CN103246921B - A kind of underwater robot environment perception method based on immune main body - Google Patents

Abstract

The invention belongs to underwater robot technical field, particularly a kind ofly can be used for solving underwater robot and the monitoring of other underwater hiding-machine equipment surrounding environment, monitoring, dangerous information process, keep away the underwater robot environment perception method of the problems such as barrier.The present invention includes following steps: immune main body and underwater robot structure are mapped; Monitoring main body obtains environmental knowledge in original state, detects the abnormal conditions in institute's monitoring range; Immunity communication agent sends a signal to dangerous discernment main body; Immune response main body performs the process to danger signal, and eliminate danger situation.The present invention is more flexible and changeable for underwater environment information processing, overcomes existing dependence traditional mathematics model, signal transacting, electronics and sensor information disposal route and lacks the defect adapting to environmental baseline complicated and changeable.Break through the limitation of traditional underwater robot environment perception method, make underwater robot environment sensing ability more effectively, flexibly.

Description

A kind of underwater robot environment perception method based on immune main body

Technical field

The invention belongs to underwater robot technical field, particularly a kind ofly can be used for solving underwater robot and the monitoring of other underwater hiding-machine equipment surrounding environment, monitoring, dangerous information process, keep away the underwater robot environment perception method of the problems such as barrier.

Background technology

Immune System especially human immune system is the result of nature long-term evolution; various immune molecule, cell, organ are combined; can comprehensively perception, utilize in external information; and process these information flexibly; various composition (intelligent agent) co-ordination completes identification, stable, defensive missions, protection health.Utilizing immune system mechanisms, has been devise computer security immune system, for defence, identifies computer virus, flame, attack etc.Underwater robot environment perception method mainly comprises visible sensation method, underwater acoustic technology, distributed information fusion etc., and the main visual information that relies on is judged the various situations that robot runs into by operator, avoids the generation of emergency condition or dangerous situation.At present, domestic and international underwater robot is not still to clear and definite method or the system of dangerous criminal, and majority utilizes the sensors such as sonar to avoid obstacle, or utilize sonar sensor data identification risk object.Immune dangerous criminal theory is not utilized to solve system or the method for this problem.Based on the Artificial Immune Network Theory of subject theory, pointing out that immune system can be seen is by various main body (corresponding B, T cell) mutual dynamic action and the network formed between main body and between main body and external environment.According to the definition of Multi-Agent Systems, a Multi-Agent Systems is a distributed computing system, common but sometimes again must the task of competition mutually in order to complete usually, is made up of and coordinates that it is movable interactive multiple main body.Immune Multiple components can meet such definition.The present invention is based on Immune System, design underwater robot dangerous criminal system, make underwater robot have Intellisense ability to a certain degree, make up the deficiency that existing method lacks adaptability, independence.

Summary of the invention

The object of the present invention is to provide a kind of underwater robot environment perception method system architecture being mapped to the different immune main body of function according to function modeling or division based on immune main body overcoming prior art deficiency.

The object of the present invention is achieved like this:

The present invention includes following steps:

(1) immune main body and underwater robot structure are mapped, in model, form immune host complex ImA={ImA by n immune main body ImA ₀, ImA ₁..., ImA _n-1, n=4, ImA ₀for immunologic surveillance main body, ImA ₁for immune communication agent, ImA ₂for dangerous discernment main body, ImA ₃for immune response main body;

(2) main body ImA is monitored ₀obtaining environmental knowledge in original state, the abnormal conditions in institute's monitoring range being detected, when monitoring underwater environment changed condition, by immune communication agent ImA ₁send a signal to dangerous discernment main body ImA ₂, described signal is expressed as X (t);

(3) prestore hazardous environment set of patterns Y={y ₁, y ₂..., y _j..., y _n, y _jfor a jth environmental pattern, j=1,2 ..., n, n are number of modes, and signal is expressed as X (t)={ x ₁(t), x ₂(t) ..., x _i(t) ..., x _m(t) | t=k}, x _it () is perceptual signal, m is for being expressed as perceptual signal number, immune communication agent ImA through filtering with after merging ₁send perceptual signal to dangerous discernment main body ImA ₂if, x _ik () ∈ X (k), Y' are the antibody set of patterns after upgrading, through dangerous discernment main body ImA ₂judge x _ik () belongs to danger signal, then learn and remember this situation, upgrading dangerous knowledge base, and Immune discrimination main body process analogue after empirically instructing, in order to second set response; If x _ik () ∈ X (k), shows there is corresponding solution in knowledge base;

(4) dangerous discernment main body ImA ₂after identifying danger signal, immune response main body ImA ₃perform the process to danger signal, eliminate danger situation.

After monitoring described danger signal, described danger signal identifies as follows:

(1) initialization: represent underwater security situation set of modes S with 16 strings of binary characters, the underwater environment signal character string of being monitored by monitoring main body is stored in R ₀in;

(2) affinity evaluation: R ₀in character string whether mate character string in S, determine whether mate by the affinity between them, adopt r to adjoin matching process;

(3) detectors set is produced: if from R ₀a character string and S in the affinity of at least one character string be greater than given threshold value, then thinks and must be eliminated this character string identification limit risk, otherwise character string introduces S set;

(4) if reach setting procedure number, then terminate, return step (2) if do not reach.

Beneficial effect of the present invention is:

The present invention is more flexible and changeable for underwater environment information processing, overcomes existing dependence traditional mathematics model, signal transacting, electronics and sensor information disposal route and lacks the defect adapting to environmental baseline complicated and changeable.Break through the limitation of traditional underwater robot environment perception method, make underwater robot environment sensing ability more effectively, flexibly.

Accompanying drawing explanation

Fig. 1 is underwater robot immunity multiagent system structure;

Fig. 2 underwater robot dangerous criminal environment;

Fig. 3 is underwater robot immunity multiagent inner structure.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described further.

The present invention, by immune main body and underwater robot function-structure mapping, forms immune host complex ImA={ImA by n immune main body ImA in model ₀, ImA ₁..., ImA _n-1, n=4, ImA ₀for immunologic surveillance main body corresponding machine environment monitoring sensor system, the signal from sonar is converted to 16 long 2 system character strings of immunologic mechanism process, ImA ₁for immune communication agent, corresponding robot communication system, is responsible for signal transmission to dangerous discernment main body.ImA ₂for dangerous discernment main body, be represent 16 of known danger pattern long 2 system character strings, store in a database.ImA ₃for immune response main body (operating personnel), the danger that dangerous discernment main body identifies is taken measures.

Immunity dangerous discernment main body adopts dangerous discernment algorithm to carry out match cognization to signal.

With underwater robot example is perceived as to dangerous information under water below, underwater robot immunity multiagent system dangerous criminal embodiment is described:

Interaction between the monitoring of danger signal and main body is as follows by immune main body algorithm realization:

(1) system initialization: environmentally knowledge sets up policy database, prestore hazardous environment set of patterns

Y＝{y ₁,y ₂,...,y _j,...,y _n}

Y _j, j=1,2 ..., n, be a jth environmental pattern, n is number of modes.

(2) danger that monitoring main body receives from underwater environment stimulates, and storehouse of refreshing one's knowledge, the antibody set of patterns after renewal is

Y'＝{y' ₁,y' ₂,...,y' _j,...,y' _n}

Y' _jfor the pattern after renewal.

(3) in the t=k moment, the perceptual signal x that monitoring main body is collected _kt (), is expressed as m mode signal through filtering with after merging, note set

X(t)＝{x ₁(t),x ₂(t),...,x _i(t),...,x _m(t)|t＝k}

(4) if x _ik () ∈ X (k), through dangerous discernment main body judges x _ik () has greater probability to belong to abnormal condition situation, then turn (7); If x _ik () ∈ X (k), shows there is corresponding solution (antibody) in knowledge base, then turns (5).

(5) immune communication agent is given an order to dangerous discernment main body.

(6) k=k+1, enters subsequent time, turns 2).

(7) learn and remember this situation, storehouse of refreshing one's knowledge, Immune discrimination main body process analogue after empirically instructing, in order to secondary immune response, and turning 5);

After monitoring danger signal, danger signal to be identified by dangerous discernment algorithm as follows:

(1) initialization: represent underwater security situation set of modes S with 16 strings of binary characters, the underwater environment signal character string of being monitored by monitoring main body is stored in R ₀in.

(2) affinity evaluation: R ₀in character string whether mate character string in S, determine whether mate by the affinity between them.R is adopted to adjoin matching process.R adjoins the binary digit symbol string that coupling adopts regular length, and the antibody in this matching process and the matching process between antigen (new model) are defined as follows: given antigen x=x ₁x ₂... x _nwith antibody d=d ₁d ₂... d _n, M represents coupling.

such x _j=d _j, j=i ..., i, i+r-1

(3) detectors set is produced: if from R ₀a character string and S in the affinity of at least one character string be greater than given threshold value, then thinks and must be eliminated this character string identification limit risk (Solid phase), otherwise character string introduces S set.

(4) do not reach setting procedure number, algorithm returns.

After dangerous main body hazard recognition signal, immune response main body is concrete operations personnel, is taken measures to eliminate danger according to the class of risk by operating personnel.

Claims (2)

1., based on a underwater robot environment perception method for immune main body, it is characterized in that, comprise the steps:

(1) immune main body and underwater robot structure are mapped, in model, form immune host complex ImA={ImA by n immune main body ImA ₀, ImA ₁..., ImA _n-1, n=4, ImA ₀for immunologic surveillance main body, ImA ₁for immune communication agent, ImA ₂for dangerous discernment main body, ImA ₃for immune response main body;

(2) main body ImA is monitored ₀obtaining environmental knowledge in original state, the abnormal conditions in institute's monitoring range being detected, when monitoring underwater environment changed condition, by immune communication agent ImA ₁send a signal to dangerous discernment main body ImA ₂, described signal is expressed as X (t);

(3) prestore hazardous environment set of patterns Y={y ₁, y ₂..., y _j..., y _n, y _jfor a jth environmental pattern, j=1,2 ..., n, n are number of modes, and signal is expressed as X (t)={ x ₁(t), x ₂(t) ..., x _i(t) ..., x _m(t) | t=k}, x _it () is perceptual signal, m is for being expressed as perceptual signal number, immune communication agent ImA through filtering with after merging ₁send perceptual signal to dangerous discernment main body ImA ₂if, x _ik () ∈ X (k), Y' are the antibody set of patterns after upgrading, through dangerous discernment main body ImA ₂judge x _ik () belongs to danger signal, then learn and remember this situation, upgrading dangerous knowledge base, and Immune discrimination main body process analogue after empirically instructing, in order to second set response; If x _ik () ∈ X (k), shows there is corresponding solution in knowledge base;

(4) dangerous discernment main body ImA ₂after identifying danger signal, immune response main body ImA ₃perform the process to danger signal, eliminate danger situation.

2. a kind of underwater robot environment perception method based on immune main body according to claim 1, it is characterized in that: after monitoring described danger signal, described danger signal identifies as follows:

(1) initialization: represent underwater security situation set of modes S with 16 strings of binary characters, the underwater environment signal character string of being monitored by monitoring main body is stored in R ₀in;

(2) affinity evaluation: R ₀in character string whether mate character string in S, determine whether mate by the affinity between them, adopt r to adjoin matching process; R adjoins the binary digit symbol string that coupling adopts regular length, and the antibody in this matching process and the matching process between antigen are defined as follows: given antigen x=x ₁x ₂... x _nwith antibody d=d ₁d ₂... d _n, M represents coupling;

DMx ≡ , such x _j=d _j, j=i ..., i, i+r-1, wherein dMx represents antibody coupling antigen

(3) detectors set is produced: if from R ₀a character string and S in the affinity of at least one character string be greater than given threshold value, then thinks and must be eliminated this character string identification limit risk, otherwise character string introduces S set;

(4) if reach setting procedure number, then terminate, return step (2) if do not reach.