CN112819190A - Method and device for predicting equipment performance, storage medium and terminal - Google Patents

Method and device for predicting equipment performance, storage medium and terminal Download PDF

Info

Publication number
CN112819190A
CN112819190A CN201911122080.7A CN201911122080A CN112819190A CN 112819190 A CN112819190 A CN 112819190A CN 201911122080 A CN201911122080 A CN 201911122080A CN 112819190 A CN112819190 A CN 112819190A
Authority
CN
China
Prior art keywords
value
subsequence
historical
performance index
detected
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.)
Granted
Application number
CN201911122080.7A
Other languages
Chinese (zh)
Other versions
CN112819190B (en
Inventor
高磊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Gener Software Technology Co ltd
Original Assignee
Gener Software Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Gener Software Technology Co ltd filed Critical Gener Software Technology Co ltd
Priority to CN201911122080.7A priority Critical patent/CN112819190B/en
Publication of CN112819190A publication Critical patent/CN112819190A/en
Application granted granted Critical
Publication of CN112819190B publication Critical patent/CN112819190B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Educational Administration (AREA)
  • Debugging And Monitoring (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

A method and a device for predicting device performance, a storage medium and a terminal are provided, the method comprises the following steps: determining a performance index for evaluating the performance of the equipment; acquiring a plurality of historical performance index sequences from historical data, calculating a historical p value to obtain a historical p value set, and determining an early warning p value according to the historical p value set; acquiring a performance index sequence to be detected from data to be detected, and calculating a p value to be detected; comparing the p value to be detected with the early warning p value, and predicting the performance of the equipment according to the comparison result; the p value to be detected and the historical p value are calculated according to the following steps: intercepting a performance index sequence with a preset length, and dividing the performance index sequence into a first subsequence and a second subsequence; constructing a statistic capable of representing a mean level difference between the first subsequence and the second subsequence, the statistic being subject to a t-distribution; the measured values of the statistics are determined and a p-value is calculated that reflects the probability of equipment failure. The scheme provided by the invention can predict the performance of the equipment so as to improve the early warning accuracy.

Description

Method and device for predicting equipment performance, storage medium and terminal
Technical Field
The invention relates to the field of electromechanical technology, in particular to a method and a device for predicting equipment performance, a storage medium and a terminal.
Background
The early warning means that before a disaster or other danger needing to be raised, an emergency signal is sent to related departments according to the rules summarized in the past or the possibility precursor obtained by observation, and the dangerous situation is reported, so that the damage is avoided under the condition of blindness or insufficient preparation, and the loss caused by the damage is relieved to the maximum extent.
In the current electromechanical system, a commonly used early warning method is an overrun early warning method based on experience. This method relies on a great deal of human experience and is susceptible to false alarms from a single outlier. Moreover, this kind of early warning mode is unfavorable for the discovery of early failure, often reports when early warning, and equipment or equipment part have already worsened to a certain extent. Besides the devices in the electromechanical system, the device can be any other device capable of running for multiple times or for a long time, and the potential fault needs to be found as early as possible, so that early warning indication is necessary.
However, due to the defects of the current early warning method, how to early warn needs to be further researched.
Disclosure of Invention
The invention solves the technical problem of how to predict the performance of equipment so as to improve the early warning accuracy.
To solve the foregoing technical problem, an embodiment of the present invention provides a method for predicting device performance, including: determining a performance index for evaluating the performance of the equipment; acquiring a plurality of historical performance index sequences from historical data, respectively calculating a historical p value of each historical performance index sequence to obtain a historical p value set, and determining an early warning p value according to the historical p value set; acquiring a to-be-detected performance index sequence from to-be-detected data, and calculating a to-be-detected p value of the to-be-detected performance index sequence; comparing the p value to be detected with the early warning p value, and predicting the performance of the equipment according to the comparison result; wherein, the p value to be detected and the historical p value are calculated according to the following steps: intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence; constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution; and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value.
Optionally, the determining the measured values of the statistics according to the first subsequence and the second subsequence may include: deleting abnormal values in the first subsequence and the second subsequence respectively to obtain an updated first subsequence and an updated second subsequence; and determining the measured value of the statistic according to the updated first subsequence and the updated second subsequence.
Optionally, the deleting the outliers in the first subsequence and the second subsequence respectively comprises: and deleting abnormal values in the first subsequence and the second subsequence respectively by using a boxplot abnormal value processing algorithm.
Optionally, the removing the outliers in the first subsequence and the second subsequence by using the outlier processing algorithm of the box plot may include: calculating respective lower quartiles and upper quartiles for the first subsequence or the second subsequence; taking the difference between the upper quartile and the lower quartile as a quartile range; deletion Interval [ Q ]1-IQR*a，Q3+IQR*a]An element other than; wherein Q is1Representing the lower quartile, Q, of the first or second subsequence3Representing the upper quartile of the first subsequence or the second subsequence, a representing a preset intensity factor, and IQR representing the quartile distance.
Optionally, the preset lengths of the historical performance index sequences are the same, the preset length is equal to the preset length of the performance index sequence to be detected, and the lengths of the divided first subsequences are the same.
Optionally, the determining an early warning p-value according to the historical p-value set includes: arranging all elements in the historical p value set from small to large to obtain a historical p value sequence; and selecting the early warning p value from the historical p value sequence according to a preset quantile.
Optionally, the length of the first subsequence is n1The length of the second subsequence is n2Constructing statistics representative of mean level differences between the first and second subsequences comprising: when n is1＝n2When n, the statistic obeying the t distribution is:
Figure BDA0002275722910000021
Figure BDA0002275722910000031
θi＝ξi-ηi(ii) a When n is1<n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000032
Figure BDA0002275722910000033
when n is1>n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000034
Figure BDA0002275722910000035
wherein ξiRepresenting the ith element, η, of said first subsequenceiRepresents the ith element of the second subsequence, i being a positive integer.
Optionally, n is determined according to the requirements for sensitivity and false alarm rate1And n2The ratio of (A) to (B): the higher the required sensitivity, the larger its ratio; the lower the required false alarm rate, the smaller its ratio.
Optionally, the p value capable of reflecting the failure probability of the equipment is calculated according to the measured value of the statistic; and calculating the p value by adopting any one of the following formulas according to the t distribution to which the statistic belongs: p ═ P (| T non-conducting phosphor)>|T0|)；p＝P(T>T0)；p＝P(T<T0) (ii) a Wherein T represents the statistic, T0Representing measured values of said statistics, P-tableAnd (6) indicating the probability.
To solve the foregoing technical problem, an embodiment of the present invention further provides an apparatus for predicting device performance, including: the determining module is used for determining a performance index for evaluating the performance of the equipment; the first calculation module is used for acquiring a plurality of historical performance index sequences from historical data, respectively calculating a historical p value of each historical performance index sequence to obtain a historical p value set, and determining an early warning p value according to the historical p value set; the second calculation module is used for acquiring a to-be-detected performance index sequence from to-be-detected data and calculating a to-be-detected p value of the to-be-detected performance index sequence; the comparison and prediction module is used for comparing the p value to be detected with the early warning p value and predicting the performance of the equipment according to the comparison result; wherein, the p value to be detected and the historical p value are calculated according to the following steps: intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence; constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution; and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value.
To solve the above technical problem, an embodiment of the present invention further provides a storage medium having stored thereon computer instructions, where the computer instructions execute the steps of the above method when executed.
In order to solve the foregoing technical problem, an embodiment of the present invention further provides a terminal, including a memory and a processor, where the memory stores computer instructions executable on the processor, and the processor executes the computer instructions to perform the steps of the foregoing method.
Compared with the prior art, the technical scheme of the embodiment of the invention has the following beneficial effects:
the embodiment of the invention provides a method for predicting equipment performance, which comprises the following steps: determining a performance index for evaluating the performance of the equipment; acquiring a plurality of historical performance index sequences from historical data, respectively calculating a historical p value of each historical performance index sequence to obtain a historical p value set, and determining an early warning p value according to the historical p value set; acquiring a to-be-detected performance index sequence from to-be-detected data, and calculating a to-be-detected p value of the to-be-detected performance index sequence; comparing the p value to be detected with the early warning p value, and predicting the performance of the equipment according to the comparison result; wherein, the p value to be detected and the historical p value are calculated according to the following steps: intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence; constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution; and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value. By the technical scheme provided by the embodiment of the invention, after the performance index for evaluating the performance of the equipment is determined, the early warning p value of the performance index can be determined by calculating a plurality of different historical performance index sequences, and the p values to be detected are determined and then compared, so that the fault probability of the equipment can be deduced and predicted. The early warning p value is obtained based on the statistical model, so that the influence of a single abnormal point can be well overcome, the false alarm probability is favorably reduced by utilizing the change of the parameter statistical rule, and the early fault is more accurately found, thereby providing a powerful basis for guaranteeing the operation and maintenance and reducing the operation cost.
Further, the determining the measured values of the statistics from the first and second subsequences comprises: deleting abnormal values in the first subsequence and the second subsequence respectively to obtain an updated first subsequence and an updated second subsequence; and determining the measured value of the statistic according to the updated first subsequence and the updated second subsequence. According to the embodiment of the invention, before the measured value of the statistic is determined, the abnormal value is deleted, so that the statistical early warning p value or the p value to be detected with higher accuracy is further ensured to be obtained.
Further, the preset lengths of the historical performance index sequences are the same, the preset lengths are equal to the preset lengths of the performance index sequences to be detected, and the lengths of the divided first subsequences are the same. According to the embodiment of the invention, each historical performance index sequence and the performance index sequence to be detected with the same length are selected, and the divided subsequences with the same length are adopted, so that the equipment performance prediction process is facilitated to be simplified, and the performance prediction accuracy can be improved.
Drawings
FIG. 1 is a flow chart illustrating a method for predicting device performance according to an embodiment of the present invention;
FIG. 2 is a flow chart illustrating a method for predicting device performance according to another embodiment of the present invention;
fig. 3 is a schematic structural diagram of an apparatus for predicting device performance according to an embodiment of the present invention.
Detailed Description
As background art, the prior art generally adopts an experience-based overrun early warning method, mainly depends on experience knowledge of technicians, has relatively low accuracy, and is easily influenced by a single abnormal point to report a false alarm.
The embodiment of the invention provides a method for predicting equipment performance, which comprises the following steps: determining a performance index for evaluating the performance of the equipment; acquiring a plurality of historical performance index sequences from historical data, respectively calculating a historical p value of each historical performance index sequence to obtain a historical p value set, and determining an early warning p value according to the historical p value set; acquiring a to-be-detected performance index sequence from to-be-detected data, and calculating a to-be-detected p value of the to-be-detected performance index sequence; comparing the p value to be detected with the early warning p value, and predicting the performance of the equipment according to the comparison result; wherein, the p value to be detected and the historical p value are calculated according to the following steps: intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence; constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution; and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value.
By the technical scheme provided by the embodiment of the invention, after the performance index for evaluating the performance of the equipment is determined, the early warning p value of the performance index can be determined by calculating a plurality of different historical performance index sequences, and the p values to be detected are determined and then compared, so that the fault probability of the equipment can be deduced and predicted. The early warning p value is obtained based on the statistical model, so that the influence of a single abnormal point can be well overcome, the false alarm probability is favorably reduced by utilizing the change of the parameter statistical rule, and the early fault is more accurately found, thereby providing a powerful basis for guaranteeing the operation and maintenance and reducing the operation cost.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail below.
Fig. 1 is a flowchart illustrating a method for predicting device performance according to an embodiment of the present invention. The device performance prediction method may be implemented by a computing device. Including but not limited to terminal devices, servers, etc. Specifically, the prediction method may include the steps of:
step S101, determining a performance index for evaluating equipment performance;
step S102, acquiring a plurality of historical performance index sequences from historical data, respectively calculating a historical p value of each historical performance index sequence to obtain a historical p value set, and determining an early warning p value according to the historical p value set;
step S103, acquiring a performance index sequence to be detected from data to be detected, and calculating a p value to be detected of the performance index sequence to be detected;
and S104, comparing the p value to be detected with the early warning p value, and predicting the performance of the equipment according to the comparison result.
Wherein, the p value to be detected and the historical p value are calculated according to the following steps:
intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence;
constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution;
and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value.
More specifically, in step S101, a performance index for evaluating the performance of the device may be determined. The equipment can be any electromechanical equipment, and can also be other equipment capable of long-term and multiple operations. The performance indicator may be a parameter indicator reflecting a certain component performance or several component performances of the apparatus. This component may be an important or critical component of the device, or may be a support component or a peripheral component of the device.
In step S102, a plurality of historical performance indicator sequences may be acquired from the historical data. In particular implementations, the historical data includes performance indicators and other information generated by the equipment during historical operations. The historical data is arranged according to the time sequence. After the complete historical performance index sequences arranged according to the time sequence are extracted from the historical data, a plurality of historical performance index sequences can be intercepted from the complete historical performance index sequences in a sliding window mode.
Thereafter, a historical p-value for each of the historical performance indicator sequences may be calculated separately. Each of the historical p-values forms a set of historical p-values.
In a specific implementation, a length of each historical performance indicator sequence is first determined, which may be referred to as a preset length and refers to the number of elements included in the historical performance indicator sequence. Secondly, the plurality of historical performance indicator sequences with the preset length can be intercepted from the complete historical performance indicator sequence. For example, assume that the complete historical performance indicator sequence length is M, and each element is represented by XiI is 1,2,3, …, M. Each historical performance index sequence is N, N<M, M, N is a positive integer. For example, one of the historical performance indicator sequences is XjJ is k, k +1, …, k + N-1, k is equal to or greater than 1; another historical performance indicator sequence is Xp，p＝k+N,k+N+1,…,k+2N-1，(k+2N-1)<And M. As another example, one of the historical performance indicator sequences is XjJ is k, k +1, …, k + N-1, k is equal to or greater than 1; another historical performance indicator sequence is Xp，p＝k+v,k+v+1,…,k+v+N-1,k<v<N-1,(k+v+N-1)<M。
Again, for each of the historical performance indicator sequences, the historical performance indicator sequence may be divided into a first historical subsequence and a second historical subsequence, and the historical performance indicator sequence includes a plurality of historical performance indicators arranged in a time sequence. In one non-limiting example, the first history subsequence has a length of N1I.e. the first history sub-sequence contains N number of elements1The length of the second history subsequence is N2I.e. the second history sub-sequence contains N number of elements2，N1And N2The sum is equal to the preset length.
In particular implementations, N may be determined based on sensitivity and false alarm rate requirements1And N2The value of (c): the higher the sensitivity required, the larger the ratio, and the easier it is to detect changes in the performance of the apparatus, in which caseIs susceptible to other factors and the false alarm rate is high. Conversely, the lower the required false alarm rate, the smaller the ratio, which is not easily affected by other factors, and the sensitivity is also reduced.
Thereafter, statistics may be constructed that are capable of representing mean level differences between the first and second historical subsequences, the statistics being subject to a t-distribution.
Further, a statistical measured value may be determined from the first and second historical subsequences, and a p-value reflecting the probability of equipment failure may be calculated from the statistical measured value. In a specific implementation, the measured value of the statistic may be calculated according to the updated first history subsequence and the updated second history subsequence. The updated first history subsequence and the updated second history subsequence are the result of deleting the abnormal value in the first history subsequence and the second history subsequence respectively.
In one non-limiting example, outliers in the first history sub-sequence or the second history sub-sequence may be deleted using a boxplot outliers processing algorithm.
Specifically, for the first history subsequence or the second history subsequence, a respective lower quartile and upper quartile may first be calculated; secondly, the difference between the upper quartile and the lower quartile can be used as a quartile distance; the section [ Q ] can be deleted again1-IQR*a，Q3+IQR*a]And (ii) other elements. Wherein Q is1Representing the lower quartile, Q, of the first or second subsequence3Representing the upper quartile of said first or second subsequence, a representing a preset intensity factor, IQR representing said quartile distance, "-" representing multiplication.
Further, statistics may be constructed that are capable of representing mean level differences between the first and second historical subsequences, the statistics being subject to a t-distribution. In one embodiment, the p-value can be calculated according to the t distribution to which the statistic belongs by using any one of the following formulas: p ═ P (| T non-conducting phosphor)>|T0|)；p＝P(T>T0)；p＝P(T<T0) (ii) a It is composed ofIn (1), T represents the statistic, T0Representing the measured value of the statistic, P representing the calculated probability.
In one non-limiting example, assume that the first history subsequence has a length of n1The length of the second history subsequence is n2，ξiRepresenting the ith element, η, of said first subsequenceiRepresents the ith element of the second subsequence, i being a positive integer. Under this condition, the statistics obeying the t-distribution are as follows: when n is1＝n2When n, the statistic obeying the t distribution is:
Figure BDA0002275722910000091
Figure BDA0002275722910000092
θi＝ξi-ηi。
when n is1<n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000093
Figure BDA0002275722910000094
when n is1>n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000095
Figure BDA0002275722910000096
in another non-limiting example, assume that the first history subsequence has a length of N1The length of the second history subsequence is N2. The respective lengths of the updated first history subsequence and the updated second history subsequence are n1And n2，ξiRepresenting the ith element, η, of said first subsequenceiRepresenting said second subsequenceThe ith element, i is a positive integer. Under the condition, when n is1＝n2When n, the statistics obeying the t distribution are:
Figure BDA0002275722910000097
Figure BDA0002275722910000098
θi＝ξi-ηi。
when n is1<n2The statistics obeying the t-distribution are:
Figure BDA0002275722910000099
Figure BDA00022757229100000910
when n is1>n2The statistics obeying the t-distribution are:
Figure BDA00022757229100000911
Figure BDA0002275722910000101
further, after the measured value of the statistic is calculated from the first and second history subsequences, a history p-value reflecting the failure probability of the equipment can be calculated from the measured value of the statistic.
And then, obtaining the historical p value of each historical performance index sequence according to a similar calculation method, and further obtaining the historical p value set. An early warning p-value may be determined based on the historical set of p-values. In one embodiment, the elements in the historical p-value set may be arranged in order from small to large to obtain a historical p-value sequence; and then, selecting the early warning p value from the historical p value sequence according to a preset quantile.
In step S103, a to-be-detected performance indicator sequence may be obtained from to-be-detected data, and a to-be-detected p-value of the to-be-detected performance indicator sequence is calculated. In practical applications, a plurality of historical performance indicator sequences with the same length can be selected. Specifically, when the lengths of the plurality of historical performance indicator sequences are the same, the length of the to-be-detected performance indicator sequence may be the same as the length of the historical performance indicator sequence. At this time, the lengths of the first to-be-detected subsequence and the second to-be-detected subsequence obtained by dividing the to-be-detected performance index sequence are respectively the same as the lengths of the first historical subsequence and the second historical subsequence obtained by dividing the historical performance index sequence.
Thereafter, statistics can be constructed that are capable of representing the mean level difference between the first suspect subsequence and the second suspect subsequence, said statistics being subject to a t-distribution.
Further, a statistical measured value can be determined according to the first sub-sequence to be detected and the second sub-sequence to be detected, and a p-value capable of reflecting the equipment fault probability can be calculated according to the statistical measured value. In a specific implementation, the measured value of the statistic may be an abnormal value of the first to-be-detected subsequence and the second to-be-detected subsequence, which are deleted, respectively obtain the updated first to-be-detected subsequence and the updated second to-be-detected subsequence, and is calculated according to the updated first to-be-detected subsequence and the updated second to-be-detected subsequence.
In a non-limiting example, the outliers in the first or second to-be-detected subsequence can be deleted using a boxplot outliers processing algorithm. Specifically, the step of deleting the abnormal value in the first to-be-detected subsequence or the second to-be-detected subsequence by using the boxplot abnormal value processing algorithm is the same as the step of deleting the first to-be-detected subsequence or the second history subsequence, and is not repeated here.
Further, statistics can be constructed that can represent either the first suspect subsequence or the second suspect subsequence to reflect a mean level difference therebetween, the statistics being subject to a t-distribution. In one embodiment, the p-value can be calculated according to the t distribution to which the statistic belongs by using any one of the following formulas: p ═ P (| T non-conducting phosphor)>|T0|)；p＝P(T>T0)；p＝P(T<T0) (ii) a Wherein T represents the statistic, T0Representing the measured value of the statistic. It should be noted that the formula of the statistic complying with the t distribution is the same as above, and is not repeated here.
Further, after the statistics measured value is obtained by calculation according to the first to-be-detected subsequence and the second to-be-detected subsequence, a to-be-detected p-value capable of reflecting the equipment fault probability can be calculated according to the statistics measured value.
It should be noted that step S102 and step S103 may be executed sequentially or in parallel. When the two sub-sequence lengths are sequentially executed, after the preset length and the two sub-sequence lengths are determined, the execution of step S102 may be before step S103 or after step S103.
With continued reference to fig. 1, in step S104, the p-value to be detected and the early warning p-value may be compared, and the performance of the device may be predicted according to the comparison result. In a non-limiting example, when the early warning p value is smaller than the p value to be detected, it can be predicted that the equipment has a failure risk, and an early warning indication can be sent.
The details of the electromechanical device are described below as a specific embodiment. Fig. 2 is a schematic flow chart of a method for detecting performance of an electromechanical device in a typical scenario according to an embodiment of the present invention.
Referring to fig. 2, first, in step S201, a performance index of the electromechanical device is extracted according to a business logic, a physical principle, or a mathematical model. When the performance index is determined, the performance index is ensured to be basically not influenced by environmental factors as much as possible, and the state of the electromechanical equipment can be depicted. Due to the existence of measurement errors or model errors, the performance index is allowed to randomly fluctuate within a certain range, but when the performance index is abnormal and does not necessarily develop into a fault, the performance index loses the original statistical rule. Based on the above, the performance index can be obtained based on statistics, and whether the electromechanical device has an abnormal risk or not can be determined.
Next, in step S202, a performance index sequence of a certain length is extracted and divided into two subsequences. For example, intercept certainA sequence of performance indicators of length, divided into a length N1And N2Two preceding and succeeding to-be-detected subsequence of (1), N1And N2The value of (c) can be determined according to the sensitivity required. If abnormal conditions occur, fast feedback is needed according to performance indexes, then N2Can be reduced appropriately; otherwise, N2May be increased appropriately. N is a radical of2If the alarm is too small, the alarm is easily affected by abnormal points caused by a few random reasons, and N is a false alarm2Is the result of a game between sensitivity and false alarm rate.
It should be noted that the performance indicator sequence may be a historical performance indicator sequence, and the historical performance indicator sequence is obtained from historical data. Alternatively, the performance indicator sequence may be a performance indicator sequence to be detected, and the performance indicator sequence to be detected may be obtained from data to be detected.
Again, in step S203, the abnormal value in the previous and next subsequences is removed by an abnormal value processing method such as a box plot. After removing the abnormal value, the updated front and back subsequences can be obtained.
For example, the abnormal value removal may be performed on the two subsequences before and after the abnormal value removal by using an abnormal value processing method of the box plot. The specific method comprises the following steps: calculating the lower quartile Q of each subsequence1And its associated upper quartile Q3And calculating the four-bit distance IQR, the IQR being Q3-Q1. Setting an intensity parameter a, deleting an interval [ Q ]1-IQR*a，Q3+IQR*a]And (ii) other elements. In practical applications, the value of a is typically 3. Those skilled in the art understand that the value of a can also be changed into other values according to actual requirements, and the value of "+" represents multiplication.
Further in step S204, a T statistic may be constructed that can be used to examine the difference in mean level of the two subsequences. Wherein the two subsequences may be two subsequences before and after the update.
For example, let the random variable ξ be a subsequence from the first half and the random variable η be a subsequence from the second half. Assuming that both the random variable ξ and the random variable η obey a normal distribution, the statistical quantity constructed at this time satisfies the t distribution. The magnitude of the statistic describes the difference of the mean levels of the performance indexes of the front part and the rear part.
In one embodiment, it is assumed that the two preceding and succeeding subsequences are each of length N1，N2. The two subsequences before and after the update are recorded as a first subsequence after the update and a second subsequence after the update, and the length of each subsequence is n1And n2。ξiAn i-th element, η, representing said updated first subsequenceiAnd i is a positive integer, and represents the ith element of the updated second subsequence. Under this condition, the statistics obeying the t-distribution are as follows: when n is1＝n2When n', the statistics obeying the t-distribution are:
Figure BDA0002275722910000121
Figure BDA0002275722910000131
θi＝ξi-ηi。
when n is1<n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000132
Figure BDA0002275722910000133
when n is1>n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000134
Figure BDA0002275722910000135
those skilled in the art will appreciate that the degree of freedom of the t-distribution varies with the construction. The T statistics all satisfy T distribution of a certain degree of freedom, and are not described herein again.
Further, in step S205, the method can be implementedAnd calculating the p value to be detected of the performance index sequence to be detected or the historical p value of the technical historical performance index sequence by the measured value of the statistic. Where p represents the probability of the statistic, T represents the probability of the statistic0Representing the measured value of the statistic. The T distribution to which the statistic belongs may be calculated as a P value corresponding to the measured value of the statistic, for example, P ═ P (| T |)>|T0|)；p＝P(T>T0)；p＝P(T<T0)。
Since the statistic is t distribution of a certain degree of freedom, the probability value can be easily calculated by a computer. The particular manner in which the p-value is calculated depends on the choice of the type of drift. If the early warning is expected to be reported no matter the current path drifts upwards or downwards, P is equal to P (| T>|T0I) calculating a p value; if only downward drift is of interest, then P ═ P (T) is used>T0) Calculating a p value; if only upward drift is of interest, then the last P ═ P (T) is used<T0) And (6) performing calculation.
Further, in step S206, a plurality of historical performance index sequences may be clipped through a window of a certain size for the historical index sequence, and each historical p-value may be calculated. Calculating the respective history p values may be done in a similar manner to step S201 to step S205.
Further, in step S207, in combination with the preset quantile b set in step S200, a value corresponding to the quantile b of all the historical p values may be calculated, and taken as the early warning p value. For example, the fixed-length historical performance index sequence may be cut out in a sliding manner, and each historical p value may be calculated in the manner from S201 to S205. And setting a quantile to control the number of the p values needing to be reported out of the early warning, and calculating the value of the historical p value under the quantile to obtain the early warning p value.
Further, in step S208, an early warning prompt is issued when the to-be-detected p value is smaller than the early warning p value.
It should be noted that the process of calculating the p value to be examined and the process of calculating the p value in the history may be executed in parallel or sequentially. When the data to be detected are sequentially executed, if the data to be detected are obtained in advance and the performance index sequence to be detected can be obtained, calculating p values to be detected according to the performance index sequence to be detected, and then calculating each historical p value; or calculating each historical p value, then determining the data to be detected, and obtaining the performance index sequence to be detected according to the data to be detected to calculate the p value to be detected.
In summary, the embodiment of the present invention can fully utilize the statistical principle to evaluate the degree of change of the performance index of the device as a whole, and compared with the overrun early warning technology in the prior art, the defect that the performance index is easily affected by a single abnormal point is overcome. According to the scheme, the state change can be recognized at the initial drifting stage of the performance index, and the state change can be recognized only when the performance index is not in a deteriorated state, so that the potential risk can be found in the bud state of the fault more easily, the healthy operation of equipment is very facilitated to be guaranteed, and the operation and maintenance cost is reduced. In addition, the early warning threshold value is determined by a statistical method, so that subjective factors of manually determining the threshold value can be reduced to a great extent, and the method is more scientific and objective.
Fig. 3 is a schematic structural diagram of an apparatus for predicting device performance according to an embodiment of the present invention. The device performance prediction device 3 (hereinafter referred to as the prediction device 3) may be used to predict various devices by using the methods shown in fig. 1 to 2, and may send out an early warning signal according to the prediction result.
In a specific implementation, the prediction device 3 may include: a determining module 31, configured to determine a performance index for evaluating the performance of the device; the first calculation module 32 is configured to obtain a plurality of historical performance index sequences from historical data, and respectively calculate a historical p value of each of the historical performance index sequences to obtain a historical p value set, and determine an early warning p value according to the historical p value set; the second calculating module 33 is configured to obtain a to-be-detected performance indicator sequence from to-be-detected data, and calculate a to-be-detected p value of the to-be-detected performance indicator sequence; the comparison and prediction module 34 is used for comparing the p value to be detected with the early warning p value and predicting the equipment performance according to the comparison result; wherein, the p value to be detected and the historical p value are calculated according to the following steps: intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence; constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution; and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value.
In a specific implementation, the determining the measured values of the statistics from the first subsequence and the second subsequence may include: deleting abnormal values in the first subsequence and the second subsequence respectively to obtain an updated first subsequence and an updated second subsequence; and determining the measured value of the statistic according to the updated first subsequence and the updated second subsequence.
In a specific implementation, the deleting the abnormal values in the first subsequence and the second subsequence respectively comprises: and deleting abnormal values in the first subsequence and the second subsequence respectively by using a boxplot abnormal value processing algorithm.
In a specific implementation, the removing the outliers in the first subsequence and the second subsequence, respectively, using the outliers processing algorithm of the box plot may include: calculating respective lower quartiles and upper quartiles for the first subsequence or the second subsequence; taking the difference between the upper quartile and the lower quartile as a quartile range; deletion Interval [ Q ]1-IQR*a，Q3+IQR*a]An element other than; wherein Q is1Representing the lower quartile, Q, of the first or second subsequence3Representing the upper quartile of the first subsequence or the second subsequence, a representing a preset intensity factor, and IQR representing the quartile distance.
In specific implementation, the preset lengths of the historical performance index sequences may be the same, the preset length is equal to the preset length of the performance index sequence to be detected, and the lengths of the divided first subsequences are the same.
In a specific implementation, the determining an early warning p-value according to the set of historical p-values includes: arranging all elements in the historical p value set from small to large to obtain a historical p value sequence; and selecting the early warning p value from the historical p value sequence according to a preset quantile.
In a specific implementation, the first subsequence has a length of n1The length of the second subsequence is n2Constructing statistics representative of mean level differences between the first and second subsequences comprising:
when n is1＝n2When n, the statistic obeying the t distribution is:
Figure BDA0002275722910000151
Figure BDA0002275722910000161
θi＝ξi-ηi(ii) a When n is1<n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000162
Figure BDA0002275722910000163
when n is1>n2Then, the statistic obeying t distribution is:
Figure BDA0002275722910000164
Figure BDA0002275722910000165
wherein ξiRepresenting the ith element, η, of said first subsequenceiRepresents the ith element of the second subsequence, i being a positive integer.
In particular implementations, n may be determined based on sensitivity and false alarm rate requirements1And n2The ratio of (A) to (B): the higher the required sensitivity, the larger its ratio; the lower the required false alarm rate, the smaller its ratio.
In specific implementation, the p value capable of reflecting the failure probability of the equipment is calculated according to the measured value of the statistic; and calculating the p value by adopting any one of the following formulas according to the t distribution to which the statistic belongs: p ═ P (| T non-conducting phosphor)>|T0|)；p＝P(T>T0)；p＝P(T<T0) (ii) a Wherein T represents the statistic, T0Representing the measured value of the statistic.
For more details of the operation principle and the operation mode of the prediction apparatus 3, reference may be made to the description in the embodiments shown in fig. 1 to fig. 2, and details are not repeated here.
Further, the embodiment of the present invention further discloses a storage medium, on which computer instructions are stored, and when the computer instructions are executed, the technical solutions of the methods in the embodiments shown in fig. 1 to fig. 2 are executed. Preferably, the storage medium may include a computer-readable storage medium such as a non-volatile (non-volatile) memory or a non-transitory (non-transient) memory. The computer readable storage medium may include ROM, RAM, magnetic or optical disks, and the like.
Further, an embodiment of the present invention further discloses a terminal, which includes a memory and a processor, where the memory stores a computer instruction capable of running on the processor, and the processor executes the technical solution of the method in the embodiment shown in fig. 1 to 2 when running the computer instruction. In particular, the terminal may be a variety of computing devices having computing capabilities.
Although the present invention is disclosed above, the present invention is not limited thereto. Various changes and modifications may be effected therein by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (12)

1. A method for predicting device performance, comprising:
determining a performance index for evaluating the performance of the equipment;
acquiring a plurality of historical performance index sequences from historical data, respectively calculating a historical p value of each historical performance index sequence to obtain a historical p value set, and determining an early warning p value according to the historical p value set;
acquiring a to-be-detected performance index sequence from to-be-detected data, and calculating a to-be-detected p value of the to-be-detected performance index sequence;
comparing the p value to be detected with the early warning p value, and predicting the performance of the equipment according to the comparison result;
wherein, the p value to be detected and the historical p value are calculated according to the following steps:
intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence;
constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution;
and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value.
2. The method of predicting according to claim 1, wherein determining the measured value of the statistic based on the first subsequence and the second subsequence comprises:
deleting abnormal values in the first subsequence and the second subsequence respectively to obtain an updated first subsequence and an updated second subsequence;
and determining the measured value of the statistic according to the updated first subsequence and the updated second subsequence.
3. The prediction method according to claim 2, wherein the deleting the outliers in the first and second subsequences, respectively, comprises:
and deleting abnormal values in the first subsequence and the second subsequence respectively by using a boxplot abnormal value processing algorithm.
4. The prediction method according to claim 3, wherein the removing the abnormal values in the first subsequence and the second subsequence, respectively, using the abnormal value processing algorithm of the box plot comprises:
calculating respective lower quartiles and upper quartiles for the first subsequence or the second subsequence;
taking the difference between the upper quartile and the lower quartile as a quartile range;
deletion Interval [ Q ]1-IQR*a，Q3+IQR*a]An element other than;
wherein Q is1Representing the lower quartile, Q, of the first or second subsequence3Representing the upper quartile of the first subsequence or the second subsequence, a representing a preset intensity factor, and IQR representing the quartile distance.
5. The prediction method according to claim 1, wherein the preset lengths of the historical performance index sequences are the same, the preset lengths are equal to the preset lengths of the performance index sequences to be detected, and the lengths of the divided first subsequences are the same.
6. The prediction method of claim 1, wherein said determining an early warning p-value from the set of historical p-values comprises:
arranging all elements in the historical p value set from small to large to obtain a historical p value sequence;
and selecting the early warning p value from the historical p value sequence according to a preset quantile.
7. The prediction method according to claim 1, wherein the first subsequence has a length of n1The length of the second subsequence is n2Said structureGenerating statistics capable of representing mean level differences between the first and second subsequences comprises:
when n is1＝n2When n, the statistic obeying the t distribution is:
Figure FDA0002275722900000021
Figure FDA0002275722900000022
θi＝ξi-ηi；
when n is1<n2Then, the statistic obeying t distribution is:
Figure FDA0002275722900000023
Figure FDA0002275722900000024
when n is1>n2Then, the statistic obeying t distribution is:
Figure FDA0002275722900000031
Figure FDA0002275722900000032
wherein ξiRepresenting the ith element, η, of said first subsequenceiRepresents the ith element of the second subsequence, i being a positive integer.
8. Prediction method according to any of claims 1 to 7, characterized in that n is determined according to the requirements for sensitivity and false alarm rate1And n2The ratio of (A) to (B): the higher the required sensitivity, the larger its ratio; the lower the required false alarm rate, the smaller its ratio.
9. The prediction method according to any one of claims 1 to 7, wherein the p-value reflecting the probability of the failure of the plant is calculated from the measured values of the statistics;
and calculating the p value by adopting any one of the following formulas according to the t distribution to which the statistic belongs:
p＝P(|T|＞|T0|)；p＝P(T＞T0)；p＝P(T＜T0)；
wherein T represents the statistic, T0Represents the measured value of the statistic, and P represents the probability of the solution.
10. An apparatus for predicting device performance, comprising:
the determining module is used for determining a performance index for evaluating the performance of the equipment;
the first calculation module is used for acquiring a plurality of historical performance index sequences from historical data, respectively calculating a historical p value of each historical performance index sequence to obtain a historical p value set, and determining an early warning p value according to the historical p value set;
the second calculation module is used for acquiring a to-be-detected performance index sequence from to-be-detected data and calculating a to-be-detected p value of the to-be-detected performance index sequence;
the comparison and prediction module is used for comparing the p value to be detected with the early warning p value and predicting the performance of the equipment according to the comparison result;
wherein, the p value to be detected and the historical p value are calculated according to the following steps:
intercepting a performance index sequence with a preset length, and dividing the intercepted performance index sequence into a first subsequence and a second subsequence, wherein the performance index sequence is a plurality of performance indexes which are arranged according to a time sequence, and the performance index sequence is the historical performance index sequence or the to-be-detected performance index sequence;
constructing a statistic representative of mean level differences between the first and second subsequences, the statistic being subject to a t-distribution;
and determining a statistic measured value according to the first subsequence and the second subsequence, and calculating a p value capable of reflecting the fault probability of equipment according to the statistic measured value, wherein the p value is the p value to be detected or the historical p value.
11. A storage medium having stored thereon computer instructions, characterized in that the computer instructions are operative to perform the steps of the method of any one of claims 1 to 9.
12. A terminal comprising a memory and a processor, the memory having stored thereon computer instructions executable on the processor, wherein the processor, when executing the computer instructions, performs the steps of the method of any one of claims 1 to 9.
CN201911122080.7A 2019-11-15 2019-11-15 Device performance prediction method and device, storage medium and terminal Active CN112819190B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911122080.7A CN112819190B (en) 2019-11-15 2019-11-15 Device performance prediction method and device, storage medium and terminal

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911122080.7A CN112819190B (en) 2019-11-15 2019-11-15 Device performance prediction method and device, storage medium and terminal

Publications (2)

Publication Number Publication Date
CN112819190A true CN112819190A (en) 2021-05-18
CN112819190B CN112819190B (en) 2024-01-26

Family

ID=75851904

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911122080.7A Active CN112819190B (en) 2019-11-15 2019-11-15 Device performance prediction method and device, storage medium and terminal

Country Status (1)

Country Link
CN (1) CN112819190B (en)

Citations (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070055477A1 (en) * 2005-09-02 2007-03-08 Microsoft Corporation Web data outlier detection and mitigation
US20080183424A1 (en) * 2007-01-25 2008-07-31 Johnson Controls Technology Company Method and system for assessing performance of control systems
US20090240366A1 (en) * 2008-03-08 2009-09-24 Tokyo Electron Limited Method and system for detection of tool performance degradation and mismatch
CN101771758A (en) * 2008-12-31 2010-07-07 北京亿阳信通软件研究院有限公司 Dynamic determine method for normal fluctuation range of performance index value and device thereof
CN102252843A (en) * 2011-07-21 2011-11-23 河南科技大学 Assessment method for rolling bearing performance variation
CN102520288A (en) * 2011-12-19 2012-06-27 江苏省电力试验研究院有限公司 Primary-rising based current differential protection synchronization performance testing system and testing method
WO2012149901A1 (en) * 2011-05-03 2012-11-08 北京中瑞泰科技有限公司 Similarity curve-based device malfunction early-warning and optimization method and system
KR20130107889A (en) * 2012-03-23 2013-10-02 삼성전자주식회사 Aparatus and method for detecting anomalous subsequence
JP5771318B1 (en) * 2014-08-26 2015-08-26 株式会社日立パワーソリューションズ Abnormality diagnosis apparatus and abnormality diagnosis method
US20160103838A1 (en) * 2014-10-09 2016-04-14 Splunk Inc. Anomaly detection
CN108260148A (en) * 2016-12-28 2018-07-06 华为技术服务有限公司 A kind of fault detection method and device
CN109587001A (en) * 2018-11-15 2019-04-05 新华三信息安全技术有限公司 A kind of performance indicator method for detecting abnormality and device
US20190188584A1 (en) * 2017-12-19 2019-06-20 Aspen Technology, Inc. Computer System And Method For Building And Deploying Models Predicting Plant Asset Failure
US20190234860A1 (en) * 2018-01-31 2019-08-01 Pratt & Whitney Canada Corp. Method and system for predicting engine performance
US20200325766A1 (en) * 2016-03-16 2020-10-15 University Of Houston System System and method for detecting, diagnosing, and correcting trips or failures of electrical submersible pumps

Patent Citations (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20070055477A1 (en) * 2005-09-02 2007-03-08 Microsoft Corporation Web data outlier detection and mitigation
US20080183424A1 (en) * 2007-01-25 2008-07-31 Johnson Controls Technology Company Method and system for assessing performance of control systems
US20090240366A1 (en) * 2008-03-08 2009-09-24 Tokyo Electron Limited Method and system for detection of tool performance degradation and mismatch
CN101771758A (en) * 2008-12-31 2010-07-07 北京亿阳信通软件研究院有限公司 Dynamic determine method for normal fluctuation range of performance index value and device thereof
WO2012149901A1 (en) * 2011-05-03 2012-11-08 北京中瑞泰科技有限公司 Similarity curve-based device malfunction early-warning and optimization method and system
CN102252843A (en) * 2011-07-21 2011-11-23 河南科技大学 Assessment method for rolling bearing performance variation
CN102520288A (en) * 2011-12-19 2012-06-27 江苏省电力试验研究院有限公司 Primary-rising based current differential protection synchronization performance testing system and testing method
KR20130107889A (en) * 2012-03-23 2013-10-02 삼성전자주식회사 Aparatus and method for detecting anomalous subsequence
JP5771318B1 (en) * 2014-08-26 2015-08-26 株式会社日立パワーソリューションズ Abnormality diagnosis apparatus and abnormality diagnosis method
US20160103838A1 (en) * 2014-10-09 2016-04-14 Splunk Inc. Anomaly detection
US20200325766A1 (en) * 2016-03-16 2020-10-15 University Of Houston System System and method for detecting, diagnosing, and correcting trips or failures of electrical submersible pumps
CN108260148A (en) * 2016-12-28 2018-07-06 华为技术服务有限公司 A kind of fault detection method and device
US20190188584A1 (en) * 2017-12-19 2019-06-20 Aspen Technology, Inc. Computer System And Method For Building And Deploying Models Predicting Plant Asset Failure
US20190234860A1 (en) * 2018-01-31 2019-08-01 Pratt & Whitney Canada Corp. Method and system for predicting engine performance
CN109587001A (en) * 2018-11-15 2019-04-05 新华三信息安全技术有限公司 A kind of performance indicator method for detecting abnormality and device

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
李波;赵洁;郭晋;: "设备故障评估新指标及基于ARMA的预测系统", 系统工程与电子技术, no. 01 *

Also Published As

Publication number Publication date
CN112819190B (en) 2024-01-26

Similar Documents

Publication Publication Date Title
US10747188B2 (en) Information processing apparatus, information processing method, and, recording medium
US20200150159A1 (en) Anomaly detection device, anomaly detection method, and storage medium
KR102025145B1 (en) Apparatus and Method for Predicting Plant Data
US20200259725A1 (en) Methods and systems for online monitoring using a variable data
CN110008096B (en) Data monitoring method, device, electronic equipment and computer readable storage medium
CN113312804B (en) Temperature early warning method, device, equipment and storage medium of transformer
JP2007328522A (en) Failure probability calculation device, method, and program
JP6521096B2 (en) Display method, display device, and program
JP2014527214A5 (en)
CN109344610B (en) Method and device for detecting sequence attack
JP6164311B1 (en) Information processing apparatus, information processing method, and program
US20210048811A1 (en) Model generation device for life prediction, model generation method for life prediction, and recording medium storing model generation program for life prediction
WO2017150286A1 (en) System analyzing device, system analyzing method, and computer-readable recording medium
TW201926041A (en) Equipment maintenance forecasting system and operation method thereof
CN111813644B (en) Evaluation method and device for system performance, electronic equipment and computer readable medium
US20190265088A1 (en) System analysis method, system analysis apparatus, and program
JP6915693B2 (en) System analysis method, system analyzer, and program
US8565910B2 (en) Manufacturing execution system (MES) including a wafer sampling engine (WSE) for a semiconductor manufacturing process
CN112819190B (en) Device performance prediction method and device, storage medium and terminal
KR20130042523A (en) Method measuring of healthy indicia of plant reflected status of sub-component and storage media thereof
WO2023184237A1 (en) Method and apparatus for calculating remaining useful life of electronic system, and computer medium
KR20130028483A (en) Method measuring of healthy indicia of plant reflected status of sub-component and storage media thereof
CN111061254B (en) PHM system performance evaluation method and system
CN113221096A (en) Method and system for analyzing correlation of random events in chaotic engineering
US20220011738A1 (en) Process management device, process management method, and process management program storage medium

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