CN113325708B - Fault estimation method of multi-unmanned aerial vehicle system based on heterogeneous multi-agent - Google Patents
Fault estimation method of multi-unmanned aerial vehicle system based on heterogeneous multi-agent Download PDFInfo
- Publication number
- CN113325708B CN113325708B CN202110524374.3A CN202110524374A CN113325708B CN 113325708 B CN113325708 B CN 113325708B CN 202110524374 A CN202110524374 A CN 202110524374A CN 113325708 B CN113325708 B CN 113325708B
- Authority
- CN
- China
- Prior art keywords
- agent
- fault
- unmanned aerial
- aerial vehicle
- intelligent agent
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The embodiment of the invention discloses a fault estimation method of a multi-unmanned aerial vehicle system based on heterogeneous multi-intelligent agents, and relates to the field of heterogeneous multi-intelligent agent fault estimation. The invention can reduce the cost of providing the fault estimator and simultaneously realize the fault estimation of the whole multi-unmanned aerial vehicle system. The invention includes: selecting a main agent in a multi-unmanned aerial vehicle system, dividing a neighbor agent to which the main agent belongs from the multi-unmanned aerial vehicle system, and deploying a fault estimator on the main agent, wherein the fault estimator comprises an F I R filter; obtaining the optimal F IR filter gain under the minimum estimation error performance index; and obtaining a fault estimation result by utilizing the output information of the sensor of the main intelligent agent and through a fault estimator arranged on the main intelligent agent. The method is suitable for unmanned aerial vehicle cluster fault estimation.
Description
Technical Field
The invention relates to the field of heterogeneous multi-agent fault estimation, in particular to a fault estimation method of a multi-unmanned aerial vehicle system based on heterogeneous multi-agents.
Background
With the continuous development of network communication technology, multi-agent technology has been a continuous focus of attention in recent years. Therefore, the method is widely applied to the practical application fields including robots, unmanned plane formation control, intelligent traffic control and the like. For example, the use of multiple drone systems in military settings for mission surveillance, surveillance and attack has been in history for decades. With the continuous advancement of technology, the cost of drones is continuously decreasing, with a number of emerging applications such as agricultural plant protection, weather detection, environmental and natural disaster monitoring, traffic control, cargo transport, public safety, etc. Along with the wide application of the method in daily life, real-time fault monitoring is the key for guaranteeing the safe operation of the method. Considering the individual difference of each object of the unmanned aerial vehicle formation, compared with a homogeneous agent, the modeling by utilizing the heterogeneous multi-agent system is a more practical and accurate modeling means. Due to the particularity of information interaction and mutual cooperation of neighbor nodes of the multi-agent system, once any sub-agent fails, fault information can be rapidly diffused and transmitted through communication topology, the performance of the whole system is further reduced, the stable operation of the system is further influenced, and great property loss and casualties are brought.
Therefore, the method has important theoretical significance and engineering value for real-time online fault diagnosis of the multi-agent system. Compared with a homogeneous multi-agent, the heterogeneous multi-agent can describe a complex actual industrial process more accurately due to different system dynamics and even state space dimensions among sub-agents of the heterogeneous multi-agent. In addition, unlike fault detection, fault estimation can obtain detailed information including fault amplitude, type and the like, so that a powerful reference can be provided for an observer to take timely countermeasures subsequently. Therefore, fault estimation for heterogeneous multi-agents is a challenging and practical direction of research. For example, the following steps are carried out: how to realize the fault estimation of the whole multi-agent system while reducing the expenditure of equipping a fault estimator; and how to reduce the computational complexity for obtaining the gain of the fault estimator and further reduce the time for fault estimation are problems to be solved.
Disclosure of Invention
The embodiment of the invention provides a fault estimation method of a multi-unmanned aerial vehicle system based on heterogeneous multi-agents, which can reduce the cost of a fault estimator and realize the fault estimation of the whole multi-unmanned aerial vehicle system.
In order to achieve the above purpose, the embodiment of the invention adopts the following technical scheme:
s1, selecting a main intelligent agent in a multi-unmanned aerial vehicle system, dividing a neighbor intelligent agent to which the main intelligent agent belongs from the multi-unmanned aerial vehicle system, and deploying a fault estimator on the main intelligent agent, wherein the fault estimator comprises an FIR filter;
s2, obtaining the optimal FIR filter gain under the minimum estimation error performance index;
s3, obtaining a fault estimation result through a fault estimator deployed on the main intelligent agent by utilizing the output information of the sensor of the main intelligent agent, wherein the output information of the sensor of the main intelligent agent comprises: when the fault estimation result is affected by the fault of the main intelligent agent and the fault of the neighbor intelligent agent to which the main intelligent agent belongs, the output information of the sensor of the main intelligent agent and the fault estimation result correspond to the fault condition affected by the fault of the main intelligent agent and the fault condition of the neighbor intelligent agent to which the main intelligent agent belongs.
The fault estimation method of the multi-unmanned aerial vehicle system based on the heterogeneous multi-agent provided by the embodiment of the invention provides a new analytic redundancy relation, and designs a new performance index in a random sense to construct a residual error generator based on an FIR filter, so that the fault estimation of the heterogeneous multi-agent with random nonlinearity and a neighbor agent thereof is realized with obvious computational advantages. The cost of equipping the fault estimator is reduced, and meanwhile, the fault estimation of the whole multi-agent is realized. The calculation amount of the online updating gain algorithm of each fault estimator is obviously improved, and the fault estimation time is shortened. In addition, the computational complexity for obtaining the gain matrix of the fault estimator is obviously reduced, and the fault estimation efficiency is further improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic diagram of a heterogeneous multi-agent system based on a multi-UAV scenario;
FIG. 2 is a communication topology undirected graph of a heterogeneous multi-agent system with five sub-agents in an embodiment of the invention;
FIG. 3 is a schematic diagram of a fault estimation waveform for the multi-agent 1 in an embodiment of the present invention;
FIG. 4 is a schematic diagram of a fault estimation waveform for a multi-agent 3 in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of a possible application of fault estimation provided by an embodiment of the present invention;
fig. 6 is a schematic flowchart of an algorithm provided in the embodiment of the present invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the present invention will be described in further detail with reference to the accompanying drawings and specific embodiments. Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention. As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or coupled. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items. It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
In the current research direction, the nonlinear phenomenon is the existence which cannot be ignored in practice. Non-linear disturbances may occur in some type of probabilistic manner, taking into account a range of phenomena such as intermittent network congestion, random failures and component repairs, changes in subsystem interconnections, sudden environmental disturbances, and adjustments to the operating point of the nonlinear system linearization model. The present embodiment is designed to design a heterogeneous multi-agent model that includes stochastic nonlinearities, which have a more general form, covering most of the uncertainty-related descriptions involved in the current invention, such as multiplicative noise, state sign-dependent noise, and nonlinear factors.
At present, among many model-based fault diagnosis methods, FIR (Finite impulse response) filters, which are typical representatives of fault diagnosis strategies with time window constraints, are widely used as residual generators because they can extract fault information from output information of Finite length for system health monitoring. In addition, it has been proved in practice that, compared with a filter with an IIR (infinite impulse response) structure (i.e., a residual generator is constructed by using all input and output information from a zero initial time to a current time based on an observer, a filter, etc.), an FIR filter has certain robustness to initial values and modeling errors, and transient uncertainty that may occur after long-time operation. While FIR filters are currently being used in the field of state estimation or signal processing, their application to fault estimation in a single system is relatively rare, let alone heterogeneous multi-agent systems. In addition, due to the time-varying characteristic of the actual industrial process, the fault estimator has to update the filter gain at every moment, so that the fault estimation algorithm based on the FIR filter with small calculation burden has certain theoretical value and application prospect for a heterogeneous multi-agent system with random nonlinearity.
In order to solve the above problems, the design idea of the present embodiment is as follows: a new analytic redundancy relation is provided, and a new performance index under the random meaning is designed to construct a residual error generator based on an FIR filter, so that the fault estimation of the heterogeneous multi-agent with random nonlinearity and the neighbor agents thereof is realized with obvious computational advantages.
Specifically, the heterogeneous multi-agent system mentioned in this embodiment may be applied to a cluster of drones that are widely used at present, for example: information interaction among multiple unmanned aerial vehicle (multi-agent) systems, fault and disturbance information in each multi-agent can be mutually transmitted in the whole multi-agent system, and the output actually acquired from a sensor corresponding to a multi-agent i is not only influenced by the fault and disturbance of the multi-agent i, but also comprises the fault and disturbance of a neighbor agent. Based on the method, if the fault estimation of a single agent is realized, resources are wasted, and meanwhile, the fault estimation of the agent i and the neighbor agents thereof is realized to be more in line with actual requirements. To achieve state of health detection for the entire multi-agent system, it is generally necessary to equip each agent with a fault estimator, which increases hardware overhead. Based on the new augmentation system, one fault estimator can realize the simultaneous fault estimation of a plurality of agents, and the hardware cost can be greatly reduced. For example, for a multi-drone system with the following topology: for example, as shown in fig. 1, the intelligent agents 2 and 4 or 2 and 3 can be selected to design the fault estimator respectively, so as to realize the fault estimation of the whole multi-intelligent-agent system, compared with the need of designing 4 fault estimators, the new augmented system provides the possibility of reducing the hardware overhead.
Based on the above heterogeneous multi-agent system that can be applied to the unmanned aerial vehicle cluster and more other scenes, an embodiment of the present invention provides a fault estimation method for a multi-unmanned aerial vehicle system based on heterogeneous multi-agents, as shown in fig. 6, including:
s1, selecting a main intelligent agent in a multi-unmanned aerial vehicle system, dividing a neighbor intelligent agent to which the main intelligent agent belongs from the multi-unmanned aerial vehicle system, and deploying a fault estimator on the main intelligent agent, wherein the fault estimator comprises an FIR filter. Specifically, in the multi-UAV system, a main agent is selected according to a known topological structure, neighbor agents to which the main agent belongs are divided from the multi-UAV system, and a fault estimator is deployed on the main agent and comprises an FIR filter.
Wherein, the fault estimator comprises an FIR filter. After the unmanned aerial vehicles in the multi-unmanned aerial vehicle system are started and networked, communication connection is established among the unmanned aerial vehicles, and the unmanned aerial vehicles which are in communication connection with the main intelligent body are used as neighbor intelligent bodies. Specifically, the method can be understood as that a heterogeneous multi-agent system with random nonlinearity selects a main agent to be monitored according to the communication topological connection relation, and derives a new augmented system state space description covering all faults of the main agent and neighbor agents thereof.
And S2, obtaining the optimal FIR filter gain under the minimum estimation error. Specifically, the optimal FIR filter gain under the minimum estimation error performance index is obtained through random matrix analysis and calculation.
S3, obtaining a fault estimation result through a fault estimator deployed on the main intelligent agent by utilizing the output information of the sensor of the main intelligent agent, wherein the output information of the sensor of the main intelligent agent comprises: when the fault estimation result is affected by the fault of the main intelligent agent and the fault of the neighbor intelligent agent to which the main intelligent agent belongs, the output information of the sensor of the main intelligent agent and the fault estimation result correspond to the fault condition affected by the fault of the main intelligent agent and the fault condition of the neighbor intelligent agent to which the main intelligent agent belongs. Specifically, the fault estimation results of the intelligent agent and the adjacent intelligent agents are obtained through a fault estimator deployed on the main intelligent agent by utilizing the output which is obtained from a sensor on the main intelligent agent and is simultaneously influenced by the faults of the intelligent agent and the adjacent intelligent agents.
In the embodiment, a new analytic redundancy relation is provided, and a new performance index in a random sense is designed to construct a residual error generator based on an FIR filter, so that fault estimation of heterogeneous multi-agent and neighbor agents thereof with random nonlinearity is realized with obvious computational advantages. On the one hand, the expense of equipping the fault estimator is reduced, and meanwhile, the fault estimation of the whole multi-agent is realized. On the other hand, the calculation amount of the online updating gain algorithm of each fault estimator is obviously improved, and the fault estimation time is shortened.
In this embodiment, after the unmanned aerial vehicles in the multi-unmanned aerial vehicle system are started and networked, the pretreatment of the system is performed, including:
and establishing a heterogeneous multi-agent model of the multi-unmanned aerial vehicle system according to the topological relation after networking of the multi-unmanned aerial vehicle system. And then, generating the heterogeneous multi-agent model into an augmentation system in the multi-unmanned aerial vehicle system under fault propagation.
Wherein the topological relation comprises: the connection relationship between each master agent and each neighbor agent. For example: the present embodiment designs a mathematical model for modeling multiple drone systems as heterogeneous multiple agents, which includes:
wherein the content of the first and second substances,
respectively representing the state, output, unknown additive disturbance and a fault signal to be detected of the intelligent agent i. A. The i (k),B d,i (k),B f,i (k),C i (k),D d,i (k),D f,i (k) Respectively, known to be a time-varying matrix of the appropriate dimension. g i (k,x i (k),δ i (k)),q i (k,x i (k),ζ i (k) Respectively, and satisfy the following random properties:
wherein, delta i (k)，ζ i (k) Is Gaussian white noise with independent intelligent agent i and zero mean and variance of 1, m 1 Is a known non-negative integer that is,
is a matrix of compatible dimensions known to be semi-positive. x is a radical of a fluorine atom 0,i Is one and g i (k,x i (k),δ i (k)),q i (k,x i (k),ζ i (k) Random vector of any appropriate dimension, regardless of the state of agent i.
As shown in FIG. 5, y i (k)∈R m Is the output information of a multi-agent i, in a multi-agent system, due to information interaction, the output information collected via a sensor is:
wherein, a ij Is the ith row and j column elements of the weighted connection matrix Λ of the multi-agent topology. The communication topology between N agents can be described by a graph G = { Ω, Λ }, where Ω represents a set of nodes (each node represents an agent), and Λ = [ a ] = ij ]∈R n×n The weighted adjacency matrix of diagram G is shown. If there is an edge from node j to node i, denoted as (i, j), then a ij >0, otherwise a ij And =0. If a is ij >0, then node i can obtain information from node j, a ij Represents the weight of the information of node j to the decision of node i, when node j is also called a neighbor of node i. It is considered that once an agent fails, the failure information is transmitted to the neighbor agents through the network communication topology, and the cooperation of the whole multi-agent system is further influenced. Because, according to the interconnection relationship, the main agent is selected, and a new augmentation system covering all dynamic information of the main agent and the neighbor agents is determined. According to the augmented system, a unified fault estimator is determined, thereby enabling simultaneous fault estimation of a master agent and its neighbor agents. The specific definition includes:
the system is converted into: in view of fault delivery, the derived new augmentation system is specifically:
wherein:
wherein
Which represents the kronecker product of,L ij the elements in row i and column j of the laplace matrix L representing the multi-agent topology. The laplace matrix of the figure is L = [ L = [ ] ij ]∈R n×n Wherein, in the process,
l ij ＝-a ij ，i≠j。
in this embodiment, the S3 includes:
the main intelligent agent obtains output information collected from the main intelligent agent in the current time window, and then the output information passes through a fault estimator to obtain a fault estimation result, wherein the output information in the current time window is represented by state information in the previous time window. The currently acquired output information comprises the output information in the current time window [ k-s, k ], and the output information in the current time window is represented by all the state information in the previous time window [ k-s-1, k-1 ].
Specifically, a certain finite time window length s needs to be determined first, and the relative output information of the agent and its neighboring agents in the time domain under the combined action of additive disturbance, fault and random nonlinearity is obtained. In the subsequent processing process, the analytical redundancy described by all states in the previous time window of the relative output information needs to be acquired, the FIR filter is utilized to estimate the faults of the main agent and the neighbor agents, and a weight matrix is introduced to construct an estimation error, wherein the estimation error is as follows: the difference between the fault estimate information and the actual fault information, and then the performance index that minimizes the fault estimate error in a random sense is determined. Finally, according to the new performance index, calculating and obtaining the optimal filter gain when the performance index obtains the minimum value; likewise, the FIR filter-based fault estimator of the selected other master agent may be constructed to enable overall system health monitoring.
Specifically, in S3, performing fault estimation by using an FIR filter includes:
wherein the content of the first and second substances,
for the sought fault estimation, p, including self and its neighbor agents i (k) Is a FIR filter gain matrix, Y i,s (k) Representing the time window k-s, k acquired from agent i]The output information of the internal memory is stored in the internal memory,
representing the set of neighbour agents to which i agents belong, i.e.
Neighbor agent representing agent i
I Mi represents the potential of the set, i.e. the number of agents in the set of neighbour agents to which the master agent corresponds,
a fault estimate for agent i is represented,
represents the fault estimation of the | Mi | th neighbor agent of the master agent i.
In this embodiment, an FIR filter may be used to estimate the fault generated by the heterogeneous random nonlinear multi-agent, specifically:
wherein the content of the first and second substances,
for the sought fault estimation, including the neighbor agent, p i (k) Is a filter gain matrix, Y i,s (k) Indicating the time of acquisition from agent iInter window [ k-s, k ]]Unlike the prior art, the output information in the present invention is not described by the initial time or the state of the current time in the current time window, but described by all the states in the previous time window, and specifically, the output information can be parsed as follows: the state quantity of the previous time window can be represented, specifically as follows:
wherein the content of the first and second substances,
specifically, this embodiment establishes a completely new type of Y i,s (k) The redundancy relationship is analyzed. It can be intuitively found that, compared with the existing scheme, the coefficient matrixes related to the new analytic redundancy provided by the embodiment all have special diagonal structures, so that the calculation complexity of the gain of the optimal filter obtained based on the analytic redundancy is greatly reduced.
In this embodiment, the weight matrix W is used f Error of structure estimation r i (k) Wherein, in the step (A),
W f f is a symbol for distinguishing the parameters. Specifically, a weight matrix W is introduced f Structure estimation error r i (k) The method comprises the following steps:
wherein, W f F i (k) Generating information relating to the current timeAll faults
True value of, setting
Is the number of possible fault dimensions for agent i and R is the real vector space.
Further, in order to ensure the accuracy of the fault estimation, the estimation error at every moment needs to be as small as possible, that is, r needs to be made i (k) As small as possible in a random sense. Based on this, E (r) is selected i T (k)r i (k) As performance indicators):
wherein:
considering W i (k) The method comprises the known variance information of random nonlinearity, and the performance index is converted into the performance index through solving through random matrix analysis
Is used to constrain the fault estimation error, the optimal filter gain is
Of the minimum value point of (a), wherein,
M qi (k) Can be obtained by mixing M gi (k-1) in
And Γ p,i (k-s-1) is replaced by
And Γ p,i (k-s) obtained, I Ni Is composed of
A square matrix of a unit of order,
is composed of
And (5) a unit square matrix of order.
In the preferred embodiment of this embodiment, the minimum value of the performance index is solved to obtain the optimal filter gain of the FIR filter as
Wherein p is i (k) Which represents the gain of the filter that is sought,
and
respectively representing time-varying matrices C of known correct dimensions i (k) And D f,i (k) The calculated result, R, was obtained i (k) Is a positive definite matrix.
Specifically, in order to minimize the fault estimation error, the optimal filter gain specifically includes:
wherein, W f Is a weight matrix, R i (k) Is a positive definite matrix. Further, the positive definite matrix R i (k) The method specifically comprises the following steps:
in this embodiment, the matrix calculations involved in the above process all have a diagonal matrix structure that is easy to obtain, specifically, the coefficient matrices of the analytical redundancy in the above description all have a diagonal structure, and the failure estimator is a failure estimation generated by using the analytical redundancy, so the solution of the obtained filter is also a diagonal matrix structure.
Compared with the existing scheme, the embodiment has the possible advantages that:
firstly, considering the randomness of nonlinear elements in an actual environment, summarizing a fault diagnosis model of a heterogeneous multi-agent into discrete time-varying state space description influenced by random nonlinearity, and combining an FIR filter to realize simultaneous fault estimation of an agent and a neighbor agent thereof. On the one hand, the expense of equipping the fault estimator is reduced, and meanwhile, the fault estimation of the whole multi-agent is realized. On the other hand, the calculation amount of the online updating gain algorithm of each fault estimator is obviously improved, and the fault estimation time is shortened.
Secondly, since variance information is coupled with state information in the random nonlinearity, the magnitude of the fault estimation error depends on the corresponding state information. Through analysis of the random matrix inequality, a new performance index completely decoupled from system virtual information is designed, and fault estimation errors are minimized in a random sense.
Moreover, by utilizing all the state information in the previous time window, new analytical redundancy of relative output information of the intelligent agent and the adjacent intelligent agents is obtained, and the calculation burden brought by updating the optimal filter gain is greatly reduced.
The embodiment generally discloses a fault estimation method of a heterogeneous random nonlinear multi-agent system based on an FIR filter. The randomness of nonlinear elements in the actual environment is considered, and the simultaneous fault estimation of the intelligent agent and the neighbor intelligent agents of the intelligent agent is realized by combining the FIR filter. On the one hand, the expense of equipping the fault estimator is reduced, and meanwhile, the fault estimation of the whole multi-agent is realized. On the other hand, the calculation amount of the online updating gain algorithm of each fault estimator is obviously improved, and the fault estimation time is shortened. Through analysis of a random matrix inequality, the variance information and the state information in the random nonlinearity are completely decoupled, so that a new performance index is designed, and the fault estimation error is minimum in a random sense. And by utilizing all the state information in the previous time window, new analytical redundancy of the relative output information of the intelligent agent and the adjacent intelligent agents is obtained, and the calculation burden brought by updating the optimal filter gain is greatly reduced.
Specifically, for example, taking a heterogeneous random nonlinear multi-agent system composed of a cluster of drones as an example, as shown in fig. 5, a set of parameters related to the multi-drone cluster system is obtained from actual engineering practice as follows:
the random nonlinear function is described as:
further, then
Γ p,i (k)＝diag{0.09，0.04,0.09},i＝1,3,5,Γ p,i (k)＝diag{0.01,0.04},i＝2,4
Initial value x of multi-agent i,0 Disturbance vector d i (k) Vectors, both set to random appropriate dimensions, assuming the failure settings of multi-agent (drone) 1 and multi-agent (drone) 3 are:
the other intelligent agents operate healthily, the length of a time window is set to be s =3, and the fault dimension and the weight matrix W are known f The method comprises the following steps:
obtaining the optimal filter gain p i (k) Optimal sensor fault estimates for multi-agent 1 and multi-agent 3 are generated as shown in fig. 3 and 4.
Fig. 2 and 3 show that the fault estimation method based on the FIR filter provided by the invention can effectively obtain the real-time fault estimation information of the main agent and the neighbor agents thereof. Further, it can be concluded that the sub-agents with high intelligence level can be selected as the main agent that needs to establish the fault estimator, so that the fault estimation of the whole heterogeneous multi-agent system affected by random nonlinearity can be realized while the number of the fault estimators that need to be equipped is reduced.
It is considered that the computational complexity is mainly caused by multiplication and division. Therefore, the number of addition and subtraction operations involved in the algorithm is ignored, and the calculation load of the algorithm is measured by using the number of multiplication and division operations as an evaluation index. The system model in the existing scheme is a single linear discrete time-varying system affected by multiplicative noise, and is a special case of the heterogeneous random nonlinear multi-agent system considered herein. For the sake of analysis convenience, consider the case of only one drone. Compared with analytical redundancy used in the existing scheme, under the performance index of the minimum estimation error, the calculation complexity for solving the optimal filter gain related by the two methods is respectively as follows:
MD our ＝(5m 2 (n 1 +n 3 +n 4 )+m(n 1 +n 3 +n 4 ) 2 +m(n d1 +n d3 +n d4 ) 2 +m(n f1 +n f3 +n f4 ) 2 +m(n 1 +n 3 +n 4 )(n d1 +n d3 +n d4 +n f1 +n f3 +n f4 )+3m 2 (n d1 +n d3 +n d4 +n f1 +n f3 +n f4 )+3m 2 )(s+1)＝2816
wherein, MD our For the calculation mode, MD ori As used by existing solutions. It is clear that the algorithm proposed by the present invention already has significant computational advantages for a single agent and in the absence of neighboring agents, let alone the application of a multi-agent system with multiple subsystems and communicating with each other.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus embodiment, since it is substantially similar to the method embodiment, it is relatively simple to describe, and reference may be made to some descriptions of the method embodiment for relevant points. The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (2)
1. A fault estimation method of a multi-unmanned aerial vehicle system based on heterogeneous multi-agent is characterized by comprising the following steps:
s1, selecting a main intelligent agent in a multi-unmanned aerial vehicle system, dividing a neighbor intelligent agent to which the main intelligent agent belongs from the multi-unmanned aerial vehicle system, and deploying a fault estimator on the main intelligent agent, wherein the fault estimator comprises an FIR filter;
after unmanned aerial vehicles in the multi-unmanned aerial vehicle system are started and networked, communication connection is established among the unmanned aerial vehicles, and the unmanned aerial vehicle which is in communication connection with a main intelligent body is used as a neighbor intelligent body;
s2, obtaining the optimal FIR filter gain under the minimum estimation error performance index;
s3, obtaining a fault estimation result through a fault estimator arranged on the main intelligent agent by utilizing the output information of the sensor of the main intelligent agent, wherein the output information of the sensor of the main intelligent agent comprises the following steps: when the fault estimation result is influenced by the fault of the main intelligent agent and the fault of the neighbor intelligent agent to which the main intelligent agent belongs, the output information of the sensor of the main intelligent agent and the fault estimation result correspond to the fault condition influenced by the fault of the main intelligent agent and the fault condition of the neighbor intelligent agent to which the main intelligent agent belongs;
the S3 comprises the following steps: the main intelligent agent obtains output information collected from the main intelligent agent in the current time window, wherein the output information in the current time window is represented by state information in the previous time window;
in the S3:
utilizing the FIR filter for fault estimation, comprising:
wherein the content of the first and second substances,
is as requestedFault estimation including self and its neighbour agents, p i (k) Is a FIR filter gain matrix, Y i,s (k) Representing the time window k-s, k acquired from agent i]The output information of the internal-use device,
representing the set of neighbour agents to which i agents belong, i.e.
Neighbor agent representing agent i
I Mi represents the potential of the set, i.e. the number of agents in the set of neighbour agents to which the master agent corresponds,
a fault estimate for agent i is represented,
fault estimation of the | Mi | th neighbor agent representing the master agent i;
the length of the time window is s =3, and the weight matrix W f ：
The optimum filter gain of the FIR filter is
Wherein p is i (k) Which represents the gain of the filter that is sought,
and
respectively representing time-varying matrices C of known correct dimensions i (k) And D f,i (k) The calculated result, R, was obtained i (k) Is a positive definite matrix.
2. The method of claim 1, wherein after the drones in the multi-drone system are started and networked, performing a pre-processing of the system, comprising:
establishing a heterogeneous multi-agent model of the multi-unmanned aerial vehicle system according to the topological relation after networking of the multi-unmanned aerial vehicle system, wherein the topological relation comprises: the connection relation between each main agent and each neighbor agent;
and generating the heterogeneous multi-agent model into an augmentation system in the multi-unmanned aerial vehicle system under fault propagation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110524374.3A CN113325708B (en) | 2021-05-13 | 2021-05-13 | Fault estimation method of multi-unmanned aerial vehicle system based on heterogeneous multi-agent |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110524374.3A CN113325708B (en) | 2021-05-13 | 2021-05-13 | Fault estimation method of multi-unmanned aerial vehicle system based on heterogeneous multi-agent |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113325708A CN113325708A (en) | 2021-08-31 |
| CN113325708B true CN113325708B (en) | 2022-11-25 |
Family
ID=77415549
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110524374.3A Active CN113325708B (en) | 2021-05-13 | 2021-05-13 | Fault estimation method of multi-unmanned aerial vehicle system based on heterogeneous multi-agent |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113325708B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114237207B (en) * | 2021-12-21 | 2024-02-02 | 西北工业大学 | Multi-agent distributed fault diagnosis method under influence of communication noise |
| CN115167147B (en) * | 2022-08-09 | 2023-04-18 | 南京航空航天大学 | Distributed fault diagnosis method for swarm unmanned aerial vehicle based on thrust residual error estimation |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106200629B (en) * | 2016-09-30 | 2018-07-13 | 山东科技大学 | A kind of detectable degree analysis method of the failure of UAV Flight Control System |
| CN108196532B (en) * | 2018-03-07 | 2020-06-26 | 山东科技大学 | Fault detection and separation method for longitudinal flight control system of unmanned aerial vehicle based on nonlinear adaptive observer |
| CN108803316B (en) * | 2018-03-09 | 2020-12-15 | 南京航空航天大学 | Active fault-tolerant control method for faults of multi-agent system actuator |
| CN109918734B (en) * | 2019-02-01 | 2022-10-11 | 济南大学 | Underwater acoustic communication system fault estimation method based on FIR filter |
| CN109634798B (en) * | 2019-02-25 | 2020-12-15 | 北京理工大学 | Design method of fault estimator of piloting-following multi-agent distributed system |
-
2021
- 2021-05-13 CN CN202110524374.3A patent/CN113325708B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113325708A (en) | 2021-08-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sathishkumar et al. | Non-fragile filtering for singular Markovian jump systems with missing measurements | |
| Ding et al. | Distributed H∞ state estimation with stochastic parameters and nonlinearities through sensor networks: the finite-horizon case | |
| CN108427401B (en) | Flight control system cooperative fault diagnosis method with joint connectivity topology | |
| CN113325708B (en) | Fault estimation method of multi-unmanned aerial vehicle system based on heterogeneous multi-agent | |
| Liu et al. | Quasi-synchronization of heterogeneous networks with a generalized Markovian topology and event-triggered communication | |
| Khan et al. | Coordinated networked estimation strategies using structured systems theory | |
| Zhao et al. | Distributed event-triggered state estimation and fault detection of nonlinear stochastic systems | |
| Zheng et al. | Consensus of multi-agent systems with faults and mismatches under switched topologies using a delta operator method | |
| Yang et al. | Finite-time synchronization of coupled Markovian discontinuous neural networks with mixed delays | |
| Duan et al. | Finite-time distributed H∞ filtering for Takagi-Sugeno fuzzy system with uncertain probability sensor saturation under switching network topology: Non-PDC approach | |
| Li et al. | Distributed state estimation for stochastic discrete-time sensor networks with redundant channels | |
| Talebi et al. | Distributed Kalman filtering in presence of unknown outer network actuations | |
| Peng et al. | Passive state estimator design for Markovian complex networks with polytopic sensor failures | |
| Khan et al. | An Explainable Intelligent Framework for Anomaly Mitigation in Cyber-Physical Inverter-based Systems | |
| CN107563103B (en) | Consistency filter design method based on local conditions | |
| Zhu et al. | Fault detection for nonlinear networked control systems based on fuzzy observer | |
| CN113992535B (en) | Network reliability assessment method and device and electronic equipment | |
| CN113411312B (en) | State estimation method of nonlinear complex network system based on random communication protocol | |
| Li et al. | Topology inference on partially observable mobile robotic networks under formation control | |
| Zhang et al. | Distributed cooperative fault detection and isolation for uncertain multi‐agent systems based on zonotope theory | |
| Qiu et al. | Mean square average generalized consensus of multi-agent systems under time-delays and stochastic disturbances | |
| CN113205204A (en) | Power distribution network outage node prediction method based on Bayesian network | |
| Wang et al. | Fault detection for complex systems with channel fadings, randomly occurring multiple delays and infinitely distributed delays | |
| Hatanaka et al. | Distributed cooperative full-state observers for discrete-time linear systems with missing data | |
| Van Scoy et al. | A tutorial on the structure of distributed optimization algorithms |
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 |