CN116149971B - Equipment fault prediction method and device, electronic equipment and storage medium - Google Patents
Equipment fault prediction method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN116149971B CN116149971B CN202310396139.1A CN202310396139A CN116149971B CN 116149971 B CN116149971 B CN 116149971B CN 202310396139 A CN202310396139 A CN 202310396139A CN 116149971 B CN116149971 B CN 116149971B
- Authority
- CN
- China
- Prior art keywords
- index data
- types
- equipment
- target
- devices
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3466—Performance evaluation by tracing or monitoring
- G06F11/3495—Performance evaluation by tracing or monitoring for systems
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The invention provides a device fault prediction method, a device, electronic equipment and a storage medium, and relates to the technical field of computers, wherein the device fault prediction method comprises the following steps: monitoring real-time index data of M devices in a service system; under the condition that the abnormality of the real-time index data of a first device in the M devices is monitored, determining a target device from the M devices based on the first device and a pre-constructed device similarity matrix; the target device is a device, of which the similarity with the first device reaches a preset threshold, in the M devices; the device similarity matrix is used to characterize the similarity between M devices. By the method, the target equipment similar to the first equipment is determined to be the equipment for predicting the faults based on the first equipment and the equipment similarity matrix, so that the accuracy of equipment fault prediction can be improved; meanwhile, modeling of huge number of service systems of equipment is avoided, manpower and material resources are saved, and efficiency of equipment fault prediction is improved.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a device failure prediction method, a device, an electronic device, and a storage medium.
Background
Along with rapid development of science and technology, computer software is deployed in a distributed cloud environment to be huge, dependency relationships among components are complicated, a large number of devices are usually used in business systems deployed by enterprises, faults caused by various reasons are unavoidable after the devices are operated for a long time, the whole business systems are greatly influenced after the faults occur, and then the enterprises are lost. Therefore, it is necessary to predict the failure of the devices in the business system.
In the related art, a method based on a reliability model, a physical model or qualitative analysis is generally used to predict faults of devices in a service system. However, for business systems deployed by enterprises, the number of devices is very large, so that effective modeling cannot be performed on the business systems with the large number of devices, which results in low accuracy of predicting device faults.
Therefore, how to accurately predict the equipment failure for the business system with huge equipment quantity is a problem to be solved at present.
Disclosure of Invention
Aiming at the problems existing in the prior art, the embodiment of the invention provides a device fault prediction method, a device, electronic equipment and a storage medium.
The invention provides a device fault prediction method, which comprises the following steps:
monitoring real-time index data of M devices in a service system;
under the condition that the abnormality of the real-time index data of a first device in the M devices is monitored, determining a target device from the M devices based on the first device and a pre-constructed device similarity matrix; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer.
Optionally, the device similarity matrix is constructed by:
generating derivative index data corresponding to N types of target historical index data in the M devices aiming at the N types of target historical index data; n is a positive integer;
and constructing the equipment similarity matrix based on the N types of target historical index data and the derivative index data in the M pieces of equipment.
Optionally, the generating derivative index data corresponding to the N types of target history index data for N types of target history index data in the M devices includes:
Determining, for each of the devices, a pair of historical index data based on a priori knowledge; the historical index data pair comprises target historical index data of a first type and target historical index data of a second type; the first type and the second type have an association relation;
nonlinear conversion is carried out on the N types of target historical index data, and first derivative index data corresponding to the N types of target historical index data are generated;
multiplying the first type of target historical index data and the second type of target historical index data in each historical index data pair to generate second derivative index data corresponding to the N types of target historical index data.
Optionally, before the generating derivative index data corresponding to the N types of target history index data for N types of target history index data in the M devices, the method further includes:
acquiring N types of original historical index data in the M devices;
for each device, performing aggregation processing on the original historical index data of each type within a preset period to generate N types of aggregation historical index data corresponding to the N types of original historical index data;
And determining the N types of target historical index data based on the N types of aggregation historical index data.
Optionally, the determining the N types of target historical index data based on the N types of aggregate historical index data includes:
calculating the similarity among N types corresponding to the aggregation history index data based on a dynamic time warping algorithm DTW for each device;
and performing de-duplication processing on the N types of aggregation history index data based on the similarity among the N types, and determining the N types of target history index data.
Optionally, the constructing the device similarity matrix based on the N types of target historical index data and the derived index data in the M devices includes:
performing standardization processing on the N types of target historical index data and the derivative index data in the M devices to generate feature vectors corresponding to the M devices;
calculating the similarity between the feature vectors by using a cosine similarity algorithm;
and constructing the equipment similarity matrix based on the similarity between the feature vectors.
Optionally, after the determining the target device as the predicted faulty device, the method further includes:
And generating an alarm event and a treatment strategy corresponding to the alarm event based on the equipment information corresponding to the predicted fault equipment.
The invention also provides a device fault prediction device, which comprises:
the monitoring module is used for monitoring real-time index data of M devices in the service system;
the determining module is used for determining target equipment from the M pieces of equipment based on the first equipment and a pre-constructed equipment similarity matrix under the condition that the real-time index data of the first equipment in the M pieces of equipment are monitored to be abnormal; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the device fault prediction method as described in any one of the above when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a device failure prediction method as described in any of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor implements a method of predicting an equipment failure as described in any one of the above.
According to the equipment fault prediction method, the equipment fault prediction device, the electronic equipment and the storage medium, the real-time index data of M pieces of equipment in the service system are monitored, and when the condition that the real-time index data of the first equipment in the service system is abnormal is monitored, the target equipment, of which the similarity with the first equipment reaches a preset threshold value, in the M pieces of equipment is determined based on the first equipment and the equipment similarity matrix, and the target equipment similar to the first equipment is determined to be the prediction fault equipment, so that the accuracy of equipment fault prediction can be improved; meanwhile, modeling of huge number of service systems of equipment is avoided, manpower and material resources are saved, and efficiency of equipment fault prediction is improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for predicting equipment failure according to the present invention;
FIG. 2 is a schematic diagram of calculating type similarity based on a DTW algorithm;
FIG. 3 is a second flow chart of the apparatus failure prediction method according to the present invention;
FIG. 4 is a schematic diagram of a device failure prediction apparatus according to the present invention;
fig. 5 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In order to facilitate a clearer understanding of the various embodiments of the present application, some relevant background knowledge is first presented below.
With the strong globalization competition, the product trend is more obvious, and the product quality and cost become the most interesting indexes of enterprises. But how to make a product cost-effective is not a small challenge for the enterprise. The improvement of the overall utilization rate of equipment becomes an important breakthrough point for enterprises to reduce the production cost. At present, in order to ensure that the system stably and safely operates, data indexes of various service system devices need to be acquired in real time, various indexes are analyzed, and after errors occur in the service system devices, the reasons of faults are found according to the indexes.
In the related art, the following method is generally adopted for predicting the equipment failure:
mode 1 adopts a failure prediction method based on a reliability model.
For example, a series of different probability density functions (Probability Density Functions, PDFs), such as an exponential distribution, a normal distribution, a Weibull distribution, are used to express the time at which each type of fault occurs or the probability of a fault occurring within a certain period of time.
Mode 2, a failure prediction method based on a physical model.
For example, a bearing life prediction model is built based on existing historical data.
Mode 3, fault diagnosis based on qualitative analysis method.
For example, fault identification and classification is achieved by using a logical reasoning method. The qualitative analysis method is commonly provided with an expert system, a graph search and qualitative simulation.
However, the above-described equipment failure prediction has the following problems:
1) In practical applications, complex systems are difficult to accurately model, so the above-mentioned methods have great limitations in practical applications.
2) Some of the physical parameters are qualitative variables and qualitative equations among the variables construct a qualitative model of the system, so that unknown faults cannot be accurately diagnosed.
3) The method is only suitable for a simple system, and for fault diagnosis of a complex system, the diagnosis precision can be reduced along with the increase of the complexity of the system.
4) Standard uniformity is poor; for a complex system with more rules, the problems of rule conflict, reasoning loopholes and the like are caused.
In summary, in order to accurately predict equipment failure for a service system with a huge number of equipment, the embodiment of the invention provides an equipment failure prediction method, an equipment failure prediction device, electronic equipment and a storage medium.
The method for predicting equipment failure provided by the invention is specifically described below with reference to fig. 1 to 2. Fig. 1 is a schematic flow chart of a method for predicting equipment failure according to the present invention, referring to fig. 1, the method includes steps 101 to 102, where:
and 101, monitoring real-time index data of M devices in a service system.
It should be noted that the execution body of the present invention may be any electronic device capable of implementing device failure prediction, for example, any one of a smart phone, a smart watch, a desktop computer, a laptop computer, and the like.
In order to accurately predict equipment faults for a service system with a huge number of equipment, in this embodiment, real-time index data of each equipment in the service system needs to be monitored respectively to determine whether abnormality occurs in each equipment in the service system.
For example, there are 100 devices in the service system, and real-time index data corresponding to the 100 devices needs to be detected respectively, so as to determine whether there is an abnormal device in the 100 devices in the service system.
In practical applications, a service system includes a plurality of devices, where a device has multiple types of index data, for example, network element index data, alarm index data, and the like.
The method for collecting the real-time index data of the M devices is various, for example, the real-time index data of each device can be collected from a local database or a cloud server of the device, or the real-time index data sent by each device can be actively received; the invention does not limit the acquisition mode of the real-time index data of the equipment.
102, determining a target device from the M devices based on the first device and a pre-constructed device similarity matrix under the condition that the abnormality of the real-time index data of the first device in the M devices is monitored; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer.
In this embodiment, when it is monitored that the real-time index data of the first device among the M devices in the service system is abnormal, the first device is indicated to be the device that is faulty or abnormal.
At this time, it is necessary to determine, from the M devices, a device whose similarity with the first device reaches a preset threshold based on the device similarity matrix.
In this embodiment, since the similarity between the target device and the first device reaches the preset threshold, the target device is considered to be potentially at risk of failure or abnormality when the first device fails or is abnormal.
Therefore, it is necessary to determine the target device as the predicted failure device, and the accuracy of device failure prediction can be improved.
According to the equipment fault prediction method provided by the invention, the real-time index data of M pieces of equipment in the service system are monitored, and under the condition that the real-time index data of the first equipment in the service system is monitored to be abnormal, the target equipment, of which the similarity with the first equipment in the M pieces of equipment reaches the preset threshold value, is determined based on the first equipment and the equipment similarity matrix, and the accuracy rate of equipment fault prediction can be improved by determining the target equipment similar to the first equipment as the prediction fault equipment; meanwhile, modeling of huge number of service systems of equipment is avoided, manpower and material resources are saved, and efficiency of equipment fault prediction is improved.
Optionally, the device similarity matrix is specifically constructed by specifically including steps 1) to 2):
step 1), aiming at N types of target historical index data in the M devices, generating derivative index data corresponding to the N types of target historical index data; n is a positive integer;
and 2) constructing the equipment similarity matrix based on the N types of target historical index data and the derivative index data in the M pieces of equipment.
In this embodiment, first, N types of target history index data of each device in the service system need to be acquired.
It should be noted that, each device includes multiple types of target historical index data, such as historical network element index data, historical alarm index data, and the like.
After N types of target history index data are acquired for each device, it is necessary to derive the target history index data based on the target history index data, thereby generating derived index data corresponding to the N types of target history index data.
It can be understood that the reason for deriving the target historical index data is that in order to increase the number of target historical index data, more useful information about the devices can be mined based on the derived historical index data, so that accuracy of similarity calculation between the devices can be improved.
After the derivative index data corresponding to the N types of target history index data are generated, a device similarity matrix is constructed based on the N types of target history index data and the derivative index data in the M devices.
In the embodiment, the target historical index data is derived, so that derived index data of more related devices are generated, more useful information about the devices can be mined based on the derived index data, available data is provided for similarity calculation between subsequent devices, and the accuracy of device similarity matrix construction is improved.
Optionally, before generating derivative index data corresponding to N types of target historical index data for N types of target historical index data in the M devices, determining N types of target historical index data in the M devices based on N types of original historical index data in the M devices; the method is realized by the following steps of (1) - (3):
step [1], obtaining N types of original historical index data in the M devices;
step [2], for each device, performing aggregation processing on the original historical index data of each type in a preset period to generate N types of aggregation historical index data corresponding to the N types of original historical index data;
And step [3], determining the N types of target historical index data based on the N types of aggregation historical index data.
In this embodiment, first, N types of original history index data in M devices of the service system need to be acquired.
It will be appreciated that each device has various types of raw historical index data, such as raw network element historical index data, raw alarm historical index data, and the like.
Because the original historical index data collected by the executing subject is often in seconds, too fine a time series is not conducive to analysis.
Therefore, after N types of original historical index data in each device are acquired, time granularity aggregation needs to be performed on the original historical index data.
Specifically, for each device, aggregation processing needs to be performed on each type of original historical index data in a preset period, so as to generate N types of aggregated historical index data corresponding to the N types of original historical index data.
For example, for device 1, by collecting raw network element history index data in device 1 over a period of 1 minute: 1,12,4,7,17, 34; original alarm history index data: 3,28,4,2,7, 42.
And then, carrying out aggregation processing on the original network element historical index data, wherein the aggregation processing is used for calculating the average value, variance and the like of the original network element historical index data and the original alarm historical index data within 1 minute.
For example, the average value is utilized to aggregate the original network element history index data and the original alarm history index data within 1 minute, so as to obtain aggregate history index data corresponding to the original network element history index data within 1 minute period: 12.5; and the aggregation history index data 14.3 corresponding to the alarm history index data in the period of 1 minute.
After generating N types of aggregation history index data corresponding to the N types of original history index data, N types of target history index data may be determined based on the N types of aggregation history index data.
In the embodiment, for each device, aggregation processing is performed on each type of original historical index data in a preset period, and N types of aggregation historical index data corresponding to the N types of original historical index data are generated, so that the calculated amount in the process of constructing a device similarity matrix can be reduced, and the efficiency of device fault prediction is improved.
Optionally, the determining the N types of target historical index data based on the N types of aggregate historical index data may be specifically implemented by the following steps [3.1] -3.2 ]:
Step [3.1], for each device, calculating the similarity among N types corresponding to the aggregation history index data based on a dynamic time warping algorithm DTW;
and step [3.2], performing de-duplication processing on the N types of aggregation history index data based on the similarity among the N types, and determining the N types of target history index data.
In this embodiment, after N types of aggregation history index data corresponding to N types of original history index data are generated, since the number of aggregation history index data is large, many types of aggregation history index data are similar, and therefore, it is necessary to calculate the similarity between N types corresponding to the aggregation history index data. And then, based on the similarity among the N types, performing de-duplication processing on the N types of aggregation history index data, and determining N types of target history index data.
For example, the aggregation indicator data corresponding to the alarm type includes a mean column and a variance column; the aggregation index data corresponding to the prompt type comprises a mean value column and a variance column;
and calculating the similarity between the alarm type and the prompt type, and under the condition that the similarity reaches a threshold value, indicating that the aggregation index data corresponding to the alarm type is similar to the aggregation index data corresponding to the prompt type.
In order to further reduce the calculated amount in the process of constructing the equipment similarity matrix and improve the equipment fault prediction efficiency, the average value column and the variance column of the aggregation index data corresponding to the alarm types can be reserved, the aggregation index data corresponding to the prompt types is subjected to de-duplication processing, namely the variance column in the aggregation index data corresponding to the prompt types is deleted, and finally the target historical index data corresponding to the alarm types and the target historical index data corresponding to the prompt types are obtained.
In practical applications, the type similarity corresponding to different types of aggregation history index data can be calculated based on a dynamic time warping algorithm (Dynamic Time Warping, DTW).
Fig. 2 is a schematic diagram of calculating type similarity based on a DTW algorithm. Referring to fig. 2, upper and lower solid lines represent the aggregation history index data of the type a and the aggregation history index data of the type B. The dashed line between type a and type B represents a point of similarity between the two types of aggregation history index data. Note that, the type a and the type B are time series.
The DTW algorithm uses the sum of the distances between all of these similarity points, also known as the canonical path distance (Warp Path Distance), to measure the similarity between the aggregate history index data for type a and the aggregate history index data for type B.
In the embodiment, for each device, the similarity between N types corresponding to the aggregation history index data is calculated based on the dynamic time warping algorithm DTW, and then the aggregation history index data of the N types is deduplicated based on the similarity between the N types, so that the calculation amount in the process of building the device similarity matrix can be further reduced, and the efficiency of device fault prediction is improved.
Optionally, in one possible implementation manner of the embodiment of the present invention, the generating, for N types of target historical index data in the M devices, derivative index data corresponding to the N types of target historical index data is specifically implemented by the following steps [1] -3):
step [1], for each of the devices, determining a pair of historical index data based on a priori knowledge; the historical index data pair comprises target historical index data of a first type and target historical index data of a second type; the first type and the second type have an association relation;
step [2], carrying out nonlinear conversion on the N types of target historical index data to generate first derivative index data corresponding to the N types of target historical index data;
And step [3], multiplying the first type of target historical index data and the second type of target historical index data in each historical index data pair to generate second derivative index data corresponding to the N types of target historical index data.
In this embodiment, after determining N types of target history index data without redundancy based on N types of aggregation history index data, in order to increase the number of target history index data, thereby mining more useful information about devices, it is necessary to derive N types of aggregation history index data for each device.
The specific implementation mode is as follows:
first, for each device, it is necessary to determine target history index data having an association relationship between types.
Specifically, it is necessary to determine a pair of historical index data based on a priori knowledge; the historical index data pair comprises target historical index data of a first type and target historical index data of a second type, and the first type and the second type have an association relation.
That is, by a priori knowledge, such as domain knowledge of expert rules (manual experience), a relational index pair, such as cpu occupancy and memory usage, is set in advance, and network IO and disk usage increases in inverse proportion.
Then, the N types of target history index data are subjected to nonlinear conversion, where the nonlinear conversion may be, for example, performing operations such as squaring, cubic, and the like on the target history index data, so as to generate first derivative index data corresponding to the N types of target history index data.
For the target historical index data with the association relationship between the types, the first type of target historical index data and the second type of target historical index data with the association relationship in each historical index data pair are required to be multiplied, so that second derivative index data corresponding to the N types of target historical index data are generated.
For example, target history index data in the apparatus 1Is 2-dimensional, i.e., there are 2 types of target historical index data: />The addition of a constant term is: />The method comprises the steps of carrying out a first treatment on the surface of the For->Derivatization, i.e. of->Nonlinear conversion (i.e. squaring operation) is performed to obtain +.>The method comprises the steps of carrying out a first treatment on the surface of the At->In the case of the corresponding type having the association relation, the methodMultiplying to obtain/>The derived target history index data may be expressed asThe method comprises the steps of carrying out a first treatment on the surface of the I.e. the target history data is +.>Derived index data of->。
The generation of derived index data corresponding to N types of target history index data is described below with reference to tables 1 to 2:
Table 1 is a feature matrix without derived index data; table 2 is a feature matrix containing derived index data.
TABLE 1
TABLE 2
It will be appreciated that the derived index B in table 2 is obtained by squaring the index B column (nonlinear conversion), and the derived index BC is obtained by multiplying the index B by the index C (combination conversion).
In the embodiment, the target historical index data is derived, so that derived index data of more related devices are generated, more useful information about the devices can be mined based on the derived index data, available data is provided for similarity calculation between subsequent devices, and the accuracy of device similarity matrix construction is improved.
Optionally, in one possible implementation manner of the embodiment of the present invention, the constructing the device similarity matrix based on the N types of target historical index data and the derived index data in the M devices may be specifically implemented by the following steps [ a ] -step [ c):
step [ a ], carrying out standardization processing on the N types of target historical index data and the derivative index data in the M devices to generate feature vectors corresponding to the M devices;
Step [ b ], calculate the similarity between every said feature vector by cosine similarity algorithm;
and step [ c ], constructing the equipment similarity matrix based on the similarity between the feature vectors.
In this embodiment, first, normalization processing needs to be performed on N types of target history index data and derivative index data in M devices, and the target history index data and the derivative index data may be normalized to the [0,1] section by performing normalization processing on the target history index data and the derivative index data.
The standardization processing of the target historical index data and the derivative index data can be realized by the following formula (1):
wherein, the liquid crystal display device comprises a liquid crystal display device,representing index data->A result after standardized treatment is carried out; />Indicating the number of index data; />Indicate->And (5) index data.
In the standardization process of the target history index data and the derived index data,can generateFeature vectors corresponding to the respective devices;
in the generation ofAfter the feature vectors corresponding to the devices are calculated, the similarity among the feature vectors is calculated by using a cosine similarity algorithm; specifically, the method can be realized by the following formula (2):
Wherein, the liquid crystal display device comprises a liquid crystal display device,the i-th element in the feature vector of the two devices needing similarity comparison is represented; />Representing the total number of feature vectors.
It should be noted that if the cosine value is closer to 1, that is, the two vectors are more similar; the two devices are illustrated to be very similar. When one of them fails, the more similar devices are more likely to fail. Accordingly, the closer the cosine value is to 0, i.e., the less similar the two vectors, i.e., the less similar the two devices. The cosine value if-1 indicates that the two vectors are diametrically opposed.
After the similarity between the feature vectors is calculated, a device similarity matrix can be constructed based on the similarity between the feature vectors. Wherein table 3 shows the similarity matrix between device 1, device 2 and device 3.
TABLE 3 Table 3
In the above embodiment, the target device similar to the first device may be determined as the prediction failure device based on the device similarity matrix, so that the accuracy of device failure prediction may be improved; meanwhile, modeling of huge number of service systems of equipment is avoided, manpower and material resources are saved, and efficiency of equipment fault prediction is improved.
Optionally, in a possible implementation manner of the embodiment of the present invention, after the determining the target device as the predicted fault device, the following steps are further performed:
And generating an alarm event and a treatment strategy corresponding to the alarm event based on the equipment information corresponding to the predicted fault equipment.
In this embodiment, when a device fails, an alarm is given, and the device similarity matrix is queried and ordered according to the similarity. Other devices for which the similarity reaches a threshold are also considered potentially risky devices (i.e., predictive failed devices).
Accordingly, it is necessary to generate an alarm event and a treatment policy corresponding to the alarm event based on device information corresponding to the predicted failure device.
For example, based on the alarm event, the cause of the failure of the device is determined, then the work order system is queried for data, the historical work orders are queried for time ordering, the best solution (i.e. the disposal strategy) is obtained, and the broadcasting is performed through the event message.
In the embodiment, the alarm event and the treatment strategy corresponding to the alarm event are generated based on the equipment information corresponding to the predicted fault equipment, so that the equipment manager can timely treat the predicted fault equipment, and loss is avoided.
Fig. 3 is a second flow chart of the device failure prediction method provided by the present invention, referring to fig. 3, the method includes steps 301 to 313, in which:
And 304, performing de-duplication processing on the N types of aggregation history index data based on the similarity among the N types, and determining N types of target history index data.
Step 306, performing nonlinear conversion on the N types of target historical index data, and generating first derivative index data corresponding to the N types of target historical index data.
And 309, calculating the similarity between the feature vectors by using a cosine similarity algorithm.
And 311, monitoring the real-time index data of the M devices.
Specifically, the first device is a device with abnormal real-time index data in the M devices; the target device is a device, of which the similarity with the first device reaches a preset threshold, of the M devices.
According to the equipment fault prediction method provided by the invention, the real-time index data of M pieces of equipment in the service system are monitored, and under the condition that the real-time index data of the first equipment in the service system is monitored to be abnormal, the target equipment, of which the similarity with the first equipment in the M pieces of equipment reaches the preset threshold value, is determined based on the first equipment and the equipment similarity matrix, and the accuracy rate of equipment fault prediction can be improved by determining the target equipment similar to the first equipment as the prediction fault equipment; meanwhile, modeling of huge number of service systems of equipment is avoided, manpower and material resources are saved, and efficiency of equipment fault prediction is improved.
The device fault prediction apparatus provided by the present invention will be described below, and the device fault prediction apparatus described below and the device fault prediction method described above may be referred to correspondingly to each other. Fig. 4 is a schematic structural diagram of an apparatus for predicting an equipment failure according to the present invention, and as shown in fig. 4, the apparatus 400 for predicting an equipment failure includes: a monitoring module 401 and a determining module 402, wherein:
the monitoring module 401 is configured to monitor real-time index data of M devices in the service system;
a first determining module 402, configured to determine, when it is monitored that the real-time index data of a first device in the M devices is abnormal, a target device from the M devices based on the first device and a device similarity matrix that is constructed in advance; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer.
According to the equipment fault prediction device, the real-time index data of M pieces of equipment in the service system are monitored, and when the condition that the real-time index data of the first equipment in the service system is abnormal is monitored, the target equipment, of which the similarity with the first equipment reaches the preset threshold value, in the M pieces of equipment is determined based on the first equipment and the equipment similarity matrix, and the target equipment similar to the first equipment is determined to be the prediction fault equipment, so that the accuracy rate of equipment fault prediction can be improved; meanwhile, modeling of huge number of service systems of equipment is avoided, manpower and material resources are saved, and efficiency of equipment fault prediction is improved.
Optionally, the apparatus further comprises:
the first generation module is used for generating derivative index data corresponding to N types of target historical index data in the M devices aiming at the N types of target historical index data; n is a positive integer;
the construction module is used for constructing the equipment similarity matrix based on the N types of target historical index data and the derivative index data in the M pieces of equipment.
Optionally, the generating module is further configured to:
determining, for each of the devices, a pair of historical index data based on a priori knowledge; the historical index data pair comprises target historical index data of a first type and target historical index data of a second type; the first type and the second type have an association relation;
Nonlinear conversion is carried out on the N types of target historical index data, and first derivative index data corresponding to the N types of target historical index data are generated;
multiplying the first type of target historical index data and the second type of target historical index data in each historical index data pair to generate second derivative index data corresponding to the N types of target historical index data.
Optionally, the apparatus further comprises:
the acquisition module is used for acquiring N types of original historical index data in the M devices;
the second generation module is used for carrying out aggregation processing on the original historical index data of each type in a preset period of time for each device to generate N types of aggregation historical index data corresponding to the N types of original historical index data;
and the second determining module is used for determining the N types of target historical index data based on the N types of aggregation historical index data.
Optionally, the second determining module is further configured to:
calculating the similarity among N types corresponding to the aggregation history index data based on a dynamic time warping algorithm DTW for each device;
And performing de-duplication processing on the N types of aggregation history index data based on the similarity among the N types, and determining the N types of target history index data.
Optionally, the building module is further configured to:
performing standardization processing on the N types of target historical index data and the derivative index data in the M devices to generate feature vectors corresponding to the M devices;
calculating the similarity between the feature vectors by using a cosine similarity algorithm;
and constructing the equipment similarity matrix based on the similarity between the feature vectors.
Optionally, the apparatus further comprises:
and the third generation module is used for generating an alarm event and a treatment strategy corresponding to the alarm event based on the equipment information corresponding to the prediction fault equipment.
Fig. 5 is a schematic structural diagram of an electronic device according to the present invention, and as shown in fig. 5, the electronic device may include: processor 510, communication interface (Communications Interface) 520, memory 530, and communication bus 540, wherein processor 510, communication interface 520, memory 530 complete communication with each other through communication bus 540. Processor 510 may invoke logic instructions in memory 530 to perform a device failure prediction method comprising: monitoring real-time index data of M devices in a service system; under the condition that the abnormality of the real-time index data of a first device in the M devices is monitored, determining a target device from the M devices based on the first device and a pre-constructed device similarity matrix; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer.
Further, the logic instructions in the memory 530 described above may be implemented in the form of software functional units and may be stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of performing the apparatus failure prediction method provided by the methods above, the method comprising: monitoring real-time index data of M devices in a service system; under the condition that the abnormality of the real-time index data of a first device in the M devices is monitored, determining a target device from the M devices based on the first device and a pre-constructed device similarity matrix; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the apparatus failure prediction method provided by the above methods, the method comprising: monitoring real-time index data of M devices in a service system; under the condition that the abnormality of the real-time index data of a first device in the M devices is monitored, determining a target device from the M devices based on the first device and a pre-constructed device similarity matrix; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (8)
1. A method for predicting equipment failure, comprising:
monitoring real-time index data of M devices in a service system;
under the condition that the abnormality of the real-time index data of a first device in the M devices is monitored, determining a target device from the M devices based on the first device and a pre-constructed device similarity matrix; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer;
wherein the device similarity matrix is constructed by:
generating derivative index data corresponding to N types of target historical index data in the M devices aiming at the N types of target historical index data; n is a positive integer;
performing standardization processing on the N types of target historical index data and the derivative index data in the M devices to generate feature vectors corresponding to the M devices;
calculating the similarity between the feature vectors by using a cosine similarity algorithm;
and constructing the equipment similarity matrix based on the similarity between the feature vectors.
2. The apparatus failure prediction method according to claim 1, wherein the generating derivative index data corresponding to N types of target history index data for N types of target history index data in the M apparatuses includes:
determining, for each of the devices, a pair of historical index data based on a priori knowledge; the historical index data pair comprises target historical index data of a first type and target historical index data of a second type; the first type and the second type have an association relation;
nonlinear conversion is carried out on the N types of target historical index data, and first derivative index data corresponding to the N types of target historical index data are generated;
multiplying the first type of target historical index data and the second type of target historical index data in each historical index data pair to generate second derivative index data corresponding to the N types of target historical index data.
3. The device failure prediction method according to claim 1, wherein before the generating derivative index data corresponding to N types of target history index data for N types of target history index data in the M devices, the method further comprises:
Acquiring N types of original historical index data in the M devices;
for each device, performing aggregation processing on the original historical index data of each type within a preset period to generate N types of aggregation historical index data corresponding to the N types of original historical index data;
and determining the N types of target historical index data based on the N types of aggregation historical index data.
4. The equipment failure prediction method according to claim 3, wherein the determining the N types of target history index data based on the N types of aggregate history index data includes:
calculating the similarity among N types corresponding to the aggregation history index data based on a dynamic time warping algorithm DTW for each device;
and performing de-duplication processing on the N types of aggregation history index data based on the similarity among the N types, and determining the N types of target history index data.
5. The device failure prediction method according to claim 1, characterized in that after the determination of a target device from the M devices, the method further comprises:
And generating an alarm event and a treatment strategy corresponding to the alarm event based on the equipment information corresponding to the target equipment.
6. An apparatus for predicting a failure of a device, comprising:
the monitoring module is used for monitoring real-time index data of M devices in the service system;
the determining module is used for determining target equipment from the M pieces of equipment based on the first equipment and a pre-constructed equipment similarity matrix under the condition that the real-time index data of the first equipment in the M pieces of equipment are monitored to be abnormal; the target device is a device, of which the similarity with the first device in the M devices reaches a preset threshold; the target equipment is predictive failure equipment; the device similarity matrix is used for representing the similarity among the M devices; m is a positive integer;
wherein the device similarity matrix is constructed by:
generating derivative index data corresponding to N types of target historical index data in the M devices aiming at the N types of target historical index data; n is a positive integer;
performing standardization processing on the N types of target historical index data and the derivative index data in the M devices to generate feature vectors corresponding to the M devices;
Calculating the similarity between the feature vectors by using a cosine similarity algorithm;
and constructing the equipment similarity matrix based on the similarity between the feature vectors.
7. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the device failure prediction method of any of claims 1 to 5 when the program is executed by the processor.
8. A non-transitory computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when executed by a processor, implements the device failure prediction method according to any one of claims 1 to 5.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310396139.1A CN116149971B (en) | 2023-04-14 | 2023-04-14 | Equipment fault prediction method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310396139.1A CN116149971B (en) | 2023-04-14 | 2023-04-14 | Equipment fault prediction method and device, electronic equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN116149971A CN116149971A (en) | 2023-05-23 |
CN116149971B true CN116149971B (en) | 2023-07-04 |
Family
ID=86339171
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310396139.1A Active CN116149971B (en) | 2023-04-14 | 2023-04-14 | Equipment fault prediction method and device, electronic equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116149971B (en) |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6107250B2 (en) * | 2013-03-12 | 2017-04-05 | 株式会社リコー | Failure prediction system, failure prediction device, and failure prediction program |
US10884404B2 (en) * | 2017-09-01 | 2021-01-05 | DOOSAN Heavy Industries Construction Co., LTD | Method of predicting plant data and apparatus using the same |
CN112330060A (en) * | 2020-11-25 | 2021-02-05 | 新智数字科技有限公司 | Equipment fault prediction method and device, readable storage medium and electronic equipment |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108022020B (en) * | 2017-12-15 | 2020-09-18 | 东软集团股份有限公司 | Equipment fault management method and device, storage medium and electronic equipment |
CN114648050A (en) * | 2020-12-02 | 2022-06-21 | 新智数字科技有限公司 | Equipment fault prediction method and device, readable storage medium and electronic equipment |
CN113255795B (en) * | 2021-06-02 | 2021-10-26 | 杭州安脉盛智能技术有限公司 | Equipment state monitoring method based on multi-index cluster analysis |
CN114676746A (en) * | 2022-02-18 | 2022-06-28 | 上海交通大学 | Method and system for predicting failure probability of power equipment under natural disasters |
-
2023
- 2023-04-14 CN CN202310396139.1A patent/CN116149971B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP6107250B2 (en) * | 2013-03-12 | 2017-04-05 | 株式会社リコー | Failure prediction system, failure prediction device, and failure prediction program |
US10884404B2 (en) * | 2017-09-01 | 2021-01-05 | DOOSAN Heavy Industries Construction Co., LTD | Method of predicting plant data and apparatus using the same |
CN112330060A (en) * | 2020-11-25 | 2021-02-05 | 新智数字科技有限公司 | Equipment fault prediction method and device, readable storage medium and electronic equipment |
Also Published As
Publication number | Publication date |
---|---|
CN116149971A (en) | 2023-05-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10914608B2 (en) | Data analytic engine towards the self-management of complex physical systems | |
CN110070461B (en) | Health degree evaluation method and system for electric power information system | |
US20190079965A1 (en) | Apparatus and method for real time analysis, predicting and reporting of anomalous database transaction log activity | |
CN111045894B (en) | Database abnormality detection method, database abnormality detection device, computer device and storage medium | |
US20170372212A1 (en) | Model based root cause analysis | |
CN110570544A (en) | method, device, equipment and storage medium for identifying faults of aircraft fuel system | |
CN113837596B (en) | Fault determination method and device, electronic equipment and storage medium | |
Si et al. | Fault prediction model based on evidential reasoning approach | |
WO2019019429A1 (en) | Anomaly detection method, device and apparatus for virtual machine, and storage medium | |
CN114138601A (en) | Service alarm method, device, equipment and storage medium | |
Wang et al. | Incremental causal graph learning for online root cause analysis | |
CN116149971B (en) | Equipment fault prediction method and device, electronic equipment and storage medium | |
CN117170915A (en) | Data center equipment fault prediction method and device and computer equipment | |
CN111783883A (en) | Abnormal data detection method and device | |
JP7062505B2 (en) | Equipment management support system | |
CN116228312A (en) | Processing method and device for large-amount point exchange behavior | |
Santos et al. | An empirical study on the influence of context in computing thresholds for Chidamber and Kemerer metrics. | |
CN114881503A (en) | Scoring determination method, device, equipment and storage medium | |
CN114722025A (en) | Data prediction method, device and equipment based on prediction model and storage medium | |
CN114760190A (en) | Service-oriented converged network performance anomaly detection method | |
CN113064812A (en) | Project development process quality defect prediction method, device and medium | |
CN115599077B (en) | Vehicle fault delimiting method and device, electronic equipment and storage medium | |
CN115001997B (en) | Extreme value theory-based smart city network equipment performance abnormal threshold evaluation method | |
Malhotra | Prediction of High-, Medium-, and Low-Severity Faults Using Software Metrics | |
CN117971536A (en) | Abnormal data processing method and device, electronic equipment and 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 |