CN106803361A - A kind of navigation method of control based on Rolling Planning strategy - Google Patents

Abstract

The present invention relates to a kind of navigation method of control based on Rolling Planning strategy, including following several steps, the real-time and historical position information of ship is obtained by sea radar first；Then in each sampling instant, the track for speculating ship in future time period is rolled according to the real-time and historical position information of ship；The current running status of ship and historical position observation sequence are based on again, obtain the numerical value of marine site wind field variable；The safety regulation collection that the ship of the running status and setting that are based on each ship again need to be met when being run in marine site, dynamic behaviour implementing monitoring to ship and for maritime traffic control centre provides timely warning information；When warning information occurs, on the premise of ship physical property and marine site traffic rules is met, by setting optimizing index function and incorporating wind field variable value, Rolling Planning is carried out to ship collision avoidance track using Adaptive Control Theory method, and program results is transferred to each ship to perform.

Description

A kind of navigation method of control based on Rolling Planning strategy

The application is Application No.：201410849264.4, invention and created name is《A kind of navigation traffic control method》, The applying date is：The divisional application of the application for a patent for invention on December 30 in 2014.

Technical field

The present invention relates to a kind of marine site traffic control method, more particularly to a kind of marine site traffic based on Rolling Planning strategy Method of control.

Background technology

With the fast development of global shipping business, the traffic in the busy marine site in part is further crowded.It is close in vessel traffic flow Collection complexity marine site, the regulation model allocated at artificial interval has still been combined for the collision scenario between ship not using sail plan Adapt to the fast development of shipping business.To ensure the personal distance between ship, implement effective conflict allotment and just handed over as marine site The emphasis of siphunculus system work.Ship conflict Resolution is a key technology in navigational field, safely and efficiently frees scheme pair It is significant in increasing marine site ship flow and ensuring that sea-freight is safe.

In order to improve the efficiency of navigation of ship, marine radar automatic plotter has been widely applied to ship monitor at present In collision prevention, the equipment is by extracting ship relevant information for the judgement of collision scenario between ship provides reference frame.Although this Kind equipment greatly reduces the load of manual monitoring, but it does not have the automatic conflict Resolution function of ship.For ship conflict Problem is freed, current processing mode mainly includes geometric deterministic algorithm and the major class scheme of Heuristic Intelligent Algorithm two, phase Close literature research and be concentrated mainly on conflict avoiding planning algorithm under unconfined condition between two ships and many with " off-line form " Be to free track in the presence of the ship planning of conflict, thereby result in each ship free the dynamic adaptable and robustness of track compared with Difference.Additionally, in ship real navigation, influenceed by various factors such as meteorological condition, navigation equipment and driver operations, it Running status often not exclusively belong to a certain specific motion state, needed during ship trajectory predictions consider it is various with The influence of machine factor, implements rolling forecast to its Future Trajectory and strengthens its rail by the newest characteristic for obtaining all kinds of enchancement factors The robustness of mark prediction.

The content of the invention

The technical problem to be solved in the present invention is to provide a kind of preferably navigation based on Rolling Planning strategy of robustness and hands over Logical method of control, the ship trajectory predictions precision of the method is higher and can effectively prevent vessel motion conflict.

Realize that the technical scheme of the object of the invention is to provide a kind of navigation method of control based on Rolling Planning strategy, including Following several steps：

1. the real-time and historical position information of ship is obtained by sea radar, the positional information of each ship is discrete two-dimensional Position sequence x'=[x₁',x₂',...,x_n'] and y'=[y₁',y₂',...,y_n'], by application wavelet transformation theory to original Discrete two-dimensional position sequence x'=[x₁',x₂',...,x_n'] and y'=[y₁',y₂',...,y_n'] preliminary treatment is carried out, so as to obtain Take the denoising discrete two-dimensional position sequence x=[x of ship₁,x₂,...,x_n] and y=[y₁,y₂,...,y_n]；

2. in each sampling instant, the real-time and historical position information of the ship 1. obtained according to step is rolled and speculates future The track of ship in period, its detailed process is as follows：

2.1) ship track data pretreatment, the original discrete two-dimensional position sequence x=[x of ship acquired in₁, x₂,...,x_n] and y=[y₁,y₂,...,y_n], treatment is carried out to it using first-order difference method and obtains new ship discrete location Sequence △ x=[△ x₁,△x₂,...,△x_n-1] and △ y=[△ y₁,△y₂,...,△y_n-1], wherein △ x_i=x_i+1-x_i,△ y_i=y_i+1-y_i(i=1,2 ..., n-1)；

2.2) ship track data is clustered, to ship discrete two-dimensional position sequence △ x and △ y new after treatment, is passed through Setting cluster number M', is clustered to it respectively using K-means clustering algorithms；

2.3) parameter training is carried out using HMM to ship track data in each sampling instant, by inciting somebody to action Vessel motion track data △ x and △ y after treatment is considered as the aobvious observation of hidden Markov models, by setting hidden status number Mesh N and parameter update period τ ', roll and obtain newest hidden Ma Erke according to T' nearest position detection value and use B-W algorithms Husband's model parameter λ '；

2.4) according to HMM parameter, obtained corresponding to current time observation using Viterbi algorithm Hidden state q；

2.5) in each sampling instant, by setting prediction time domain W, the hidden state q based on ship current time, obtain not Carry out the position prediction value O of period ship；

3. in each sampling instant, based on the current running status of ship and historical position observation sequence, marine site wind is obtained The numerical value of field variable；

4. in each sampling instant, the ship of running status and setting based on each ship needs to meet when being run in marine site Safety regulation collection, when being possible to occur violating the situation of safety regulation between ship, to its dynamic behaviour implementing monitoring and be Maritime traffic control centre provides timely warning information；

5. when warning information occurs, on the premise of ship physical property and marine site traffic rules is met, by setting Optimizing index function and wind field variable value is incorporated, ship collision avoidance track is rolled using Adaptive Control Theory method Planning, and program results is transferred to each ship execution, its detailed process is as follows：

5.1) termination reference point locations P, collision avoidance policy control time domain Θ, the trajectory predictions of ship collision avoidance trajectory planning are set Time domain W；

5.2) on the premise of being set in given optimizing index function, based on cooperative collision avoidance trajectory planning thought, by giving Each ship assigns different weight and incorporates real-time wind field variable filtering numerical value, obtains the collision avoidance track of each ship and keeps away Hit control strategy and program results is transferred to each ship and perform, and each ship only implements its first in Rolling Planning interval Optimal Control Strategy；

5.3) in next sampling instant, repeat step 5.2) until each ship reaches it and frees terminal.

Further, the step 1. in, by application wavelet transformation theory to original discrete two-dimensional position sequence x'= [x₁',x₂',...,x_n'] and y'=[y₁',y₂',...,y_n'] preliminary treatment is carried out, so as to obtain the denoising discrete two-dimensional of ship Position sequence x=[x₁,x₂,...,x_n] and y=[y₁,y₂,...,y_n]：For the original two dimensional sequence data x'=for giving [x₁',x₂',...,x_n'], it is carried out approximately respectively using the linear representation of following form：

Wherein：

F'(x' the function expression to being obtained after data smoothing processing) is represented, ψ (x') represents female ripple, and δ, J and K are small Wave conversion constant, ψ_J,K(x') transition form of female ripple, c are represented_J,KThe function coefficients that expression is obtained by wavelet transform procedure, its body Wavelet ψ is showed_J,K(x') to the weight size of whole approximation to function, if this coefficient very little, then it means wavelet ψ_J,K(x') Weight it is also smaller, thus can be on the premise of not influence function key property, from during approximation to function by wavelet ψ_J,K (x') remove；In real data processing procedure, " threshold transition " is implemented by given threshold χ, work as c_J,K<During χ, c is set_J,K =0；The selection of threshold function table uses the following two kinds mode：

With

For y'=[y₁',y₂',...,y_n'], being also adopted by the above method carries out denoising；

The detailed process that 3. step obtains the numerical value of marine site wind field variable is as follows：

3.1) stop position of ship is set as track reference coordinate origin and is set up axis of abscissas in the horizontal plane and is indulged Reference axis；

3.2) when ship is in straight running condition and at the uniform velocity turning running status, marine site wind field linear filtering mould is built Type x₁(t+ △ t)=F (t) x₁(t)+w (t) and z (t)=H (t) x₁T ()+v (t) obtains wind field variable value, wherein △ t are represented Sampling interval, x₁T () represents the state vector of t, z (t) represents the observation vector of t, and x₁(t)=[x (t), y (t), v_x(t),v_y(t),w_x(t),w_y(t)]^T, wherein x (t) and y (t) represent t vessel position in axis of abscissas and ordinate respectively Component on axle, v_x(t) and v_yT () represents component of the t speed of the ship in metres per second on axis of abscissas and axis of ordinates, w respectively_x(t) And w_yT () represents component of the t wind field numerical value on axis of abscissas and axis of ordinates respectively, F (t) and H (t) represent shape respectively State transfer matrix and output calculation matrix, w (t) and v (t) represent system noise vector sum measurement noise vector respectively：

When ship is in speed change turning running status, marine site wind field nonlinear filtering wave pattern x is built₁(t+ △ t)=Ψ (t,x₁(t), u (t))+w (t), z (t)=Ω (t, x₁(t))+v (t) and u (t)=[ω_a(t),γ_a(t)]^T, wherein Ψ () and Ω () represents state-transition matrix and output calculation matrix, ω respectively_a(t) and γ_aT () represents turning rate and rate of acceleration respectively：

Wherein：△ t represent sampling time interval,

3.3) Filtering Model according to constructed by obtains the numerical value of wind field variable.

Further, the step 2. in, step 2.3) in determine flight path HMM parameter lambda '=(π, A, B) Process it is as follows：

2.3.1) variable assigns initial value：Variable π is given using being uniformly distributed_i, a_ijAnd b_j(o_k) assign initial value WithAnd It is set to meet constraints：WithThus To λ₀=(π₀,A₀,B₀), wherein o_kRepresent a certain aobvious observation, π₀、A₀And B₀It is respectively by elementWithThe square of composition Battle array, makes parameter l=0, o=(o_t-T'+1,...,o_t-1,o_t) it is T' historical position observation before current time t；

2.3.2 E-M algorithms) are performed：

2.3.2.1) E- steps：By λ_lCalculate ξ_e(i, j) and γ_e(s_i)；

VariableSo

Wherein s represents a certain hidden state；

2.3.2.2) M- steps：WithRespectively Estimate π_i, a_ijAnd b_j(o_k) and thus obtain λ_l+1；

2.3.2.3) circulate：L=l+1, repeats E- steps and M- steps, until π_i、a_ijAnd b_j(o_k) convergence, i.e.,

|P(o|λ_l+1)-P(o|λ_l)|<ε, wherein parameter ε=0.00001, return to step 2.3.2.4)；

2.3.2.4)：Make λ '=λ_l+1, algorithm terminates.

Further, the step 2. in, step 2.4) determine the iterative process of the optimal hidden status switch of ship track It is as follows：

2.4.1) variable assigns initial value：Make g=2, β_T'(s_i(the s of)=1_i∈ S), δ₁(s_i)=π_ib_i(o₁), ψ₁(s_i)=0, its In,

, wherein variable ψ_g(s_j) represent make variable δ_g-1(s_i)a_ijTake the hidden state s of ship track of maximum_i, parameter S represents The set of hidden state；

2.4.2) recursive process：

2.4.3) moment renewal：G=g+1 is made, if g≤T', return to step 2.4.2), otherwise iteration ends and go to step 2.4.4)；

2.4.4)Go to step 2.4.5)；

2.4.5) optimal hidden status switch is obtained：

2.4.5.1) variable assigns initial value：Make g=T'-1；

2.4.5.2) backward recursion：

2.4.5.3) moment renewal：G=g-1 is made, if g >=1, return to step 2.4.5.2), otherwise terminate.

Further, the step 2. in, the value of cluster number M' is 4, and the value of hidden state number N is 3, when parameter updates Section τ ' is 30 seconds, and T' is 10, and prediction time domain W is 300 seconds.

Further, the step 4. in the dynamic behaviour implementing monitoring of each ship and for maritime traffic control centre carries Detailed process for timely warning information is as follows：

4.1) the safety regulation collection D that need to be met when construction ship runs in marine site_mr(t)≥D_min, wherein D_mrT () represents Any two ship m and ship r t distance, D_minRepresent the minimum safe distance between ship；

4.2) according to the sampling time, set up by the observer Λ of the continuous running status of ship to discrete sampling state:Γ→ Ξ, wherein Γ represent the continuous running status of ship, and Ξ represents the discrete sampling state of ship；

4.3) as the observer Λ of ship m and r_mAnd Λ_rDiscrete observation numerical value Ξ_mAnd Ξ_rShow the vector not in t When safety regulation is concentrated, i.e. relational expression D_mr(t)≥D_minWhen invalid, alarm letter is sent to maritime traffic control centre at once Breath.

Further, step 5. in, step 5.2) detailed process be：Order

WhereinRepresent distance between t ship R present positions and next navigation channel point square, P_R(t)= (x_Rt,y_Rt),The priority index of so t ship R may be set to：

Wherein Z_tRepresent the ship number for existing in t marine site and conflicting, from the implication of priority index, ship away from From its next navigation channel point more close to, its priority is higher；

Setting optimizing index

, wherein R ∈ I (t) expressions ship code and I (t)={ 1,2 ..., Z_t, P_R(t+h △ t) represents ship at the moment The position vector of (t+h △ t),Represent that ship R's frees terminating point, u_RThe optimal control sequence of ship R to be optimized is represented, Q_RtIt is positive definite diagonal matrix, its diagonal element is priority index Ls of the ship R in t_Rt, and

Further, 5. middle termination reference point locations P is set as next navigation channel point of vessel motion, collision avoidance to the step Policy control time domain Θ is 300 seconds；Trajectory predictions time domain W is 300 seconds.

The present invention has positive effect：(1) present invention during the real-time estimate of ship track, incorporated it is random because The influence of element, the rolling track prediction scheme for being used can in time extract the changing condition of extraneous enchancement factor, improve ship The accuracy of oceangoing ship trajectory predictions.

(2) present invention has incorporated the influence of wind field in marine site during ship conflict Resolution, and the rolling for being used is freed Trajectory planning scheme can track be freed in adjustment in time according to the change of wind field in marine site, improves the robust of ship conflict Resolution Property.

(3) present invention is based on different performance index, can free trajectory planning side for the multiple ships in the presence of conflict are provided Case, improves the economy of vessel motion and the utilization rate of sea area resources.

Brief description of the drawings

Fig. 1 is the short-term Track Pick-up schematic flow sheet of vessel motion in the present invention；

Fig. 2 is the Wind filter method flow schematic diagram in the present invention；

Fig. 3 is the vessel motion situation monitoring schematic flow sheet in the present invention；

Fig. 4 is the ship collision avoidance track optimizing method schematic flow sheet in the present invention.

Specific embodiment

(embodiment 1)

A kind of navigation method of control based on Rolling Planning strategy of the present embodiment includes following several steps：

1. the real-time and historical position information of ship is obtained by sea radar, the positional information of each ship is discrete two-dimensional Position sequence x'=[x₁',x₂',...,x_n'] and y'=[y₁',y₂',...,y_n'], by application wavelet transformation theory to original Discrete two-dimensional position sequence x'=[x₁',x₂',...,x_n'] and y'=[y₁',y₂',...,y_n'] preliminary treatment is carried out, so as to obtain Take the denoising discrete two-dimensional position sequence x=[x of ship₁,x₂,...,x_n] and y=[y₁,y₂,...,y_n]：It is original for what is given Two-dimensional sequence data x'=[x₁',x₂',...,x_n'], it is carried out approximately respectively using the linear representation of following form：

Wherein：

F'(x' the function expression to being obtained after data smoothing processing) is represented, ψ (x') represents female ripple, and δ, J and K are small Wave conversion constant, ψ_J,K(x') transition form of female ripple, c are represented_J,KThe function coefficients that expression is obtained by wavelet transform procedure, its body Wavelet ψ is showed_J,K(x') to the weight size of whole approximation to function, if this coefficient very little, then it means wavelet ψ_J,K(x') Weight it is also smaller, thus can be on the premise of not influence function key property, from during approximation to function by wavelet ψ_J,K (x') remove；In real data processing procedure, " threshold transition " is implemented by given threshold χ, work as c_J,K<During χ, c is set_J,K =0；The selection of threshold function table uses the following two kinds mode：

With

For y'=[y₁',y₂',...,y_n'], being also adopted by the above method carries out denoising.

2. in each sampling instant, the real-time and historical position information of the ship 1. obtained according to step is rolled and speculates future The track of ship, sees Fig. 1 in period, and its detailed process is as follows：

2.1) ship track data pretreatment, the original discrete two-dimensional position sequence x=[x of ship acquired in₁, x₂,...,x_n] and y=[y₁,y₂,...,y_n], treatment is carried out to it using first-order difference method and obtains new ship discrete location Sequence △ x=[△ x₁,△x₂,...,△x_n-1] and △ y=[△ y₁,△y₂,...,△y_n-1], wherein △ x_i=x_i+1-x_i,△ y_i=y_i+1-y_i(i=1,2 ..., n-1)；

2.2) ship track data is clustered, to ship discrete two-dimensional position sequence △ x and △ y new after treatment, is passed through Setting cluster number M', is clustered to it respectively using K-means clustering algorithms；

2.3) parameter training is carried out using HMM to ship track data in each sampling instant, by inciting somebody to action Vessel motion track data △ x and △ y after treatment is considered as the aobvious observation of hidden Markov models, by setting hidden status number Mesh N and parameter update period τ ', roll and obtain newest hidden Ma Erke according to T' nearest position detection value and use B-W algorithms Husband's model parameter λ '；Determine flight path HMM parameter lambda '=the process of (π, A, B) is as follows：

2.3.1) variable assigns initial value：Variable π is given using being uniformly distributed_i, a_ijAnd b_j(o_k) assign initial value WithAnd It is set to meet constraints：WithThus To λ₀=(π₀,A₀,B₀), wherein o_kRepresent a certain aobvious observation, π₀、A₀And B₀It is respectively by elementWithThe square of composition Battle array, makes parameter l=0, o=(o_t-T'₊₁,...,o_t-1,o_t) it is T' historical position observation before current time t；

2.3.2 E-M algorithms) are performed：

2.3.2.1) E- steps：By λ_lCalculate ξ_e(i, j) and γ_e(s_i)；

VariableSo

Wherein s represents a certain hidden state；

2.3.2.2) M- steps：WithRespectively Estimate π_i, a_ijAnd b_j(o_k) and thus obtain λ_l+1；

2.3.2.3) circulate：L=l+1, repeats E- steps and M- steps, until π_i、a_ijAnd b_j(o_k) convergence, i.e.,

|P(o|λ_l+1)-P(o|λ_l)|<ε, wherein parameter ε=0.00001, return to step 2.3.2.4)；

2.3.2.4)：Make λ '=λ_l+1, algorithm terminates.

2.4) according to HMM parameter, obtained corresponding to current time observation using Viterbi algorithm Hidden state q；Determine that the iterative process of the optimal hidden status switch of ship track is as follows：

2.4.1) variable assigns initial value：Make g=2, β_T'(s_i(the s of)=1_i∈ S), δ₁(s_i)=π_ib_i(o₁), ψ₁(s_i)=0, its In,

, wherein variable ψ_g(s_j) represent make variable δ_g-1(s_i)a_ijTake the hidden state s of ship track of maximum_i, parameter S represents The set of hidden state；

2.4.2) recursive process：

2.4.3) moment renewal：G=g+1 is made, if g≤T', return to step 2.4.2), otherwise iteration ends and go to step 2.4.4)；

2.4.4)Go to step 2.4.5)；

2.4.5) optimal hidden status switch is obtained：

2.4.5.1) variable assigns initial value：Make g=T'-1；

2.4.5.2) backward recursion：

2.4.5.3) moment renewal：G=g-1 is made, if g >=1, return to step 2.4.5.2), otherwise terminate.

2.5) in each sampling instant, by setting prediction time domain W, the hidden state q based on ship current time, obtain not Carry out the position prediction value O of period ship.

The value of above-mentioned cluster number M' is 4, and the value of hidden state number N is 3, and parameter updated period τ ' for 30 seconds, and T' is 10, Prediction time domain W is 300 seconds.

3. in each sampling instant, based on the current running status of ship and historical position observation sequence, marine site wind is obtained The numerical value of field variable, is shown in Fig. 2, and its detailed process is as follows：

3.1) stop position of ship is set as track reference coordinate origin and is set up axis of abscissas in the horizontal plane and is indulged Reference axis；

3.2) when ship is in straight running condition and at the uniform velocity turning running status, marine site wind field linear filtering mould is built Type x₁(t+ △ t)=F (t) x₁(t)+w (t) and z (t)=H (t) x₁T ()+v (t) obtains wind field variable value, wherein △ t are represented Sampling interval, x₁T () represents the state vector of t, z (t) represents the observation vector of t, and x₁(t)=[x (t), y (t), v_x(t),v_y(t),w_x(t),w_y(t)]^T, wherein x (t) and y (t) represent t vessel position in axis of abscissas and ordinate respectively Component on axle, v_x(t) and v_yT () represents component of the t speed of the ship in metres per second on axis of abscissas and axis of ordinates, w respectively_x(t) And w_yT () represents component of the t wind field numerical value on axis of abscissas and axis of ordinates respectively, F (t) and H (t) represent shape respectively State transfer matrix and output calculation matrix, w (t) and v (t) represent system noise vector sum measurement noise vector respectively：

When ship is in speed change turning running status, marine site wind field nonlinear filtering wave pattern x is built₁(t+ △ t)=Ψ (t,x₁(t), u (t))+w (t), z (t)=Ω (t, x₁(t))+v (t) and u (t)=[ω_a(t),γ_a(t)]^T, wherein Ψ () and Ω () represents state-transition matrix and output calculation matrix, ω respectively_a(t) and γ_aT () represents turning rate and rate of acceleration respectively：

Wherein：△ t represent sampling time interval,

3.3) Filtering Model according to constructed by obtains the numerical value of wind field variable.

4. in each sampling instant, the ship of running status and setting based on each ship needs to meet when being run in marine site Safety regulation collection, when being possible to occur violating the situation of safety regulation between ship, to its dynamic behaviour implementing monitoring and be Maritime traffic control centre provides timely warning information, sees Fig. 3, and its detailed process is as follows：

4.1) the safety regulation collection D that need to be met when construction ship runs in marine site_mr(t)≥D_min, wherein D_mrT () represents Any two ship m and ship r t distance, D_minRepresent the minimum safe distance between ship；

4.2) according to the sampling time, set up by the observer Λ of the continuous running status of ship to discrete sampling state:Γ→ Ξ, wherein Γ represent the continuous running status of ship, and Ξ represents the discrete sampling state of ship；

4.3) as the observer Λ of ship m and r_mAnd Λ_rDiscrete observation numerical value Ξ_mAnd Ξ_rShow the vector not in t When safety regulation is concentrated, i.e. relational expression D_mr(t)≥D_minWhen invalid, alarm letter is sent to maritime traffic control centre at once Breath.

5. when warning information occurs, on the premise of ship physical property and marine site traffic rules is met, by setting Optimizing index function and wind field variable value is incorporated, ship collision avoidance track is rolled using Adaptive Control Theory method Planning, and program results is transferred to each ship execution, see Fig. 4, its detailed process is as follows：

5.1) termination reference point locations P, collision avoidance policy control time domain Θ, the trajectory predictions of ship collision avoidance trajectory planning are set Time domain W；

5.2) on the premise of being set in given optimizing index function, based on cooperative collision avoidance trajectory planning thought, by giving Each ship assigns different weight and incorporates real-time wind field variable filtering numerical value, obtains the collision avoidance track of each ship and keeps away Hit control strategy and program results is transferred to each ship and perform, and each ship only implements its first in Rolling Planning interval Optimal Control Strategy：Order

WhereinRepresent distance between t ship R present positions and next navigation channel point square, P_R(t)= (x_Rt,y_Rt),The priority index of so t ship R may be set to：

Wherein Z_tRepresent the ship number for existing in t marine site and conflicting, from the implication of priority index, ship away from From its next navigation channel point more close to, its priority is higher；

Setting optimizing index

, wherein R ∈ I (t) expressions ship code and I (t)={ 1,2 ..., Z_t, P_R(t+h △ t) represents ship at the moment The position vector of (t+h △ t),Represent that ship R's frees terminating point, u_RThe optimal control sequence of ship R to be optimized is represented, Q_RtIt is positive definite diagonal matrix, its diagonal element is priority index Ls of the ship R in t_Rt, and

5.3) in next sampling instant, repeat step 5.2 is until each ship reaches it and frees terminal.

Above-mentioned termination reference point locations P is set as next navigation channel point of vessel motion, and collision avoidance policy control time domain Θ is 300 seconds；Trajectory predictions time domain W is 300 seconds.

Obviously, above-described embodiment is only intended to clearly illustrate example of the present invention, and is not to of the invention The restriction of implementation method.For those of ordinary skill in the field, it can also be made on the basis of the above description The change or variation of its multi-form.There is no need and unable to be exhaustive to all of implementation method.And these belong to this hair Obvious change that bright spirit is extended out or among changing still in protection scope of the present invention.

Claims (1)

1. a kind of navigation method of control based on Rolling Planning strategy, it is characterised in that including following several steps：

1. the real-time and historical position information of ship is obtained by sea radar, the positional information of each ship is discrete two-dimensional position Sequence x'=[x₁',x₂',...,x_n'] and y'=[y₁',y₂',...,y_n'], by application wavelet transformation theory to original discrete Two-dimensional position sequence x'=[x₁',x₂',...,x_n'] and y'=[y₁',y₂',...,y_n'] preliminary treatment is carried out, so as to obtain ship The denoising discrete two-dimensional position sequence x=[x of oceangoing ship₁,x₂,...,x_n] and y=[y₁,y₂,...,y_n]；

2. in each sampling instant, the real-time and historical position information of the ship 1. obtained according to step is rolled and speculates future time period The track of interior ship, its detailed process is as follows：

2.1) ship track data pretreatment, the original discrete two-dimensional position sequence x=[x of ship acquired in₁,x₂,..., x_n] and y=[y₁,y₂,...,y_n], treatment is carried out to it using first-order difference method and obtains new ship discrete location sequence △ x =[△ x₁,△x₂,...,△x_n-1] and △ y=[△ y₁,△y₂,...,△y_n-1], wherein △ x_i=x_i+1-x_i,△y_i=y_i+1- y_i(i=1,2 ..., n-1)；

2.2) ship track data is clustered, to ship discrete two-dimensional position sequence △ x and △ y new after treatment, by setting Cluster number M', is clustered to it respectively using K-means clustering algorithms；

2.3) parameter training is carried out using HMM to ship track data in each sampling instant, by that will process Vessel motion track data △ x and △ y afterwards is considered as the aobvious observation of hidden Markov models, by setting hidden state number N Period τ ' is updated with parameter, is rolled and is obtained newest Hidden Markov according to T' nearest position detection value and use B-W algorithms Model parameter λ '；

2.4) according to HMM parameter, the hidden shape corresponding to current time observation is obtained using Viterbi algorithm State q；

2.5) in each sampling instant, by setting prediction time domain W, the hidden state q based on ship current time, when obtaining following The position prediction value O of section ship；

3. in each sampling instant, based on the current running status of ship and historical position observation sequence, obtain marine site wind field and become The numerical value of amount；

4. in each sampling instant, the peace that the ship of running status and setting based on each ship need to meet when being run in marine site Full rule set, when being possible to occur violating the situation of safety regulation between ship, to its dynamic behaviour implementing monitoring and for marine Traffic control center provides timely warning information；

5. when warning information occurs, on the premise of ship physical property and marine site traffic rules is met, optimized by setting Target function and wind field variable value is incorporated, rolling rule are carried out to ship collision avoidance track using Adaptive Control Theory method Draw, and program results is transferred to each ship and perform, its detailed process is as follows：

5.1) termination reference point locations P, collision avoidance policy control time domain Θ, the trajectory predictions time domain of ship collision avoidance trajectory planning are set W；

5.2) on the premise of being set in given optimizing index function, based on cooperative collision avoidance trajectory planning thought, by each Ship assigns different weight and incorporates real-time wind field variable filtering numerical value, obtains collision avoidance track and the collision avoidance control of each ship Program results is simultaneously transferred to each ship execution, and each ship only implements its first optimization in Rolling Planning interval by system strategy Control strategy；

5.3) in next sampling instant, repeat step 5.2) until each ship reaches it and frees terminal；

The detailed process that 3. step obtains the numerical value of marine site wind field variable is as follows：

3.1) stop position of ship is set as track reference coordinate origin and sets up axis of abscissas and ordinate in the horizontal plane Axle；

3.2) when ship is in straight running condition and at the uniform velocity turning running status, structure marine site wind field linear filtering model x₁ (t+ △ t)=F (t) x₁(t)+w (t) and z (t)=H (t) x₁T ()+v (t) obtains wind field variable value, wherein △ t represent sampling Interval, x₁T () represents the state vector of t, z (t) represents the observation vector of t, and x₁(t)=[x (t), y (t), v_x (t),v_y(t),w_x(t),w_y(t)]^T, wherein x (t) and y (t) represent t vessel position in axis of abscissas and ordinate respectively Component on axle, v_x(t) and v_yT () represents component of the t speed of the ship in metres per second on axis of abscissas and axis of ordinates, w respectively_x(t) And w_yT () represents component of the t wind field numerical value on axis of abscissas and axis of ordinates respectively, F (t) and H (t) represent shape respectively State transfer matrix and output calculation matrix, w (t) and v (t) represent system noise vector sum measurement noise vector respectively：

$F\left(t\right)=\left[\begin{array}{cccccc}1& 0& \frac{\sin \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}& \frac{1-\cos \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}& \mathrm{\&Delta;}t& 0\\ 0& 1& \frac{\cos \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)-1}{{\mathrm{\&omega;}}_a\left(t\right)}& \frac{\sin \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}& 0& \mathrm{\&Delta;}t\\ 0& 0& \cos \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)& \sin \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)& 0& 0\\ 0& 0& -\sin \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)& \cos \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)& 0& 0\\ 0& 0& 0& 0& 1& 0\\ 0& 0& 0& 0& 0& 1\end{array}\right]$

$H\left(k\right)=\left[\begin{array}{cccccc}1& 0& 0& 0& 0& 0\\ 0& 1& 0& 0& 0& 0\\ 0& 0& 1& 0& 0& 0\\ 0& 0& 0& 1& 0& 0\end{array}\right];$

When ship is in speed change turning running status, marine site wind field nonlinear filtering wave pattern x is built₁(t+ △ t)=Ψ (t, x₁ (t), u (t))+w (t), z (t)=Ω (t, x₁(t))+v (t) and u (t)=[ω_a(t),γ_a(t)]^T, wherein Ψ () and Ω () represents state-transition matrix and output calculation matrix, ω respectively_a(t) and γ_aT () represents turning rate and rate of acceleration respectively：

$\mathrm{\&Psi;}=\left[\begin{array}{c}x\left(t\right)+v_x\left(t\right)(\frac{\sin \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}+\frac{{\mathrm{\&gamma;}}_a\left(t\right)\mathrm{\&Delta;}t}{\sqrt{v_x^2\left(t\right)+v_y^2\left(t\right)}}{C}_5)+v_y\left(t\right)(\frac{1-\cos \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}+\frac{{\mathrm{\&gamma;}}_a\left(t\right)\mathrm{\&Delta;}t}{\sqrt{v_x^2\left(t\right)+v_y^2\left(t\right)}}{C}_6)+w_x\left(t\right)\\ y\left(t\right)-v_x\left(t\right)(\frac{1-\cos \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}+\frac{{\mathrm{\&gamma;}}_a\left(t\right)\mathrm{\&Delta;}t}{\sqrt{v_x^2\left(t\right)+v_y^2\left(t\right)}}{C}_6)+v_y\left(t\right)(\frac{\sin \left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}+\frac{{\mathrm{\&gamma;}}_a\left(t\right)\mathrm{\&Delta;}t}{\sqrt{v_x^2\left(t\right)+v_y^2\left(t\right)}}{C}_5)+w_y\left(t\right)\\ \left(\right(1+\frac{{\mathrm{\&gamma;}}_a\left(t\right)\mathrm{\&Delta;}t}{\sqrt{v_x^2\left(t\right)+v_y^2\left(t\right)}}\left)\right(v_x\left(t\right)\cos \left({\mathrm{\&omega;}}_a\left(t\right)\mathrm{\&Delta;}t\right)+v_y\left(t\right)\sin \left({\mathrm{\&omega;}}_a\left(t\right)\mathrm{\&Delta;}t\right)\left)\right)\\ \left(\right(1+\frac{{\mathrm{\&gamma;}}_a\left(t\right)\mathrm{\&Delta;}t}{\sqrt{v_x^2\left(t\right)+v_y^2\left(t\right)}}\left)\right(v_y\left(t\right)\cos \left({\mathrm{\&omega;}}_a\left(t\right)\mathrm{\&Delta;}t\right)-v_x\left(t\right)\sin \left({\mathrm{\&omega;}}_a\left(t\right)\mathrm{\&Delta;}t\right)\left)\right)\\ w_x\left(t\right)\\ w_y\left(t\right)\end{array}\right],$

Wherein：△ t represent sampling time interval,

$C_5=(\frac{sin\left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)}-\frac{1-cos\left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a^2\left(t\right)\mathrm{\&Delta;}t}),$

$C_6=(\frac{sin\left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a^2\left(t\right)\mathrm{\&Delta;}t}-\frac{cos\left({\mathrm{\&omega;}}_a\right(t\left)\mathrm{\&Delta;}t\right)}{{\mathrm{\&omega;}}_a\left(t\right)});$

3.3) Filtering Model according to constructed by obtains the numerical value of wind field variable.