CN113962251A - Unmanned aerial vehicle fault detection method and device, electronic equipment and storage medium - Google Patents
Unmanned aerial vehicle fault detection method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN113962251A CN113962251A CN202111076651.5A CN202111076651A CN113962251A CN 113962251 A CN113962251 A CN 113962251A CN 202111076651 A CN202111076651 A CN 202111076651A CN 113962251 A CN113962251 A CN 113962251A
- Authority
- CN
- China
- Prior art keywords
- fault
- data
- bearing
- unmanned aerial
- aerial vehicle
- 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
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Abstract
The invention discloses a fault detection method and device of an unmanned aerial vehicle, electronic equipment and a storage medium, wherein flight state data of the unmanned aerial vehicle are acquired in real time, so that detection results of different parts of the unmanned aerial vehicle are obtained according to different types of flight state data, namely, the attitude detection result of the unmanned aerial vehicle is obtained by using flight attitude data; meanwhile, inputting the operation data of the actuator and the vibration data of the bearing into the trained deep learning network so as to obtain the detection result of the actuator and the detection result of the bearing by means of the trained deep learning network, thereby obtaining whether the actuator and the bearing have faults or not; finally, according to the three detection results, the fault type of the unmanned aerial vehicle can be obtained; therefore, the fault detection method and the fault detection device can realize the detection of various fault types of the unmanned aerial vehicle, have higher accuracy and stronger real-time performance compared with the traditional fault detection method using expert experience, and are more suitable for the fault detection of the unmanned aerial vehicle.
Description
Technical Field
The invention belongs to the technical field of unmanned aerial vehicle fault detection, and particularly relates to a fault detection method and device for an unmanned aerial vehicle, electronic equipment and a storage medium.
Background
The unmanned aerial vehicle can fly in a remote control mode or fly autonomously, has the advantages of light weight, small size, good maneuverability, no physiological constraint of operators, no limitation of flying environment and the like, and is widely applied to the fields of military, civil use and the like; in the flight process of the unmanned aerial vehicle, the unmanned aerial vehicle is easily influenced by complex environmental factors, and inevitably fails, so that the unmanned aerial vehicle cannot normally fly and even has crash accidents; therefore, the fault detection of the unmanned aerial vehicle has important significance for the reliable operation of the unmanned aerial vehicle.
With the improvement of computer capability and the progress of miniaturization, artificial intelligence and ultra-precision technology, the number of airborne electronic devices of the unmanned aerial vehicle is increased, the system is also more and more complex, and higher requirements are provided for the fault detection of the unmanned aerial vehicle; at present, traditional unmanned aerial vehicle detects mainly with data acquisition, and carries out fault diagnosis then mainly relies on expert's experience to unmanned aerial vehicle according to the data of gathering and judges, and this has just proposed higher requirement to the diagnostician, and it has following not enough: the method has the advantages that the accuracy is low, time and labor are wasted, meanwhile, real-time fault diagnosis of the unmanned aerial vehicle and diagnosis of various fault types cannot be realized, and the fault diagnosis requirements of modern unmanned aerial vehicles cannot be met; therefore, it is urgent to provide a method capable of diagnosing a fault of an unmanned aerial vehicle in real time and from multiple aspects.
Disclosure of Invention
The invention aims to provide a fault detection method and device of an unmanned aerial vehicle, electronic equipment and a storage medium, and aims to solve the problems that the existing method for diagnosing faults of the unmanned aerial vehicle by adopting expert experience is low in accuracy rate and cannot diagnose the faults of the unmanned aerial vehicle in real time in multiple aspects.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention provides a fault detection method of an unmanned aerial vehicle, which comprises the following steps:
acquiring flight state data of the unmanned aerial vehicle, wherein the flight state data comprises flight attitude data, actuator operation data and fuselage bearing vibration data;
carrying out flight attitude detection on the unmanned aerial vehicle by using the flight attitude data to obtain an attitude detection result;
inputting the actuator operation data and the bearing vibration data into a trained deep learning network to obtain an actuator detection result and a bearing detection result;
and obtaining a fault detection result of the unmanned aerial vehicle according to the attitude detection result, the actuator detection result and the bearing detection result.
Based on the above disclosure, the flight state data of the unmanned aerial vehicle is obtained in real time, so that the detection results of different parts of the unmanned aerial vehicle are obtained according to different types of flight state data, namely, the attitude detection result of the unmanned aerial vehicle is obtained by using the flight attitude data, and whether the unmanned aerial vehicle deviates from a steady state is judged; meanwhile, inputting the operation data of the actuator and the vibration data of the bearing into the trained deep learning network so as to obtain the detection result of the actuator and the detection result of the bearing by means of the trained deep learning network, thereby obtaining whether the actuator and the bearing have faults or not; and finally, according to the three detection results, the fault type of the unmanned aerial vehicle can be obtained, so that the detection of various fault types of the unmanned aerial vehicle is realized.
Through the design, the flight attitude of the unmanned aerial vehicle can be analyzed in real time through the flight state data of the unmanned aerial vehicle, and whether the bearing and the actuator of the unmanned aerial vehicle break down or not is detected in real time, so that the detection of various fault types of the unmanned aerial vehicle can be realized.
In one possible design, the flight attitude data includes angular velocity, acceleration, and altitude, wherein the flight attitude data is utilized to perform flight attitude detection on the unmanned aerial vehicle, and obtaining the attitude detection result includes:
calculating to obtain the flight angle of the unmanned aerial vehicle by using the angular velocity;
calculating the altitude change rate of the unmanned aerial vehicle by using the altitude;
and obtaining the attitude detection result according to the acceleration, the angular velocity, the flight angle and the altitude change rate.
Based on the above disclosure, the flight angle and the altitude change rate of the unmanned aerial vehicle can be calculated through the flight attitude data, so that the flight attitude of the unmanned aerial vehicle can be obtained according to the flight angle, the altitude change rate, the angular velocity and the acceleration in the flight attitude data, that is, whether the unmanned aerial vehicle deviates from a steady state or not is obtained by judging whether the parameters exceed the threshold values or not.
In one possible design, the trained deep learning network comprises a trained SDA deep learning network;
correspondingly, inputting the actuator operation data into the trained deep learning network to obtain an actuator detection result, including:
inputting the actuator operation data into the trained SDA deep learning network to obtain a constant deviation fault coefficient, a constant gain fault coefficient and a stuck fault coefficient of the actuator through the trained SDA deep learning network;
and obtaining the detection result of the actuator according to the constant deviation fault coefficient, the constant gain fault coefficient and the stuck fault coefficient.
Based on the above disclosure, the invention can obtain the constant deviation fault coefficient, the constant gain fault coefficient and the stuck fault coefficient of the actuator through the trained SDA deep learning network, so as to obtain the actuator detection result according to the three coefficients, which substantially is: and judging whether the actuator has a constant deviation fault, a constant gain fault or a stuck fault according to the three coefficients.
In one possible design, the trained deep learning network comprises a trained BP neural network, and the fuselage bearing vibration data comprises a plurality of bearing vibration signals, wherein the bearing vibration signals are sequenced according to sequence numbers, and the sequence numbers are serial numbers of bearings in the unmanned aerial vehicle;
correspondingly, the bearing vibration data is input into the deep learning network after training, and a bearing detection result is obtained, wherein the bearing detection result comprises the following steps:
resampling each bearing vibration signal in the plurality of bearing vibration signals to obtain a vibration signal section corresponding to each bearing vibration signal;
carrying out wavelet packet decomposition on a vibration signal section corresponding to each bearing vibration signal to obtain a frequency band characteristic corresponding to each bearing vibration signal;
performing fast Fourier transform on a vibration signal section corresponding to each bearing vibration signal to obtain a frequency spectrum characteristic corresponding to each bearing vibration signal;
and inputting the frequency band characteristic and the frequency spectrum characteristic corresponding to each bearing vibration signal into the trained BP neural network to obtain the bearing detection result.
Based on the above-mentioned disclosure, through the BP neural network after the training, can realize the fault detection of all bearings in the unmanned aerial vehicle, and can learn the position of the bearing that breaks down through the serial number to be favorable to operating personnel to the judgement of fault level.
In one possible design, before inputting the actuator operation data into the trained deep learning network to obtain the actuator detection result, the method further includes:
acquiring fault data of an actuator of the unmanned aerial vehicle, wherein the fault data comprises actuator constant deviation fault data, actuator constant gain fault data and actuator stuck fault data;
and inputting the fault data into an SDA deep learning network for training to obtain the trained SDA deep learning network.
In one possible design, the SDA deep learning network includes a multi-layer self-encoder, wherein inputting the fault data into the SDA deep learning network for training to obtain the trained SDA deep learning network includes:
s01, for a first-layer self-encoder, encoding the fault data by using an encoding function of an SDA deep learning network to obtain encoded fault data;
s02, decoding the coding fault data by using a decoding function of the SDA deep learning network to obtain decoding output data;
s03, performing similarity fitting on the decoded output data and the fault data to obtain a data similarity function;
s04, obtaining a loss function of the SDA deep learning network by using the data similarity function;
s05, minimizing the loss function so as to finish the training of the first layer of self-encoder after the loss function is minimized;
and S06, inputting the coding fault data into a second-layer self-encoder, and repeating the steps S01-S06 until the training of the multi-layer self-encoder is finished so as to obtain the trained SDA deep learning network.
In one possible design, before inputting the bearing vibration data into the trained deep learning network to obtain the bearing detection result, the method further includes:
acquiring vibration signals of all bearings in the unmanned aerial vehicle during normal work and vibration signals during faults;
resampling the vibration signal when the bearing normally works and the vibration signal when the bearing fails to obtain a normal vibration signal section corresponding to the vibration signal when the bearing normally works and a fault vibration signal section corresponding to the vibration signal when the bearing fails;
extracting the amplitudes of the normal vibration signal section and the fault vibration signal section;
dividing the fault vibration signal into a plurality of fault levels according to the difference between the amplitude of the normal vibration signal section and the amplitude of the fault vibration signal section;
respectively carrying out wavelet packet decomposition and fast Fourier transform on the fault vibration signal segment to respectively obtain frequency band characteristics and frequency spectrum characteristics corresponding to the fault vibration signal segment;
dividing the frequency band characteristics corresponding to the fault vibration signal section into a plurality of frequency band characteristic sections according to the fault level, and dividing the frequency spectrum characteristics corresponding to the fault vibration signal section into a plurality of frequency spectrum characteristic sections according to the fault level;
and inputting the plurality of frequency band characteristic segments and the plurality of frequency spectrum characteristic segments into a BP neural network as sample data for training to obtain the trained BP neural network.
Based on the disclosure, the invention discloses a specific training method of the BP neural network, namely, the training of the BP neural network is carried out by using a plurality of frequency band characteristic segments and frequency spectrum characteristic segments which are divided into fault levels as sample data; therefore, when fault identification is carried out subsequently, the specific fault grade of each bearing can be directly obtained, and the fault severity degree can be conveniently divided by an operator so that the operator can take emergency measures of corresponding grades.
In a second aspect, the present invention provides a fault detection device for an unmanned aerial vehicle, including: an acquisition unit, an attitude detection unit, a body component detection unit, and a failure result generation unit;
the acquiring unit is used for acquiring flight state data of the unmanned aerial vehicle, wherein the flight state data comprises flight attitude data, actuator operation data and fuselage bearing vibration data;
the attitude detection unit is used for detecting the flight attitude of the unmanned aerial vehicle by using the flight attitude data to obtain an attitude detection result;
the machine body component detection unit is used for inputting the actuator operation data and the bearing vibration data into the trained deep learning network to obtain an actuator detection result and a bearing detection result;
the fault result generation unit is used for obtaining a fault detection result of the unmanned aerial vehicle according to the attitude detection result, the actuator detection result and the bearing detection result.
In a third aspect, the present invention provides an electronic device, including a memory, a processor, and a transceiver, which are communicatively connected in sequence, where the memory is used to store a computer program, the transceiver is used to transmit and receive messages, and the processor is used to read the computer program and execute the method for detecting a fault of the drone according to the first aspect or any one of the possible designs of the first aspect.
In a fourth aspect, the present invention provides a storage medium having stored thereon instructions that, when executed on a computer, perform a method of fault detection for a drone as described in the first aspect or any one of the possible designs of the first aspect.
In a fifth aspect, the present invention provides a computer program product comprising instructions which, when run on a computer, cause the computer to perform a method of fault detection for a drone as described in the first aspect or any one of the possible designs of the first aspect.
Drawings
Fig. 1 is a schematic flow chart illustrating steps of a method for detecting a fault of an unmanned aerial vehicle according to the present invention;
FIG. 2 is a schematic diagram of a network structure of a layer of self-encoder provided in the present invention;
fig. 3 is a schematic structural diagram of a fault detection device of an unmanned aerial vehicle provided by the invention;
fig. 4 is a schematic structural diagram of an electronic device provided in the present invention.
Detailed Description
The invention is further described with reference to the following figures and specific embodiments. It should be noted that the description of the embodiments is provided to help understanding of the present invention, but the present invention is not limited thereto. Specific structural and functional details disclosed herein are merely illustrative of example embodiments of the invention. This invention may, however, be embodied in many alternate forms and should not be construed as limited to the embodiments set forth herein.
It will be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of example embodiments of the present invention.
It should be understood that, for the term "and/or" as may appear herein, it is merely an associative relationship that describes an associated object, meaning that three relationships may exist, e.g., a and/or B may mean: a exists alone, B exists alone, and A and B exist at the same time; for the term "/and" as may appear herein, which describes another associative object relationship, it means that two relationships may exist, e.g., a/and B, may mean: a exists independently, and A and B exist independently; in addition, for the character "/" that may appear herein, it generally means that the former and latter associated objects are in an "or" relationship.
Examples
As shown in fig. 1, according to the fault detection method for the unmanned aerial vehicle provided by the first aspect of the present embodiment, the flight attitude of the unmanned aerial vehicle can be analyzed in real time through the flight state data of the unmanned aerial vehicle, and whether the bearing and the actuator of the unmanned aerial vehicle have faults or not can be detected in real time.
In this embodiment, the fault detection method provided by the embodiment can be implemented by, but not limited to, a Computer device with certain computing resources, for example, an electronic device such as a Personal Computer (PC, which refers to a multipurpose Computer with a size, price and performance suitable for Personal use; a desktop Computer, a notebook Computer, a mini-notebook Computer, a tablet Computer, a super book, etc. all belong to a Personal Computer), a Personal digital assistant (PAD), etc.; aforementioned computer equipment can set up on unmanned aerial vehicle, also can set up in unmanned aerial vehicle's control chamber to the realization is to real-time and many-sided fault detection of unmanned aerial vehicle.
The fault detection of the drone provided in the first aspect of this embodiment may include, but is not limited to, the following steps S1 to S4.
S1, acquiring flight state data of the unmanned aerial vehicle, wherein the flight state data comprise flight attitude data, actuator operation data and fuselage bearing vibration data.
Step S1, acquiring flight data of the unmanned aerial vehicle in the flight process so as to provide a data basis for subsequent fault detection; in this embodiment, the flight attitude data is data such as an angular velocity, an acceleration, and an altitude of the unmanned aerial vehicle during flight; the operation data of the actuator is data such as lift force, roll moment, pitch moment and yaw moment of the unmanned aerial vehicle; and fuselage bearing vibration data is the vibration signal of the bearing used in various mechanical structures of the unmanned aerial vehicle.
Certainly, in order to locate the position of the failed bearing in the unmanned aerial vehicle in time when the bearing is detected to be failed subsequently, in this embodiment, the bearings in the unmanned aerial vehicle can be numbered, so that the bearing vibration signals after being numbered are sorted, that is, the serial number of the signal is the serial number of the bearing; for example, there are 3 bearings in the drone; respectively numbered as: the bearing comprises a No. 1 bearing, a No. 2 bearing and a No. 3 bearing; wherein, the No. 1 bearing is a motor bearing; the No. 2 bearing is a rotor bearing correspondingly; the No. 3 bearing corresponds to a holder bearing of the unmanned aerial vehicle; then the vibration signal corresponding to bearing No. 1 is: the vibration signal that first bearing vibration signal, No. 2 bearings correspond is: the vibration signal of the second bearing, the vibration signal that No. 3 bearings correspond is: a third bearing vibration signal; therefore, the bearings can be distinguished and positioned according to the serial numbers of the vibration signals.
In the present embodiment, the bearing vibration data may be acquired, but not limited to, by a vibration sensor installed at the periphery of the bearing, for example.
After acquiring the flight state data of the drone, the flight state data can be used to detect multiple fault types of the drone, as shown in steps S2 and S3 below.
And S2, carrying out flight attitude detection on the unmanned aerial vehicle by using the flight attitude data to obtain an attitude detection result.
Step S2 is to determine whether the flight of the unmanned aerial vehicle deviates from the steady state by using the flight attitude data, that is, determine whether the flight altitude, acceleration, and angular velocity of the unmanned aerial vehicle suddenly change according to the flight attitude data, thereby determining whether the flight attitude of the unmanned aerial vehicle is normal, and if the sudden change occurs, it indicates that the flight attitude of the unmanned aerial vehicle deviates from the steady state, and a fault occurs in the attitude.
Since the types of data included in the flight attitude data have been described above, a detailed attitude detection method is exemplified below as shown in the following steps S20 to S22.
And S20, calculating to obtain the flight angle of the unmanned aerial vehicle by utilizing the angular velocity.
In this embodiment, the real-time flight angle of the unmanned aerial vehicle, that is, the flight angle at each moment, needs to be obtained through the angular velocity, and therefore, assuming that the current time of flight of the unmanned aerial vehicle is t, the flight angle α can be calculated according to the following formula:
in the formula, β represents the angular velocity of the drone; namely, by the formula, the flight angle (namely the attitude angle) of the unmanned aerial vehicle changing along with the time can be calculated.
And S21, calculating to obtain the height change rate of the unmanned aerial vehicle by utilizing the height.
Similarly, the same principle can be used to calculate the altitude change rate of the drone, that is, the change rate of the altitude with time, and the calculation formula is as follows:
in the foregoing formula, H represents the altitude change rate, and H represents the altitude of the unmanned aerial vehicle.
After the flight angle and the altitude change rate of the drone are obtained, the flight attitude data and the flight attitude data can be combined to obtain the attitude detection result of the drone, as shown in the following step S22.
And S22, obtaining the attitude detection result according to the acceleration, the angular velocity, the flight angle and the altitude change rate.
In this embodiment, for example, when the acceleration is greater than a first preset value (e.g. 6g), the angular velocity is greater than a second preset value (e.g. 220 °/s), the flight angle is greater than a third preset value (e.g. 75 °), and the altitude change rate is greater than a fourth preset value (0.7), it can be determined that the drone deviates from the steady state.
For example, when the acceleration is greater than 6g, the unmanned aerial vehicle can be considered to receive an impact, such as a bird-hitting accident; and when the altitude change rate is greater than 0.7 and the flight angle is greater than 75 degrees, the unmanned aerial vehicle attitude controller can be considered to be in fault.
Through the aforesaid elaboration to unmanned aerial vehicle gesture detection step, can realize the real-time detection of unmanned aerial vehicle gesture through flight attitude data to whether the flight attitude that reachs unmanned aerial vehicle breaks down, whether skew stable state promptly.
Meanwhile, in the embodiment, the trained deep learning network is further used for performing fault recognition on the actuator operation data and the bearing vibration data, so that fault detection of the actuator and the bearing of the unmanned aerial vehicle is realized, as shown in the following step S3.
And S3, inputting the actuator operation data and the bearing vibration data into the trained deep learning network to obtain an actuator detection result and a bearing detection result.
In this embodiment, the example trained deep learning network includes a trained SDA (Stacked automatic encoder) deep learning network and a trained bp (back propagation) neural network; namely, the actuator is subjected to fault detection by using the trained SDA deep learning network, and the bearing in the unmanned aerial vehicle is subjected to fault detection by using the trained BP neural network.
Therefore, the following describes the training process for both:
the training process of the exemplary SDA deep learning network may be, but is not limited to, as shown in steps S001 and S002 described below.
And S001, acquiring fault data of an actuator of the unmanned aerial vehicle, wherein the fault data comprises actuator constant deviation fault data, actuator constant gain fault data and actuator stuck fault data.
And S002, inputting the fault data into the SDA deep learning network for training to obtain the trained SDA deep learning network.
Namely, the constant deviation fault data of the actuator, the constant gain fault data of the actuator and the stuck fault data of the actuator are used as sample data and input into the SDA deep learning network for continuous iterative training, so as to obtain the trained SDA deep learning network.
In this embodiment, the SDA deep learning network includes multiple layers of self-encoders, that is, a network formed by stacking multiple layers of self-encoders, and meanwhile, each layer of self-encoder is a neural network formed by an input layer, a hidden layer, and an output layer, and the structure of the self-encoder can be seen in fig. 2, where data is encoded from the input layer to the hidden layer, and decoded from the hidden layer to the output layer; therefore, the following describes a specific procedure of training on the structure of the neural network, and may be, but is not limited to, the following steps S01 to S06.
And S01, for the first-layer self-encoder, encoding the fault data by using an SDA deep learning network encoding function to obtain encoded fault data.
And S02, decoding the coding fault data by using a decoding function of the SDA deep learning network to obtain decoding output data.
And S03, performing similarity fitting on the decoded output data and the fault data to obtain a data similarity function.
Steps S01, S02, and S03 are processes of inputting the failure data of the actuator to the hidden layer, then outputting the failure data from the hidden layer through the output layer, thereby performing data encoding to obtain encoded failure data, then decoding the encoded failure data to obtain decoded output data, and finally, comparing the decoded output data with the failure data to judge the similarity between the two, thereby realizing the similarity fit between the two.
In the present embodiment, the encoded failure data and the output decoded data can be expressed by the following formulas:
z＝f(x)＝sf(Wx+p) (1)
y＝g(z)＝sg(WTz+q) (2)
in the above equation (1), z represents encoded data, x represents failure data, f represents an encoding function, and sfThen represents the activation function of the input layer in the self-encoder; in the formula (2), y represents output decoded data, g represents a decoding function, and sgRepresenting an activation function of an output layer, and W represents a weight matrix of an input layer and a hidden layer; and WTThen represents the weight matrix between the hidden layer and the output layer, p represents the bias vector on the hidden layer, and q represents the bias vector on the output layer.
For example, but not limited to, the fitting of the similarity between the decoded output data and the failure data may be implemented by using an error function, that is, the closeness of the two is represented by using the error function, and the smaller the value of the error function, the closer the two are, the higher the similarity is, the more reliable the training result is, and vice versa.
The similarity function of the output decoded output data to the fault data is given below, as shown below:
in the above formula (3), xiIndicating the ith fault data, yiAnd the output decoding data corresponding to the ith fault data is shown, n represents the total number of the fault data, and L (x, y) represents the similarity of the decoding output data and the fault data.
After the similarity function is obtained, the similarity function may be used to obtain a loss function of the SDA deep learning network, that is, a loss function of the first-layer self-encoder, so as to perform minimization processing on the loss function, and make the first-layer self-encoder converge, thereby implementing training of the first-layer self-encoder, as shown in steps S04 and S05 below.
And S04, obtaining a loss function of the SDA deep learning network by using the data similarity function.
And S05, minimizing the loss function so as to finish the training of the first layer of self-encoder after the loss function is minimized.
The loss function based on the similarity function is provided as follows:
in the above-mentioned formula (4),
represents the weight coefficient of the first layer self-encoder, and R is a fault data set composed of n fault data.
Therefore, the steps S01 to S05 are to train the recognition of the fault data, and finally obtain the weight coefficient of the first-layer self-encoder occupying the whole SDA deep learning network through the minimization process of the loss function, so as to realize the weight setting of the SDA deep learning network through data training.
Therefore, the encoded fault data is input to the second layer self-encoder, and the above steps are repeated, so as to loop through the loop, and continuously enter the next layer self-encoder for training, so as to obtain the weight of each layer self-encoder, and thus, a trained SDA deep learning network can be obtained, as shown in the following step S06.
And S06, inputting the coding fault data into a second-layer self-encoder, and repeating the steps S01-S06 until the training of the multi-layer self-encoder is finished so as to obtain the trained SDA deep learning network.
After the foregoing steps S01-S06, a trained SDA deep learning network is obtained, which can be represented by, but is not limited to, the following expression:
yout＝m+byin+c (5)
in the above formula (5), yin＝[E1 E2 E3 E4]TRepresents the desired output of the actuator, and E1、E2、E3And E4The lift, roll, pitch and yaw moments that the actuator is expected to output are represented; y isout＝[E′1 E′2 E′3 E′4]TThen represents the actual output of the actuator, and E'1、E′2、E′3And E'4Respectively representing the lift force, the roll moment, the pitch moment and the yaw moment which are actually output by the actuator; m, b and c respectively represent a constant deviation fault coefficient, a constant gain fault coefficient and a stuck fault coefficient; in addition, the term]TRepresenting a transpose operation.
Therefore, when the actuator fault is identified subsequently, the values of the constant deviation fault coefficient, the constant gain fault coefficient and the stuck fault coefficient can be obtained according to the trained SDA deep learning network, so that the actuator detection result is obtained.
Similarly, the training process of the BP neural network is described below, as shown in steps S21-S27.
And S21, acquiring vibration signals of all bearings in the unmanned aerial vehicle during normal work and vibration signals during faults.
S22, resampling is carried out on the vibration signal when the bearing normally works and the vibration signal when the bearing fails, and a normal vibration signal section corresponding to the vibration signal when the bearing normally works and a fault vibration signal section corresponding to the vibration signal when the bearing fails are obtained.
Step S22 is to perform signal processing on the vibration signal in normal operation and the vibration signal in fault operation through resampling, so as to obtain a normal vibration signal segment and a fault vibration signal segment, so as to provide a data basis for subsequent network training.
In this embodiment, the resampling may be performed by using, but not limited to, nearest neighbor interpolation, bilinear interpolation, or cubic convolution interpolation.
And S23, extracting the amplitudes of the normal vibration signal section and the fault vibration signal section.
And S24, dividing the fault vibration signal into a plurality of fault grades according to the difference value between the amplitude of the normal vibration signal section and the amplitude of the fault vibration signal section.
Step S24 is to divide the level of the fault vibration signal according to the difference between the amplitudes, so as to divide the normal vibration signal segment and the fault vibration signal segment according to the fault level, thereby facilitating the training of the subsequent network, and meanwhile, directly obtaining the fault level of the bearing during the fault detection.
In the present embodiment, the dividing of the fault vibration signal into a plurality of fault levels according to the aforementioned difference values may be, but is not limited to, adopting the following method: the first step is as follows: acquiring the maximum difference value between the amplitude of the normal vibration signal section and the amplitude of the fault vibration signal section; the second step is that: equally dividing the maximum difference value into a plurality of data segments; a third step; and dividing the fault grade according to the corresponding numerical value interval of the data section.
For example, 0-4 is a first failure level (mild in severity); 4-8 is a second failure level (medium severity); above 8 is the third failure level (severity is severe).
After the fault grade is divided, the frequency band and the frequency spectrum of the fault vibration signal segment can be extracted and divided according to the fault grade, so that a plurality of frequency band characteristic segments and frequency spectrum characteristic segments are obtained to serve as sample data; and finally, inputting the sample data into the BP neural network for iterative training to obtain the trained BP neural network, as shown in the following steps S25-S27.
And S25, respectively carrying out wavelet packet decomposition and fast Fourier transform on the fault vibration signal section to respectively obtain frequency band characteristics and frequency spectrum characteristics corresponding to the fault vibration signal section.
And S26, dividing the frequency band characteristics corresponding to the fault vibration signal section into a plurality of frequency band characteristic sections according to the fault level, and dividing the frequency spectrum characteristics corresponding to the fault vibration signal section into a plurality of frequency spectrum characteristic sections according to the fault level.
And S27, inputting the plurality of frequency band characteristic segments and the plurality of frequency spectrum characteristic segments into a BP neural network as sample data for training to obtain the trained BP neural network.
Therefore, through the steps S21-S27, the trained BP neural network can be obtained, so that when the fault of the unmanned aerial vehicle is detected, the vibration signals corresponding to the bearings are input into the trained BP neural network, and whether the bearings are in fault or not and the fault grades corresponding to the faulted bearings are obtained.
After the trained SDA deep learning network and the trained BP neural network are obtained, the actuator and bearing detection may be performed, as shown in steps S31 to S32 and steps S33 to S36.
Inputting the actuator operation data into the trained deep learning network to obtain an actuator detection result, wherein the method comprises the following steps of S31-S32:
and S31, inputting the actuator operation data into the trained SDA deep learning network to obtain a constant deviation fault coefficient, a constant gain fault coefficient and a stuck fault coefficient of the actuator through the trained SDA deep learning network.
And S32, obtaining the detection result of the actuator according to the constant deviation fault coefficient, the constant gain fault coefficient and the stuck fault coefficient.
As already described above, the trained SDA deep learning network can be represented by the foregoing equation (5), and therefore, after the actuator operation data is input to the trained SDA deep learning network, the constant deviation fault coefficient, the constant gain fault coefficient, and the stuck fault coefficient of the unmanned aerial vehicle can be obtained; thus, the actuator detection result can be obtained according to the coefficient.
In this embodiment, for example, when the constant deviation fault coefficient is not equal to 0, the constant gain fault coefficient is equal to 1, and the stuck fault coefficient is equal to 0, it is determined that the actuator has the constant deviation fault; when the constant deviation fault coefficient is equal to 0, the constant gain fault coefficient is not equal to 1, and the stuck fault coefficient is equal to 0, determining that the actuator has a constant gain fault; and when the constant deviation fault coefficient is equal to 0, the constant gain fault coefficient is equal to 0, and the stuck fault coefficient is not equal to 0, determining that the actuator has stuck fault.
Similarly, in the present embodiment, the bearing vibration data is input to the trained deep learning network to obtain the bearing detection result, but the method may include, but is not limited to, the following steps S33 to S36.
S33, resampling is carried out on each bearing vibration signal in the plurality of bearing vibration signals, and a vibration signal section corresponding to each bearing vibration signal is obtained.
And S34, carrying out wavelet packet decomposition on the vibration signal section corresponding to each bearing vibration signal to obtain the frequency band characteristic corresponding to each bearing vibration signal.
S35, carrying out fast Fourier transform on the vibration signal section corresponding to each bearing vibration signal to obtain the frequency spectrum characteristic corresponding to each bearing vibration signal.
And S36, inputting the frequency band characteristic and the frequency spectrum characteristic corresponding to each bearing vibration signal into the trained BP neural network to obtain the bearing detection result.
Step S33, step S34 and step S35 are consistent with the principles of step S22, step S25 and step S26, and the signals are resampled, and then the frequency band and the frequency spectrum of each bearing vibration signal pair are obtained by wavelet packet decomposition and fast fourier transform; and finally, inputting the frequency band and the frequency spectrum into the trained BP neural network to obtain a bearing detection result.
As described above, the BP neural network is trained by using a plurality of spectrum feature segments and band feature segments divided according to fault levels; therefore, after the frequency band characteristic and the frequency spectrum characteristic of the bearing vibration signal are input into the trained BP neural network, the fault grade of the bearing can be obtained, and the serial number of the bearing is obtained according to the serial number of the bearing vibration signal, so that the position of the bearing is obtained.
Finally, the fault detection result of the unmanned aerial vehicle can be obtained comprehensively according to the attitude detection result, the actuator detection result and the bearing detection result, as shown in step S4.
And S4, obtaining a fault detection result of the unmanned aerial vehicle according to the attitude detection result, the actuator detection result and the bearing detection result.
The following is illustrated by way of an example:
assuming that the attitude detection result is that the unmanned aerial vehicle deviates from a steady state; the detection result of the actuator is as follows: the actuator has a stuck fault; and the bearing detection result is: the 3 rd bearing vibration signal has a fault, and the fault grade is a second fault grade; therefore, the fault detection result is: the acceleration of the unmanned aerial vehicle is larger than 6g, the acceleration is larger than 220 degrees/s, the flight angle is larger than 75 degrees, and the altitude change rate is larger than 0.7; the actuator has a stuck fault; the cradle head bearing of the unmanned aerial vehicle has medium faults.
Therefore, by the fault detection method of the unmanned aerial vehicle, which is described in detail by the steps S1-S4 and the sub-steps, the flight attitude of the unmanned aerial vehicle can be analyzed in real time through the flight state data of the unmanned aerial vehicle, and whether the bearing and the actuator of the unmanned aerial vehicle break down or not can be detected in real time, so that the detection of various fault types of the unmanned aerial vehicle can be realized.
As shown in fig. 3, a second aspect of the present embodiment provides a hardware device for implementing the method for detecting a failure of an unmanned aerial vehicle in the first aspect of the embodiment, including: the device comprises an acquisition unit, an attitude detection unit, a body component detection unit and a fault result generation unit.
The acquiring unit is used for acquiring flight state data of the unmanned aerial vehicle, wherein the flight state data comprises flight attitude data, actuator operation data and fuselage bearing vibration data.
And the attitude detection unit is used for detecting the flight attitude of the unmanned aerial vehicle by using the flight attitude data to obtain an attitude detection result.
And the machine body component detection unit is used for inputting the actuator operation data and the bearing vibration data into the trained deep learning network to obtain an actuator detection result and a bearing detection result.
The fault result generation unit is used for obtaining a fault detection result of the unmanned aerial vehicle according to the attitude detection result, the actuator detection result and the bearing detection result.
For the working process, the working details, and the technical effects of the hardware apparatus provided in this embodiment, reference may be made to the first aspect of the embodiment, which is not described herein again.
As shown in fig. 4, a third aspect of the present embodiment provides an electronic device, including: the unmanned aerial vehicle fault detection device comprises a memory, a processor and a transceiver which are sequentially connected in a communication mode, wherein the memory is used for storing computer programs, the transceiver is used for transceiving messages, and the processor is used for reading the computer programs and executing the unmanned aerial vehicle fault detection method according to the first aspect of the embodiment.
For example, the Memory may include, but is not limited to, a Random Access Memory (RAM), a Read Only Memory (ROM), a Flash Memory (Flash Memory), a First In First Out (FIFO), and/or a First In Last Out (FILO), and the like; the processor may not be limited to a microprocessor of a model number STM32F105 series, a reduced instruction set computer (RSIC) microprocessor, an architecture processor such as X86, or a processor integrated with a neural-Network Processing Unit (NPU); the transceiver may be, but is not limited to, a wireless fidelity (WIFI) wireless transceiver, a bluetooth wireless transceiver, a General Packet Radio Service (GPRS) wireless transceiver, a ZigBee wireless transceiver (ieee802.15.4 standard-based low power local area network protocol), a 3G transceiver, a 4G transceiver, and/or a 5G transceiver, etc. In addition, the device may also include, but is not limited to, a power module, a display screen, and other necessary components.
For the working process, the working details, and the technical effects of the computer main device provided in this embodiment, reference may be made to the first aspect of the embodiment, which is not described herein again.
A fourth aspect of the present embodiment provides a storage medium storing instructions including the method for detecting a failure of a drone according to the first aspect of the present embodiment, that is, the storage medium stores instructions that, when executed on a computer, perform the method for detecting a failure of a drone according to the first aspect.
The storage medium refers to a carrier for storing data, and may include, but is not limited to, a floppy disk, an optical disk, a hard disk, a flash Memory, a flash disk and/or a Memory Stick (Memory Stick), etc., and the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.
For the working process, the working details, and the technical effects of the storage medium provided in this embodiment, reference may be made to the first aspect of the embodiment, which is not described herein again.
A fifth aspect of the present embodiments provides a computer program product containing instructions which, when run on a computer, cause the computer to perform the method for detecting a failure of a drone according to the first aspect of the present embodiments, wherein the computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable device.
Finally, it should be noted that: the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A fault detection method of an unmanned aerial vehicle is characterized by comprising the following steps:
acquiring flight state data of the unmanned aerial vehicle, wherein the flight state data comprises flight attitude data, actuator operation data and fuselage bearing vibration data;
carrying out flight attitude detection on the unmanned aerial vehicle by using the flight attitude data to obtain an attitude detection result;
inputting the actuator operation data and the bearing vibration data into a trained deep learning network to obtain an actuator detection result and a bearing detection result;
and obtaining a fault detection result of the unmanned aerial vehicle according to the attitude detection result, the actuator detection result and the bearing detection result.
2. The method of claim 1, wherein the flight attitude data comprises angular velocity, acceleration, and altitude, and wherein using the flight attitude data to perform flight attitude detection for the drone results in an attitude detection result comprises:
calculating to obtain the flight angle of the unmanned aerial vehicle by using the angular velocity;
calculating the altitude change rate of the unmanned aerial vehicle by using the altitude;
and obtaining the attitude detection result according to the acceleration, the angular velocity, the flight angle and the altitude change rate.
3. The method of claim 1, wherein the trained deep learning network comprises a trained SDA deep learning network;
correspondingly, inputting the actuator operation data into the trained deep learning network to obtain an actuator detection result, including:
inputting the actuator operation data into the trained SDA deep learning network to obtain a constant deviation fault coefficient, a constant gain fault coefficient and a stuck fault coefficient of the actuator through the trained SDA deep learning network;
and obtaining the detection result of the actuator according to the constant deviation fault coefficient, the constant gain fault coefficient and the stuck fault coefficient.
4. The method of claim 1, wherein the trained deep learning network comprises a trained BP neural network, the fuselage bearing vibration data comprises a plurality of bearing vibration signals, wherein the plurality of bearing vibration signals are ordered by sequence number, and the sequence number is a number of a bearing in the drone;
correspondingly, the bearing vibration data is input into the deep learning network after training, and a bearing detection result is obtained, wherein the bearing detection result comprises the following steps:
resampling each bearing vibration signal in the plurality of bearing vibration signals to obtain a vibration signal section corresponding to each bearing vibration signal;
carrying out wavelet packet decomposition on a vibration signal section corresponding to each bearing vibration signal to obtain a frequency band characteristic corresponding to each bearing vibration signal;
performing fast Fourier transform on a vibration signal section corresponding to each bearing vibration signal to obtain a frequency spectrum characteristic corresponding to each bearing vibration signal;
and inputting the frequency band characteristic and the frequency spectrum characteristic corresponding to each bearing vibration signal into the trained BP neural network to obtain the bearing detection result.
5. The method of claim 3, wherein before inputting the actuator operating data into the trained deep learning network to obtain actuator detection results, the method further comprises:
acquiring fault data of an actuator of the unmanned aerial vehicle, wherein the fault data comprises actuator constant deviation fault data, actuator constant gain fault data and actuator stuck fault data;
and inputting the fault data into an SDA deep learning network for training to obtain the trained SDA deep learning network.
6. The method of claim 5, wherein the SDA deep learning network comprises a multi-layer self-encoder, and wherein inputting the fault data into the SDA deep learning network for training to obtain the trained SDA deep learning network comprises:
s01, for a first-layer self-encoder, encoding the fault data by using an encoding function of an SDA deep learning network to obtain encoded fault data;
s02, decoding the coding fault data by using a decoding function of the SDA deep learning network to obtain decoding output data;
s03, performing similarity fitting on the decoded output data and the fault data to obtain a data similarity function;
s04, obtaining a loss function of the SDA deep learning network by using the data similarity function;
s05, minimizing the loss function so as to finish the training of the first layer of self-encoder after the loss function is minimized;
and S06, inputting the coding fault data into a second-layer self-encoder, and repeating the steps S01-S06 until the training of the multi-layer self-encoder is finished so as to obtain the trained SDA deep learning network.
7. The method of claim 4, wherein before inputting the bearing vibration data into the trained deep learning network to derive a bearing test result, the method further comprises:
acquiring vibration signals of all bearings in the unmanned aerial vehicle during normal work and vibration signals during faults;
resampling the vibration signal when the bearing normally works and the vibration signal when the bearing fails to obtain a normal vibration signal section corresponding to the vibration signal when the bearing normally works and a fault vibration signal section corresponding to the vibration signal when the bearing fails;
extracting the amplitudes of the normal vibration signal section and the fault vibration signal section;
dividing the fault vibration signal into a plurality of fault levels according to the difference between the amplitude of the normal vibration signal section and the amplitude of the fault vibration signal section;
respectively carrying out wavelet packet decomposition and fast Fourier transform on the fault vibration signal segment to respectively obtain frequency band characteristics and frequency spectrum characteristics corresponding to the fault vibration signal segment;
dividing the frequency band characteristics corresponding to the fault vibration signal section into a plurality of frequency band characteristic sections according to the fault level, and dividing the frequency spectrum characteristics corresponding to the fault vibration signal section into a plurality of frequency spectrum characteristic sections according to the fault level;
and inputting the plurality of frequency band characteristic segments and the plurality of frequency spectrum characteristic segments into a BP neural network as sample data for training to obtain the trained BP neural network.
8. The utility model provides an unmanned aerial vehicle's fault detection device which characterized in that includes: an acquisition unit, an attitude detection unit, a body component detection unit, and a failure result generation unit;
the acquiring unit is used for acquiring flight state data of the unmanned aerial vehicle, wherein the flight state data comprises flight attitude data, actuator operation data and fuselage bearing vibration data;
the attitude detection unit is used for detecting the flight attitude of the unmanned aerial vehicle by using the flight attitude data to obtain an attitude detection result;
the machine body component detection unit is used for inputting the actuator operation data and the bearing vibration data into the trained deep learning network to obtain an actuator detection result and a bearing detection result;
the fault result generation unit is used for obtaining a fault detection result of the unmanned aerial vehicle according to the attitude detection result, the actuator detection result and the bearing detection result.
9. An electronic device, comprising: the unmanned aerial vehicle fault detection device comprises a memory, a processor and a transceiver which are connected in sequence, wherein the memory is used for storing a computer program, the transceiver is used for transceiving a message, and the processor is used for reading the computer program and executing the unmanned aerial vehicle fault detection method according to any one of claims 1 to 7.
10. A storage medium having stored thereon instructions for performing the method of fault detection of a drone according to any one of claims 1 to 7 when the instructions are run on a computer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111076651.5A CN113962251B (en) | 2021-09-14 | 2021-09-14 | Unmanned aerial vehicle fault detection method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111076651.5A CN113962251B (en) | 2021-09-14 | 2021-09-14 | Unmanned aerial vehicle fault detection method and device, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113962251A true CN113962251A (en) | 2022-01-21 |
| CN113962251B CN113962251B (en) | 2022-07-22 |
Family
ID=79461397
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111076651.5A Active CN113962251B (en) | 2021-09-14 | 2021-09-14 | Unmanned aerial vehicle fault detection method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113962251B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116304695A (en) * | 2023-03-07 | 2023-06-23 | 珠海紫燕无人飞行器有限公司 | Method and system for predicting vibration faults of unmanned aerial vehicle based on convolutional neural network |
| CN117520950A (en) * | 2024-01-04 | 2024-02-06 | 贵州大学 | Multi-target UAV fault diagnosis method based on attention knowledge sharing network |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108883824A (en) * | 2016-03-23 | 2018-11-23 | 冯春魁 | The method and system of acquisition, the processing and flight condition monitoring of the data of aircraft |
| CN109765935A (en) * | 2019-03-05 | 2019-05-17 | 广州极飞科技有限公司 | The malfunction monitoring method for early warning and unmanned vehicle of unmanned vehicle |
| US20190250067A1 (en) * | 2016-12-27 | 2019-08-15 | South China University Of Technology | Fault diagnosis and life testing machine for flexible precision thin-wall bearing |
| CN110307983A (en) * | 2019-06-19 | 2019-10-08 | 电子科技大学 | Unmanned plane Method for Bearing Fault Diagnosis based on CNN-Bagging |
| CN111619806A (en) * | 2020-06-03 | 2020-09-04 | 深圳市道通智能航空技术有限公司 | Unmanned aerial vehicle fault detection method, unmanned aerial vehicle and unmanned aerial vehicle system |
| CN112348124A (en) * | 2021-01-05 | 2021-02-09 | 北京航空航天大学 | Data-driven micro fault diagnosis method and device |
| CN112857348A (en) * | 2020-12-31 | 2021-05-28 | 北京航空航天大学 | Angular velocity measuring method using magnetic suspension bearing |
| CN113191240A (en) * | 2021-04-23 | 2021-07-30 | 石家庄铁道大学 | Multi-task deep neural network method and device for bearing fault diagnosis |
-
2021
- 2021-09-14 CN CN202111076651.5A patent/CN113962251B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108883824A (en) * | 2016-03-23 | 2018-11-23 | 冯春魁 | The method and system of acquisition, the processing and flight condition monitoring of the data of aircraft |
| US20190250067A1 (en) * | 2016-12-27 | 2019-08-15 | South China University Of Technology | Fault diagnosis and life testing machine for flexible precision thin-wall bearing |
| CN109765935A (en) * | 2019-03-05 | 2019-05-17 | 广州极飞科技有限公司 | The malfunction monitoring method for early warning and unmanned vehicle of unmanned vehicle |
| CN110307983A (en) * | 2019-06-19 | 2019-10-08 | 电子科技大学 | Unmanned plane Method for Bearing Fault Diagnosis based on CNN-Bagging |
| CN111619806A (en) * | 2020-06-03 | 2020-09-04 | 深圳市道通智能航空技术有限公司 | Unmanned aerial vehicle fault detection method, unmanned aerial vehicle and unmanned aerial vehicle system |
| CN112857348A (en) * | 2020-12-31 | 2021-05-28 | 北京航空航天大学 | Angular velocity measuring method using magnetic suspension bearing |
| CN112348124A (en) * | 2021-01-05 | 2021-02-09 | 北京航空航天大学 | Data-driven micro fault diagnosis method and device |
| CN113191240A (en) * | 2021-04-23 | 2021-07-30 | 石家庄铁道大学 | Multi-task deep neural network method and device for bearing fault diagnosis |
Non-Patent Citations (4)
| Title |
|---|
| K.S. AMELIN等: "An algorithm for refinement of the position of a light UAV on the basis of Kalman filtering of bearing measurements", 《JOURNAL OF COMMUNICATIONS TECHNOLOGY AND ELECTRONICS》 * |
| YU-LONG WANG等: "Network-Based Fault Detection Filter and Controller Coordinated Design for Unmanned Surface Vehicles in Network Environments", 《IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS》 * |
| 孟宏鹏: "基于机器学习的飞行器舵面系统故障诊断研究", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
| 田皓洋: "无人机用活塞式航空发动实时在线测振系统设计与实现", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116304695A (en) * | 2023-03-07 | 2023-06-23 | 珠海紫燕无人飞行器有限公司 | Method and system for predicting vibration faults of unmanned aerial vehicle based on convolutional neural network |
| CN116304695B (en) * | 2023-03-07 | 2024-04-09 | 珠海紫燕无人飞行器有限公司 | Method and system for predicting vibration faults of unmanned aerial vehicle based on convolutional neural network |
| CN117520950A (en) * | 2024-01-04 | 2024-02-06 | 贵州大学 | Multi-target UAV fault diagnosis method based on attention knowledge sharing network |
| CN117520950B (en) * | 2024-01-04 | 2024-03-19 | 贵州大学 | Multi-target UAV fault diagnosis method based on attention knowledge sharing network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113962251B (en) | 2022-07-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113962251B (en) | Unmanned aerial vehicle fault detection method and device, electronic equipment and storage medium | |
| CN107941537A (en) | A kind of mechanical equipment health state evaluation method | |
| Liu et al. | An audio-based fault diagnosis method for quadrotors using convolutional neural network and transfer learning | |
| CN110216680B (en) | Cloud-ground cooperative fault diagnosis system and method for service robot | |
| CN111259737B (en) | Method and device for predicting failure of steering wheel of vehicle, electronic equipment and storage medium | |
| CN103674468A (en) | Method of determining a turbulent condition in an aircraft | |
| CN108764122A (en) | A kind of optimization of AUV control systems and fault monitoring method based on deep learning | |
| CN106777622A (en) | The method and system of the electromechanical equipment on-line fault diagnosis based on artificial intelligence | |
| CN111159924A (en) | Method and apparatus for predicting drift trajectory | |
| CN106585979A (en) | Paddle angle adjustment method based on unmanned aerial vehicle propeller and unmanned aerial vehicle | |
| Windau et al. | Inertial machine monitoring system for automated failure detection | |
| CN110263836B (en) | Bad driving state identification method based on multi-feature convolutional neural network | |
| CN109144272B (en) | Quad-rotor unmanned aerial vehicle control method based on data glove gesture recognition | |
| Wang et al. | USVs‐Sim: A general simulation platform for unmanned surface vessels autonomous learning | |
| CN111598161A (en) | Engine gas circuit state diagnosis system based on CNN transfer learning | |
| CN111694880A (en) | Unmanned ship platform health management method and system based on multi-source data | |
| CN116243683A (en) | Method for diagnosing faults of propulsion system based on torque and multi-head self-encoder | |
| CN115469679A (en) | Unmanned aerial vehicle flight state parameter prediction method and system | |
| WO2023211499A1 (en) | Machine learning real property object detection and analysis apparatus, system, and method | |
| CN115526227A (en) | Helicopter flight data instantaneous anomaly detection method | |
| KR20220106347A (en) | System and method for forming vibration image data, and computer-readable recording medium including the same | |
| CN113934159A (en) | Unmanned ship reliability test environment model construction method | |
| CN113110547B (en) | Flight control method, device and equipment of miniature aviation aircraft | |
| CN114556069A (en) | State detection method, device and system for movable platform and movable platform | |
| CN115871890B (en) | Unmanned ship autonomous navigation capability test evaluation system |
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 |