CN113486564B - Unmanned ship propeller fault diagnosis system and method - Google Patents
Unmanned ship propeller fault diagnosis system and method Download PDFInfo
- Publication number
- CN113486564B CN113486564B CN202110649578.XA CN202110649578A CN113486564B CN 113486564 B CN113486564 B CN 113486564B CN 202110649578 A CN202110649578 A CN 202110649578A CN 113486564 B CN113486564 B CN 113486564B
- Authority
- CN
- China
- Prior art keywords
- value
- state
- unmanned ship
- propeller
- sampling
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000003745 diagnosis Methods 0.000 title claims abstract description 21
- 239000002245 particle Substances 0.000 claims abstract description 161
- 238000005070 sampling Methods 0.000 claims abstract description 142
- 230000033001 locomotion Effects 0.000 claims abstract description 141
- 239000011159 matrix material Substances 0.000 claims description 92
- 238000005259 measurement Methods 0.000 claims description 30
- 238000012952 Resampling Methods 0.000 claims description 20
- 238000006073 displacement reaction Methods 0.000 claims description 16
- 238000013016 damping Methods 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 13
- 238000012546 transfer Methods 0.000 claims description 13
- 230000007704 transition Effects 0.000 claims description 7
- 238000012937 correction Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000004891 communication Methods 0.000 description 12
- 238000001914 filtration Methods 0.000 description 12
- 230000001133 acceleration Effects 0.000 description 6
- 230000009471 action Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000013095 identification testing Methods 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/25—Design optimisation, verification or simulation using particle-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Geometry (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- General Engineering & Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Aviation & Aerospace Engineering (AREA)
- Automation & Control Theory (AREA)
- Operations Research (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
- Feedback Control In General (AREA)
Abstract
The invention relates to a fault diagnosis system and a fault diagnosis method for an unmanned ship propeller, wherein a ship end thrust loss sub-estimator is used for carrying out forward prediction on the movement state of the unmanned ship and the expected output value of the propeller in each calculation period to obtain the movement state predicted value of the unmanned ship and the expected output predicted value of the propeller; inputting a desired output sampling value and a predicted value of the thrust of the unmanned ship propeller corresponding to the current calculation period into a Kalman filter, and calculating by the Kalman filter to obtain a rough thrust loss estimated value; if the event driving condition is met, sending the sampling value and the predicted value of the unmanned ship motion state corresponding to the current calculation period to a main thrust loss estimator at the shore end; a main estimator of the thrust loss at the shore end adopts a particle filter to estimate the accurate estimated value of the thrust loss in real time; and grading the fault degree according to the magnitude of the accurate estimated value of the thrust loss, and transmitting the grading result and the accurate estimated value of the thrust loss to the boat end.
Description
Technical Field
The invention relates to a fault diagnosis system and method for an unmanned ship propeller, and belongs to the field of state estimation of ocean intelligent aircrafts.
Background
The working range of the unmanned ship is often in a water area with complex and changeable environment, and the unmanned ship is often influenced in an unpredictable way. With the improvement of unmanned ship operation capability, the complexity is improved, and the safety guarantee is also deeply focused. It is desirable that unmanned ships can discover possible faults as soon as possible, so as to reduce the potential risks of unmanned ships, and facilitate the subsequent fault-tolerant process, so that the realization of autonomous fault diagnosis is the core. The state equation is used for carrying out state estimation of the unmanned ship and correcting by using known observables so as to predict unknown system state variables, thereby diagnosing potential faults of an unmanned ship actuator and being beneficial to further fault-tolerant control for action adjustment.
The particle filter has higher estimation accuracy on the thrust loss of the propeller by utilizing the motion state of the unmanned ship, but has the defect of large calculation amount caused by the particle filter, and can generate larger burden on the limited calculation resource of the small unmanned ship. If the unmanned aerial vehicle is placed at the shore, the motion state of the unmanned aerial vehicle is transmitted to the shore in real time in a radio communication mode, so that the calculation load of a computer of the unmanned aerial vehicle can be reduced, the calculation resources of the unmanned aerial vehicle can be more effectively distributed, the operation capacity of the unmanned aerial vehicle is enhanced, the problem of occupying the bandwidth of a communication channel and the problem of time delay caused by long-distance data transmission are also caused.
Disclosure of Invention
The invention solves the technical problems that: aiming at the problem that the unmanned ship needs to diagnose faults existing in the propeller in order to keep the working capacity of the unmanned ship, the defects of the prior art are overcome, the unmanned ship propeller fault diagnosis system and method are provided, the fault existence can be detected as early as possible, the fault degree of the propeller is estimated, and conditions are provided for the follow-up fault-tolerant process.
The solution of the invention is as follows: a failure diagnosis system for an unmanned ship propeller, the system comprising a ship end thrust loss sub-estimator and a shore end thrust loss main estimator, wherein:
The ship end thrust loss sub-estimator acquires a movement state sampling value of the unmanned ship and a thrust expected output sampling value of the unmanned ship propeller in real time in each calculation period; forward predicting the unmanned ship movement state and the propeller expected output value by using a gray prediction method to obtain the unmanned ship movement state predicted value and the propeller expected output predicted value; inputting an unmanned ship propeller thrust expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value corresponding to a current calculation period into a Kalman filter, and calculating by the Kalman filter to obtain a thrust loss rough estimated value; if the absolute value of the difference between any one of the unmanned ship motion state sampling values and the unmanned ship motion state sampling value exceeds a corresponding preset threshold value or the rough thrust loss estimated value exceeds a second threshold value, the unmanned ship motion state sampling value, the propeller expected output sampling value, the unmanned ship motion state predicted value and the propeller expected output predicted value corresponding to the current calculation period are regarded as meeting the event driving condition, and the unmanned ship motion state sampling value, the propeller expected output predicted value and the unmanned ship motion state predicted value are used as fault state quantities to be sent to a shore-end thrust loss main estimator;
a main estimator for the thrust loss at the shore end, which utilizes the fault state quantity and adopts a particle filter to estimate the accurate estimated value of the thrust loss in real time; and grading the fault degree according to the accurate estimated value of the thrust loss, and transmitting the grading result and the accurate estimated value of the thrust loss to the boat end so as to carry out fault tolerance control on the subsequent unmanned boat.
The gray prediction method comprises the steps of performing one-step gray prediction according to a current calculation period unmanned aerial vehicle motion state sampling value X state (k) and previous m calculation period unmanned aerial vehicle motion state sampling values X state(k-1),...,Xstate(k-m+1),Xstate (k-m) to obtain a next calculation period unmanned aerial vehicle motion state prediction value X state (k+1); according to a current calculation period propeller expected output sampling value ctrl (k) and previous m calculation period unmanned plane motion state sampling values ctrl (k-1),. The ctrl (k-m+1), the ctrl (k-m) carries out one-step gray prediction on the propeller expected output value to obtain a next calculation period propeller expected output prediction value ctrl (k+1), and the m value is more than or equal to 10.
The state variable X (k) of the kalman filter comprises two parts: the unmanned ship motion state X state (k) and the thrust loss value X loss (k) of the current calculation period, wherein the unmanned ship motion state X state (k) comprises an unmanned ship longitudinal moving speed u (k), a transverse moving speed v (k) and a heading angular speed r (k); the thrust loss value X loss (k) includes a propeller forward thrust loss L u (k) and a yaw moment loss L r (k), that is:
X(k)＝[Xstate(k);Xloss(k)]＝[u(k),v(k),r(k),Lu(k),Lr(k)]T；
The input variable U (k) of the kalman filter is the propeller desired output value ctrl (k), including the forward thrust T u (k) and the yaw moment T r (k), i.e.:
U(k)＝[Tu(k),Tr(k)]T；
the state transfer equation of the Kalman filter is established based on the ship dynamics model, when the ship dynamics model is of the following form:
Wherein x is the motion state of the unmanned ship, and x= [ u, v, r ] T, wherein u, v, r are the longitudinal moving speed, the transverse moving speed and the heading angular speed of the unmanned ship respectively; m is an inertial parameter matrix of the under-actuated unmanned ship; g (x) is Coriolis Li Juzhen; h is a damping coefficient matrix; f= [ T u,Tr]T ] is a control input matrix, wherein T u is forward thrust, and T r is yaw moment; k is a control input configuration matrix;
the state transition equation is specifically:
X(k+1)＝AX(k)+BU(k)
wherein A is a coefficient matrix:
Delta T is a calculation period, M is an inertial parameter matrix of the under-actuated unmanned ship, G (x) is a Coriolis matrix, and H is a damping coefficient matrix.
B is a control matrix:
K is the control input configuration matrix,
Wherein k 11,k22,k32 is the determination coefficient of forward thrust to longitudinal displacement speed, the determination coefficient of yaw moment to transverse displacement speed and the determination coefficient of yaw moment to heading angular speed respectively;
The measurement variable Z (k+1) of the Kalman filter comprises two parts: the predicted value X state (k+1) of the unmanned ship motion state and the predicted value ctrl (k+1) of the propeller expected output are obtained through grey prediction;
the measurement equation of the Kalman filter is:
Z(k+1)＝CX(k+1)+DU(K+1)
Wherein C is an output matrix; d is a direct feed matrix;
The specific process of calculating the rough thrust loss estimated value by the Kalman filter is as follows:
S1, estimating a state: according to the state transition equation, the state variable prior estimated value of the next calculation period can be calculated by the state variable X (k)
S2, calculating a priori covariance matrix P - (k+1) of the state variables:
P-(k+1)＝AP(k)AT+Q
Wherein Q is a process noise covariance matrix, P (k) is a state variable covariance matrix, and the initial value of the state variable covariance matrix P (k) is: p (0) =diag [0, 0];
s3, calculating Kalman gain K (k+1):
K(k+1)＝P-(k+1)CT(CP-(k+1)CT+R)-1
wherein R is a measurement noise covariance matrix;
s4, taking the predicted value of the unmanned ship motion state and the predicted value of the propeller expected output in the fault state quantity as a measurement variable Z (k+1), and calculating the optimal estimated value of the state variable at the next moment
S5, updating a state covariance matrix:
P(k+1)＝(I-K(k+1)H)P-(k+1)
and S6, returning to the step S1 when the next calculation period comes, and executing the loop iteration calculation of the steps S1 to S5 again.
The state variable X' of the particle filter comprises two parts: the unmanned ship comprises an unmanned ship motion state X ' state and a thrust loss value X ' loss, wherein the unmanned ship motion state X ' state comprises an unmanned ship longitudinal moving speed u, a transverse moving speed v and a heading angular speed r; the thrust loss value X' loss includes a propeller forward thrust loss L u and a yaw moment loss L r, namely:
X'＝[u,v,r,Lu,Lr]T；
The input variable U' of the kalman filter is the desired output value ctrl of the propeller, including the forward thrust T u and the yaw moment T r, namely:
U'＝[Tu,Tr]T；
the state transfer equation of the particle filter is established based on a ship dynamics model, and specifically comprises the following steps:
X'pred＝AX+BU'
X' pred is a state variable predictor.
Wherein A is a coefficient matrix:
Delta T is a calculation period, M is an inertial parameter matrix of the under-actuated unmanned ship, G (x) is a Coriolis matrix, and H is a damping coefficient matrix.
B is a control matrix:
K is a control input configuration matrix;
Wherein k 11,k22,k32 is a determination coefficient of forward thrust to longitudinal displacement speed, a determination coefficient of yaw moment to transverse displacement speed, and a determination coefficient of yaw moment to heading angular speed, respectively.
The measured variable Z' of the particle filter comprises two parts: the unmanned ship motion state predicted value X 'state_grey and the propeller expected output predicted value ctrl_grey' obtained through grey prediction;
The measurement equation of the particle filter is:
Z'pred＝C·X'pred+D·ctrl_grey'
wherein Z' pred is the predicted value of the measured variable, and C is the output matrix; d is a direct feed matrix.
The particle filter is used for estimating the accurate estimated value of the thrust loss in real time, and comprises the following specific steps:
S2.1, acquiring a fault state quantity, wherein the fault state quantity comprises an unmanned ship motion state sampling value, a propeller expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value which are received from a ship end thrust loss sub-estimator in a fault calculation period as the fault state quantity;
S2.2, initializing: setting the total number N of particles, taking a state variable formed by an unmanned ship motion state sampling value in a fault state quantity and a propeller expected output sampling value as a mean value of the state variable, presetting the distribution of the particles in a preset value space, and simulating the state variable particle space to obtain N particles conforming to a distribution rule in the value space;
S2.3, prediction: substituting each particle into a state transfer equation and a measurement equation to generate N predicted particles Z' pred (i):
Z'pred(i)＝C·X'pred(i)+D·ctrl_grey'(i) i＝1,2...,N
s2.4, correction: the predicted value of the motion state of the unmanned ship in the fault state quantity and the predicted value expected to be output by the propeller are used as the updated value Z' of the measured variable;
S2.5, weight calculation: calculating the Euclidean distance W i between each predicted particle Z 'pred (i) and the measured updated value Z', and assigning a weight to each particle according to the difference of the distances, wherein the predicted particle with smaller distance has larger weight;
s2.6, resampling: setting the resampling quantity M, and resampling according to the sampling weight of each particle to obtain a new particle group consisting of M particles;
s2.7, carrying out normalization processing on the sampling weights of M particles in the particle group obtained in the step S2.6 again to obtain respective weights alpha i, and carrying out weighted average on state variable predicted values corresponding to the particles according to the weights alpha i to obtain state variable predicted values Thereby obtaining the accurate estimated value/>, of the thrust loss prediction
The sampling weight of each particle of the particle filter is obtained by calculating the following method:
A. Absolute sampling weights d' i are calculated:
d'i＝1/Wi
B. The absolute sampling weight d' i is normalized to obtain the sampling weight The sampling weight d i of each particle is limited to the (0, 1) range, and
The other technical scheme of the invention is as follows: a method of fault diagnosis of an unmanned aerial vehicle propeller, the method comprising:
S1, executing the following steps at each calculation period of the boat end:
S1.1, acquiring a sampling value of the motion state of the unmanned ship and a sampling value of the thrust expected output of the unmanned ship propeller;
s1.2, forward prediction is carried out on the unmanned ship motion state and the propeller expected output value by using a gray prediction method, so that the unmanned ship motion state prediction value and the propeller expected output prediction value are obtained;
S1.3, inputting a propeller thrust expected output sampling value of the unmanned ship, a motion state predicted value of the unmanned ship and a propeller expected output predicted value corresponding to a current calculation period into a Kalman filter;
s1.4, calculating by a Kalman filter to obtain a rough thrust loss estimated value;
S1.5, if the absolute value of the difference between any one of the unmanned ship motion state sampling values and the unmanned ship motion state sampling value exceeds a corresponding preset threshold value or the thrust loss rough estimated value exceeds a second threshold value, the unmanned ship motion state sampling value, the propeller expected output sampling value, the unmanned ship motion state predicted value and the propeller expected output predicted value corresponding to the current calculation period are regarded as meeting the event driving condition, and the unmanned ship motion state sampling value, the propeller expected output predicted value and the unmanned ship motion state sampling value are transmitted to a shore as fault state quantities;
S2, after the boat end receives the fault state quantity, sequentially executing the following steps:
s2.1, a main estimator of the thrust loss at the shore end, wherein a particle filter is adopted to estimate an accurate estimated value of the thrust loss in real time by utilizing a fault state quantity;
s2.2, grading the fault degree according to the size of the accurate estimated value of the thrust loss, and transmitting the grading result and the accurate estimated value of the thrust loss to the boat end so as to carry out fault tolerance control on the subsequent unmanned boat.
The step S2.1 of the particle filter is to estimate the accurate estimated value of the thrust loss in real time, and the specific steps are as follows:
S2.1.1, acquiring a fault state quantity, wherein the fault state quantity comprises an unmanned ship motion state sampling value, a propeller expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value which are received from a ship end thrust loss sub-estimator in a fault calculation period as the fault state quantity;
S2.1.2, initializing: setting the total number N of particles, taking a state variable formed by an unmanned ship motion state sampling value and a propeller expected output sampling value in a fault state quantity as a mean value of the state variable according to preset particle distribution, and simulating a state variable particle space in a preset value space to obtain N particles conforming to a distribution rule in the value space;
s2.1.3, predicting: substituting each particle into a state transfer equation and a measurement equation to generate N predicted particles Z' pred (i);
Z'pred(i)＝C·X'pred(i)+D·ctrl_grey'(i) i＝1,2...,N
S2.1.4, correction: the predicted value of the motion state of the unmanned ship in the fault state quantity and the predicted value expected to be output by the propeller are used as the updated value Z' of the measured variable;
S2.1.5, weight calculation: calculating the Euclidean distance W i between each predicted particle Z 'pred (i) and the measured updated value Z', and assigning a weight to each particle according to the difference of the distances, wherein the predicted particle with smaller distance has larger weight;
s2.1.6, resampling: setting the resampling quantity M, and resampling according to the sampling weight of each particle to obtain a new particle group consisting of M particles;
S2.1.7, carrying out normalization processing on sampling weights of M particles in the particle group obtained in the step S2.6 again to obtain respective weights alpha i, carrying out weighted average on state variable predicted values corresponding to the particles according to the weights alpha i to obtain state variable predicted values Thereby obtaining the accurate estimated value/>, of the thrust loss prediction
The sampling weight of each particle of the particle filter is obtained by calculating the following method:
A. Absolute sampling weights d' i are calculated:
d'i＝1/Wi
B. The absolute sampling weight d' i is normalized to obtain the sampling weight The sampling weight d i of each particle is limited to the (0, 1) range, and
Compared with the prior art, the invention has the beneficial effects that:
(1) According to the invention, the particle filter is arranged at the shore end, so that the high estimation precision of the thrust loss value can be ensured, the Kalman estimator which occupies less calculation resources and has low estimation precision is arranged at the boat end, the occupation of the calculation resources of too many boat end computers can be avoided, and the running reliability of the boat end computers is ensured.
(2) According to the invention, an event-driven mode is adopted, the shore side carries out Kalman filtering estimation on the predicted value of the thrust loss to obtain the rough estimated value of the thrust loss, the information of the shore side is transmitted to the shore side in a wireless communication mode when the event-driven condition is met to carry out particle filtering estimation to obtain the accurate estimated value of the thrust loss, so that the fault is accurately judged, the unmanned ship does not always transmit data to the shore side, and the problems of occupation of communication channel bandwidth, time delay caused by long-distance data transmission and the like are avoided.
(3) The invention fully considers various conditions of occurrence of thrust loss faults, the event driving condition comprises that the absolute value of the difference between the unmanned ship motion state sampling value and the unmanned ship motion state sampling value exceeds a first threshold value or the rough thrust loss estimation value exceeds a second threshold value, and the two conditions meet one condition to be regarded as meeting the event driving condition, so that the condition that larger thrust loss is possible to occur at the next moment is indicated, and the more accurate estimation value is ensured under the transition states of unmanned ship acceleration, bow turning and the like, so that the condition of missed diagnosis or misdiagnosis is avoided.
(4) The particle filter of the invention assigns a weight to each particle according to the difference of Euclidean distance, and the predicted particle with smaller distance has larger weight.
Drawings
FIG. 1 is a flow chart of the operation of an embodiment of the present invention.
Detailed Description
The invention will be described in detail below with reference to the drawings and the specific embodiments.
Aiming at the problem that the unmanned ship needs to diagnose faults existing in the propeller in order to keep the working capacity of the unmanned ship, the invention provides a fault diagnosis method for the unmanned ship propeller based on event driving and grey prediction. The method can detect the existence of faults as early as possible, and simultaneously estimate the fault degree of the propeller so as to complete the fault tolerance process of the unmanned ship.
As shown in fig. 1, the present invention provides a failure diagnosis system of an unmanned ship propeller, the system comprising a ship end thrust loss sub-estimator and a shore end thrust loss main estimator, wherein:
The ship end thrust loss sub-estimator periodically acquires a sampling value of the unmanned ship motion state and a sampling value of the unmanned ship propeller thrust expected output in real time at a fixed sampling frequency by using a sensor carried by a ship body in each calculation period; forward predicting the unmanned ship movement state and the propeller expected output value by using a gray prediction method to obtain the unmanned ship movement state predicted value and the propeller expected output predicted value; inputting an unmanned ship propeller thrust expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value corresponding to a current calculation period into a Kalman filter, and calculating by the Kalman filter to obtain a thrust loss rough estimated value; and if the absolute value of the difference between any one of the unmanned ship motion state sampling values and the unmanned ship motion state sampling value exceeds a corresponding preset threshold value or the rough thrust loss estimated value exceeds a second threshold value, the unmanned ship motion state sampling value, the propeller expected output sampling value, the unmanned ship motion state predicted value and the propeller expected output predicted value corresponding to the current calculation period are regarded as meeting the event driving condition, and the unmanned ship motion state sampling value, the propeller expected output predicted value and the unmanned ship motion state sampling value are used as fault state quantities and are sent to a shore-end thrust loss main estimator in a wireless communication mode.
A main estimator for the thrust loss at the shore end, which utilizes the fault state quantity and adopts a particle filter to estimate the accurate estimated value of the thrust loss in real time; and grading the fault degree according to the accurate estimated value of the thrust loss, and transmitting the grading result and the accurate estimated value of the thrust loss to the boat end through a wireless communication mode so as to carry out fault tolerance control on the subsequent unmanned boat. The classification result may be expressed by an integer of 0 to 5.
The above diagnostic method can find possible faults of the propeller earlier by using grey prediction and detect the existence of faults earlier by grey prediction in the early stage of amplification. After the boat receives the fault degree parameters, a corresponding fault-tolerant control strategy can be selected according to the fault degree parameters, and a control instruction is issued to the propeller to perform action adjustment, so that the unmanned boat is ensured to operate in an optimal state. The fault tolerance control method can be used for compensating the expected output of the ship end propeller through the fault grade and the thrust loss value estimated by the particle filter in the fault compensation process.
And the boat end thrust loss sub-estimator and the shore end thrust loss main estimator are interacted in a wireless communication mode.
The gray prediction method comprises the steps of performing one-step gray prediction according to a current calculation period unmanned aerial vehicle motion state sampling value X state (k) and previous m calculation period unmanned aerial vehicle motion state sampling values X state(k-1),...,Xstate(k-m+1),Xstate (k-m) to obtain a next calculation period unmanned aerial vehicle motion state prediction value X state (k+1); according to a current calculation period propeller expected output sampling value ctrl (k) and previous m calculation period unmanned plane motion state sampling values ctrl (k-1),. The ctrl (k-m+1), the ctrl (k-m) carries out one-step gray prediction on the propeller expected output value to obtain a next calculation period propeller expected output prediction value ctrl (k+1), and the m value is more than or equal to 10.
Preferably, the preset threshold is selected as follows: the increment threshold value of the longitudinal movement speed is a longitudinal movement speed increment corresponding to a forward acceleration average value acquired by an acceleration navigation test in an unmanned ship parameter identification test; the increment threshold of the transverse moving speed is a longitudinal moving speed increment corresponding to a transverse linear acceleration mean value acquired by a sailing test under the 3-level sea condition in the unmanned ship parameter identification test; the increment threshold value of the heading angular velocity is a heading angular velocity increment corresponding to the average value of the heading angular acceleration obtained in the unmanned ship rotation test acceleration heading stage.
The second threshold value is 10% of the expected output sampling value of the current propeller.
The state variable X (k) of the kalman filter comprises two parts: the unmanned ship motion state X state (k) and the thrust loss value X loss (k) of the current calculation period, wherein the unmanned ship motion state X state (k) comprises an unmanned ship longitudinal moving speed u (k), a transverse moving speed v (k) and a heading angular speed r (k); the thrust loss value X loss (k) includes a propeller forward thrust loss L u (k) and a yaw moment loss L r (k), that is:
X(k)＝[Xstate(k);Xloss(k)]＝[u(k),v(k),r(k),Lu(k),Lr(k)]T；
The input variable U (k) of the kalman filter is the propeller desired output value ctrl (k), including the forward thrust T u (k) and the yaw moment T r (k), i.e.:
U(k)＝[Tu(k),Tr(k)]T；
the state transfer equation of the Kalman filter is established based on the ship dynamics model, when the ship dynamics model is of the following form:
Wherein x is the motion state of the unmanned ship, and x= [ u, v, r ] T, wherein u, v, r are the longitudinal moving speed, the transverse moving speed and the heading angular speed of the unmanned ship respectively; m is an inertial parameter matrix of the under-actuated unmanned ship; g (x) is Coriolis Li Juzhen; h is a damping coefficient matrix; f= [ T u,Tr]T ] is a control input matrix, wherein T u is forward thrust, and T r is yaw moment; k is a control input configuration matrix;
the state transition equation is specifically:
X(k+1)＝AX(k)+BU(k)
wherein A is a coefficient matrix:
Delta T is a calculation period, M is an inertial parameter matrix of the under-actuated unmanned ship, G (x) is a Coriolis matrix, and H is a damping coefficient matrix.
B is a control matrix:
K is the control input configuration matrix,
Wherein k 11,k22,k32 is the determination coefficient of forward thrust to longitudinal displacement speed, the determination coefficient of yaw moment to transverse displacement speed and the determination coefficient of yaw moment to heading angular speed respectively;
The measurement variable Z (k+1) of the Kalman filter comprises two parts: the predicted value X state (k+1) of the unmanned ship motion state and the predicted value ctrl (k+1) of the propeller expected output are obtained through grey prediction;
the measurement equation of the Kalman filter is:
Z(k+1)＝CX(k+1)+DU(K+1)
Wherein C is an output matrix; d is a direct feed matrix;
The specific process of calculating the rough thrust loss estimated value by the Kalman filter is as follows:
S1, estimating a state: according to the state transition equation, the state variable prior estimated value of the next calculation period can be calculated by the state variable X (k)
S2, calculating a priori covariance matrix P - (k+1) of the state variables:
P-(k+1)＝AP(k)AT+Q
Wherein Q is a process noise covariance matrix, P (k) is a state variable covariance matrix, and the initial value of the state variable covariance matrix P (k) is: p (0) =diag [0, 0];
s3, calculating Kalman gain K (k+1):
K(k+1)＝P-(k+1)CT(CP-(k+1)CT+R)-1
wherein R is a measurement noise covariance matrix;
s4, taking the predicted value of the unmanned ship motion state and the predicted value of the propeller expected output in the fault state quantity as a measurement variable Z (k+1), and calculating the optimal estimated value of the state variable at the next moment
S5, updating a state covariance matrix:
P(k+1)＝(I-K(k+1)H)P-(k+1)
and S6, returning to the step S1 when the next calculation period comes, and executing the loop iteration calculation of the steps S1 to S5 again.
From the above, kalman filtering method, fusion state variable estimation valueAnd measuring and updating the information of Y (k+1), and calculating to obtain a state variable predicted value X (k+1) of the (k+1) calculation period, so as to obtain a thrust loss predicted value estimation X loss (k+1) of the next calculation period.
The Kalman filtering method can select a Kalman filtering method, an extended Kalman filtering method, a lossless Kalman filtering method and the like according to the properties of a system state transfer equation and a measurement equation.
The state variable X' of the particle filter comprises two parts: the unmanned ship comprises an unmanned ship motion state X ' state and a thrust loss value X ' loss, wherein the unmanned ship motion state X ' state comprises an unmanned ship longitudinal moving speed u, a transverse moving speed v and a heading angular speed r; the thrust loss value X' loss includes a propeller forward thrust loss L u and a yaw moment loss L r, namely:
X'＝[u,v,r,Lu,Lr]T；
The input variable U' of the kalman filter is the desired output value ctrl of the propeller, including the forward thrust T u and the yaw moment T r, namely:
U'＝[Tu,Tr]T；
the state transfer equation of the particle filter is established based on a ship dynamics model, and specifically comprises the following steps:
X'pred＝AX'+BU'
Wherein X' pred is the state variable predictor.
Wherein A is a coefficient matrix:
Delta T is a calculation period, M is an inertial parameter matrix of the under-actuated unmanned ship, G (x) is a Coriolis matrix, and H is a damping coefficient matrix.
B is a control matrix:
K is a control input configuration matrix;
Wherein k 11,k22,k32 is a determination coefficient of forward thrust to longitudinal displacement speed, a determination coefficient of yaw moment to transverse displacement speed, and a determination coefficient of yaw moment to heading angular speed, respectively.
The measured variable Z' of the particle filter comprises two parts: the unmanned ship motion state predicted value X 'state_grey and the propeller expected output predicted value ctrl_grey' obtained through grey prediction;
The measurement equation of the particle filter is:
Z'pred＝C·X'pred+D·ctrl_grey'
wherein Z' pred is the predicted value of the measured variable, and C is the output matrix; d is a direct feed matrix.
The particle filter is used for estimating the accurate estimated value of the thrust loss in real time, and comprises the following specific steps:
S2.1, acquiring a fault state quantity, wherein the fault state quantity comprises an unmanned ship motion state sampling value, a propeller expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value which are received from a ship end thrust loss sub-estimator in a fault calculation period as the fault state quantity;
S2.1, initializing: setting the total number N of particles, taking a state variable formed by an unmanned ship motion state sampling value and a propeller expected output sampling value in a fault state quantity as a mean value of the state variable according to preset particle distribution, and simulating a state variable particle space in a preset value space to obtain N particles conforming to a distribution rule in the value space;
s2.2, prediction: substituting each particle into a state transfer equation and a measurement equation to generate N predicted particles Z' pred (i);
Z'pred(i)＝C·X'pred(i)+D·ctrl_grey'(i) i＝1,2...,N
S2.3, correction: the predicted value of the motion state of the unmanned ship in the fault state quantity and the predicted value expected to be output by the propeller are used as the updated value Z' of the measured variable;
S2.4, weight calculation: calculating the Euclidean distance W i between each predicted particle Z 'pred (i) and the measured updated value Z', and assigning a weight to each particle according to the difference of the distances, wherein the predicted particle with smaller distance has larger weight;
s2.5, resampling: setting the resampling quantity M, and resampling according to the sampling weight of each particle to obtain a new particle group consisting of M particles;
S2.6, carrying out normalization processing on the sampling weights of M particles in the particle group obtained in the step S2.6 again to obtain respective weights alpha i, and carrying out weighted average on state variable predicted values corresponding to the particles according to the weights alpha i to obtain state variable predicted values Thereby obtaining the accurate estimated value/>, of the thrust loss prediction
The sampling weight of each particle of the particle filter is obtained by calculating the following method:
A. Absolute sampling weights d' i are calculated:
d'i＝1/Wi
B. The absolute sampling weight d' i is normalized to obtain the sampling weight The sampling weight d i of each particle is limited to the (0, 1) range, and
The other technical scheme of the invention is as follows: a method of fault diagnosis of an unmanned aerial vehicle propeller, the method comprising:
S1, executing the following steps at each calculation period of the boat end:
S1.1, acquiring a sampling value of the motion state of the unmanned ship and a sampling value of the thrust expected output of the unmanned ship propeller;
s1.2, forward prediction is carried out on the unmanned ship motion state and the propeller expected output value by using a gray prediction method, so that the unmanned ship motion state prediction value and the propeller expected output prediction value are obtained;
S1.3, inputting a propeller thrust expected output sampling value of the unmanned ship, a motion state predicted value of the unmanned ship and a propeller expected output predicted value corresponding to a current calculation period into a Kalman filter;
s1.4, calculating by a Kalman filter to obtain a rough thrust loss estimated value;
S1.5, if the absolute value of the difference between any one of the unmanned ship motion state sampling values and the unmanned ship motion state sampling value exceeds a corresponding preset threshold value or the thrust loss rough estimated value exceeds a second threshold value, the unmanned ship motion state sampling value, the propeller expected output sampling value, the unmanned ship motion state predicted value and the propeller expected output predicted value corresponding to the current calculation period are regarded as meeting the event driving condition, and the unmanned ship motion state sampling value, the propeller expected output predicted value and the unmanned ship motion state sampling value are transmitted to a shore as fault state quantities;
s2, executing the following steps at each calculation period of the boat end:
S2.1, a main estimator of the thrust loss at the shore end, wherein a particle filter is adopted to estimate an accurate estimated value of the thrust loss in real time by utilizing a fault state quantity; the method is realized by adopting the specific steps of estimating the accurate estimated value of the thrust loss in real time by the particle filter.
S2.2, grading the fault degree according to the size of the accurate estimated value of the thrust loss, and transmitting the grading result and the accurate estimated value of the thrust loss to the boat end so as to carry out fault tolerance control on the subsequent unmanned boat.
Examples:
in one embodiment of the present invention, the fault diagnosis method includes: the boat end thrust loss sub estimator and the shore end thrust loss main estimator are implemented as follows:
(1) Loading a thrust loss sub-estimator on a control computer of the unmanned ship, and loading a shore-end thrust loss main estimator on a shore-end monitoring computer;
(2) Unmanned ship control computer utilizes the sensor that the hull carried to periodically obtain unmanned ship motion state X state (k) and propeller expected output value ctrl (k) with fixed sampling frequency (1 Hz), wherein: the motion state X state (k) comprises a longitudinal moving speed u (k), a transverse moving speed v (k) and a heading angular speed r (k) of the current unmanned ship; the propeller expected output value ctrl (k) comprises forward thrust T u (k) and thrust yaw moment T r (k) output by the unmanned ship propeller, and the state is updated by the ship end sensor at each sampling time.
(3) The motion state X state(k),Xstate(k-1),...,Xstate(k-9),Xstate (k-10) of the unmanned ship and the expected output value ctrl (k) of the propeller at the current moment and the previous 10 moments are utilized, ctrl (k-1), ctrl (k-9), ctrl (k-10) are utilized, the motion state of the unmanned ship and the expected output value of the propeller are respectively subjected to one-step forward prediction based on a gray prediction method GM (1, 1), and the motion state predicted value X state (k+1) of the next moment and the expected output predicted value ctrl (k+1) of the propeller are obtained;
(4) The state variables, state equations and observed quantity and measurement equations adopted by the Kalman filtering thrust loss estimator at the boat end and the particle filtering thrust loss estimator at the shore end are the same, and are as follows:
The equation of state:
kinetic model equation for unmanned boat:
Wherein x= [ u, v, r ] T, wherein u, v, r are respectively the longitudinal movement speed, the transverse movement speed and the heading angular speed of the unmanned ship; m is an inertial parameter matrix of the under-actuated unmanned ship; g (x) is a coriolis/centripetal force matrix; h is a damping coefficient matrix; f= [ T u,Tr]T ] is a control input matrix, wherein T u is forward thrust, and T r is yaw moment; k is a control input configuration matrix, and is used for mapping the control input into corresponding forces and moments acting on the hull, M, G (x) and K have the following structures:
Wherein: m 11、m22、m33 is the unmanned ship longitudinal movement inertia coefficient, the transverse movement inertia coefficient and the bow turning inertia coefficient; h 11、h22、h33 is a longitudinal displacement damping coefficient, a transverse displacement damping coefficient and a bow turning damping coefficient of the unmanned ship; k 11、k22、k33 is the action coefficient of forward thrust to longitudinal displacement speed, the action coefficient of yaw moment to transverse displacement speed and the action coefficient of yaw moment to heading angular speed.
The state variables X= [ u, v, r, L u,Lr]T are selected, wherein L u,Lr is a forward thrust loss value and a yaw moment loss value of the propeller respectively, and a sampling time interval (namely, a calculation period) is set to be delta T. The input variable U is the propeller desired output value (including forward thrust output values ctrl u and ctrl r), i.e., U (k) = [ ctrl u(k),ctrlr(k)]T).
Based on the hull dynamics model, a state equation is established:
X(k+1)＝AX(k)+BU(k)，
wherein A is a coefficient matrix; representing the relationship between the state variables of the system; b is a control matrix representing how the input variables control the state variables, in this example:
measurement equation:
the measurement variable Y= [ u, v, r, ctrl u,ctrlr]T is selected.
The system measurement equation is:
Y(k+1)＝CX(k+1)+DU(K+1)，
wherein C is an output matrix representing how the output variables reflect the state variables; d is a direct feed matrix representing the direct effect of the input on the output. In this example of the present invention,
And taking the current propeller expected output value as a state estimation input, taking the unmanned ship motion state and the propeller expected output forward predicted value obtained by the gray prediction method as measurement update, and inputting the measurement update into a Kalman filter to roughly estimate the thrust loss value at the next moment.
If the change increment of the gray predicted value of the unmanned ship motion state exceeds a set threshold or the thrust loss value of the Kalman filter exceeds a preset threshold, the unmanned ship is considered to meet event driving conditions when larger thrust loss possibly occurs at the next moment, the unmanned ship transmits the current motion state, the propeller expected output value and the propeller expected output predicted value obtained through the gray prediction method to a shore end computer in a wireless communication mode, and a shore end thrust loss main estimator inputs the received information into a particle filter of the shore end thrust loss main estimator to estimate the accurate value of the thrust loss.
The two thrust loss estimation methods are as follows:
The calculation process of the thrust loss sub-estimator based on Kalman filtering at the end of the ship comprises the following steps:
Setting an initial value: initial value X (0) = [0, 0] T of state variable; state variable covariance matrix P (0) =diag [0, 0]; process noise covariance matrix q=diag (0.1,0.1,0.1,0.1,0.1); the measurement noise covariance matrix R is set according to the on-board sensor longitude, in this example r=diag (0.2,0.2,0.1,0.5,0.5).
(1) Taking the expected input value of the current propeller as an input variable U (k), and calculating the state priori predicted value of the next moment according to a state equation from the state variable value X (k) of the current moment
(2) A priori covariance matrix P - (k+1) of state variables is calculated:
P-(k+1)＝AP(k)AT+Q
(3) Calculate the kalman gain K (k+1):
K(k+1)＝P-(k+1)HT(HP-(k+1)HT+R)-1
(4) Taking the unmanned ship motion state obtained by the gray prediction method and the one-step forward predicted value expected to be output by the propeller as a measurement variable Z (k+1), and calculating the optimal estimated value at the next moment
(5) Calculating a state covariance matrix:
P(k+1)＝(I-K(k+1)H)P-(k+1)
And (3) returning to the step (1) and repeating the steps (1) - (5) again for iterative computation when the next sampling time arrives. The state variable predicted value X (k+1) at the moment (k+1) can be calculated by using a Kalman filtering method, and then the thrust loss predicted value estimation at the next moment can be obtained
The calculation process of the main thrust loss estimator based on particle filtering by the shore end comprises the following steps:
(1) Initializing: initializing: setting the total number N of particles, taking a state variable formed by an unmanned ship motion state sampling value and a propeller expected output sampling value in a fault state quantity as a mean value of the state variable according to preset particle distribution, and simulating a state variable particle space in a preset value space to obtain N particles conforming to a distribution rule in the value space;
In this embodiment, the total number of particles N is set to 300, and the distribution of the initial values of the state variables X (k) is set according to the initial value of the unmanned ship motion state, where the initial distribution of the particles is set as follows: and 0.1 is a random normal distribution with [0, 0] T as the mean and 0.1 as the variance. Each particle is numbered in turn, the initial sampling weight d i of each particle is the same,
(2) And (3) predicting: substituting each particle into a state transfer equation and a measurement equation to generate N predicted particles Z' pred (i);
Z'pred(i)＝C·X'pred(i)+D·ctrl_grey'(i) i＝1,2...,N
(3) Correcting: the predicted value of the motion state of the unmanned ship in the fault state quantity and the predicted value expected to be output by the propeller are used as the updated value Z' of the measured variable;
(4) And (5) calculating weights: calculating the Euclidean distance W i between each predicted particle Z 'pred (i) and the measured updated value Z', and assigning a weight to each particle according to the difference of the distances, wherein the predicted particle with smaller distance has larger weight;
The sampling weight of each particle of the particle filter is obtained by calculating the following method:
A. Absolute sampling weights d' i are calculated:
d'i＝1/Wi
B. The absolute sampling weight d' i is normalized to obtain the sampling weight The sampling weight d i of each particle is limited to the (0, 1) range, and
The smaller the Euclidean distance, the larger the predicted particle weight, and the sampling weight is normalized, so that the sampling weight d i of each particle is limited in the (0, 1) range,
(5) Resampling: the resampling number n=50 is set, and resampling particles are selected according to the distribution of the sampling weight sizes. The method comprises the following steps: all particles are formed into a pie chart, the larger the weight is, the larger the area of the cake occupied by the particles is, then a random decimal between (0 and 1) is generated, the random decimal is sampled on the pie chart, and the particles represented by the particles are taken as resampled particles when the particles fall in the weight interval. The particle with the greatest weight may then be selected multiple times as the resampling particle.
(6) Weighted average: after the sampling reaches the set resampling quantity n, the sampling weights of the particles are normalized again to obtain the respective weights alpha i, namelyPredicted value/>, of state variable corresponding to these particlesWeighted average is carried out according to the weight alpha i, so as to obtain the state variable predicted valueThereby obtaining accurate estimation/>, of the thrust loss prediction
According to the predicted value of the thrust loss at the next moment calculated by the particle filter, fault degree parameters (represented by numerals 1-5) are set, the main estimator of the shore-side thrust loss transmits the parameters representing the fault degree and the predicted value of the thrust loss calculated by the particle filter back to the unmanned ship side in a wireless communication mode, and a control computer of the unmanned ship side performs fault-tolerant control by using the thrust loss values. The fault tolerance control method is a fault compensation process, and the expected output of the ship end propeller is compensated through the fault grade and the thrust loss value estimated by the particle filter.
In summary, the invention provides a fault diagnosis method which can ensure higher estimation precision and avoid occupying excessive boat end computer resources and wireless communication channel bandwidth resources. The particle filter can be arranged in the shore-end computer and is connected with the boat-end computer through wireless communication, the event driving method is adopted, the boat-end information can be transmitted to the shore-end particle filter when the event driving condition is met, the shore-end particle filter is used for accurately estimating the thrust loss value of the unmanned ship, and the unmanned ship does not always transmit data to the shore end, so that the resources required by communication are reduced.
Although the present invention has been described in terms of the preferred embodiments, it is not intended to be limited to the embodiments, and any person skilled in the art can make any possible variations and modifications to the technical solution of the present invention by using the methods and technical matters disclosed above without departing from the spirit and scope of the present invention, so any simple modifications, equivalent variations and modifications to the embodiments described above according to the technical matters of the present invention are within the scope of the technical matters of the present invention.
Claims (7)
1. The fault diagnosis system of the unmanned ship propeller is characterized by comprising a ship end thrust loss sub-estimator and a shore end thrust loss main estimator, wherein:
The ship end thrust loss sub-estimator acquires a movement state sampling value of the unmanned ship and a thrust expected output sampling value of the unmanned ship propeller in real time in each calculation period; forward predicting the unmanned ship movement state and the propeller expected output value by using a gray prediction method to obtain the unmanned ship movement state predicted value and the propeller expected output predicted value; inputting an unmanned ship propeller thrust expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value corresponding to a current calculation period into a Kalman filter, and calculating by the Kalman filter to obtain a thrust loss rough estimated value; if the absolute value of the difference between any one of the unmanned ship motion state sampling values and the average value of the unmanned ship motion state sampling values exceeds a corresponding preset threshold value or the rough thrust loss estimated value exceeds a second threshold value, the unmanned ship motion state sampling values, the propeller expected output sampling values, the unmanned ship motion state predicted value and the propeller expected output predicted value corresponding to the current calculation period are regarded as meeting event driving conditions, and the unmanned ship motion state sampling values, the propeller expected output predicted value and the unmanned ship motion state predicted value are sent to a shore-end thrust loss main estimator as fault state quantities;
A main estimator for the thrust loss at the shore end, which utilizes the fault state quantity and adopts a particle filter to estimate the accurate estimated value of the thrust loss in real time; grading the fault degree according to the magnitude of the accurate estimated value of the thrust loss, and transmitting the grading result and the accurate estimated value of the thrust loss to the boat end so as to carry out fault tolerance control on the subsequent unmanned boat;
the state variable X' of the particle filter comprises two parts: the unmanned ship comprises an unmanned ship motion state X ' state and a thrust loss value X ' loss, wherein the unmanned ship motion state X ' state comprises an unmanned ship longitudinal moving speed u, a transverse moving speed v and a heading angular speed r; the thrust loss value X' loss includes a propeller forward thrust loss L u and a yaw moment loss L r, namely:
X'＝[u,v,r,Lu,Lr]T；
The input variable U' of the kalman filter is the desired output value ctrl of the propeller, including the forward thrust T u and the yaw moment T r, namely:
U'＝[Tu,Tr]T；
the state transfer equation of the particle filter is established based on a ship dynamics model, and specifically comprises the following steps:
X′pred＝AX+BU
x' pred is a state variable predictor;
wherein A is a coefficient matrix:
delta T is a calculation period, M is an inertia parameter matrix of the under-actuated unmanned ship, G (x) is a Coriolis matrix, and H is a damping coefficient matrix;
B is a control matrix:
K is a control input configuration matrix;
wherein k 11,k22,k32 is the determination coefficient of forward thrust to longitudinal displacement speed, the determination coefficient of yaw moment to transverse displacement speed and the determination coefficient of yaw moment to heading angular speed respectively;
The measured variable Z' of the particle filter comprises two parts: the unmanned ship motion state predicted value X 'state_grey and the propeller expected output predicted value ctrl_grey' obtained through grey prediction;
The measurement equation of the particle filter is:
Z′pred＝C·X′pred+D·ctrl_grey′
Wherein Z' pred is the predicted value of the measured variable, and C is the output matrix; d is a direct feed matrix;
The specific steps of adopting the particle filter to estimate the accurate estimated value of the thrust loss in real time are as follows:
S2.1, acquiring a fault state quantity, wherein the fault state quantity comprises an unmanned ship motion state sampling value, a propeller expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value which are received from a ship end thrust loss sub-estimator in a fault calculation period as the fault state quantity;
S2.2, initializing: setting the total number N of particles, taking a state variable formed by an unmanned ship motion state sampling value in a fault state quantity and a propeller expected output sampling value as a mean value of the state variable, presetting the distribution of the particles in a preset value space, and simulating the state variable particle space to obtain N particles conforming to a distribution rule in the value space;
S2.3, prediction: substituting each particle into a state transfer equation and a measurement equation to generate N predicted particles Z' pred (i):
Z′pred(i)＝C·X′pred(i)+D·ctrl_grey′(i) i＝1,2...,N
s2.4, correction: the predicted value of the motion state of the unmanned ship in the fault state quantity and the predicted value expected to be output by the propeller are used as the updated value Z' of the measured variable;
S2.5, weight calculation: calculating the Euclidean distance W i between each predicted particle Z 'pred (i) and the measured updated value Z', and assigning a weight to each particle according to the difference of the distances, wherein the predicted particle with smaller distance has larger weight;
s2.6, resampling: setting the resampling quantity M, and resampling according to the sampling weight of each particle to obtain a new particle group consisting of M particles;
s2.7, carrying out normalization processing on the sampling weights of M particles in the particle group obtained in the step S2.6 again to obtain respective weights alpha i, and carrying out weighted average on state variable predicted values corresponding to the particles according to the weights alpha i to obtain state variable predicted values Thereby obtaining the accurate estimated value/>, of the thrust loss prediction
2. The system for diagnosing faults of the unmanned ship propeller according to claim 1, wherein the grey prediction method is characterized in that the grey prediction is carried out according to a current calculation period unmanned ship motion state sampling value X state (k) and previous m calculation period unmanned ship motion state sampling values X state(k-1),...,Xstate(k-m+1),Xstate (k-m) to obtain a next calculation period unmanned ship motion state prediction value X state (k+1); according to a current calculation period propeller expected output sampling value ctrl (k) and previous m calculation period unmanned plane motion state sampling values ctrl (k-1),. The ctrl (k-m+1), the ctrl (k-m) carries out one-step gray prediction on the propeller expected output value to obtain a next calculation period propeller expected output prediction value ctrl (k+1), and the m value is more than or equal to 10.
3. A failure diagnosis system for an unmanned aerial vehicle propeller according to claim 2, wherein the state variable X (k) of the kalman filter comprises two parts: the unmanned ship motion state X state (k) and the thrust loss value X loss (k) of the current calculation period, wherein the unmanned ship motion state X state (k) comprises an unmanned ship longitudinal moving speed u (k), a transverse moving speed v (k) and a heading angular speed r (k); the thrust loss value X loss (k) includes a propeller forward thrust loss L u (k) and a yaw moment loss L r (k), that is:
X(k)＝[Xstate(k);Xloss(k)]＝[u(k),v(k),r(k),Lu(k),Lr(k)]T；
The input variable U (k) of the kalman filter is the propeller desired output value ctrl (k), including the forward thrust T u (k) and the yaw moment T r (k), i.e.:
U(k)＝[Tu(k),Tr(k)]T；
the state transfer equation of the Kalman filter is established based on the ship dynamics model, when the ship dynamics model is of the following form:
Wherein x is the motion state of the unmanned ship, and x= [ u, v, r ] T, wherein u, v, r are the longitudinal moving speed, the transverse moving speed and the heading angular speed of the unmanned ship respectively; m is an inertial parameter matrix of the under-actuated unmanned ship; g (x) is Coriolis Li Juzhen; h is a damping coefficient matrix; f= [ T u,Tr]T ] is a control input matrix, wherein T u is forward thrust, and T r is yaw moment; k is a control input configuration matrix;
the state transition equation is specifically:
X(k+1)＝AX(k)+BU(k)
wherein A is a coefficient matrix:
delta T is a calculation period, M is an inertia parameter matrix of the under-actuated unmanned ship, G (x) is a Coriolis matrix, and H is a damping coefficient matrix;
B is a control matrix:
K is the control input configuration matrix,
Wherein k 11,k22,k32 is the determination coefficient of forward thrust to longitudinal displacement speed, the determination coefficient of yaw moment to transverse displacement speed and the determination coefficient of yaw moment to heading angular speed respectively;
The measurement variable Z (k+1) of the Kalman filter comprises two parts: the predicted value X state (k+1) of the unmanned ship motion state and the predicted value ctrl (k+1) of the propeller expected output are obtained through grey prediction;
the measurement equation of the Kalman filter is:
Z(k+1)＝CX(k+1)+DU(K+1)
Wherein C is an output matrix; d is a direct feed matrix;
4. a fault diagnosis system for an unmanned ship propeller according to claim 3, wherein the kalman filter calculates a rough estimate of thrust loss as follows:
S1, estimating a state: according to the state transition equation, the state variable prior estimated value of the next calculation period can be calculated by the state variable X (k)
S2, calculating a priori covariance matrix P - (k+1) of the state variables:
P-(k+1)＝AP(k)AT+Q
Wherein Q is a process noise covariance matrix, P (k) is a state variable covariance matrix, and the initial value of the state variable covariance matrix P (k) is: p (0) =diag [0, 0];
s3, calculating Kalman gain K (k+1):
K(k+1)＝P-(k+1)CT(CP-(k+1)CT+R)-1
wherein R is a measurement noise covariance matrix;
s4, taking the predicted value of the unmanned ship motion state and the predicted value of the propeller expected output in the fault state quantity as a measurement variable Z (k+1), and calculating the optimal estimated value of the state variable at the next moment
S5, updating a state covariance matrix:
P(k+1)＝(I-K(k+1)H)P-(k+1)
and S6, returning to the step S1 when the next calculation period comes, and executing the loop iteration calculation of the steps S1 to S5 again.
5. The unmanned ship propeller fault diagnosis system according to claim 1, wherein the sampling weight of each particle of the particle filter is calculated by the following method:
A. Absolute sampling weights d' i are calculated:
d′i＝1/Wi
B. The absolute sampling weight d' i is normalized to obtain the sampling weight The sampling weight d i of each particle is limited to the (0, 1) range, and
6. A method of fault diagnosis of an unmanned aerial vehicle propeller based on the fault diagnosis system of an unmanned aerial vehicle propeller according to claim 1, comprising:
S1, executing the following steps at each calculation period of the boat end:
S1.1, acquiring a sampling value of the motion state of the unmanned ship and a sampling value of the thrust expected output of the unmanned ship propeller;
s1.2, forward prediction is carried out on the unmanned ship motion state and the propeller expected output value by using a gray prediction method, so that the unmanned ship motion state prediction value and the propeller expected output prediction value are obtained;
S1.3, inputting a propeller thrust expected output sampling value of the unmanned ship, a motion state predicted value of the unmanned ship and a propeller expected output predicted value corresponding to a current calculation period into a Kalman filter;
s1.4, calculating by a Kalman filter to obtain a rough thrust loss estimated value;
S1.5, if the absolute value of the difference between any one of the unmanned ship motion state sampling values and the average value of the unmanned ship motion state sampling values exceeds a corresponding preset threshold value or the rough estimated value of the thrust loss exceeds a second threshold value, the unmanned ship motion state sampling values, the propeller expected output sampling values, the unmanned ship motion state predicted values and the propeller expected output predicted values corresponding to the current calculation period are regarded as meeting event driving conditions, and the unmanned ship motion state sampling values, the propeller expected output predicted values and the unmanned ship motion state sampling values are sent to a shore as fault state quantities;
S2, after the bank terminal receives the fault state quantity, sequentially executing the following steps:
s2.1, a main estimator of the thrust loss at the shore end, wherein a particle filter is adopted to estimate an accurate estimated value of the thrust loss in real time by utilizing a fault state quantity;
S2.2, grading the fault degree according to the size of the accurate estimated value of the thrust loss, and transmitting the grading result and the accurate estimated value of the thrust loss to the boat end so as to facilitate the subsequent unmanned boat to perform fault-tolerant control;
the particle filter is used for estimating the accurate estimated value of the thrust loss in real time, and comprises the following specific steps:
S2.1.1, acquiring a fault state quantity, wherein the fault state quantity comprises an unmanned ship motion state sampling value, a propeller expected output sampling value, an unmanned ship motion state predicted value and a propeller expected output predicted value which are received from a ship end thrust loss sub-estimator in a fault calculation period as the fault state quantity;
S2.1.2, initializing: setting the total number N of particles, taking a state variable formed by an unmanned ship motion state sampling value and a propeller expected output sampling value in a fault state quantity as a mean value of the state variable according to preset particle distribution, and simulating a state variable particle space in a preset value space to obtain N particles conforming to a distribution rule in the value space;
s2.1.3, predicting: substituting each particle into a state transfer equation and a measurement equation to generate N predicted particles Z' pred (i);
Z′pred(i)＝C·X′pred(i)+D·ctrl_grey′(i) i＝1,2...,N
S2.1.4, correction: the predicted value of the motion state of the unmanned ship in the fault state quantity and the predicted value expected to be output by the propeller are used as the updated value Z' of the measured variable;
S2.1.5, weight calculation: calculating the Euclidean distance W i between each predicted particle Z 'pred (i) and the measured updated value Z', and assigning a weight to each particle according to the difference of the distances, wherein the predicted particle with smaller distance has larger weight;
s2.1.6, resampling: setting the resampling quantity M, and resampling according to the sampling weight of each particle to obtain a new particle group consisting of M particles;
S2.1.7, carrying out normalization processing on sampling weights of M particles in the particle group obtained in the step S2.6 again to obtain respective weights alpha i, carrying out weighted average on state variable predicted values corresponding to the particles according to the weights alpha i to obtain state variable predicted values Thereby obtaining the accurate estimated value/>, of the thrust loss prediction
7. The method for diagnosing faults of the unmanned ship propeller according to claim 6, wherein the sampling weight of each particle of the particle filter is calculated by the following method:
A. Absolute sampling weights d' i are calculated:
d′i＝1/Wi
B. The absolute sampling weight d' i is normalized to obtain the sampling weight The sampling weight d i of each particle is limited to the (0, 1) range, and
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110649578.XA CN113486564B (en) | 2021-06-10 | 2021-06-10 | Unmanned ship propeller fault diagnosis system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110649578.XA CN113486564B (en) | 2021-06-10 | 2021-06-10 | Unmanned ship propeller fault diagnosis system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113486564A CN113486564A (en) | 2021-10-08 |
| CN113486564B true CN113486564B (en) | 2024-05-14 |
Family
ID=77934627
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110649578.XA Active CN113486564B (en) | 2021-06-10 | 2021-06-10 | Unmanned ship propeller fault diagnosis system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113486564B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114491819B (en) * | 2022-02-16 | 2022-08-02 | 哈尔滨逐宇航天科技有限责任公司 | Rocket carrying capacity intelligent solving method based on speed loss calculation |
| CN115576184B (en) * | 2022-08-27 | 2024-08-02 | 华中科技大学 | Online fault diagnosis and fault tolerance control method for propeller of underwater robot |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101920762A (en) * | 2009-06-09 | 2010-12-22 | 同济大学 | Ship dynamic positioning method based on real-time correction of noise matrix |
| CN106643723A (en) * | 2016-11-07 | 2017-05-10 | 哈尔滨工程大学 | Calculation method of safe navigation position of unmanned ship |
| CN108319138A (en) * | 2018-01-29 | 2018-07-24 | 哈尔滨工程大学 | A kind of sliding formwork of drive lacking unmanned boat-contragradience double loop Trajectory Tracking Control method |
| KR101907589B1 (en) * | 2018-01-22 | 2018-10-12 | 연세대학교 산학협력단 | Structural system identification using extended kalman filter and genetic algorithm |
| CN112099065A (en) * | 2020-08-11 | 2020-12-18 | 智慧航海（青岛）科技有限公司 | Redundant fault-tolerant system for state estimator of ship dynamic positioning system |
-
2021
- 2021-06-10 CN CN202110649578.XA patent/CN113486564B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101920762A (en) * | 2009-06-09 | 2010-12-22 | 同济大学 | Ship dynamic positioning method based on real-time correction of noise matrix |
| CN106643723A (en) * | 2016-11-07 | 2017-05-10 | 哈尔滨工程大学 | Calculation method of safe navigation position of unmanned ship |
| KR101907589B1 (en) * | 2018-01-22 | 2018-10-12 | 연세대학교 산학협력단 | Structural system identification using extended kalman filter and genetic algorithm |
| CN108319138A (en) * | 2018-01-29 | 2018-07-24 | 哈尔滨工程大学 | A kind of sliding formwork of drive lacking unmanned boat-contragradience double loop Trajectory Tracking Control method |
| CN112099065A (en) * | 2020-08-11 | 2020-12-18 | 智慧航海（青岛）科技有限公司 | Redundant fault-tolerant system for state estimator of ship dynamic positioning system |
Non-Patent Citations (1)
| Title |
|---|
| 胡常青等.《 无人水面艇自主导航技术》.《导航与控制》.2019,全文. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113486564A (en) | 2021-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113486564B (en) | Unmanned ship propeller fault diagnosis system and method | |
| CN110703742B (en) | Event-driven and output quantization-based fault detection method for unmanned surface vehicle system | |
| CN109975839B (en) | Joint filtering optimization method for vehicle satellite positioning data | |
| CN113178098B (en) | Unmanned ship event triggering layered cooperative control system | |
| CN110954132A (en) | Method for carrying out navigation fault identification through GRNN (generalized regression neural network) assisted adaptive Kalman filtering | |
| CN112432644A (en) | Unmanned ship integrated navigation method based on robust adaptive unscented Kalman filtering | |
| CN113175926B (en) | Self-adaptive horizontal attitude measurement method based on motion state monitoring | |
| CN110933597A (en) | Bluetooth-based multi-unmanned vehicle collaborative fault-tolerant navigation positioning method and system | |
| CN109682383B (en) | Real-time filtering positioning method for measuring distance and data by using deep space three-way | |
| CN111930094A (en) | Unmanned aerial vehicle actuator fault diagnosis method based on extended Kalman filtering | |
| CN116954137B (en) | Method, system, device, equipment and medium for controlling movement of medium-crossing aircraft | |
| CN110703205A (en) | Ultrashort baseline positioning method based on adaptive unscented Kalman filtering | |
| CN117268381A (en) | Spacecraft state judging method | |
| Liu et al. | Navigation algorithm based on PSO-BP UKF of autonomous underwater vehicle | |
| Wang et al. | Physics-informed data-driven approach for ship docking prediction | |
| CN113989327B (en) | Single UUV target state estimation method based on convolutional neural network particle filter algorithm | |
| CN111273671A (en) | Non-periodic communication remote observer of intelligent ship | |
| CN114061592B (en) | Adaptive robust AUV navigation method based on multiple models | |
| CN113504555B (en) | Air route path updating method, system and storage medium | |
| CN114035154A (en) | Motion parameter assisted single-station radio frequency signal positioning method | |
| CN117367410B (en) | State estimation method, unmanned underwater vehicle and computer readable storage medium | |
| CN112924936B (en) | Multi-AUV (autonomous Underwater vehicle) cooperative positioning method based on sliding observation | |
| Hemakumara et al. | UAV parameter estimation with multi-output local and global Gaussian process approximations | |
| CN110926496B (en) | Method, device and system for detecting motion abnormity of underwater vehicle | |
| Sushchenko et al. | Application of Neural Networks for Processing Information in Redundant Rate Gyroscopes |
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 | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20220413 Address after: 266200 aoshanwei sub district office, Jimo District, Qingdao City, Shandong Province Applicant after: Aerospace Times (Qingdao) marine equipment technology development Co.,Ltd. Address before: 142 box 403, box 100854, Beijing, Beijing, Haidian District Applicant before: BEIJIGN INSTITUTE OF AEROSPACE CONTROL DEVICES |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |