CN114611616A - Unmanned aerial vehicle intelligent fault detection method and system based on integrated isolated forest - Google Patents
Unmanned aerial vehicle intelligent fault detection method and system based on integrated isolated forest Download PDFInfo
- Publication number
- CN114611616A CN114611616A CN202210259150.9A CN202210259150A CN114611616A CN 114611616 A CN114611616 A CN 114611616A CN 202210259150 A CN202210259150 A CN 202210259150A CN 114611616 A CN114611616 A CN 114611616A
- Authority
- CN
- China
- Prior art keywords
- unmanned aerial
- aerial vehicle
- flight
- data
- isolated forest
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B64—AIRCRAFT; AVIATION; COSMONAUTICS
- B64F—GROUND OR AIRCRAFT-CARRIER-DECK INSTALLATIONS SPECIALLY ADAPTED FOR USE IN CONNECTION WITH AIRCRAFT; DESIGNING, MANUFACTURING, ASSEMBLING, CLEANING, MAINTAINING OR REPAIRING AIRCRAFT, NOT OTHERWISE PROVIDED FOR; HANDLING, TRANSPORTING, TESTING OR INSPECTING AIRCRAFT COMPONENTS, NOT OTHERWISE PROVIDED FOR
- B64F5/00—Designing, manufacturing, assembling, cleaning, maintaining or repairing aircraft, not otherwise provided for; Handling, transporting, testing or inspecting aircraft components, not otherwise provided for
- B64F5/60—Testing or inspecting aircraft components or systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Manufacturing & Machinery (AREA)
- Transportation (AREA)
- Aviation & Aerospace Engineering (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to the technical field of unmanned aerial vehicle fault detection, and discloses an unmanned aerial vehicle intelligent fault detection method and system based on an integrated isolated forest, wherein an unmanned aerial vehicle intelligent fault detection model based on the integrated isolated forest is developed by utilizing internal and external multisource original flight data of an unmanned aerial vehicle in a safe flight state, and the method comprises the following steps: the method comprises the steps of collecting multisource original flight data inside and outside the unmanned aerial vehicle, extracting basic features by using a flight log of the original unmanned aerial vehicle, digitizing characteristic data of the unmanned aerial vehicle in specification, integrating various machine learning technologies to form an integrated isolated forest model, detecting the flight condition of the unmanned aerial vehicle through a safe flight mode, deeply analyzing the internal incidence relation among various flight data features of the unmanned aerial vehicle in each flight period, and timely finding abnormal conditions in flight. Compared with the prior art, the method and the device have the advantages that the abnormal information of the faults of the unmanned aerial vehicle can be intuitively and quickly captured, and the guarantee is provided for the safe flight of the unmanned aerial vehicle and the safety of the low-altitude airspace region.
Description
Technical Field
The invention relates to the technical field of unmanned aerial vehicle fault detection, in particular to an unmanned aerial vehicle intelligent fault detection method and system based on an integrated isolated forest.
Background
In recent years, unmanned aerial vehicles have been rapidly developed to hopefully create great benefits for society, but since unmanned aerial vehicles rely heavily on sensors providing information about the state of external environment or internal systems, actuators physically realizing required movement, and controllers driving the actuators according to measurement results and tasks to be completed, they are vulnerable to various attacks, and if communication signals are invaded by malicious attackers or sudden hardware failures, unmanned aerial vehicles lose control, which results in failure to stabilize the body in the air, and once the unmanned aerial vehicles fail, huge damage, including commercial and non-commercial losses, can be caused. When the unmanned aerial vehicle operates abnormally, the safety of the unmanned aerial vehicle is influenced, and the larger influence is the spread of public safety and even the interference on the management of the whole low-altitude airspace.
The unmanned aerial vehicle system mainly comprises three parts, including unmanned aerial vehicle, ground station and the communication link of transmission information. The abnormality of the unmanned aerial vehicle generally refers to the loss of control and loss of connection of the unmanned aerial vehicle. On one hand, the abnormality is caused by hardware faults of the unmanned aerial vehicle or interference of an external environment, and the abnormality can be solved by replacing an unmanned aerial vehicle component; on the other hand, the unmanned aerial vehicle System comprises attack to a communication network of the unmanned aerial vehicle, Global Positioning System (GPS) deception, GPS interference and the like, once the unmanned aerial vehicle is invaded, the unmanned aerial vehicle loses control, so that the airframe cannot be stabilized in the air, and safety accidents are caused. Therefore, how to quickly and accurately identify the abnormal state of the unmanned aerial vehicle becomes a key issue for improving the operation safety of the unmanned aerial vehicle.
Since the attacks suffered by drones in real environments are complex, problems and challenges are often faced with respect to the detection of anomalies by drone systems: the attack is unknown; the manual marking cost is high, and the data is lack of labels; under the operation of low computational load, high-precision detection and the like of multiple abnormal types are realized. At present, unmanned aerial vehicle fault detection methods can be divided into four broad categories, namely model-based methods, knowledge-based methods, data-driven methods and discrete event-based methods. While the discrete event method is in the theoretical research stage, the first three methods have been developed and form more systematic branches due to the accumulation of time, and new methods have been derived in the research and application aspects in the field of flight control.
Based on a model analysis method, the faults of the unmanned aerial vehicle are detected by constructing a complex model, the situations of system misjudgment or missing report caused by computation redundancy and easy error generation exist, and the problems of noise signal interference, robustness and the like are to be improved and solved. The knowledge-based method depends on well-defined features and priori supervision knowledge to a great extent, and has some inevitable disadvantages, for example, a training data set needs to be finely marked, fault types which can be identified by a model are limited, fault types which are not trained cannot be identified, and the requirement of unmanned aerial vehicle flight processing on various types of data and continuous operation for avoiding time delay is difficult to meet.
The method utilizes unsupervised machine learning, training set data do not need artificial labeling results, belongs to a data driving-based method, utilizes characteristic information extracted from original flight data of the unmanned aerial vehicle to carry out effective judgment, and can solve the problems of uncertainty, strong coupling and the like of an unmanned aerial vehicle system
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems in the prior art, the invention provides an unmanned aerial vehicle intelligent fault detection method and system based on an integrated isolated forest.
The technical scheme is as follows: the invention provides an unmanned aerial vehicle intelligent fault detection method based on an integrated isolated forest, which comprises the following steps:
step 1: acquiring internal and external multi-source original flight data of the unmanned aerial vehicle in a safe and stable state, wherein the internal and external multi-source original flight data comprise flight control parameters, engine parameters, electrical parameters, positioning parameters and radio parameters, so as to form an unmanned aerial vehicle data sample set;
step 2: carrying out digital processing and time measurement unification on the unmanned aerial vehicle data sample set in the step 1 to form a normalized unmanned aerial vehicle flight data set;
and step 3: training an integrated isolated forest according to the unmanned aerial vehicle flight data set normalized in the step 2, and constructing a compliance integrated isolated forest model based on the unmanned aerial vehicle flight data, wherein the specific process is that for each unmanned aerial vehicle flight data subsample, a compliance model tree is established by respectively using three different branching criteria:
1) branching criteria 1: slicing the data by utilizing a hyperplane with a random slope, and randomly selecting a slope to cut branches according to a branch criterion 1 in a training stage;
2) branching criterion 2: randomly selecting two columns of characteristics from a sample, mapping two columns of characteristic values on a hyperplane to form a column of composite attribute mapping values, and establishing a compliant isolated tree according to the mapping values;
3) branching criteria 3: selecting the hyperplane corresponding to the minimum H (X) for segmentation;
performing branching operation on each sub-sample according to 3 types of branching criteria, dividing a sample smaller than a division hyperplane in a normalized data sample of the unmanned aerial vehicle into a left branch and a right branch larger than the division hyperplane in the sample, and repeating the binary branching operation on the left branch and the right branch until a certain condition is met;
and 4, step 4: and completing safety detection of the flight state of the unmanned aerial vehicle by utilizing the compliance integration isolated forest model.
Further, the branch formula rule of the branch criterion 1 in step 2 is as follows:
wherein the content of the first and second substances,the vector is a normal vector, and the vector is a vector,is a random intercept vector.
Further, the branch formula rule of the branch criterion 2 in step 3 is as follows:
wherein Q is all characteristic attributes of normalized data of all unmanned aerial vehicles, j is randomly selected attribute of unmanned aerial vehicle, and cjIs [ -1,1 [ ]]A randomly selected value, X' is a sample set of secondary sampling of the unmanned aerial vehicle, XjThe j-th characteristic attribute value of 'X' is a random division point, p is a random division point, tau hyperplanes are created, the hyperplane corresponding to the maximum S (Y) is selected for division, the optimal hyperplane is selected from the hyperplanes, Y is a real value set obtained by f (X) projection of Xl∪YrDividing Y at random by plAnd YrAnd (4) separating.
Further, the partition rule and branch formula rule of the branch rule 3 in step 3 are as follows:
wherein, p (x)i) Represents X is XiThe probability of (c).
Further, the conditions satisfied in step 3 are as follows:
condition 1: the attribute data set of the unmanned aerial vehicle is not divisible;
condition 2: the scaled tree reaches a defined maximum depth of 8.
Further, the normalization processing process of the flight data of the unmanned aerial vehicle in the step 2 is as follows:
step 2.1: acquiring the flight data of the unmanned aerial vehicle in the step 1 for a time length at least longer than 120 hours, acquiring and processing the data of the unmanned aerial vehicle in the time period to form an unmanned aerial vehicle flight data set in a safe state under the condition that the unmanned aerial vehicle is in a normal flight state in the time length, wherein the record set is used as a training set for training a safe flight mode of the unmanned aerial vehicle;
step 2.2: digitally converting non-digital records in the training set safety flight data record set into a digital record set which can be learned by a machine;
step 2.3: and carrying out time measurement unification treatment on the training set safety flight data record set after the data is converted into a normalized unmanned aerial vehicle flight data set.
Further, the process of obtaining the flight data subsample of the unmanned aerial vehicle in the step 2 is as follows: and randomly selecting the normalized data subsamples of the unmanned aerial vehicle for 100 times through the Bagging technology for the normalized unmanned aerial vehicle flight data set.
Further, the set of compliant model trees in the compliant integrated isolated forest model in step 3 is composed of 300 flying compliance-scale trees.
Further, the specific process of completing the safety detection of the flight state of the unmanned aerial vehicle by utilizing the compliance integration isolated forest model in the step 4 is as follows:
for each detected unmanned aerial vehicle compliance data, traversing all the established unmanned aerial vehicle compliance model trees to obtain the path length generated after each tree is traversed, then calculating the average path length of the path length of each tree by using a statistical method, judging the unmanned aerial vehicle compliance model trees to be abnormal flight state data when the average path is shorter and the abnormal degree is larger, calculating the abnormal score of each unmanned aerial vehicle compliance model tree according to the average path length, and judging whether the flight state of the unmanned aerial vehicle is abnormal or not according to the abnormal score.
The invention also discloses unmanned aerial vehicle intelligent fault detection based on the integrated isolated forest, which comprises the following steps:
the flight data acquisition module is used for acquiring and processing unmanned aerial vehicle data to form an unmanned aerial vehicle flight data set in a safe state;
the flight data normalization processing module is used for preprocessing the flight data of the original unmanned aerial vehicle and converting the flight data into normalized data of the unmanned aerial vehicle which can be learned by a machine;
the integrated isolated forest algorithm training module is used for carrying out unsupervised automatic learning on the safe flight state of the unmanned aerial vehicle and constructing a compliant integrated isolated forest model;
and the fault detection module is used for completing safety detection on the flight state of the unmanned aerial vehicle by the compliance integration isolated forest model.
Has the advantages that:
1. the machine learning technology utilized in the invention belongs to unsupervised machine learning, and the training set data does not need the result of artificial marking. Compared with a supervised method which relies on an accurately labeled data set to perform model training, sufficient positive and negative samples are needed, and the unsupervised machine learning method can be used for landing in unmanned aerial vehicle intelligent fault analysis with less cost.
2. According to the distribution characteristics of flight data of the unmanned aerial vehicle, the unmanned aerial vehicle safety four-element power system, the main controller, the communication link module and the sensors are surrounded, a safety flight state model based on the unmanned aerial vehicle four-element is established through a machine learning technology, an isolated forest model is integrated based on compliance, the unmanned aerial vehicle system abnormity detection is realized, the attack behavior of the unmanned aerial vehicle can be found and early warned in time, the unmanned aerial vehicle is prevented from flying against instructions, and the unmanned aerial vehicle operation safety can be effectively protected.
3. According to the invention, the set of the flight compliance model trees is composed of 300 flight compliance type trees, after 300 trees are obtained in the unmanned aerial vehicle compliance integration isolated forest model, the number of the flight compliance model trees is increased, and the result of anomaly detection is not further improved, so that the set of 300 flight compliance type trees can meet the anomaly detection precision, and the condition of energy waste is avoided.
4. When unmanned aerial vehicle data presents a plurality of data distribution cluster distributions or data wave curve distributions, the branching criterion of the traditional isolated forest model is covered by abnormity, while the branching criterion 1 of the invention allows data to be sliced to use a hyperplane with random slope, and as is apparent from the attached figure 6 in the specification, the branching criterion 1 of the invention detects the abnormity which is not detected by the branching criterion of the traditional isolated forest model.
5. Since the attacks suffered by drones in real environments are complex, the drone exceptions may be generated by different mechanisms, and the cluster exceptions facing drones may have their own distribution. The unmanned aerial vehicle combined-scale tree established by the branching criterion 2 isolates the outliers by introducing a new cutting surface, surpasses the randomly-fitted tree, and can effectively separate the aggregation abnormity from the normal points.
6. The branch criterion 3 of the invention judges according to the information entropy, namely, the information gain criterion of the decision tree is utilized for branching, when the data of the unmanned aerial vehicle with correction measures, which is more close to the normal data distribution of the unmanned aerial vehicle, is faced, the phenomenon of overall backward shift occurs in abnormal score performance, the branch criterion is integrated, the essence is the correction of the branch criterion 1 and the branch criterion 2, namely, when the data of the unmanned aerial vehicle with correction measures are detected, the unmanned aerial vehicle synthetic-scale tree established according to the branch criterion 3 carries out abnormal score correction on the unmanned aerial vehicle synthetic-scale tree established under the branch criterion 1 and the branch criterion 2. In conclusion, the integrated isolated forest combines three branch criteria, makes up the defects of the traditional isolated forest model, keeps the low time load of the algorithm, and meets the requirement of the abnormal detection of the unmanned aerial vehicle.
Drawings
FIG. 1 is a schematic flow chart of steps of an unmanned aerial vehicle intelligent fault detection model based on an integrated isolated forest, provided by the invention;
FIG. 2 is a schematic view of a digital process for a set of flight data records according to the present invention;
FIG. 3 is a schematic view of a process for unifying time metrics of a flight data record set according to the present invention;
FIG. 4 is a schematic diagram of specific steps for constructing a compliant integrated isolated forest model according to the present invention;
FIG. 5 is a schematic diagram of a safety detection process of a compliance integration isolated forest model for a flight state of an unmanned aerial vehicle, provided by the invention;
fig. 6 is a schematic diagram of the abnormal value segmentation of the branching criterion of the conventional isolated forest model and the branching criterion 1 of the present invention in the data distribution for 2 kinds of unmanned aerial vehicles.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
Referring to the attached drawings 1-5, the invention discloses an unmanned aerial vehicle intelligent fault detection method and system based on an integrated isolated forest.
According to the schematic flow chart of the steps of the unmanned aerial vehicle intelligent fault detection method based on the integrated isolated forest in fig. 1, the unmanned aerial vehicle intelligent fault detection method based on the integrated isolated forest comprises the following specific steps:
step 1: the unmanned aerial vehicle safety four-element power system, the main controller, the communication link module and the sensor are connected with the ground ATE automatic test system by utilizing embedded application software to complete real-time data information acquisition, ground data transmission and storage of the unmanned aerial vehicle. The data acquisition program is mainly used for interrupting a kernel of a CPU service program from a software layer and controlling digital signals and analog signals in a certain process. For example, data acquisition of flight routes, heights, speeds, postures, steering engine positions and the like is in charge of AD soft interruption, data acquisition of positioning of GPS and the like is in charge of Uart interruption, and finally, the acquired data are packaged and stored in a buffer area, wherein the data comprise internal and external multisource original flight data of the unmanned aerial vehicle, such as flight control parameters, engine parameters, electrical parameters, positioning parameters, radio parameters and the like, and the data are written into a mobile memory through the buffer area to form layered and structured unmanned aerial vehicle flight data.
Step 2: and (3) carrying out digital processing and time measurement unification on the unmanned aerial vehicle data sample set in the step (1) to form a normalized unmanned aerial vehicle flight data set.
Step 2.1: the data acquisition of the step 1 is carried out, the flight data of the unmanned aerial vehicle with the time length at least larger than 120 hours are acquired, the unmanned aerial vehicle in the time length is ensured to be in a normal flight state, an unmanned aerial vehicle flight data set in a safe state is formed by acquiring and processing the data of the unmanned aerial vehicle in the time period, the record set is used as a training set for training a safe flight mode of the unmanned aerial vehicle, and therefore a compliance integration isolated forest model of the unmanned aerial vehicle based on normalized flight data is established.
Step 2.2: and (3) performing digital processing on the training set safety flight data record set, wherein the digital processing is to complete the conversion work of the relevant characteristic attribute values from non-numbers to numbers according to the set conversion rule on each characteristic attribute value in the flight data (see fig. 2). The digitization rule is that after the non-digital attributes are subjected to de-duplication, the non-digital characteristic values are sequenced, each non-digital characteristic value is mapped to a corresponding serial number, each characteristic value is guaranteed to correspond to a unique digital value, a characteristic attribute digitization mapping table is constructed, and then the non-digital flight data are converted into a digital record set which can be learned by a machine through the digitization mapping table.
Step 2.3: since the sensor data are independent from each other, the attribute of each type of unmanned aerial vehicle feature is recorded in different time periods, and it is difficult to describe the state of the unmanned aerial vehicle in a specific time period, so that the time measurement unification processing is performed on the safety flight data recording set of the training set after the data is digitalized, that is, a representative value is randomly selected from each unit time period, so that each feature has the same number of values in the same time window, and a normalized unmanned aerial vehicle flight data set is formed, for example, the number of each feature of unmanned aerial vehicle feature 1, feature 2, and feature 3 is 7, 3, and 5 in the same time window, and the time measurement unification processing randomly selects a representative value from each unit time period, and ensures that the number of feature values in a single time window is the same (see fig. 3).
And step 3: and (3) training an integrated isolated forest according to the unmanned aerial vehicle flight data set normalized in the step (2), and constructing a compliant integrated isolated forest model based on the unmanned aerial vehicle flight data.
The method comprises the steps of completing training and learning of a normalized unmanned aerial vehicle flight data set by using an integrated isolated forest algorithm, wherein the training result is that flight compliance model trees are established according to various flight characteristic attribute values, each flight compliance tree forms a compliance integrated isolated forest model, and each compliance tree in the compliance integrated isolated forest model is used for detecting abnormal flight modes. The specific process of completing the training and learning of the rule behavior event data set by using the integrated isolated forest-based algorithm is as follows (see fig. 4):
1) randomly selecting normalized data subsamples of the unmanned aerial vehicles for 100 times from the normalized unmanned aerial vehicle flight data set through Bagging technology, wherein each normalized data subsample of the unmanned aerial vehicles comprises 235 pieces of unmanned aerial vehicle flight data;
2) for each subsample, a compliance model tree is built using three different types of branching criteria:
branching criteria 1: slicing data by utilizing a hyperplane with a random slope, and randomly selecting a slope to cut branches according to a branch rule 1 in a training stage, wherein the branch formula rule of the branch rule 1 is as follows:
wherein the content of the first and second substances,the vector is a normal vector, and the vector is a vector,is a random intercept vector.
Branching criteria 2: randomly selecting two columns of characteristics from a sample, mapping two columns of characteristic values on a hyperplane to form a column of composite attribute mapping values, and establishing a compliance isolated tree according to the mapping values, wherein the branch formula rule of the branch criterion 2 is as follows:
wherein Q is all characteristic attributes of normalized data of all unmanned aerial vehicles, j is randomly selected attribute of unmanned aerial vehicle, and cjIs [ -1,1 [ ]]A randomly selected value, X' is a sample set of secondary sampling of the unmanned aerial vehicle, XjThe j characteristic attribute value of X 'is, p is a random dividing point, tau candidate hyperplanes are created, the hyperplane corresponding to the maximum S (Y) is selected for division, the optimal hyperplane is selected from the candidate hyperplanes, Y is a real value set obtained by f (X) projection of X', Y isl∪YrDividing Y at random by plAnd YrAnd (4) separating.
Branching criteria 3: the partition criterion is as follows, and the hyperplane corresponding to the minimum H (X) is selected for segmentation:
wherein, p (x)i) Represents X is XiThe probability of (c).
Performing branching operation on each subsample according to 3 types of branching criteria, dividing a sample smaller than a segmentation hyperplane in the normalized data sample of the unmanned aerial vehicle into a left branch, and dividing a sample larger than the segmentation hyperplane in the sample into a right branch, and then repeating the binary branching operation on the left branch and the right branch until the following conditions are met:
condition 1: the attribute data set of the unmanned aerial vehicle is not divisible;
condition 2: the scaled tree reaches a defined maximum depth of 8.
The set of the compliance model trees of the unmanned aerial vehicle in the safe flight state is formed by 300 unmanned aerial vehicle compliance integration isolated forest model trees.
And step 3: and completing safety detection of the flight state of the unmanned aerial vehicle by utilizing the compliance integration isolated forest model.
The specific process of completing the safety detection of the flight state of the unmanned aerial vehicle by utilizing the compliance integration isolated forest model is as follows (see fig. 5):
step 3.1: for each detected unmanned aerial vehicle compliance data, traversing all established unmanned aerial vehicle compliance model trees to obtain path length generated after each tree is traversed, then calculating the average path length of the path length of each tree by using a statistical method, determining abnormal flight state data when the average path is shorter and the abnormal degree is larger, calculating an abnormal score of each unmanned aerial vehicle compliance data according to the average path length, judging whether the flight state of the unmanned aerial vehicle is abnormal according to the abnormal score, wherein the flight data with the abnormal score lower than an abnormal threshold value are the normal flight data of the unmanned aerial vehicle, and otherwise, the flight data with the abnormal score higher than the abnormal threshold value are the abnormal flight data of the unmanned aerial vehicle;
step 3.2: and informing an alarm system of the abnormal flight state data to finish alarm operation.
Aiming at the unmanned aerial vehicle intelligent fault detection method based on the integrated isolated forest, the unmanned aerial vehicle intelligent fault detection system based on the integrated isolated forest comprises a flight data acquisition module, a data acquisition module and a data processing module, wherein the flight data acquisition module is used for acquiring and processing unmanned aerial vehicle data to form an unmanned aerial vehicle flight data set in a safe state; the flight data normalization processing module is used for preprocessing the flight data of the original unmanned aerial vehicle and converting the flight data into normalized data of the unmanned aerial vehicle which can be learned by a machine; the integrated isolated forest algorithm training module is used for carrying out unsupervised automatic learning on the safe flight state of the unmanned aerial vehicle and constructing a compliant integrated isolated forest model; and the fault detection module is used for completing safety detection on the flight state of the unmanned aerial vehicle by the compliance integration isolated forest model. The system is only used for executing the unmanned aerial vehicle intelligent fault detection method based on the integrated isolated forest.
Aiming at the unmanned aerial vehicle intelligent fault detection method based on the integrated isolated forest, the branching criterion 1 of the invention allows data to be sliced to use a hyperplane with random slope, fig. 6 is a schematic diagram of the branching criterion of the traditional isolated forest model and the segmentation of the branching criterion 1 of the invention in terms of abnormal values of 2 kinds of unmanned aerial vehicle data distribution, wherein a black point is an abnormal data point, a gray point is a data point in normal distribution, fig. 6(a) (b) is the branching segmentation generated by the branching criterion of the traditional isolated forest model, and (c) (d) is the branching segmentation generated by the branching criterion 1 of the invention, so that the branching criterion 1 of the invention obviously detects the abnormality which is not detected by the branching criterion of the traditional isolated forest model.
In addition, the invention respectively detects the unmanned aerial vehicle flight abnormity (UA data set) without corrective measures and the unmanned aerial vehicle flight abnormity (ALFA data set) with corrective measures, and compares the invention (IIF algorithm) with the traditional isolated Forest (iForest), the branching criterion 1, the branching criterion 2, the branching criterion 3 and the traditional unsupervised abnormity detection algorithm: local Outlier Factors (LOF) and a Class of support vector machines (One-Class SVM, OC SVM) are compared, and the comparison results of different algorithms are realized, which is shown in tables 1 and 2:
table 1 anomaly detection algorithm performance comparison for UA datasets
It can be easily found from table 1 that for unmanned aerial vehicle flight anomalies (UA data sets) without corrective measures, the F-measure values of the present invention for GPS interference, GPS spoofing, and mixed anomalies are all kept above 0.9. Compared with the traditional LOF and One-Class SVM, the algorithm accuracy is respectively improved by 68% and 12.6%. For the traditional iForest, the algorithm accuracy and the recall rate are respectively and averagely improved by 1.3% and 62.6%.
TABLE 2 comparison of anomaly detection algorithm Performance for ALFA datasets
As can be seen from Table 2, the F-measure value of the conventional unsupervised anomaly detection algorithm is between 0.33 and 0.69 for the anomaly data of the unmanned aerial vehicle with the corrective measures. In other algorithms, the branching criterion 3 algorithm F-measure value is the highest and reaches 0.92, 0.83, 0.84 and 0.88 respectively, and further analysis on the accuracy and the recall rate can find that the accuracy of detecting four types of abnormalities is lower than that of the algorithm of the invention, but the recall rate of the algorithm is 100% in an adverse view. The situation that overfitting occurs to the abnormal data of the unmanned aerial vehicle with the corrective measures in the branch criterion 3 algorithm is difficult to see, that is, all the data of the detection set are classified into abnormal data. The F-measure values of the invention for detecting the engine abnormality, the rudder abnormality, the elevator abnormality and the aileron abnormality are respectively 0.86, 0.76, 0.75 and 0.75, and the four kinds of F-measure values are ranked in the second place. Followed by iForest and branching criteria 1, while branching criteria 2 performance ranks last.
The above embodiments are merely illustrative of the technical concepts and features of the present invention, and the purpose of the embodiments is to enable those skilled in the art to understand the contents of the present invention and implement the present invention, and not to limit the protection scope of the present invention. All equivalent changes and modifications made according to the spirit of the present invention should be covered within the protection scope of the present invention.
Claims (10)
1. An unmanned aerial vehicle intelligent fault detection method based on an integrated isolated forest is characterized by comprising the following steps:
step 1: acquiring internal and external multisource original flight data of the unmanned aerial vehicle in a safe and stable state, wherein the internal and external multisource original flight data comprise flight control parameters, engine parameters, electrical parameters, positioning parameters and radio parameters, so as to form an unmanned aerial vehicle data sample set;
step 2: carrying out digital processing and time measurement unification on the unmanned aerial vehicle data sample set in the step 1 to form a normalized unmanned aerial vehicle flight data set;
and step 3: training an integrated isolated forest according to the unmanned aerial vehicle flight data set normalized in the step 2, and constructing a compliance integrated isolated forest model based on the unmanned aerial vehicle flight data, wherein the specific process is that for each unmanned aerial vehicle flight data subsample, a compliance model tree is established by respectively using three different branching criteria:
1) branching criteria 1: slicing the data by utilizing a hyperplane with a random slope, and randomly selecting a slope to cut branches according to a branch criterion 1 in a training stage;
2) branching criterion 2: randomly selecting two columns of characteristics from a sample, mapping two columns of characteristic values on a hyperplane to form a column of composite attribute mapping values, and establishing a compliant isolated tree according to the mapping values;
3) branching criteria 3: selecting the hyperplane corresponding to the minimum information entropy H (X) for segmentation;
performing branching operation on each sub-sample according to 3 types of branching criteria, dividing a sample smaller than a division hyperplane in a normalized data sample of the unmanned aerial vehicle into a left branch and a right branch larger than the division hyperplane in the sample, and repeating the binary branching operation on the left branch and the right branch until a certain condition is met;
and 4, step 4: and completing safety detection of the flight state of the unmanned aerial vehicle by utilizing the compliance integration isolated forest model.
2. The intelligent unmanned aerial vehicle fault detection method based on the integrated isolated forest as claimed in claim 1, wherein the branching formula rule of the branching criterion 1 in the step 2 is as follows:
3. The intelligent unmanned aerial vehicle fault detection method based on the integrated isolated forest as claimed in claim 1, wherein the branching formula rule of the branching criterion 2 in the step 3 is as follows:
wherein Q is all characteristic attributes of normalized data of all unmanned aerial vehicles, j is randomly selected attribute of unmanned aerial vehicle, and cjIs [ -1,1 [ ]]A randomly selected value, X' is a sample set of secondary sampling of the unmanned aerial vehicle, XjThe j characteristic attribute value of X 'is, p is a random dividing point, tau candidate hyperplanes are created, the hyperplane corresponding to the maximum S (Y) is selected for division, the optimal hyperplane is selected from the candidate hyperplanes, Y is a real value set obtained by f (X) projection of X', Y isl∪YrDividing Y at random by plAnd YrAnd (4) separating.
4. The intelligent unmanned aerial vehicle fault detection method based on the integrated isolated forest as claimed in claim 1, wherein the division criterion branching formula rule of the branching criterion 3 in the step 3 is as follows:
wherein, p (x)i) Representative of X being XiThe probability of (c).
5. An intelligent unmanned aerial vehicle fault detection method based on an integrated isolated forest according to any one of claims 2 to 4, wherein the conditions met in the step 3 are as follows:
condition 1: the attribute data set of the unmanned aerial vehicle is not divisible;
condition 2: the scaled tree reaches a defined maximum depth of 8.
6. The intelligent unmanned aerial vehicle fault detection method based on the integrated solitary forest as claimed in claim 1, wherein the normalization processing process of the flight data of the unmanned aerial vehicle in the step 2 is as follows:
step 2.1: acquiring the flight data of the unmanned aerial vehicle in the step 1 for a time length at least longer than 120 hours, acquiring and processing the data of the unmanned aerial vehicle in the time period to form an unmanned aerial vehicle flight data set in a safe state under the condition that the unmanned aerial vehicle is in a normal flight state in the time length, wherein the record set is used as a training set for training a safe flight mode of the unmanned aerial vehicle;
step 2.2: digitally converting non-digital records in the training set safety flight data record set into a digital record set which can be learned by a machine;
step 2.3: and carrying out time measurement unification treatment on the training set safety flight data record set after the data is converted into a normalized unmanned aerial vehicle flight data set.
7. The intelligent unmanned aerial vehicle fault detection method based on the integrated isolated forest as claimed in claim 1, wherein the process of obtaining the sub-samples of the flight data of the unmanned aerial vehicle in the step 2 is as follows: and randomly selecting the normalized data subsamples of the unmanned aerial vehicle for 100 times through the Bagging technology for the normalized unmanned aerial vehicle flight data set.
8. An intelligent unmanned aerial vehicle fault detection method based on integrated isolated forest according to claim 1, wherein the set of compliant model trees in the compliant integrated isolated forest model in the step 3 is composed of 300 flying compliance-type trees.
9. The intelligent unmanned aerial vehicle fault detection method based on the integrated isolated forest as claimed in claim 1, wherein the specific process of completing the safety detection of the flight state of the unmanned aerial vehicle by using the compliant integrated isolated forest model in the step 4 is as follows:
for each detected unmanned aerial vehicle compliance data, traversing all the established unmanned aerial vehicle compliance model trees to obtain the path length generated after each tree is traversed, then calculating the path length average path length of each tree by using a statistical method, judging that the average path is shorter and the abnormal degree is larger as the average path is shorter as the average path is longer as the abnormal flight state data is obtained, calculating the abnormal score of each unmanned aerial vehicle compliance data according to the average path length, and judging whether the flight state of the unmanned aerial vehicle is abnormal or not according to the abnormal score.
10. The utility model provides an unmanned aerial vehicle intelligent fault detection based on integrated isolated forest which characterized in that includes:
the flight data acquisition module is used for acquiring and processing unmanned aerial vehicle data to form an unmanned aerial vehicle flight data set in a safe state;
the flight data normalization processing module is used for preprocessing the flight data of the original unmanned aerial vehicle and converting the flight data into normalized data of the unmanned aerial vehicle which can be learned by a machine;
the integrated isolated forest algorithm training module is used for carrying out unsupervised automatic learning on the safe flight state of the unmanned aerial vehicle and constructing a compliant integrated isolated forest model;
and the fault detection module is used for completing safety detection on the flight state of the unmanned aerial vehicle by the compliance integration isolated forest model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210259150.9A CN114611616B (en) | 2022-03-16 | 2022-03-16 | Unmanned aerial vehicle intelligent fault detection method and system based on integrated isolated forest |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210259150.9A CN114611616B (en) | 2022-03-16 | 2022-03-16 | Unmanned aerial vehicle intelligent fault detection method and system based on integrated isolated forest |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114611616A true CN114611616A (en) | 2022-06-10 |
CN114611616B CN114611616B (en) | 2023-02-07 |
Family
ID=81863225
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210259150.9A Active CN114611616B (en) | 2022-03-16 | 2022-03-16 | Unmanned aerial vehicle intelligent fault detection method and system based on integrated isolated forest |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114611616B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115185303A (en) * | 2022-09-14 | 2022-10-14 | 南开大学 | Unmanned aerial vehicle patrol path planning method for national parks and natural protected areas |
CN115294674A (en) * | 2022-10-09 | 2022-11-04 | 南京信息工程大学 | Unmanned ship navigation state monitoring and evaluating method |
CN115457414B (en) * | 2022-09-15 | 2023-05-05 | 西华大学 | Unmanned aerial vehicle abnormal behavior identification method based on improved residual error network |
Citations (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7127633B1 (en) * | 2001-11-15 | 2006-10-24 | Xiotech Corporation | System and method to failover storage area network targets from one interface to another |
US20070294068A1 (en) * | 2006-05-24 | 2007-12-20 | Jones Jeffrey P | Line-walking recursive partitioning method for evaluating molecular interactions and questions relating to test objects |
CN107453709A (en) * | 2017-07-03 | 2017-12-08 | 重庆大学 | The photovoltaic hot spot method for diagnosing faults that a kind of isolation mech isolation test merges with intersecting measurement |
CN108765961A (en) * | 2018-08-03 | 2018-11-06 | 东华理工大学 | A kind of floating car data processing method based on modified amplitude limit average filter |
CN108877949A (en) * | 2018-06-11 | 2018-11-23 | 吉林大学 | Down's syndrome screening method based on isolated forest algorithm and voting mechanism |
US20190080142A1 (en) * | 2017-09-13 | 2019-03-14 | X Development Llc | Backup Navigation System for Unmanned Aerial Vehicles |
CN109684311A (en) * | 2018-12-06 | 2019-04-26 | 中科恒运股份有限公司 | Abnormal deviation data examination method and device |
CN109753960A (en) * | 2018-12-24 | 2019-05-14 | 西安理工大学 | The underwater unnatural object detection method of isolated forest based on fractal theory |
CN110149258A (en) * | 2019-04-12 | 2019-08-20 | 北京航空航天大学 | A kind of automobile CAN-bus network data method for detecting abnormality based on isolated forest |
CN110322349A (en) * | 2019-06-25 | 2019-10-11 | 阿里巴巴集团控股有限公司 | A kind of processing method of data, device and equipment |
CN110399935A (en) * | 2019-08-02 | 2019-11-01 | 哈工大机器人(合肥)国际创新研究院 | The real-time method for monitoring abnormality of robot and system based on isolated forest machine learning |
WO2020010701A1 (en) * | 2018-07-11 | 2020-01-16 | 平安科技(深圳)有限公司 | Pollutant anomaly monitoring method and system, computer device, and storage medium |
CN110929643A (en) * | 2019-11-21 | 2020-03-27 | 西北工业大学 | Hyperspectral anomaly detection method based on multiple features and isolated trees |
CN111222556A (en) * | 2019-12-31 | 2020-06-02 | 中国南方电网有限责任公司 | Method and system for identifying electricity utilization category based on decision tree algorithm |
CN111279370A (en) * | 2017-09-29 | 2020-06-12 | 甲骨文国际公司 | Guiding trajectories through communication decision trees using iterative artificial intelligence |
WO2020140678A1 (en) * | 2019-01-04 | 2020-07-09 | 深圳壹账通智能科技有限公司 | Abnormal application detection method and apparatus, and computer device and storage medium |
CN111950585A (en) * | 2020-06-29 | 2020-11-17 | 广东技术师范大学 | XGboost-based underground comprehensive pipe gallery safety condition assessment method |
US20200374720A1 (en) * | 2018-06-04 | 2020-11-26 | Jiangnan University | Method for Detecting Abnormal Data in Sensor Network |
WO2021114821A1 (en) * | 2019-12-12 | 2021-06-17 | 支付宝(杭州)信息技术有限公司 | Isolation forest model construction and prediction method and device based on federated learning |
CN113077165A (en) * | 2021-04-15 | 2021-07-06 | 广东电力交易中心有限责任公司 | Method for judging market force abuse of generator set |
CN113268835A (en) * | 2021-06-21 | 2021-08-17 | 河南工业大学 | Pumped storage hydroelectric generating set health assessment method and device based on extended isolation forest and cloud model |
CN113327172A (en) * | 2021-05-07 | 2021-08-31 | 河南工业大学 | Grain condition data outlier detection method based on isolated forest |
AU2021103976A4 (en) * | 2021-03-22 | 2021-09-09 | Jiangsu University | Asthma diagnosis system based on decision tree and improved SMOTE algorithm |
CN113556254A (en) * | 2021-08-02 | 2021-10-26 | 北京天融信网络安全技术有限公司 | Abnormal alarm method and device, electronic equipment and readable storage medium |
WO2021218314A1 (en) * | 2020-04-27 | 2021-11-04 | 深圳壹账通智能科技有限公司 | Event identification method and apparatus based on position locating, and device and storage medium |
CN113627521A (en) * | 2021-08-09 | 2021-11-09 | 西华大学 | Intelligent logistics unmanned aerial vehicle abnormal behavior identification method based on isolated forest method |
CN114124482A (en) * | 2021-11-09 | 2022-03-01 | 中国电子科技集团公司第三十研究所 | Access flow abnormity detection method and device based on LOF and isolated forest |
-
2022
- 2022-03-16 CN CN202210259150.9A patent/CN114611616B/en active Active
Patent Citations (27)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7127633B1 (en) * | 2001-11-15 | 2006-10-24 | Xiotech Corporation | System and method to failover storage area network targets from one interface to another |
US20070294068A1 (en) * | 2006-05-24 | 2007-12-20 | Jones Jeffrey P | Line-walking recursive partitioning method for evaluating molecular interactions and questions relating to test objects |
CN107453709A (en) * | 2017-07-03 | 2017-12-08 | 重庆大学 | The photovoltaic hot spot method for diagnosing faults that a kind of isolation mech isolation test merges with intersecting measurement |
US20190080142A1 (en) * | 2017-09-13 | 2019-03-14 | X Development Llc | Backup Navigation System for Unmanned Aerial Vehicles |
CN111279370A (en) * | 2017-09-29 | 2020-06-12 | 甲骨文国际公司 | Guiding trajectories through communication decision trees using iterative artificial intelligence |
US20200374720A1 (en) * | 2018-06-04 | 2020-11-26 | Jiangnan University | Method for Detecting Abnormal Data in Sensor Network |
CN108877949A (en) * | 2018-06-11 | 2018-11-23 | 吉林大学 | Down's syndrome screening method based on isolated forest algorithm and voting mechanism |
WO2020010701A1 (en) * | 2018-07-11 | 2020-01-16 | 平安科技(深圳)有限公司 | Pollutant anomaly monitoring method and system, computer device, and storage medium |
CN108765961A (en) * | 2018-08-03 | 2018-11-06 | 东华理工大学 | A kind of floating car data processing method based on modified amplitude limit average filter |
CN109684311A (en) * | 2018-12-06 | 2019-04-26 | 中科恒运股份有限公司 | Abnormal deviation data examination method and device |
CN109753960A (en) * | 2018-12-24 | 2019-05-14 | 西安理工大学 | The underwater unnatural object detection method of isolated forest based on fractal theory |
WO2020140678A1 (en) * | 2019-01-04 | 2020-07-09 | 深圳壹账通智能科技有限公司 | Abnormal application detection method and apparatus, and computer device and storage medium |
CN110149258A (en) * | 2019-04-12 | 2019-08-20 | 北京航空航天大学 | A kind of automobile CAN-bus network data method for detecting abnormality based on isolated forest |
CN110322349A (en) * | 2019-06-25 | 2019-10-11 | 阿里巴巴集团控股有限公司 | A kind of processing method of data, device and equipment |
CN110399935A (en) * | 2019-08-02 | 2019-11-01 | 哈工大机器人(合肥)国际创新研究院 | The real-time method for monitoring abnormality of robot and system based on isolated forest machine learning |
CN110929643A (en) * | 2019-11-21 | 2020-03-27 | 西北工业大学 | Hyperspectral anomaly detection method based on multiple features and isolated trees |
WO2021114821A1 (en) * | 2019-12-12 | 2021-06-17 | 支付宝(杭州)信息技术有限公司 | Isolation forest model construction and prediction method and device based on federated learning |
CN111222556A (en) * | 2019-12-31 | 2020-06-02 | 中国南方电网有限责任公司 | Method and system for identifying electricity utilization category based on decision tree algorithm |
WO2021218314A1 (en) * | 2020-04-27 | 2021-11-04 | 深圳壹账通智能科技有限公司 | Event identification method and apparatus based on position locating, and device and storage medium |
CN111950585A (en) * | 2020-06-29 | 2020-11-17 | 广东技术师范大学 | XGboost-based underground comprehensive pipe gallery safety condition assessment method |
AU2021103976A4 (en) * | 2021-03-22 | 2021-09-09 | Jiangsu University | Asthma diagnosis system based on decision tree and improved SMOTE algorithm |
CN113077165A (en) * | 2021-04-15 | 2021-07-06 | 广东电力交易中心有限责任公司 | Method for judging market force abuse of generator set |
CN113327172A (en) * | 2021-05-07 | 2021-08-31 | 河南工业大学 | Grain condition data outlier detection method based on isolated forest |
CN113268835A (en) * | 2021-06-21 | 2021-08-17 | 河南工业大学 | Pumped storage hydroelectric generating set health assessment method and device based on extended isolation forest and cloud model |
CN113556254A (en) * | 2021-08-02 | 2021-10-26 | 北京天融信网络安全技术有限公司 | Abnormal alarm method and device, electronic equipment and readable storage medium |
CN113627521A (en) * | 2021-08-09 | 2021-11-09 | 西华大学 | Intelligent logistics unmanned aerial vehicle abnormal behavior identification method based on isolated forest method |
CN114124482A (en) * | 2021-11-09 | 2022-03-01 | 中国电子科技集团公司第三十研究所 | Access flow abnormity detection method and device based on LOF and isolated forest |
Non-Patent Citations (7)
Title |
---|
JULIEN LESOUPLE等: ""Generalized isolation forest for anomaly detection"", 《PATTERN RECOGNITION LETTERS》 * |
KEIPOUR A等: ""ALFA:A Dataset for UAV Fault and Anomaly Detection"", 《THE INTERNATIONAL JOURNAL OF ROBOTICS REASEARCH》 * |
WAHID SAKMAN AI FARIZI等: ""Isolation Forest Based Anomaly Detection:A Systematic Literature Review"", 《2021 18TH INTERNATIONAL CONFERENCE ON INFORMATION TECHNOLOGY,COMPUTER AND ELECTRICAL ENGINEERING》 * |
夏火松等: ""基于自编码器和集成学习的半监督异常检测算法"", 《计算机工程与科学》 * |
李国成等: ""基于Bagging二次加权集成的孤立森林窃电检测算法"", 《电力系统自动化》 * |
谢雨等: ""基于随机子空间的扩展隔离林算法"", 《计算机应用》 * |
陈冲等: ""基于孤立森林方法的催化裂化装置排污数据异常识别"", 《西安石油大学学报(自然科学版)》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115185303A (en) * | 2022-09-14 | 2022-10-14 | 南开大学 | Unmanned aerial vehicle patrol path planning method for national parks and natural protected areas |
CN115457414B (en) * | 2022-09-15 | 2023-05-05 | 西华大学 | Unmanned aerial vehicle abnormal behavior identification method based on improved residual error network |
CN115294674A (en) * | 2022-10-09 | 2022-11-04 | 南京信息工程大学 | Unmanned ship navigation state monitoring and evaluating method |
Also Published As
Publication number | Publication date |
---|---|
CN114611616B (en) | 2023-02-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114611616B (en) | Unmanned aerial vehicle intelligent fault detection method and system based on integrated isolated forest | |
CN109766992B (en) | Industrial control abnormity detection and attack classification method based on deep learning | |
CN111027615B (en) | Middleware fault early warning method and system based on machine learning | |
US7716152B2 (en) | Use of sequential nearest neighbor clustering for instance selection in machine condition monitoring | |
CN112415331B (en) | Power grid secondary system fault diagnosis method based on multi-source fault information | |
Zheng et al. | Fault diagnosis method based on supervised particle swarm optimization classification algorithm | |
Perera | Marine engine centered localized models for sensor fault detection under ship performance monitoring | |
CN115797749B (en) | Method and system for detecting state of autonomous underwater vehicle | |
CN111271183B (en) | Method and system for self-adaptive online prediction of state of diesel engine | |
CN113157732A (en) | Underground scraper fault diagnosis method based on PSO-BP neural network | |
CN115561546A (en) | Abnormity detection and alarm system for power system | |
CN109902731B (en) | Performance fault detection method and device based on support vector machine | |
CN115952067A (en) | Database operation abnormal behavior detection method and readable storage medium | |
CN114416423A (en) | Root cause positioning method and system based on machine learning | |
Côme et al. | Aircraft engine health monitoring using self-organizing maps | |
CN116168019B (en) | Power grid fault detection method and system based on machine vision technology | |
CN116563690A (en) | Unmanned aerial vehicle sensor type unbalanced data anomaly detection method and detection system | |
CN111598161A (en) | Engine gas circuit state diagnosis system based on CNN transfer learning | |
CN115964883A (en) | State monitoring and maintenance support system of unmanned ship | |
CN112884027B (en) | Cutting process real-time state monitoring method and device based on pattern recognition | |
CN115293236A (en) | Hybrid clustering-based parallel fault diagnosis method and device for power equipment | |
CN114459760A (en) | Rolling bearing fault diagnosis method and system under strong noise environment | |
Duan et al. | Anomaly Detection Method for Operation and Maintenance Data Based on One-Class Learning | |
Mughaid et al. | Enhancing cybersecurity in scada iot systems: A novel machine learning-based approach for man-in-the-middle attack detection | |
CN108520005A (en) | The wrong report removing method for network active monitoring system based on machine learning |
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 |