CN108922139B - A kind of online test method that industry alarm is spread unchecked - Google Patents
A kind of online test method that industry alarm is spread unchecked Download PDFInfo
- Publication number
- CN108922139B CN108922139B CN201810808869.7A CN201810808869A CN108922139B CN 108922139 B CN108922139 B CN 108922139B CN 201810808869 A CN201810808869 A CN 201810808869A CN 108922139 B CN108922139 B CN 108922139B
- Authority
- CN
- China
- Prior art keywords
- alarm
- variable
- spread unchecked
- moment
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B29/00—Checking or monitoring of signalling or alarm systems; Prevention or correction of operating errors, e.g. preventing unauthorised operation
- G08B29/18—Prevention or correction of operating errors
- G08B29/185—Signal analysis techniques for reducing or preventing false alarms or for enhancing the reliability of the system
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B29/00—Checking or monitoring of signalling or alarm systems; Prevention or correction of operating errors, e.g. preventing unauthorised operation
- G08B29/02—Monitoring continuously signalling or alarm systems
- G08B29/06—Monitoring of the line circuits, e.g. signalling of line faults
Abstract
The invention belongs to field of signal processing, more particularly to a kind of online test method that industry alarm is spread unchecked, the following steps are included: (1) obtains the measurement data of alarm variable, each alarm variable corresponding duration and alarm interval are calculated, rejecting pretreatment is carried out to the shake alarm in measurement data by Delay timer;(2) pretreated measurement data is divided by several data sets according to the sliding size of the time window of setting and time window, calculates the indexed set of each data set;(3) radix for calculating each indexed set determines whether to have alarm and spread unchecked according to the radix of indexed set, and introduces the Delay timer that parameter is m in decision process and eliminate shake alarm;(4) operation is successively iterated to the data set of all divisions, testing result is analyzed by time-tendency graph.The present invention, which can solve, to be carried out being influenced excessive problem by alarm is created disturbances to during the detection spread unchecked of alarm using the statistical result of alarm frequency of occurrence.
Description
Technical field
The invention belongs to field of signal processing more particularly to a kind of online test methods that industry alarm is spread unchecked.
Background technique
The major function of industrial alarm system is monitored to industrial processes, it is ensured that production process is safe and efficient
Operation, can be generated, records and convey warning message to give production run personnel, production run personnel provide according to alarm system
Warning message, perceive abnormality rapidly and simultaneously take corresponding treatment measures, guarantee the safe and efficient fortune of production process to reach
Capable purpose.
The generation of warning message is to be compared the real time value of process variable with its alarm threshold value, when process variable
Triggering alarm when real time value is greater than (or being less than) alarm threshold value.The a large amount of reports for referring to and generating in a short period of time are spread unchecked in alarm
Alert, alarm quantity seriously has exceeded the processing capacity of operations staff, so be possible to will lead to important warning message be submerged and
It cannot handle in time, until the later period generates serious production accident.When alarm, which is spread unchecked, to be occurred, production run personnel face huge
Big operational management pressure, may result in that many emerging alarms are artificially ignored or even certain and severely subnormal is raw
The relevant alarm of occurrence condition is also ignored, and seriously affects the performance of the action and function of alarm system.
Detect it is emerging alarm spread unchecked and from historical data concentrate detect alarm spread unchecked there are extremely important.
It is timely detected out if emerging alarm is spread unchecked, it is general that production run personnel can be concerned about emerging alarm
Indiscriminate and relevant abnormal conditions;To history alarm spread unchecked data analysis be equally it is critically important, find it is similar alarm it is general
Indiscriminate rule can make operations staff make timely, effective reaction when similar alarm is spread unchecked and occurred.
Currently, about the detection spread unchecked of alarm usually in accordance with the statistical result of alarm frequency of occurrence in alarm system.So
And it is very big by the influence for creating disturbances to alarm to spread unchecked detection method by this simple alarm, can not obtain detect it is true
Alarm is spread unchecked.So propose it is a kind of by create disturbances to alarm influenced small alarm spread unchecked detection method have it is highly important reality anticipate
Justice.
Summary of the invention
According to the above-mentioned deficiencies of the prior art, the present invention provides a kind of online test method for spreading unchecked of industry alarm,
Can solve is influenced during carrying out the detection that alarm is spread unchecked using the statistical result of alarm frequency of occurrence at present by alarm is created disturbances to
Excessive problem.
Present invention solves the technical problem that the technical solution used are as follows: include the following steps:
(1) alarm variable x is obtained1,x2,···,xMMeasurement data
X (t-N+1:t)=[x1(t-N+1:t),...,xM(t-N+1:t)] when, each alarm variable of calculating is corresponding lasting
Between and alarm interval, and by Delay timer in measurement data shake alarm carry out rejecting pretreatment;
(2) according to the sliding size 1s of the time window 10min of setting and time window by pretreated measurement data
It is divided into n data set, the data in a time window before the corresponding sampling instant of each data set record calculate every
The indexed set of a data set;
Specific steps include:
A. the indexed set I that definition record alarm occurs1(t), there is alarm condition for being recorded in time window
Alarm variable;
Index of definition collection I2(t), alarm variable more than or equal to time window is lasted for recording alarm condition;
In view of resident alarm, index of definition collection I3(t), for recording the alarm change for being greater than resident time of fire alarming window
Amount;
Index of definition collection I (t), for recording the alarm variable for being recently entered alarm condition;
B. for k-th of alert data collection X (kT-599:kT), X (kT-599:kT) is indicated before k-th of sampling instant
The alert data collection recorded in one time window, indexed set are divided into two parts, first part I1(kT), second part is
Entered alarm condition in last window and the duration of alarming is greater than time window, the alarm less than resident time of fire alarming window becomes
Amount, as I3(kT) supplementary set and I ((k-1) T) and I2(kT) intersection, specially
(3) the corresponding radix of each indexed set is calculated, determines whether to have alarm according to the radix of indexed set and spread unchecked, specifically
Are as follows:
The corresponding radix of indexed set for calculating k-th of alert data collection, i.e., the quantity of element in indexed set, is denoted as | I (kT)
|;The threshold value that setting alarm is spread unchecked, if the radix of indexed set | I (kT) | it is greater than the threshold value, determines that there is alarm spreads unchecked, and remembers
xAF(kT)=1, if the radix of indexed set | I (kT) | it is less than the threshold value, determines that there is no alarms to spread unchecked, remembers xAF(kT)=
0;And the Delay timer that parameter is m is introduced in the decision process that alarm is spread unchecked and eliminates shake alarm;
(4) operation is successively iterated to the data set of all divisions, testing result is divided by time-tendency graph
Analysis.
Further, the specific implementation process of the step (1) are as follows:
Set xi(t-N+1:t)=[xi((t-N+1)h),...,xi(th)]T, i ∈ [1, M] is i-th of alarm variable xi's
Data set, t are the sampling interval that a value is positive integer, and N is the quantity and a positive integer of sample, in order to simplify table
Show, it is assumed that h value is 1s, then the physical interpretation of t is instant sampling instant, and unit is the second, if alarm variable xiInto alarm
State, then xiValue becomes 1, on the contrary then be 0;
Define the duration T of alarm1, T1Indicate alarm variable xiCurrent moment of alarming out from one is clear to next alarm
Except the time continuous between the moment, alarm occurs being defined as alarm variable xiEvent from " 0 " to " 1 ", alarm are removed and are defined as
Alarm variable xiEvent from " 1 " to " 0 ", duration T of alarming1=t2-t1+ 1, wherein xi(t1- 1)=0, xi(t2+ 1)=0,t1It alarms out current moment for one, t2For t1Adjacent next alarm disappears the moment;
Define the interval time T of alarm0, T0Indicate alarm variable xiIt alarms out from an alarm disappearance moment to next
Continuous time between current moment, alarm interval are specially T0=t2-t1+ 1, wherein xi(t1- 1)=1, xi(t2+ 1)=1,t1It disappears the moment for an alarm, t2For t1Adjacent next alarm
Current moment out.
Further, the shake alarm is less than the alarm variable of y for alarm duration and alarm interval, wherein y
Size is determined according to industrial actual conditions.
In another China that one of present inventor is done, invention (Patent No. is disclosed
CN201710305676.5, entitled " a kind of online elimination method and system of shake alarm and repetition of alarms ") inner, in detail
The method and principle that shake alarm is eliminated using Warning delay device are described, the Delay timer used herein is to shake alarm
It is identical with the elimination for using Warning delay device to alarm shake in patent CN201710305676.5 to eliminate Principle of Process.?
Have in the dependent claim of patent CN201710305676.5: " the following steps are included: (1) chooses time of fire alarming section ts extremely
The alarm signal of te, the alarm signal is converted to 0 for the normal condition of representative and with 1 be representative alarm condition
One alarm sample, wherein the corresponding alarm signal of ts and te is normal condition;(2) the parameter m of Warning delay device is initialized, it is right
Above-mentioned first alarm sample carries out prediction assignment, obtains the second alarm sample ", " the first alarm sample is carried out in step (2) pre-
Survey assignment specifically: and if only if there is continuous m sample to be alarm condition, alarm system just generates alarm signal 1;And if only if
Having continuous m sample is normal condition, and alarm system just generates normal signal 0 ".In the explanation of patent CN201710305676.5
Above-mentioned steps (2) are explained in detail again in book: " when t-m moment alarm signal is in normal condition 0, if when t-m+1
It carves, the t-m+2 moment ..., t moment alarm signal is all in alarm condition 1, then the second alarm sample is 1 in the value of t moment;
When t-m moment alarm signal is in normal condition 0, if the t-m+1 moment, t-m+2 moment ..., t-m+a (a < m) moment alarms
Signal is all in alarm condition 1, i.e., continuously the sample number in alarm condition is less than m, then the second alarm sample is at the t-m moment,
T-m+1 moment ..., the value at t-m+a moment are 0;When t-m moment alarm signal is in alarm condition 1, if the t-1 moment,
At the t-2 moment ..., t moment alarm signal is all in normal condition 0, then the second alarm sample is 0 in the value of t moment;Work as t-m
Moment alarm signal is in alarm condition 1, if the t-m+1 moment, at the t-m+2 moment ..., t-m+a (a < m) moment alarm signal is all
In normal condition 0, i.e., continuously the sample number in normal condition is less than m, then the second alarm sample is at the t-m moment, t-m+1
It carves ..., the value at t-m+a moment is 1."
Further, the length of the resident time of fire alarming window in step (2) is 30min.
Further, it is contemplated that shake alarm can not be completely eliminated in actual conditions, so general in alarm in step (3)
The sampling Delay timer that parameter is m is introduced in indiscriminate decision process, detection variable is spread unchecked in definition alarmIf
Detect variableThe x within m sampling timeAFIt (kT) is all 1, i.e. xAF((k-m+1) T:kT)=1, andThen rememberIt is 1, is considered as alarm and spreads unchecked appearance;If the x within m sampling timeAF(kT) all it is
0, i.e. xAF((k-m+1) T:kT)=0, andThen rememberIt is 0, is considered as alarm and spreads unchecked disappearance.
Above-mentioned m sampling Delay timer has solved the elimination Principle of Process of shake alarm in detail in the elaboration of step (1)
It releases, details are not described herein again.
Finally, be successively iterated operation to all alert data collection, the case where for on-line checking, the selection clothes of m
Postpone from maximum admissible detection.
The invention has the following advantages: when the present invention passes through the duration for calculating the alarm variable obtained and alarm
Between be spaced and shake alarm eliminated using Delay timer, and the sliding size of setting time window and time window is to above-mentioned
Data that treated are divided several indexed sets for not resided alarm, and are spread unchecked by the radix and alarm of indexed set
Decision threshold be compared, and introduce the Delay timer that parameter is m and eliminate shake alarm, determine that alarm is spread unchecked with accurate
Appearing and subsiding.The present invention overcomes the presence for alarming and residing alarm when cause shake, what error detection alarming was spread unchecked is asked
Topic, ensure that the safety and efficiency of plant produced.
Detailed description of the invention
Fig. 1 is that on-line checking design flow diagram is spread unchecked in the alarm of industry described in the embodiment of the present invention;
Fig. 2 be the embodiment of the present invention a certain concrete application scene in alarm the testing result spread unchecked;
Fig. 3 is the time trend of the alarm variable in a certain concrete application scene of the embodiment of the present invention in indexed set
Figure;
Fig. 4 is the time-tendency graph of several failure relative alarms in a certain concrete application scene of the embodiment of the present invention.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.
Embodiment one:
Such as Fig. 1, a kind of online test method that industry alarm is spread unchecked of the present invention includes the following steps:
S1 obtains alarm variable x1,x2,···,xMMeasurement data:
X (t-N+1:t)=[x1(t-N+1:t),...,xM(t-N+1:t)] when, each alarm variable of calculating is corresponding lasting
Between and alarm interval, and by Delay timer in measurement data shake alarm carry out rejecting pretreatment;
S2, according to the sliding size 1s of the time window 10min of setting and time window by pretreated measurement data
It is divided into n data set, the data in a time window before the corresponding sampling instant of each data set record calculate every
The indexed set of a data set;
Specific steps include:
A. the indexed set I that definition record alarm occurs1(t), there is alarm condition for being recorded in time window
Alarm variable;
Index of definition collection I2(t), alarm variable more than or equal to time window is lasted for recording alarm condition;
In view of resident alarm, index of definition collection I3(t), for recording the alarm change for being greater than resident time of fire alarming window
Amount;
Index of definition collection I (t), for recording the alarm variable for being recently entered alarm condition;
B. for k-th of alert data collection X (kT-599:kT), X (kT-599:kT) is indicated before k-th of sampling instant
The alert data collection recorded in one time window 10min, indexed set are divided into two parts, first part I1(kT), second
Part be last window in enter alarm condition and alarm the duration be greater than time window, less than resident time of fire alarming window
Alarm variable, as I3(kT) supplementary set and I ((k-1) T) and I2(kT) intersection, specially
S3 calculates the corresponding radix of each indexed set, determines whether to have alarm according to the radix of indexed set and spread unchecked, and
The Delay timer that parameter is m is introduced in the decision process spread unchecked of alarming eliminates shake alarm;
S4, is successively iterated operation to the data set of all divisions, is divided by time-tendency graph testing result
Analysis.
In step sl, it sets
xi(t-N+1:t)=[xi((t-N+1)h),...,xi(th)]T, i ∈ [1, M] is i-th of alarm variable xiData
Collection, h are a real value sampling interval, and t is the sampling interval that a value is positive integer, and N is the sample that a value is positive integer
This quantity indicates to simplify, it is assumed that h value is 1s, then the physical interpretation of t is instant sampling instant, and unit is the second, if
Alarm variable xiInto alarm condition, then xiValue becomes 1, on the contrary then be 0;
The duration of definition alarm is T1, T1Indicate alarm variable xiCurrent moment is alarmed out from one to next alarm
It removes the continuous time between the moment, alarm occurs being defined as alarm variable xiDefinition is removed in event from " 0 " to " 1 ", alarm
For the variable x that alarmsiEvent from " 1 " to " 0 ",
Alarm duration T1=t2-t1+ 1, wherein xi(t1- 1)=0, xi(t2+ 1)=0,t1It alarms out current moment for one, t2For t1Adjacent next alarm disappears the moment;
The alarm interval of definition alarm is T0, T0Indicate alarm variable xiFrom an alarm disappearance moment to next alarm
Continuous time between current moment, alarm interval are specially T out0=t2-t1+ 1, wherein xi(t1- 1)=1, xi(t2+ 1)=1,t1It disappears the moment for an alarm, t2For t1Adjacent next alarm
Current moment out.
Wherein, the shake alarm in measurement data is eliminated using Delay timer, the shake alarm is alarm
Duration and alarm interval are less than the alarm variable of y, and wherein the size of y can be set as 20 seconds according to industrial actual conditions.
Shake alarm is generally caused by external disturbances such as some noises, and there is no reference values, belong to interference alarm, can be tight
Ghost image rings the performance of industrial alarm system, is not only easy to cause useful warning message to be submerged, but also be easy to cause behaviour
Make personnel's long-standing neglect warning message, distrust is even switched off industrial alarm system.
In another China that one of present inventor is done, invention (Patent No. is disclosed
CN201710305676.5, entitled " a kind of online elimination method and system of shake alarm and repetition of alarms ") inner, in detail
The method and principle that shake alarm is eliminated using Warning delay device are described, the Delay timer used herein is to shake alarm
It is identical with the elimination for using Warning delay device to alarm shake in patent CN201710305676.5 to eliminate Principle of Process.?
Have in the dependent claim of patent CN201710305676.5: " the following steps are included: (1) chooses time of fire alarming section ts extremely
The alarm signal of te, the alarm signal is converted to 0 for the normal condition of representative and with 1 be representative alarm condition
One alarm sample, wherein the corresponding alarm signal of ts and te is normal condition;(2) the parameter m of Warning delay device is initialized, it is right
Above-mentioned first alarm sample carries out prediction assignment, obtains the second alarm sample ", " the first alarm sample is carried out in step (2) pre-
Survey assignment specifically: and if only if there is continuous m sample to be alarm condition, alarm system just generates alarm signal 1;And if only if
Having continuous m sample is normal condition, and alarm system just generates normal signal 0 ".In the explanation of patent CN201710305676.5
Above-mentioned steps (2) are explained in detail again in book: " when t-m moment alarm signal is in normal condition 0, if when t-m+1
It carves, the t-m+2 moment ..., t moment alarm signal is all in alarm condition 1, then the second alarm sample is 1 in the value of t moment;
When t-m moment alarm signal is in normal condition 0, if the t-m+1 moment, t-m+2 moment ..., t-m+a (a < m) moment alarms
Signal is all in alarm condition 1, i.e., continuously the sample number in alarm condition is less than m, then the second alarm sample is at the t-m moment,
T-m+1 moment ..., the value at t-m+a moment are 0;When t-m moment alarm signal is in alarm condition 1, if the t-1 moment,
At the t-2 moment ..., t moment alarm signal is all in normal condition 0, then the second alarm sample is 0 in the value of t moment;Work as t-m
Moment alarm signal is in alarm condition 1, if the t-m+1 moment, at the t-m+2 moment ..., t-m+a (a < m) moment alarm signal is all
In normal condition 0, i.e., continuously the sample number in normal condition is less than m, then the second alarm sample is at the t-m moment, t-m+1
It carves ..., the value at t-m+a moment is 1."
Determine that the time window of resident alarm can be selected as 30min, resident alarm refers to persistently keeping very long after occurring
The alarm of time, these alarms are not fallen clearly still after operator takes movement usually, will affect operator to system work shape
The judgement of state, so needing to reject resident alarm, therefore needing setting time window when judgement alarm spreads unchecked and comes across disappearance
Resident alarm is determined in alarm variable;About resident alarm decision time window can according to the needs of operator into
Row selection will lead to some resident alarms and be not removed and then interfere with alarm and spread unchecked but if time window selection is excessive
Judgement.
In addition, in actual use, even if some alarm variable occurred recently has not been alarmed at current time
State, operator still need to find the basic reason for triggering the alarm.
In step s3, radix corresponding to the indexed set of k-th of alert data collection, i.e., the number of element in indexed set are calculated
Amount | I (kT) |.It is 10 that the decision threshold that spreads unchecked of alarm can be set in the present embodiment, i.e., by element number in indexed set and 10 works
Compare, if being not less than 10, is judged to alarming and spreads unchecked appearance, remember xAF(kT)=1, if the radix of indexed set | I (kT) | it is small
In the threshold value, then determines that there is no alarms to spread unchecked, remember xAF(kT)=0.
M is introduced in view of shake alarm can not be completely eliminated in actual conditions, therefore in the stage that judgement is spread unchecked in alarm
Sample Delay timer.Detection variable is spread unchecked in definition alarmIf the x within m sampling timeAFIt (kT) is all 1, i.e.,
xAF((k-m+1) T:kT)=1, andThen rememberIt is 1, is considered as alarm and spreads unchecked appearance.If in m
X in a sampling timeAFIt (kT) is all 0, i.e. xAF((k-m+1) T:kT)=0, andThen rememberFor
0, it is considered as alarm and spreads unchecked disappearance.
Above-mentioned m sampling Delay timer has solved the elimination Principle of Process of shake alarm in detail in the elaboration of step S1
It releases, details are not described herein again.
In step s 4, operation is successively iterated to all alert data collection, the case where for on-line checking, m's
Maximum admissible detection delay is obeyed in selection, and for generating set, the delay of 1min is allowed, therefore in this feelings
Under condition, the default value of m is selected as 60s;Testing result is verified finally by the time-tendency graph of overview alarm variable.
The present invention is by the duration of the alarm variable of calculating acquisition and alarm time interval and using retardation timing
Device eliminates shake alarm, and the sliding size of setting time window and time window divide to treated data
To several indexed sets of not resident alarm, and it is compared by the radix of indexed set with the decision threshold that alarm is spread unchecked, and
It introduces the Delay timer that parameter is m and eliminates shake alarm, with the accurate appearing and subsiding for determining alarm and spreading unchecked.The present invention overcomes
When cause shakes alarm and resident there are problems that for alarm and error detection alarming is spread unchecked, the safety of plant produced ensure that
And efficiency.
In a certain concrete application scene, the applicable cases for the online test method that industry alarm of the present invention is spread unchecked
It is as follows:
It provides one and is collected in 10 AM on June 25th, 2015 to ten one points of one hour data set of length, wherein
Contain 2226 alarm variables.The sliding size T of time window is defined as 1 second.The parameter m of Delay timer is set to 60.
The first step first pre-processes the shake alarm that alert data is concentrated.
Second step divides pretreated data set according to the sliding size of time window and time window, and
Calculate the indexed set of each data set.
Third step, the radix of computation index collection, and alarmed shake into one according to preset Delay timer parameter m=60
Step is eliminated.It finally calculates and alarm has occurred in time interval t ∈ [2530,3023] spreads unchecked.As shown in Figure 2.Fig. 2 (a) is should
The radix of the indexed set of data set, red dotted line represent the threshold value 10 that alarm is spread unchecked.Fig. 2 (b) blue solid lines are to prolong without reference
The testing result of stagnant timer, red dotted line are the testing result for referring to Delay timer.
4th step, observing time tendency chart simultaneously analyze result.Fig. 3 is time-tendency graph, it can be seen that in set I (t=
2530) in, with the presence of 13 alarm variables, that is to say, that be implicitly present in alarm and spread unchecked.The alarm detected spread unchecked mainly by
Caused by blender terminal Bearing Temperature caused by the exception of blender C.The terminal bearing temperature configuration of blender
Alarm variable is #517 and #514.Fig. 4 (a) is the measurement data of the two temperature sensors.When temperature is higher than 51 degrees Celsius,
Two sensors will trigger alarm.When one of temperature is higher than 55 degrees Celsius, another is higher than 60 degrees Celsius, another
The variable #2090 that alarms will enter alarm condition, as shown in Fig. 4 (b) blue solid lines.This abnormal conditions was by operator at that time
It ignores, after 22 minutes, it is tight to trigger safety protecting mechanism because two sensor detection temperature are both greater than 60 degree by blender C
Anxious delay machine, corresponding alarm are #2128, and as shown in the red dotted line of Fig. 4 (b), and the delay machine of blender C leads to plant produced power
245MW is fallen below from 275, as shown in Fig. 4 (c), this abnormal conditions results in the heavy losses of generating set.
Claims (6)
1. a kind of online test method that industry alarm is spread unchecked, which comprises the steps of:
(1) alarm variable x is obtained1,x2,…,xMMeasurement data
X (t-N+1:t)=[x1(t-N+1:t),...,xM(t-N+1:t)], calculate each alarm variable corresponding duration with
Alarm interval, and rejecting pretreatment is carried out to the shake alarm in measurement data by Delay timer;
Specific steps include:
Set xi(t-N+1:t)=[xi((t-N+1)h),...,xi(th)]T, i ∈ [1, M] is i-th of alarm variable xiData
Collection, h are a real value sampling interval, and t is the sampling interval that a value is positive integer, and N is the sample that a value is positive integer
This quantity indicates to simplify, it is assumed that h value is 1s, then the physical interpretation of t is instant sampling instant, and unit is the second, if
Alarm variable xiInto alarm condition, then xiValue becomes 1, on the contrary then be 0;
The duration of definition alarm is T1, T1Indicate alarm variable xiCurrent moment of alarming out from one removes to next alarm
Continuous time between moment, alarm occur being defined as alarm variable xiEvent from " 0 " to " 1 ", alarm remove and are defined as reporting
Alert variable xiEvent from " 1 " to " 0 ", duration T of alarming1=t2-t1+ 1, wherein xi(t1- 1)=0, xi(t2+ 1)=0,t2> t1, t1It alarms out current moment for one, t2For t1Adjacent next alarm disappears the moment;
The alarm interval of definition alarm is T0, T0Indicate alarm variable xiOccur from an alarm disappearance moment to next alarm
Continuous time between moment, alarm interval are specially T0=t2-t1+ 1, wherein xi(t1- 1)=1, xi(t2+ 1)=1,t2> t1, t1It disappears the moment for an alarm, t2For t1It is adjacent next to alarm out
Current moment;
(2) pretreated measurement data is divided by n number according to the sliding size of the time window of setting and time window
According to collection, data in a time window before the corresponding sampling instant of each data set record calculate each data set
Indexed set;
Specific steps include:
A. the indexed set I that definition record alarm occurs1(t), become for being recorded in the alarm for occurring alarm condition in time window
Amount;
Index of definition collection I2(t), alarm variable more than or equal to time window is lasted for recording alarm condition;
In view of resident alarm, index of definition collection I3(t), for recording the alarm variable for being greater than resident time of fire alarming window;
Index of definition collection I (t), for recording the alarm variable for being recently entered alarm condition;
B. the alert data for being recorded in a time window before k-th of alert data set representations, k-th of sampling instant
Collection, indexed set are divided into two parts, first part I1(kT), second part is to enter alarm condition and alarm in last window
Duration is greater than time window, less than the alarm variable of resident time of fire alarming window, as I3(kT) supplementary set and I ((k-1)
) and I T2(kT) intersection, specially
(3) the corresponding radix of each indexed set is calculated, determines whether to have alarm according to the radix of indexed set and spread unchecked, specifically:
The corresponding radix of indexed set for calculating k-th of alert data collection, i.e., the quantity of element in indexed set, is denoted as | I (kT) |;If
It alarms calmly the threshold value spread unchecked, if the radix of indexed set | I (kT) | it is greater than the threshold value, determines that there is alarm spreads unchecked, and remembers xAF
(kT)=1, if the radix of indexed set | I (kT) | it is less than the threshold value, determines that there is no alarms to spread unchecked, remembers xAF(kT)=0;And
The Delay timer that parameter is m is introduced in the decision process that alarm is spread unchecked eliminates shake alarm;(4) to the data of all divisions
Collection is successively iterated operation, is analyzed by time-tendency graph testing result.
2. the online test method that industry alarm according to claim 1 is spread unchecked, which is characterized in that the shake, which is alarmed, is
Alarm duration and alarm interval are less than the alarm variable of y.
3. the online test method that industry alarm according to claim 1 is spread unchecked, which is characterized in that in the step (2),
The length of time window is 10min, and the sliding size of time window is T=1~600s.
4. the online test method that industry alarm according to claim 3 is spread unchecked, which is characterized in that the sliding of time window
Size is T=1s.
5. the online test method that industry alarm according to claim 1 is spread unchecked, which is characterized in that the institute in step (2)
The length for stating resident time of fire alarming window is 30min.
6. the online test method that industry alarm according to claim 1 is spread unchecked, which is characterized in that in step (3), reporting
The sampling Delay timer that parameter is m is introduced in the alert decision process spread unchecked, detection variable is spread unchecked in definition alarmSuch as
Fruit detects variableThe x within m sampling timeAFIt (kT) is all 1, i.e. xAF((k-m+1) T:kT)=1, andThen rememberIt is 1, is considered as alarm and spreads unchecked appearance;If the x within m sampling timeAF(kT) all it is
0, i.e. xAF((k-m+1) T:kT)=0, andThen rememberIt is 0, is considered as alarm and spreads unchecked disappearance.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810808869.7A CN108922139B (en) | 2018-07-23 | 2018-07-23 | A kind of online test method that industry alarm is spread unchecked |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810808869.7A CN108922139B (en) | 2018-07-23 | 2018-07-23 | A kind of online test method that industry alarm is spread unchecked |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108922139A CN108922139A (en) | 2018-11-30 |
CN108922139B true CN108922139B (en) | 2019-09-17 |
Family
ID=64415788
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810808869.7A Active CN108922139B (en) | 2018-07-23 | 2018-07-23 | A kind of online test method that industry alarm is spread unchecked |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108922139B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111028487B (en) * | 2019-12-27 | 2021-07-13 | 杭州青泓科技有限公司 | Water treatment monitoring method and system |
CN112346421B (en) * | 2020-11-10 | 2024-03-26 | 北京化工大学 | Alarm flooding sequence data pattern extraction method based on improved PreFixSpan algorithm |
CN117118811A (en) * | 2023-10-25 | 2023-11-24 | 南京邮电大学 | Alarm analysis method for industrial alarm flooding |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102436720B (en) * | 2011-09-28 | 2013-07-03 | 清华大学 | Repeated alarm processing method based on data filtering |
CN105006119B (en) * | 2015-06-30 | 2017-12-29 | 中国寰球工程公司 | A kind of alarm system optimization method based on Bayesian network |
CN106971521B (en) * | 2017-05-03 | 2018-09-21 | 山东科技大学 | A kind of online elimination method and system of shake alarm and repetition of alarms |
-
2018
- 2018-07-23 CN CN201810808869.7A patent/CN108922139B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN108922139A (en) | 2018-11-30 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108922139B (en) | A kind of online test method that industry alarm is spread unchecked | |
EP0843244B1 (en) | Diagnostic trend analysis for aircraft engines | |
US4295128A (en) | Apparatus for measuring the degradation of a sensor time constant | |
CN107832896B (en) | Power plant equipment slow-changing fault early warning method and device | |
CN110286656B (en) | False alarm filtering method and device for tolerance of error data | |
CN109469896B (en) | Industrial boiler fault diagnosis method and system based on time series analysis | |
CN110648480B (en) | Single variable alarm system and method based on change rate | |
CN110469496B (en) | Intelligent early warning method and system for water pump | |
KR100997009B1 (en) | The method for dynamic detection and on-time warning of industrial process | |
US11620539B2 (en) | Method and device for monitoring a process of generating metric data for predicting anomalies | |
CN113109051B (en) | Fault early warning method and system based on vibration data range sequence | |
CN109991956B (en) | Method for predicting steady-state fault of liquid rocket engine | |
CN114664063A (en) | Equipment grading alarm method based on general index trend analysis | |
CN109087490B (en) | A kind of interference alarm recognition methods based on alarm duration features | |
CN117176199B (en) | HPLC communication unit fault diagnosis method and device | |
CN112326246A (en) | Bearing safety state online monitoring method based on periodic data and nuclear density estimation | |
CN117314020A (en) | Wetland carbon sink data monitoring system of plankton | |
CN111207306B (en) | Heating power pipe network leakage monitoring method based on digital pressure signal processing | |
EP3414631B1 (en) | Detection of temperature sensor failure in turbine systems | |
CN115034094B (en) | Prediction method and system for operation state of metal processing machine tool | |
KR101982235B1 (en) | Method and device for similar equipment clustering using feature extraction method based on survival function | |
CN104731056B (en) | The method of the quick operation stability for judging chemical production device | |
Galicia et al. | Statistics Pattern Analysis based fault detection and diagnosis | |
US11131985B2 (en) | Noise generation cause estimation device | |
CN113778044A (en) | Monitoring method and device for blower system of thermal power plant |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |