

[0065] The accompanying drawings are included to provide a further understanding of the present application, and constitute a part of the specification, illustrate the present application together with the embodiments thereof, and explain the present application, and do not constitute a limitation of the present application. In the drawings:

[0066] Figure 1 is a structural schematic diagram of a fault monitoring and early warning system based on big data. DETAILED DESCRIPTION

[0067] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0068] Please refer to Figure 1 The present application provides a technical solution: a fault monitoring and early warning method based on big data, comprising the following analysis steps:

[0069] Step S1: obtaining state monitoring data of a to-be-monitored device in an effective monitoring period, the effective monitoring period being a continuous monitoring period recorded by a sensor when a state monitoring parameter of the to-be-monitored device is within a corresponding parameter threshold interval; screening effective state parameters output in the effective monitoring period; for example, in an intelligent power grid monitoring system, an intelligent power device is a device that uses wireless transmission network technology, takes an intelligent microprocessor as a core, and uses precise sensing instruments to continuously and real-timely monitor key parameters of various power devices such as switch cabinets, transformers, circuit breakers, GIS, etc.

[0070] Step S2: based on the effective state parameters, constructing a time-state parameter deviation linear regression equation of the to-be-monitored device;

[0071] Step S3: Based on the time-state parameter deviation linear regression equation, predict the state parameter deviation value within adjacent effective monitoring periods, and obtain the actual state parameter deviation value within adjacent effective monitoring periods to divide the fault monitoring set of the equipment to be monitored.

[0072] Step S4: Based on different fault monitoring sets, implement different fault monitoring priorities and monitoring cycles.

[0073] Step S1 includes the following analysis steps:

[0074] Step S11: Obtain the status monitoring parameter a of the i-th device to be monitored, with any identical monitoring time as the monitoring start point. i And mark the status monitoring parameter a corresponding to the i-th device to be monitored. i Belongs to the parameter threshold range a i0 Continuous monitoring duration h i Compare the continuous monitoring duration h corresponding to m devices to be monitored. i Output the minimum value min[h] i The corresponding continuous monitoring duration is the effective monitoring period, where m represents the total number of devices to be monitored. The minimum value is chosen because the monitoring system needs to ensure that the parameters of all devices to be monitored are within the threshold range within the monitoring period. The effective monitoring period without limiting the monitoring start point can adapt to various state moments of the devices to be monitored, and makes parameter analysis more accurate when controlling different devices within the same monitoring interval.

[0075] Step S12: Obtain the usage duration H of the i-th device to be monitored before the effective monitoring period. i The usage duration refers to the time from the first recording of status monitoring parameters by the monitored device to the start of the effective monitoring cycle; extract the minimum value min[H] of the usage duration corresponding to the monitored device. i ] and maximum value max[H i If max[H] i ]-min[H i If the difference is greater than or equal to the threshold, then the device to be monitored that is less than or equal to the average usage time H0 is marked as an abnormal monitoring device, and H0 = (1 / m)∑H i ;

[0076] The purpose of analyzing the abnormal monitoring devices corresponding to the usage time is to filter out devices whose usage time differs significantly from that of other devices to be monitored. This is because the fluctuation range of the status parameters corresponding to the initial use and the use after a longer period of time are different, making it impossible to perform effective analysis on the same dimension.

[0077] Step S13: Mark the device corresponding to the device maintenance record existing in the average use time length of the to-be-monitored device before the effective monitoring period as an abnormal monitoring device; the device maintenance record refers to the record of repairing or replacing the to-be-monitored device; remove the abnormal monitoring device from the to-be-monitored device and output the effective monitoring device and the effective state parameter in the effective monitoring period. Analyzing the maintenance record is because the maintenance may change the state parameter of the device.

[0078] The construction of the time-state parameter deviation linear regression equation of the to-be-monitored device includes the following analysis steps:

[0079] Step S21: Calculate the difference A between the effective state parameter corresponding to the jth effective monitoring device and the parameter threshold value j According to the time development order in the effective monitoring period, generate a sequence P j , j≤n, n represents the number of effective monitoring devices, n<m; the time development order in the effective monitoring period refers to the development order in days as the monitoring unit, and the effective state parameter represents the average state parameter in the monitoring unit;

[0080] Generate a sequence T j based on the time length value of the monitoring unit in the effective monitoring period in chronological order; based on the time length value means based on the time length from each monitoring unit of the effective monitoring period to the first record of the state monitoring parameter of the to-be-monitored device;

[0081] Step S22: Construct the time-state parameter deviation linear regression equation as y=ax+b, wherein the dependent variable y represents the state parameter deviation value, the independent variable x represents the time, a represents the slope of the regression equation, and b represents the intercept of the regression equation;

[0082] Step S23: Record the loss function , wherein k≥1 and k is an ordered positive integer set, k represents the number of columns in the time value xk and the state parameter deviation value yk, and w represents the total number of columns; according to the formula:

[0083]

[0084] Solve to get , ;

[0085] Step S24: Substitute the sequence P j and the sequence T j , and solve to get the slope a and the intercept b of the regression equation; and substitute them into the regression equation y=ax+b to get the regression equation determined by the slope and the intercept.

[0086] Step S3 includes the following analysis steps:

[0087] Step S31: Obtain the time length of the pth target period after the effective monitoring period of the effective monitoring device, input the time length into the time-state parameter deviation linear regression equation, obtain the state parameter deviation prediction value corresponding to each monitoring unit in the pth target period and form the prediction set Qp of the pth target period p ; the target period refers to a period length less than the effective monitoring period, and the target period contains at least two monitoring units;

[0088] Step S32: Calculate the abnormal value C in the pth prediction set Qp p , C p =U p / V p , wherein U p represents the number of state parameter deviation prediction values less than the actual state parameter deviation value in the pth prediction set Qp, and the actual state parameter deviation value does not belong to the parameter threshold interval; V p represents the number of state parameter deviation prediction values in the pth prediction set Qp;

[0089] Step S33: Traverse and analyze the number z1 of target periods corresponding to the same abnormal value of any two effective monitoring devices in z target periods, calculate the correlation index g, g=z0*(z1 / z), p≤z; mark the effective monitoring device corresponding to g>g0 as the first correlation device, g0 represents the correlation index threshold, and z0 represents the average abnormal value;

[0090] Step S34: Extract all the effective monitoring devices marked as the first correlation device, and store the effective monitoring devices constructing the closed-loop correlation as the first fault monitoring set; the closed-loop correlation refers to the first first correlation device E1, the second first correlation device E2,..., the Nth first correlation device En N Any two effective monitoring devices are first correlation devices; when there is an effective monitoring device with the number of target periods corresponding to the same abnormal value in z target periods being 0, the output is an independent monitoring device, which is stored in the fourth fault monitoring set;

[0091] If there are effective monitoring device 1, effective monitoring device 2 and effective monitoring device 3, the effective monitoring device 1 and the effective monitoring device 2 are first correlation devices, the effective monitoring device 2 and the effective monitoring device 3 are first correlation devices, and the effective monitoring device 1 and the effective monitoring device 3 are first correlation devices, it is indicated that the effective monitoring device 1, the effective monitoring device 2 and the effective monitoring device 3 can constitute a closed-loop correlation;

[0092] Step S35: When there is only one marked record for the two effective monitoring devices marked as the first associated device, output the effective monitoring device as a linearly associated device and store it in the third fault monitoring set; when the two effective monitoring devices marked as the first associated device are marked several times and cannot form a closed-loop association, output it as an open associated device and store it in the second fault monitoring set; and store the effective monitoring device marked as the first associated device in the data according to the analysis order corresponding to the first fault monitoring set to the fourth fault monitoring set;

[0093] Step S36: When the effective monitoring device stored in any fault monitoring set again meets the storage condition of another fault monitoring set; then calculate the influence index R of the two effective monitoring devices corresponding to the fault monitoring set,

[0094] R=s1*g+s2*(a0 / a r )+s3*(b0 / b r )

[0095] Wherein g represents the association index of the two effective monitoring devices meeting the storage condition of the fault monitoring set, a r represents the absolute value of the slope difference of the linear regression equation corresponding to the two effective monitoring devices meeting the storage condition of the fault monitoring set, a0 represents the average value of the slope of the linear regression equation corresponding to the two effective monitoring devices, b r represents the absolute value of the intercept difference of the linear regression equation corresponding to the two effective monitoring devices meeting the storage condition of the fault monitoring set, b0 represents the average value of the intercept of the linear regression equation corresponding to the two effective monitoring devices; s1, s2 and s3 respectively represent the corresponding reference coefficients and the values are all greater than 0 and less than 1;

[0096] Mark the first stored fault monitoring set as the parent set and the second stored fault monitoring set as the child set;

[0097] Compare the influence index R1 corresponding to the parent set with the influence index R2 corresponding to the child set;

[0098] If R1≥R2, keep the same effective monitoring device stored in the parent set and the different effective monitoring device stored in the child set;

[0099] If R1<R2, store all the effective monitoring devices involved in the above analysis in the parent set.

[0100] As shown in the embodiment: there are effective monitoring devices a, b, c, d, e;

[0101] And there is no effective monitoring device with the same abnormal value corresponding to the target period number of 0 in the z target periods; that is, any two effective monitoring devices can calculate the association index which is not 0; 10 groups of association indexes are obtained;

[0102] If ac, ad, bc, cd and ce are all marked as the first associated device based on the correlation index analysis;

[0103] As can be seen from the above, devices a, c and d constitute a closed-loop association, and are stored in the first fault monitoring set;

[0104] And there is no linearly associated device, and the third fault monitoring set is not stored;

[0105] When bc and cd are first associated devices, the storage condition of the second fault monitoring set is met, but at this time the effective monitoring device has been stored in the first fault monitoring set, so the storage path of the effective monitoring device b needs to be determined;

[0106] The correlation index corresponding to bc and the correlation index corresponding to ac are obtained respectively, as well as the difference between the linear regression equation corresponding to devices b and c in slope and intercept and the difference between the linear regression equation corresponding to devices a and c in slope and intercept, to calculate the corresponding influence index;

[0107] If the influence index corresponding to the first fault monitoring set is greater than or equal to the influence index corresponding to the second fault monitoring set, it means that the association of devices a and c is more influential; if the influence index corresponding to the first fault monitoring set is less than the influence index corresponding to the second fault monitoring set, it means that the association of devices b and c is more influential, and then device b associated with device c is stored in the first fault monitoring set with higher priority.

[0108] The monitoring devices are divided into different fault monitoring sets in order to effectively monitor a large number of devices reasonably, and different monitoring sets are set to distinguish the devices and quickly limit and troubleshoot the influence range that the fault device can reach when any device fails, thereby achieving a prior warning effect and avoiding greater losses.

[0109] Step S4 includes the following analysis steps:

[0110] When the device to be monitored is maintained or the monitoring system warns of abnormal parameters of non-monitoring devices, the processing priority of the first fault monitoring set is greater than that of the second fault monitoring set, which is greater than that of the third fault monitoring set, which is greater than that of the fourth fault monitoring set;

[0111] And the maintenance period corresponding to the first fault monitoring set is less than that of the second fault monitoring set, which is less than that of the third fault monitoring set, which is less than that of the fourth fault monitoring set.

[0112] Because the devices stored in the first fault monitoring set have strong association and large influence range, when corresponding to abnormal fault troubleshooting, the first fault monitoring set is prioritized;

[0113] The device corresponding to the device stored in the first fault monitoring set has similar device life cycle and possibly large use history, and the limited time corresponding to the maintenance period is the shortest maintenance period of the first period, because in the device with large life cycle, the closer to the end of life, the greater the possibility of abnormality; therefore, different maintenance periods are set to improve the safety and stability of the monitoring device in the first fault monitoring set to the overall system.

[0114] The fault monitoring and early warning system comprises a state monitoring data acquisition module, an effective state parameter analysis module, a linear regression equation construction module, a fault monitoring set division module and a monitoring parameter implementation module.

[0115] The state monitoring data acquisition module is used to extract the state monitoring data of the device to be monitored within the effective monitoring period.

[0116] The effective state parameter analysis module is used to screen different state monitoring data and output effective state parameters.

[0117] The linear regression equation construction module is used to construct a time-state parameter deviation linear regression equation of the device to be monitored.

[0118] The fault monitoring set division module is used to divide the fault monitoring set of the device to be monitored based on the regression equation.

[0119] The effective state parameter analysis module comprises an effective monitoring period output unit, an abnormal monitoring device marking unit and an effective state parameter output unit.

[0120] The effective monitoring period output unit is used to compare the continuous monitoring time corresponding to the device to be monitored, and output the continuous monitoring time corresponding to the minimum value as the effective monitoring period.

[0121] The abnormal monitoring device marking unit is used to extract the minimum value and the maximum value of the use time corresponding to the device to be monitored; if the difference between the maximum value and the minimum value is greater than or equal to the difference threshold value, the device to be monitored corresponding to the use time less than or equal to the average use time is marked as an abnormal monitoring device; and the device corresponding to the device maintenance record existing in the average use time before the effective monitoring period is marked as an abnormal monitoring device.

[0122] The effective state parameter output unit is used to eliminate the abnormal monitoring device in the device to be monitored and output the effective monitoring device, and the effective state parameter within the effective monitoring period.

[0123] The linear regression equation construction module comprises a sequence generation unit, a linear regression equation setting unit and a parameter calculation unit.

[0124] The number sequence generating unit is configured to generate a number sequence by ordering the differences Aj between the jth effective monitoring device corresponding effective state parameter and the parameter threshold value according to the time sequence in the effective monitoring period, and generate a number sequence by ordering the monitoring units in the effective monitoring period according to the time sequence based on the time length value.

[0125] The linear regression equation setting unit is configured to set a time-state parameter deviation linear regression equation.

[0126] The parameter calculating unit is configured to substitute the number sequence and solve the loss function to calculate the slope and intercept of the linear regression equation, and output the linear regression equation with the determined slope and intercept.

[0127] The fault monitoring set division module includes a prediction set constructing unit, an abnormal value calculating unit, a correlation index calculating unit, a fault monitoring set generating unit, and a storage abnormality analyzing unit.

[0128] The prediction set constructing unit is configured to input the time length into the time-state parameter deviation linear regression equation to obtain the state parameter deviation prediction value corresponding to each monitoring unit in the target period and form a prediction set of the target period.

[0129] The abnormal value calculating unit is configured to calculate the abnormal value in the prediction set.

[0130] The correlation index calculating unit is configured to traverse and analyze the number of target periods in which any two effective monitoring devices have the same abnormal value in the target period, and calculate the correlation index.

[0131] The fault monitoring set generating unit is configured to generate fault monitoring sets that meet different storage conditions.

[0132] The storage abnormality analyzing unit is configured to analyze the repetition abnormality when the effective monitoring devices store the fault monitoring sets.

[0133] The storage abnormality analyzing unit includes an influence index calculating unit and a storage path executing unit.

[0134] The influence index calculating unit is configured to calculate the influence index of the two effective monitoring devices corresponding to the fault monitoring sets when the effective monitoring devices meet the storage condition of another fault monitoring set after being stored in any fault monitoring set.

[0135] The storage path executing unit is configured to retain the same effective monitoring devices in the parent set and the different effective monitoring devices in the child set when the influence index corresponding to the parent set is greater than the influence index corresponding to the child set, and store all the effective monitoring devices involved in the above analysis in the parent set when the influence index corresponding to the parent set is less than the influence index corresponding to the child set. Mark the first stored fault monitoring set as the parent set and the second stored fault monitoring set as the child set.

[0136] It should be noted that the relationship terms, such as first and second, and the like, are used only to differentiate one entity or action from another, and do not necessarily require or imply any actual such relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0137] Finally, it should be noted that the above-described embodiments are merely possible implementations of the present application, but not to be taken to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, the technical solutions recorded in the foregoing embodiments can still be modified by those skilled in the art, or some technical features thereof can be replaced by equivalent replacements. Any modifications, equivalent replacements, improvements, and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for fault monitoring and early warning based on big data, characterized in that, The step S1 comprises the following analysis steps: Step S1: obtaining state monitoring data of the to-be-monitored equipment in an effective monitoring period, the effective monitoring period being a continuous monitoring period in which the sensor records state monitoring parameters of the to-be-monitored equipment within a corresponding parameter threshold interval; screening effective state parameters output by different state monitoring data within the effective monitoring period; Step S2: constructing a time-state parameter deviation linear regression equation of the to-be-monitored equipment based on the effective state parameters; Step S3: predicting a state parameter deviation prediction value in a neighboring effective monitoring period based on the time-state parameter deviation linear regression equation, and obtaining an actual state parameter deviation value in the neighboring effective monitoring period, to divide a fault monitoring set of the to-be-monitored equipment; The step S3 comprises the following analysis steps: Step S31: Obtain the time length of the pth target period after the effective monitoring period of the effective monitoring device, input the time length into the time-state parameter deviation linear regression equation, obtain the state parameter deviation prediction value corresponding to each monitoring unit in the pth target period, and form a prediction set Q of the pth target period p ; the target period refers to a period length less than the effective monitoring period, and the target period contains at least two monitoring units; Step S32: Calculate the transaction value C in the pth prediction set Qp p , C p = U p / V p , wherein U p represents the number of state parameter deviation prediction values in the pth prediction set that are less than the actual state parameter deviation value and the actual state parameter deviation value does not belong to the parameter threshold interval; V p represents the number of state parameter deviation prediction values in the pth prediction set. Step S33: traversing and analyzing a target period number z1 corresponding to the same abnormal value of any two effective monitoring equipment in z target periods, calculating a correlation index g, g=z0*(z1 / z), p≤z; marking the effective monitoring equipment corresponding to g>g0 as a first correlation equipment, g0 representing a correlation index threshold, and z0 representing an abnormal value average; Step S34: extract all the effective monitoring devices marked the first associated device, and store the effective monitoring devices constructing the closed-loop association as the first fault monitoring set; the closed-loop association refers to the 1st first associated device E1, the 2nd first associated device E2,..., the Nth first associated device EN N Any two effective monitoring devices are first associated devices; when there is an effective monitoring device whose corresponding target period number is 0 in the same abnormal value in z target periods, the output is an independent monitoring device, and it is stored in the fourth fault monitoring set; Step S35: when the two effective monitoring equipment marked as the first correlation equipment only has one marking record, outputting the effective monitoring equipment as a linear correlation equipment, and storing it as a third fault monitoring set; when the two effective monitoring equipment marked as the first correlation equipment is marked several times and cannot form a closed-loop correlation, outputting it as an open correlation equipment, and storing it in a second fault monitoring set; and storing data of the effective monitoring equipment marked as the first correlation equipment according to an analysis sequence corresponding to the first fault monitoring set to the fourth fault monitoring set; Step S36: when the effective monitoring equipment is stored in any fault monitoring set and meets another fault monitoring set storage condition; calculating an influence index R of the two effective monitoring equipment corresponding to the fault monitoring set, R = s1*g + s2*(a0 / a r ) + s3*(b0 / b r ) wherein g represents the correlation index of two effective monitoring devices meeting the fault monitoring set storage condition, a r represents the absolute value of the slope difference of the linear regression equation corresponding to the two effective monitoring devices meeting the fault monitoring set storage condition, a0 represents the average value of the slope of the linear regression equation corresponding to the two effective monitoring devices, b r represents the absolute value of the intercept difference of the linear regression equation corresponding to the two effective monitoring devices meeting the fault monitoring set storage condition, b0 represents the average value of the intercept of the linear regression equation corresponding to the two effective monitoring devices; s1, s2 and s3 respectively represent the corresponding reference coefficients and the values are all greater than 0 and less than 1. marking a fault monitoring set stored first as a parent set, and marking a fault monitoring set stored later as a child set; comparing an influence index R1 corresponding to the parent set with an influence index R2 corresponding to the child set; if R1≥R2, retaining the same effective monitoring equipment in the parent set, and storing different effective monitoring equipment in the child set; if R1<R2, storing all effective monitoring equipment involved in the above analysis in the parent set; Step S4: implementing monitoring priorities and monitoring periods of different fault monitoring based on different fault monitoring sets. 2.The big data-based fault monitoring and early warning method according to claim 1, characterized in that: The step S1 comprises the following analysis steps: Step S11: acquiring the state monitoring parameter a of the ith to-be-monitored device with any same monitoring moment as the monitoring starting point i , and marking the state monitoring parameter a corresponding to the ith to-be-monitored device i when the parameter threshold interval a i0 is continuously monitored for h i , comparing the continuous monitoring time lengths h i corresponding to the m to-be-monitored devices, and outputting the continuous monitoring time length corresponding to the minimum value min[h i ] as the effective monitoring period, wherein m represents the total number of to-be-monitored devices Step S12: obtaining the use duration H of the i-th to-be-monitored device before the effective monitoring period i , wherein the use duration refers to the duration corresponding to the first recording of the state monitoring parameter of the to-be-monitored device to the monitoring starting point of the effective monitoring period; extracting minimum value min[H i ] and maximum value max[H i ] of the usage time length corresponding to the to-be-monitored device; if max[H i ]-min[H i ] is greater than or equal to a difference threshold value, marking the to-be-monitored device corresponding to the usage time length less than or equal to the average usage time length H0 as an abnormal monitoring device, H0=(1 / m)∑H i ; Step S13: marking an equipment existing in an equipment maintenance record within an average use duration before the effective monitoring period of the to-be-monitored equipment as an abnormal monitoring equipment; The equipment maintenance record is a record of repairing or replacing the to-be-monitored equipment; the abnormal monitoring equipment in the to-be-monitored equipment is excluded, and the effective monitoring equipment and the effective state parameters in the effective monitoring period are output. 3.The method of claim 2, wherein: The step of constructing the time-state parameter deviation linear regression equation of the to-be-monitored equipment comprises the following analysis steps: Step S21: obtaining the difference A between the jth effective monitoring device corresponding effective state parameter and the parameter threshold value j generating the sequence P according to the time development order in the effective monitoring period j , j≤n, n represents the number of effective monitoring devices, n<m; the time development order in the effective monitoring period refers to the development order in days as the monitoring unit, and the effective state parameter represents the average state parameter in the monitoring unit; The monitoring units in the effective monitoring period are generated in time sequence based on time length values to form a sequence T j The time length values are the time length from each monitoring unit of the effective monitoring period to the first recording of the state monitoring parameter by the device to be monitored. Step S22: constructing a time-state parameter deviation linear regression equation as y=ax+b, wherein the dependent variable y represents the state parameter deviation value, the independent variable x represents the time, a represents the slope of the regression equation, and b represents the intercept of the regression equation; Step S23: record loss function where k≥1 and k is a set of ordered positive integers, k represents the column number of the time value xk and the state parameter deviation value yk, and w represents the total number of the sequence; according to the formula: Solving to obtain , ; Step S24: Substituting the number series P j and the number series T j , to solve the slope a and the intercept b of the regression equation; and substituting the regression equation y=ax+b, to obtain the regression equation determined by the slope and the intercept.

4. The method of claim 3, wherein the method further comprises: The step S4 comprises the following analysis steps: When the to-be-monitored equipment is maintained or the monitoring system warns of an abnormal parameter of the non-monitored equipment, the processing priority of the first fault monitoring set is greater than that of the second fault monitoring set, greater than that of the third fault monitoring set, and greater than that of the fourth fault monitoring set; And the maintenance cycle corresponding to the first fault monitoring set is less than that of the second fault monitoring set, less than that of the third fault monitoring set, and less than that of the fourth fault monitoring set.

5. A fault monitoring and early warning system applying the fault monitoring and early warning method based on big data according to any one of claims 1-4, characterized in that, The state monitoring data acquisition module, the effective state parameter analysis module, the linear regression equation construction module, the fault monitoring set division module, and the monitoring parameter implementation module are included. The state monitoring data acquisition module is configured to extract the state monitoring data of the to-be-monitored equipment within the effective monitoring period. The effective state parameter analysis module is configured to screen different state monitoring data and output effective state parameters. The linear regression equation construction module is configured to construct a time-state parameter deviation linear regression equation of the to-be-monitored equipment. The fault monitoring set division module is configured to divide the fault monitoring set of the to-be-monitored equipment based on the regression equation. The monitoring parameter implementation module is configured to implement the monitoring priority and the monitoring cycle of different fault monitoring.

6. The fault monitoring and alerting system of claim 5, wherein: The effective state parameter analysis module comprises an effective monitoring period output unit, an abnormal monitoring equipment marking unit, and an effective state parameter output unit. The effective monitoring period output unit is configured to compare the continuous monitoring time lengths corresponding to the to-be-monitored equipment, and output the continuous monitoring time length corresponding to the minimum value as the effective monitoring period. The abnormal monitoring equipment marking unit is configured to extract the minimum value and the maximum value of the use time length corresponding to the to-be-monitored equipment. If the difference between the maximum value and the minimum value is greater than or equal to a difference threshold value, the to-be-monitored equipment corresponding to the use time length less than or equal to the average use time length is marked as an abnormal monitoring equipment. If the to-be-monitored equipment has a device maintenance record within the average use time length before the effective monitoring period, the device corresponding to the record is marked as an abnormal monitoring equipment. The effective state parameter output unit is configured to exclude the abnormal monitoring equipment from the to-be-monitored equipment and output the effective monitoring equipment and the effective state parameters within the effective monitoring period.

7. The fault monitoring and alerting system of claim 6, wherein: The linear regression equation construction module comprises a sequence generation unit, a linear regression equation setting unit, and a parameter calculation unit. The sequence generation unit is configured to sort the difference Aj between the effective state parameter corresponding to the jth effective monitoring equipment and the parameter threshold value in the time development order within the effective monitoring period to generate a sequence, and generate a sequence in the time order based on the time length value of the monitoring unit within the effective monitoring period. The linear regression equation setting unit is configured to set the time-state parameter deviation linear regression equation. The parameter calculation unit is configured to substitute the sequence and solve the loss function to calculate the slope and the intercept of the linear regression equation, and output the linear regression equation with the determined slope and intercept.

8. The fault monitoring and alerting system of claim 7, wherein: The fault monitoring set division module comprises a prediction set construction unit, an abnormal value calculation unit, a correlation index calculation unit, a fault monitoring set generation unit and a storage abnormality analysis unit; The prediction set construction unit is configured to input a time length into a time-state parameter deviation linear regression equation to obtain a state parameter deviation prediction value corresponding to each monitoring unit in a target period and form a prediction set of the target period; The abnormal value calculation unit is configured to calculate abnormal values in the prediction set; The correlation index calculation unit is configured to traverse and analyze a target period number corresponding to the same abnormal value of any two effective monitoring devices in the target period to calculate a correlation index; The fault monitoring set generation unit is configured to generate fault monitoring sets meeting different storage conditions; The storage abnormality analysis unit is configured to analyze a repeated abnormality when the effective monitoring devices store the fault monitoring sets.

9. The fault monitoring and alerting system of claim 8, wherein: The storage abnormality analysis unit comprises an influence index calculation unit and a storage path execution unit; The influence index calculation unit is configured to calculate an influence index of two effective monitoring devices corresponding to a fault monitoring set when the effective monitoring devices meet storage conditions of another fault monitoring set after being stored in any fault monitoring set; The storage path execution unit is configured to retain the same effective monitoring devices in a parent set and the different effective monitoring devices in a child set when the influence index corresponding to the parent set is greater than the influence index corresponding to the child set; or store all the effective monitoring devices involved in the above analysis in the parent set when the influence index corresponding to the parent set is less than the influence index corresponding to the child set; mark the fault monitoring set stored first as the parent set and the fault monitoring set stored later as the child set.

Images