CN114996110B - Deep inspection optimization method and system based on micro-service architecture - Google Patents
Deep inspection optimization method and system based on micro-service architecture Download PDFInfo
- Publication number
- CN114996110B CN114996110B CN202210701767.1A CN202210701767A CN114996110B CN 114996110 B CN114996110 B CN 114996110B CN 202210701767 A CN202210701767 A CN 202210701767A CN 114996110 B CN114996110 B CN 114996110B
- Authority
- CN
- China
- Prior art keywords
- inspection
- data
- micro
- layer
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F11/3447—Performance evaluation by modeling
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F11/3452—Performance evaluation by statistical analysis
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Probability & Statistics with Applications (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a deep inspection optimization method and system based on a micro-service architecture, and belongs to the technical field of artificial intelligence. The method comprises the following steps: building a health inspection model, and transversely layering a micro-service architecture system; inputting the fault history data into the health inspection model to obtain the probability of fault in inspection in the next period of physical equipment corresponding to each module of the transverse layer; preprocessing the inspection data through secondary interpolation to construct a depth inspection model; inputting the pretreated inspection data into the deep inspection model to obtain an actual value; and comparing the fitting value with the actual value generated by the deep inspection model operation to evaluate possible data defects and hidden functional faults of the micro-service architecture system. The invention can integrally evaluate possible data defects and hidden functional faults of the micro-service system architecture, so that the machine can take the place of people to make decisions, thereby realizing full automation and improving the efficiency of fault detection.
Description
Technical Field
The invention belongs to the technical field of artificial intelligence, and particularly relates to a deep inspection optimization method and system based on a micro-service architecture.
Background
With the rapid development of computer technology, information networks have become an important guarantee of social development. In the micro-service architecture system, in order to ensure that the functions of each layer are normal, each layer is monitored and regularly inspected in an inspection mode, so that the health problems of a physical equipment layer and an application layer of the micro-service architecture system are effectively solved. However, a great improvement space is still provided, and only the problem of deeper dimension of health inspection is well solved, the operation and maintenance data can be used as a drive, an AI algorithm is used as a core, professional operation and maintenance service modules such as infrastructure monitoring, fault accurate positioning and intelligent processing, 3D digital twin, management cockpit and the like are covered, and the huge monomer type application is decomposed into a plurality of services through a micro service architecture mode.
With large-scale application of big data and AI, how to ensure stable running environment of each layer of the micro-service architecture system is a technical bottleneck faced today.
Disclosure of Invention
The invention aims to overcome the defects and shortcomings of the prior art, and provides a depth inspection optimization method and a system based on a micro-service architecture, which can automatically realize the depth inspection of each layer in the micro-service architecture by transversely layering the micro-service architecture system and constructing a health inspection model and a depth inspection model, and can integrally evaluate possible data defects and hidden function faults of the micro-service architecture by comparing actual values generated by the operation of the depth inspection model with fitting values calculated by a ridge regression method, so that a machine can take a decision instead of a person, thereby realizing full automation and improving the fault detection efficiency.
According to one aspect of the invention, the invention provides a deep inspection optimization method based on a micro-service architecture, which comprises the following steps:
S1: building a health inspection model, and transversely layering a micro-service architecture system; acquiring physical equipment information corresponding to each module of a transverse layer and fault history data of the physical equipment, and inputting the fault history data into the health inspection model to obtain the probability of fault occurrence of inspection in the next period of the physical equipment corresponding to each module of the transverse layer;
S2: preprocessing the inspection data through secondary interpolation to construct a depth inspection model; inputting the pretreated inspection data into the deep inspection model to obtain an actual value; calculating to obtain a fitting value by a ridge regression method;
s3: and comparing the fitting value with the actual value generated by the deep inspection model operation to evaluate possible data defects and hidden functional faults of the micro-service architecture system.
Preferably, the building the health inspection model, and the horizontally layering the micro-service architecture system includes:
the micro-service architecture system is transversely divided into seven layers: the system comprises a basic environment layer, a supervision and management layer, a cloud platform, an information resource layer, a service layer, an interface layer and a U I display layer;
the health inspection model comprises a Markov transfer matrix algorithm model: x (k+1) =x (k) ×p
Wherein: x (k) represents a state vector of the trend analysis and prediction object at time t=k, P represents a one-step transition probability matrix, and X (k+1) represents a state vector of the trend analysis and prediction object at time t=k+1.
Preferably, the preprocessing the inspection data through secondary interpolation includes:
and carrying out differential processing on data which are unevenly sampled by different nodes, and carrying out interpolation by adopting a secondary differential method and every 3 adjacent points to obtain secondary interpolation, wherein the formula is as follows:
wherein x is the current value of the classified object; y is 3 adjacent points of the classified object; i is the sequence number of the classification object.
Preferably, the constructing the depth inspection model includes:
For the time dimension, carrying out time sequence coding by PositionEncoding, and utilizing the attribute to discover the characteristic association of the time dimension, wherein the formula is as follows:
PositionEncoding＝cos2(pos/N)
for the space dimension, different multi-space dimension features are extracted, and the multi-space dimension features are fused through MultiHead, wherein the formula is as follows:
Headi＝Attention(Qi,Ki,Vi)
MultiHead(Q,K,V)＝Concat(head1,...,headh)*WO
Wherein d k represents the dimension of K; n is the adjustable length; q is query feature map; k is feature mapping to be matched; v is the mapping of the monitoring data; headi is the result of time attention, qi is the i-th set of query feature maps; ki is the i-th set of feature mapping to be matched; vi is the i-th group monitoring data mapping; WO is a feature fusion matrix; concat is feature cascade fusion; pos is a data sequence number; positionEncoding is position sequence coding; multiHead is a multi-headed feature fusion.
Preferably, the calculating the fitting value by the ridge regression method includes:
The model formula of the ridge regression method is as follows: x theta-y| 2+||Γθ||2
The formula of the anti-overfitting operation is as follows: θ (a) = (X TX+aI)-1XT y)
Wherein X represents an input; y represents the output prediction result; i represents a canonical operation; i represents an identity matrix; θ is a fitting hyper-parameter; Γ is a weight constant; a is the weight of the identity matrix; θ (a) represents a value of θ obtained when a is determined.
According to another aspect of the present invention, there is also provided a deep inspection optimization system based on a micro-service architecture, the system comprising:
The first prediction module is used for constructing a health inspection model and transversely layering the micro-service architecture system; acquiring physical equipment information corresponding to each module of a transverse layer and fault history data of the physical equipment, and inputting the fault history data into the health inspection model to obtain the probability of fault occurrence of inspection in the next period of the physical equipment corresponding to each module of the transverse layer;
The second prediction module is used for preprocessing the inspection data through secondary interpolation and constructing a depth inspection model; inputting the pretreated inspection data into the deep inspection model to obtain an actual value; calculating to obtain a fitting value by a ridge regression method;
and the comparison and evaluation module is used for comparing the fitting value with the actual value generated by the deep inspection model operation and evaluating possible data defects and hidden function faults of the micro-service architecture system.
Preferably, the first prediction module builds a health inspection model, and the lateral layering of the micro-service architecture system includes:
The micro-service architecture system is transversely divided into seven layers: the system comprises a basic environment layer, a supervision and management layer, a cloud platform, an information resource layer, a service layer, an interface layer and a UI display layer;
the health inspection model comprises a Markov transfer matrix algorithm model: x (k+1) =x (k) ×p
Wherein: x (k) represents a state vector of the trend analysis and prediction object at time t=k, P represents a one-step transition probability matrix, and X (k+1) represents a state vector of the trend analysis and prediction object at time t=k+1.
Preferably, the preprocessing the patrol data by the second prediction module through quadratic interpolation includes:
and carrying out differential processing on data which are unevenly sampled by different nodes, and carrying out interpolation by adopting a secondary differential method and every 3 adjacent points to obtain secondary interpolation, wherein the formula is as follows:
Wherein x is the current value of the classified object, y is 3 adjacent points of the classified object, and i is the sequence number of the classified object.
Preferably, the second prediction module builds a depth inspection model including:
For the time dimension, carrying out time sequence coding by PositionEncoding, and utilizing the attribute to discover the characteristic association of the time dimension, wherein the formula is as follows:
PositionEncoding＝cos2(pos/N)
for the space dimension, different multi-space dimension features are extracted, and the multi-space dimension features are fused through MultiHead, wherein the formula is as follows:
Headi＝Attention(Qi,Ki,Vi)
MultiHead(Q,K,V)＝Concat(head1,...,headh)*WO
Wherein d k represents the dimension of K; n is the adjustable length; q is query feature map; k is feature mapping to be matched; v is the mapping of the monitoring data; head i is the result of time attention, Q i is the i-th set of query feature map; ki is the i-th set of feature mapping to be matched; vi is the i-th group monitoring data mapping; WO is a feature fusion matrix; concat is feature cascade fusion; pos is a data sequence number; positionEncoding is position sequence coding; multiHead is a multi-headed feature fusion.
Preferably, the calculating, by the second prediction module, the fitting value by the ridge regression method includes:
The model formula of the ridge regression method is as follows: x theta-y| 2+||Γθ||2
The formula of the anti-overfitting operation is as follows: θ (a) = (X TX+aI)-1XT y)
Wherein X represents an input; y represents the output prediction result; i represents a canonical operation; i represents an identity matrix; θ is a fitting hyper-parameter; Γ is a weight constant; a is the weight of the identity matrix; θ (a) represents a value of θ obtained when a is determined.
The beneficial effects are that: the invention carries out deep health inspection on the major classes and subclasses of a micro-service architecture system transverse basic environment layer, a supervision management layer, a cloud platform, an information resource layer, a service layer, an interface layer and a UI display layer; meanwhile, through the combination technology of a plurality of algorithm models in the artificial intelligence field, the defect that the micro-service system architecture possibly exists and the loss possibly caused by hidden function faults to an application platform are comprehensively evaluated by making up the angles that the prior art of inspection is too much in importance to physical equipment and application detection and lacks the comparison between fitting values of all levels tested through simulation data and actual values generated by the operation of the deep inspection model on all transverse layers, so that a machine can take a decision instead of a person, and the realization of full automation and intelligence is possible in a real sense.
Features and advantages of the present invention will become apparent by reference to the following drawings and detailed description of embodiments of the invention.
Drawings
FIG. 1 is a flow chart of a deep inspection optimization method based on a microservice architecture;
Fig. 2 is a schematic diagram of a deep inspection optimization system based on a micro-service architecture.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention. 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.
Example 1
Fig. 1 is a flow chart of a deep patrol optimization method based on a micro-service architecture. As shown in fig. 1, the invention provides a deep inspection optimization method based on a micro-service architecture, which comprises the following steps:
S1: building a health inspection model, and transversely layering a micro-service architecture system; and acquiring physical equipment information corresponding to each module of the transverse layer and fault history data of the physical equipment, and inputting the fault history data into the health inspection model to obtain the probability of fault occurrence of inspection in the next period of the physical equipment corresponding to each module of the transverse layer.
Preferably, the building the health inspection model, and the horizontally layering the micro-service architecture system includes:
the micro-service architecture system is transversely divided into seven layers: the system comprises a basic environment layer, a supervision and management layer, a cloud platform, an information resource layer, a service layer, an interface layer and a U I display layer;
the health inspection model comprises a Markov transfer matrix algorithm model: x (k+1) =x (k) ×p
Wherein: x (k) represents a state vector of the trend analysis and prediction object at time t=k, P represents a one-step transition probability matrix, and X (k+1) represents a state vector of the trend analysis and prediction object at time t=k+1.
Specifically, first, the micro-service architecture system is laterally divided into seven layers: the inspection model of the basic environment layer, the supervision and management layer, the cloud platform, the information resource layer, the business layer, the interface layer and the U I display layer is constructed.
And secondly, obtaining the physical equipment information corresponding to each layer of module and the sub-module through the relational database of the information resource layer.
And then, acquiring fault history data of the physical equipment corresponding to each sub-module of each transverse level of the technical architecture from a history fault database.
And finally, the data are participated in the operation of the health inspection model, and the operation result is the probability of the inspection failure of the current transverse layer module corresponding to the physical equipment in the lower period.
Wherein, the Markov transfer matrix algorithm model formula: x (k+1) =x (k) ×p
In the formula: x (k) represents a state vector of the trend analysis and prediction object at time t=k, P represents a one-step transition probability matrix, and X (k+1) represents a state vector of the trend analysis and prediction object at time t=k+1.
The current layer control index of the inspection is 100, wherein 30 percent of the current layer control index is normal and 70 percent of the current layer control index is abnormal, 60 percent of the normal monitoring index is likely to continue to be normal, and 40 percent of the current layer control index is likely to be converted into 0.6 and 0.4 percent of the current layer control index is abnormal
70% Of the normal monitoring indexes of the current layer of the inspection are probably still normal, 30% of the normal monitoring indexes are probably converted into anomalies [0.3 and 0.7], the abnormal index transition probabilities [0.6 and 0.4] of the current layer of the inspection, the normal index transition probabilities [0.3 and 0.7] of the current layer of the inspection, and the fault occurrence probability of the current layer of the next inspection
=30X0.6+30x0.7=39, probability of normal occurrence=30x0.4+70x0.7=61.
And calculating 39% of fault occurrence probability and 61% of normal occurrence probability of each transverse layer of inspection in the next period through a model. Repeating the model to obtain the occurrence probability of the next inspection of other transverse layers.
S2: preprocessing the inspection data through secondary interpolation to construct a depth inspection model; inputting the pretreated inspection data into the deep inspection model to obtain an actual value; fitting values are calculated by a ridge regression method.
Preferably, the preprocessing the inspection data through secondary interpolation includes:
and carrying out differential processing on data which are unevenly sampled by different nodes, and carrying out interpolation by adopting a secondary differential method and every 3 adjacent points to obtain secondary interpolation, wherein the formula is as follows:
Wherein x is the current value of the classified object, y is 3 adjacent points of the classified object, and i is the sequence number of the classified object.
Specifically, the depth inspection model based on the AI algorithm is carried out on six layers in the transverse direction except for the basic environment layer (mainly composed of physical devices), and whether the core technology of each module has stronger generalization capability is detected by combining the fitting algorithm. Generalization capability (generalization ability) refers to the ability of a machine learning algorithm to adapt to fresh samples. The purpose of learning is to learn the law underlying the data, and for data outside the learning set with the same law, the trained network can also give appropriate output, which is called generalization capability.
A depth inspection model is constructed by adopting a transducer technology, and inspection data can be more accurate before operation by adopting a secondary interpolation technology carried by the transducer technology.
Firstly, for adapting to model processing, differential processing is carried out on data which are unevenly sampled by different nodes. And (3) performing interpolation by adopting a secondary difference method and every 3 adjacent points to obtain secondary interpolation. I.e. data after the artificial intelligence algorithm is optimized. The method has the advantages that:
1. The intervals are uniform and more matched with the transducer timing process.
2. And the simulation real (current module input type data, longitudinal correlation module input type data and current module history monitoring normal acquisition data) scene missing data of each horizontal level in the inspection process is truly restored.
Preferably, the constructing the depth inspection model includes:
For the time dimension, carrying out time sequence coding by PositionEncoding, and utilizing the attribute to discover the characteristic association of the time dimension, wherein the formula is as follows:
PositionEncoding＝cos2(pos/N)
for the space dimension, different multi-space dimension features are extracted, and the multi-space dimension features are fused through MultiHead, wherein the formula is as follows:
Headi＝Attention(Qi,Ki,Vi)
MultiHead(Q,K,V)＝Concat(head1,...,headh)*WO
Wherein d k represents the dimension of K; n is the adjustable length; q is query feature map; k is feature mapping to be matched; v is the mapping of the monitoring data; headi is the result of time attention, qi is the i-th set of query feature maps; ki is the i-th set of feature mapping to be matched; vi is the i-th group monitoring data mapping; WO is a feature fusion matrix; concat is feature cascade fusion; pos is a data sequence number; positionEncoding is position sequence coding; multiHead is a multi-headed feature fusion.
Specifically, the data after the secondary interpolation is accurately input into a depth inspection model for simulation operation, and the result is expressed as a predicted fault occurrence probability value after the depth inspection of the current transverse layer module in the micro-service technology architecture system.
Preferably, the calculating the fitting value by the ridge regression method includes:
The model formula of the ridge regression method is as follows: x theta-y| 2+||Γθ||2
The formula of the anti-overfitting operation is as follows: θ (a) = (X TX+aI)-1XT y)
Wherein, X represents input, y represents output prediction result, I represents regular operation, and I represents identity matrix; θ is a fitting hyper-parameter; Γ is a weight constant; a is the weight of the identity matrix; θ (a) represents a value of θ obtained when a is determined.
Specifically, whether the functions of the modules at each level are abnormal or not and whether the modules have stronger generalization capability or not are detected by a ridge regression method. The reason for the overfitting in machine learning may be the following: (1) the data is noisy (obsolete data); (2) insufficient training data, limited training data; (3) training the model excessively results in a very complex model. Model overfitting is thus prevented by the ridge regression method.
In order to solve the above problem, the present embodiment converts the unfit problem into the proper problem: a regularization term is added to the loss function.
S3: and comparing the fitting value with the actual value generated by the deep inspection model operation to evaluate possible data defects and hidden functional faults of the micro-service architecture system.
Specifically, the fitting value is compared with the total inspection ratio (hereinafter referred to as actual value) of the actual service history inspection normal data of the current level- > submodule, that is, the fitting value-actual value=difference value, if the difference value is within 10%, the error of the operation result of the current model simulation inspection data and the actual service inspection data passing through the depth inspection model is smaller. If more than 10% is judged to be in the overfitting state, relevant technicians are required to be informed of functional inspection of the level and the module where the inspected object is located, including background inspection of recorded data, analysis of historical data, whether related other modules or level business are abnormal or not, and the like.
Preferably, the present embodiment further includes the steps of: and generating an intelligent inspection report.
Specifically, the inspection results of the sub-modules in each layer are summarized, and a report is generated, as shown in table 1. Thereby completing the whole intelligent deep inspection process. The intelligent report can comprehensively understand the monitoring index of the micro-service technology architecture from the equipment to the comprehensive health monitoring of whether the functions of each layer normally run. The method makes up for the fact that in the inspection prior art, the physical equipment and application detection are emphasized, and the fact that the angle that fitting values of all levels tested through simulation data are compared with actual values generated by operation of a depth inspection model of all transverse layers is lacked, so that possible data defects of a micro-service technology architecture are evaluated integrally, and possible loss of functional faults to an application platform is hidden.
Table 1 inspection report
According to the embodiment, the health inspection model and the depth inspection model are constructed by transversely layering the micro-service architecture system, the depth inspection of each layer in the micro-service architecture system can be automatically realized, meanwhile, the data defects and hidden function faults possibly existing in the micro-service architecture can be integrally evaluated by comparing the actual value generated by the operation of the depth inspection model with the fitting value calculated by the ridge regression method, so that a machine can take a decision instead of a person, the full automation is realized, and the fault detection efficiency is improved.
Example 2
Fig. 2 is a schematic diagram of a deep inspection optimization system based on a micro-service architecture. As shown in fig. 2, the present invention further provides a deep inspection optimization system based on a micro-service architecture, where the system includes:
The first prediction module 201 is configured to construct a health inspection model, and perform lateral layering on the micro-service architecture system; acquiring physical equipment information corresponding to each module of a transverse layer and fault history data of the physical equipment, and inputting the fault history data into the health inspection model to obtain the probability of fault occurrence of inspection in the next period of the physical equipment corresponding to each module of the transverse layer;
the second prediction module 202 is configured to pre-process the inspection data through secondary interpolation, and construct a depth inspection model; inputting the pretreated inspection data into the deep inspection model to obtain an actual value; calculating to obtain a fitting value by a ridge regression method;
And the comparison and evaluation module 203 is configured to compare the fitting value with the actual value generated by the operation of the deep inspection model, and evaluate possible data defects and hidden functional faults of the micro-service architecture system.
Preferably, the first prediction module 201 constructs a health inspection model, and the lateral layering of the micro-service architecture system includes:
the micro-service architecture system is transversely divided into seven layers: the system comprises a basic environment layer, a supervision and management layer, a cloud platform, an information resource layer, a service layer, an interface layer and a U I display layer;
the health inspection model comprises a Markov transfer matrix algorithm model: x (k+1) =x (k) ×p
Wherein: x (k) represents a state vector of the trend analysis and prediction object at time t=k, P represents a one-step transition probability matrix, and X (k+1) represents a state vector of the trend analysis and prediction object at time t=k+1.
Preferably, the preprocessing the patrol data by the second prediction module 202 includes:
and carrying out differential processing on data which are unevenly sampled by different nodes, and carrying out interpolation by adopting a secondary differential method and every 3 adjacent points to obtain secondary interpolation, wherein the formula is as follows:
Wherein x is the current value of the classified object, y is 3 adjacent points of the classified object, and i is the sequence number of the classified object.
Preferably, the second prediction module 202 constructs a depth inspection model including:
For the time dimension, carrying out time sequence coding by PositionEncoding, and utilizing the attribute to discover the characteristic association of the time dimension, wherein the formula is as follows:
PositionEncoding＝cos2(pos/N)
for the space dimension, different multi-space dimension features are extracted, and the multi-space dimension features are fused through MultiHead, wherein the formula is as follows:
Headi＝Attention(Qi,Ki,Vi)
MultiHead(Q,K,V)＝Concat(head1,...,headh)*WO
Wherein d k represents the dimension of K; n is the adjustable length; q is query feature map; k is feature mapping to be matched; v is the mapping of the monitoring data; headi is the result of time attention, qi is the i-th set of query feature maps; ki is the i-th set of feature mapping to be matched; vi is the i-th group monitoring data mapping; WO is a feature fusion matrix; concat is feature cascade fusion; pos is a data sequence number; positionEncoding is position sequence coding; multiHead is a multi-headed feature fusion.
Preferably, the calculating the fitting value by the second prediction module 202 through a ridge regression method includes:
The model formula of the ridge regression method is as follows: x theta-y| 2+||Γθ||2
The formula of the anti-overfitting operation is as follows: θ (a) = (X TX+aI)-1XT y)
Wherein, X represents input, y represents output prediction result, I represents regular operation, and I represents identity matrix; θ is a fitting hyper-parameter; Γ is a weight constant; a is the weight of the identity matrix; θ (a) represents a value of θ obtained when a is determined.
The specific implementation process of the functions implemented by each module in this embodiment 2 is the same as that in embodiment 1, and will not be described here again.
The foregoing description is only of the preferred embodiments of the present invention and is not intended to limit the scope of the invention, and all equivalent structural changes made by the specification and drawings of the present invention or direct/indirect application in other related technical fields are included in the scope of the present invention.
Claims (6)
1. The deep inspection optimization method based on the micro-service architecture is characterized by comprising the following steps of:
S1: building a health inspection model, and transversely layering a micro-service architecture system; acquiring physical equipment information corresponding to each module of a transverse layer and fault history data of the physical equipment, and inputting the fault history data into the health inspection model to obtain the probability of fault occurrence of inspection in the next period of the physical equipment corresponding to each module of the transverse layer;
S2: preprocessing the inspection data through secondary interpolation to construct a depth inspection model; inputting the pretreated inspection data into the deep inspection model to obtain an actual value; calculating to obtain a fitting value by a ridge regression method;
s3: comparing the fitting value with the actual value generated by the deep inspection model operation to evaluate possible data defects and hidden functional faults of the micro-service architecture system;
the constructing the depth inspection model comprises the following steps:
For the time dimension, carrying out time sequence coding by PositionEncoding, and utilizing the attribute to discover the characteristic association of the time dimension, wherein the formula is as follows:
PositionEncoding＝cos2(pos/N)
for the space dimension, different multi-space dimension features are extracted, and the multi-space dimension features are fused through MultiHead, wherein the formula is as follows:
Headi＝Attention(Qi,Ki,Vi)
MultiHead(Q,K,V)＝Concat(head1,...,headh)*WO
Wherein d k represents the dimension of K; n is the adjustable length; q is query feature map; k is feature mapping to be matched; v is the mapping of the monitoring data; headi is the result of time attention, qi is the i-th set of query feature maps; ki is the i-th set of feature mapping to be matched; vi is the i-th group monitoring data mapping; WO is a feature fusion matrix; concat is feature cascade fusion; pos is a data sequence number; positionEncoding is position sequence coding; multiHead is multi-head feature fusion;
the fitting value obtained by calculation through the ridge regression method comprises the following steps:
The model formula of the ridge regression method is as follows: x theta-y| 2+||Γθ||2
The formula of the anti-overfitting operation is as follows: θ (a) = (X TX+aI)-1XT y)
Wherein X represents an input; y represents the output prediction result; i represents a canonical operation; i represents an identity matrix; θ is a fitting hyper-parameter; Γ is a weight constant; a is the weight of the identity matrix; θ (a) represents a value of θ obtained when a is determined.
2. The method of claim 1, wherein constructing the health inspection model and laterally layering the micro-service architecture system comprises:
The micro-service architecture system is transversely divided into seven layers: the system comprises a basic environment layer, a supervision and management layer, a cloud platform, an information resource layer, a service layer, an interface layer and a UI display layer;
The health inspection model comprises a Markov transfer matrix algorithm model: x (k+1) =x (k) ×p wherein: x (k) represents a state vector of the trend analysis and prediction object at time t=k, P represents a one-step transition probability matrix, and X (k+1) represents a state vector of the trend analysis and prediction object at time t=k+1.
3. The method of claim 1, wherein preprocessing the patrol data by quadratic interpolation comprises:
and carrying out differential processing on data which are unevenly sampled by different nodes, and carrying out interpolation by adopting a secondary differential method and every 3 adjacent points to obtain secondary interpolation, wherein the formula is as follows:
wherein x is the current value of the classified object; y is 3 adjacent points of the classified object; i is the sequence number of the classification object.
4. A micro-service architecture-based deep inspection optimization system, the system comprising:
The first prediction module is used for constructing a health inspection model and transversely layering the micro-service architecture system; acquiring physical equipment information corresponding to each module of a transverse layer and fault history data of the physical equipment, and inputting the fault history data into the health inspection model to obtain the probability of fault occurrence of inspection in the next period of the physical equipment corresponding to each module of the transverse layer;
The second prediction module is used for preprocessing the inspection data through secondary interpolation and constructing a depth inspection model; inputting the pretreated inspection data into the deep inspection model to obtain an actual value; calculating to obtain a fitting value by a ridge regression method;
the comparison and evaluation module is used for comparing the fitting value with the actual value generated by the operation of the deep inspection model and evaluating possible data defects and hidden function faults of the micro-service architecture system;
the second prediction module constructing a depth inspection model comprises:
For the time dimension, carrying out time sequence coding by PositionEncoding, and utilizing the attribute to discover the characteristic association of the time dimension, wherein the formula is as follows:
PositionEncoding＝cos2(pos/N)
for the space dimension, different multi-space dimension features are extracted, and the multi-space dimension features are fused through MultiHead, wherein the formula is as follows:
Headi＝Attention(Qi,Ki,Vi)
MultiHead(Q,K,V)＝Concat(head1,...,headh)*WO
Wherein d k represents the dimension of K; n is the adjustable length; q is query feature map; k is feature mapping to be matched; v is the mapping of the monitoring data; headi is the result of time attention, qi is the i-th set of query feature maps; ki is the i-th set of feature mapping to be matched; vi is the i-th group monitoring data mapping; WO is a feature fusion matrix; concat is feature cascade fusion; pos is a data sequence number; positionEncoding is position sequence coding; multiHead is multi-head feature fusion;
The second prediction module calculates a fitting value through a ridge regression method, wherein the fitting value comprises the following steps:
The model formula of the ridge regression method is as follows: x theta-y| 2+||Γθ||2
The formula of the anti-overfitting operation is as follows: θ (a) = (X TX+aI)-1XT y)
Wherein X represents an input; y represents the output prediction result; i represents a canonical operation; i represents an identity matrix; θ is a fitting hyper-parameter; Γ is a weight constant; a is the weight of the identity matrix; θ (a) represents a value of θ obtained when a is determined.
5. The system of claim 4, wherein the first predictive module builds a health inspection model, and wherein laterally layering the micro-service architecture comprises:
The micro-service architecture system is transversely divided into seven layers: the system comprises a basic environment layer, a supervision and management layer, a cloud platform, an information resource layer, a service layer, an interface layer and a UI display layer;
the health inspection model comprises a Markov transfer matrix algorithm model: x (k+1) =x (k) ×p
Wherein: x (k) represents a state vector of the trend analysis and prediction object at time t=k, P represents a one-step transition probability matrix, and X (k+1) represents a state vector of the trend analysis and prediction object at time t=k+1.
6. The system of claim 4, wherein the second prediction module pre-processes the inspection data by quadratic interpolation comprising:
and carrying out differential processing on data which are unevenly sampled by different nodes, and carrying out interpolation by adopting a secondary differential method and every 3 adjacent points to obtain secondary interpolation, wherein the formula is as follows:
wherein x is the current value of the classified object; y is 3 adjacent points of the classified object; i is the sequence number of the classification object.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210701767.1A CN114996110B (en) | 2022-06-20 | 2022-06-20 | Deep inspection optimization method and system based on micro-service architecture |
| PCT/CN2022/114210 WO2023245851A1 (en) | 2022-06-20 | 2022-08-23 | Deep inspection optimization method and system based on micro-service architecture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210701767.1A CN114996110B (en) | 2022-06-20 | 2022-06-20 | Deep inspection optimization method and system based on micro-service architecture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114996110A CN114996110A (en) | 2022-09-02 |
| CN114996110B true CN114996110B (en) | 2024-08-09 |
Family
ID=83036573
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210701767.1A Active CN114996110B (en) | 2022-06-20 | 2022-06-20 | Deep inspection optimization method and system based on micro-service architecture |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN114996110B (en) |
| WO (1) | WO2023245851A1 (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115880086A (en) * | 2022-11-02 | 2023-03-31 | 中电信数智科技有限公司 | A Method for Predicting Abnormal Financial Expenditure Behavior Based on Distributed Model Training |
| CN115630709A (en) * | 2022-11-08 | 2023-01-20 | 广西壮族自治区公众信息产业有限公司 | Method for dynamically analyzing and optimizing AI energy consumption threshold of data center |
| CN116017595A (en) * | 2022-11-30 | 2023-04-25 | 广西壮族自治区公众信息产业有限公司 | Method for analyzing and early warning network slice combined with cell signal reselection |
| CN115915215A (en) * | 2022-12-30 | 2023-04-04 | 天翼物联科技有限公司 | A method, device, equipment and medium based on 6GNR air interface signal optimization |
| CN118096116A (en) * | 2024-02-19 | 2024-05-28 | 北京凯隆分析仪器有限公司 | Environment-friendly monitoring equipment operation state analysis method and system |
| CN118095812B (en) * | 2024-04-29 | 2024-07-23 | 山东汇通创软信息技术有限公司 | Intelligent patrol management method and system for power plant |
| CN118886070B (en) * | 2024-10-08 | 2025-02-25 | 浪潮云洲工业互联网有限公司 | A method and device for safe operation of energy equipment based on blockchain |
| CN119806992B (en) * | 2025-03-17 | 2025-05-23 | 瀚高基础软件股份有限公司 | Intelligent inspection method and device for database |
| CN119991098B (en) * | 2025-04-16 | 2025-07-11 | 铁正检测科技有限公司 | A bridge and tunnel inspection and maintenance management system and method based on three-dimensional panoramic view and GIS map |
| CN120692277A (en) * | 2025-06-04 | 2025-09-23 | 广东树华科技咨询服务有限公司 | Community-level cloud-edge collaborative microservice architecture and privacy computing system |
| CN120297697A (en) * | 2025-06-11 | 2025-07-11 | 国能浙江舟山发电有限责任公司 | Intelligent scheduling method for multi-view inspection tasks based on BIM data integration |
| CN120763821B (en) * | 2025-08-28 | 2025-12-09 | 北京中天路通智控科技有限公司 | Unmanned aerial vehicle-based subway protection area intelligent patrol system |
| CN120933512A (en) * | 2025-10-10 | 2025-11-11 | 合肥瑞曼达电子科技有限公司 | Battery health management system and method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111242428A (en) * | 2019-12-31 | 2020-06-05 | 华为技术有限公司 | Microservice processing method, microservice processing device, microservice processing apparatus, and storage medium |
| CN114462644A (en) * | 2022-02-11 | 2022-05-10 | 阿里巴巴（中国）有限公司 | Method and system for micro-service operation and maintenance |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6393386B1 (en) * | 1998-03-26 | 2002-05-21 | Visual Networks Technologies, Inc. | Dynamic modeling of complex networks and prediction of impacts of faults therein |
| CN106327013A (en) * | 2016-08-24 | 2017-01-11 | 广东工业大学 | Power transmission line inspection path planning method and system |
| EP3850513A1 (en) * | 2018-09-14 | 2021-07-21 | Telefonaktiebolaget LM Ericsson (publ) | Malchain detection |
| CN109038832A (en) * | 2018-09-19 | 2018-12-18 | 上海许继电气有限公司 | Substation's auxiliary monitoring system based on micro services framework |
| CN110334740A (en) * | 2019-06-05 | 2019-10-15 | 武汉大学 | The electrical equipment fault of artificial intelligence reasoning fusion detects localization method |
| US11573848B2 (en) * | 2020-11-10 | 2023-02-07 | International Business Machines Corporation | Identification and/or prediction of failures in a microservice architecture for enabling automatically-repairing solutions |
-
2022
- 2022-06-20 CN CN202210701767.1A patent/CN114996110B/en active Active
- 2022-08-23 WO PCT/CN2022/114210 patent/WO2023245851A1/en not_active Ceased
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111242428A (en) * | 2019-12-31 | 2020-06-05 | 华为技术有限公司 | Microservice processing method, microservice processing device, microservice processing apparatus, and storage medium |
| CN114462644A (en) * | 2022-02-11 | 2022-05-10 | 阿里巴巴（中国）有限公司 | Method and system for micro-service operation and maintenance |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2023245851A1 (en) | 2023-12-28 |
| CN114996110A (en) | 2022-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114996110B (en) | Deep inspection optimization method and system based on micro-service architecture | |
| CN110943857B (en) | Power communication network fault analysis and positioning method based on convolutional neural network | |
| CN110703057B (en) | Partial discharge diagnosis method of power equipment based on data augmentation and neural network | |
| CN118133203A (en) | A method for fault diagnosis of electric energy metering detection information | |
| CN110929918A (en) | 10kV feeder line fault prediction method based on CNN and LightGBM | |
| CN119167805B (en) | Power failure calculation prediction method and system based on digital twin | |
| CN111460728A (en) | Method and device for predicting residual life of industrial equipment, storage medium and equipment | |
| CN117556347A (en) | A power equipment fault prediction and health management method based on industrial big data | |
| CN115099335A (en) | Abnormal identification and feature screening method and system for multi-source heterogeneous data | |
| He et al. | Fault diagnosis and location based on graph neural network in telecom networks | |
| CN110636066A (en) | Network Security Threat Situation Assessment Method Based on Unsupervised Generative Reasoning | |
| CN111586728B (en) | Small sample characteristic-oriented heterogeneous wireless network fault detection and diagnosis method | |
| CN118964900B (en) | Causal knowledge guidance-based power equipment fault diagnosis method and causal knowledge guidance-based power equipment fault diagnosis system | |
| CN114553756A (en) | Equipment fault detection method based on joint generation countermeasure network and electronic equipment | |
| CN120412314B (en) | Traffic signal light fault diagnosis method and system based on artificial intelligence | |
| CN109870306A (en) | A kind of tractor engine Fault diagnosis and forecast method | |
| CN119135568A (en) | Intelligent operation and maintenance fault diagnosis system based on communication network | |
| CN113191556A (en) | Nuclear power Loca event fault prediction and diagnosis method | |
| CN120573556A (en) | An elevator fault prediction method and system based on big data technology | |
| CN120087546A (en) | A gas pipeline network operation data intelligent management system | |
| CN119476009A (en) | A power transformer fault diagnosis method based on improved graph convolutional network | |
| CN119482923A (en) | Analysis and management system of telecontrol information data | |
| CN118551296A (en) | GIS-based power equipment fault diagnosis and early warning system and method | |
| CN118278512B (en) | Knowledge tracking visual analysis system based on deep learning model interpretability | |
| CN120847624B (en) | Method and System for Fault Diagnosis and Early Warning of Battery Modules in Dynamically Reconfigurable Energy Storage Systems |
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 |