CN103303298B - Automatic processing device for emergency braking signal of high-speed train based on optimal control - Google Patents

Abstract

The invention discloses an automatic processing device for an emergency braking signal of a high-speed train based on optimal control. The automatic processing device comprises a train speed sensor, a dangerous case distance/processing time input unit, a high-speed train central control MCU (micro controller unit), a brake unit and emergency braking alarm and state display equipment. After the train speed sensor is started for measuring current train speed in real time, a train driver inputs a dangerous case distance and the dangerous case processing time into the dangerous case distance/processing time input unit; and the high-speed train central control MCU is used for performing an internal optimal control method, calculating a braking strategy capable of enabling the train to safely pass through a dangerous case position and enabling the train delay time to be shortest at the same time, converting the braking strategy obtained by the calculation into a braking instruction, sending the braking instruction to the brake unit, and sending an emergency braking alarm signal at the same time. According to the automatic processing device, the high-speed train can be guaranteed to safely pass through the dangerous case position and the train delay time can be also enabled to be shortest at the same time.

Description

A kind of high speed train emergency brake signal automatic processing device based on optimal control

Technical field

The present invention relates to track traffic security fields, mainly a kind of high speed train emergency brake signal automatic processing device based on method for optimally controlling.Train of sening as an envoy to can be calculated when there is emergency in train front to waste time the shortest braking strategy, and it can be used as speed-slackening signal to be implemented.

Background technology

In the process of moving, due to various enchancement factor, paroxysmal emergency may be there is in front side in high speed train.If process not in time, serious accident will be led to.

In Japan of technology maturation, Germany and French, high speed train has an accident unrare.A typical case is: on April 25th, 2005, Japan's one row high speed train is through Ni Qi city, Bingku county, cause derailed because driver has little time deceleration on bend for recovering the overdue moment, after train and a train colliding, pour a housing block, cause the first compartment and the second compartment entirely to ruin, cause 107 people dead, 555 people are injured.This plays tragic incident and causes Japanese government and Congress to have modified " railway cause method ", specifies that each railroad must bear and installs obligatioies such as " ATS Automatic Train Stopper (ATS) " along the railway.

China " 7.23 " Wenzhou rear end collision of motor train accident causes the great attention of people to train safe especially.The speed car of domestic independent research needs exploitation promptly to avoid braking technology and Related product equally.

Summary of the invention

The object of the invention is to for the deficiencies in the prior art, a kind of high speed train emergency brake signal automatic processing device based on method for optimally controlling is provided, this device can calculate the braking strategy meeting above-mentioned requirements, and it can be used as speed-slackening signal to be implemented.

The math modeling of high-speed train braking process can be described as

${\stackrel{\·}{x}}_1\left(t\right)=x_2\left(t\right)$

${\stackrel{\·}{x}}_2\left(t\right)=F\left(t\right)$

x ₁(t ₀)=0

x ₂(t ₀)=x ₂₀

x ₁(t _f)≤s _b

Wherein t represents the time, x ₁t () represents the distance of train driving, x ₁the first derivative of (t), x ₂t () represents the moving velocity of train, x ₂the first derivative of (t), t ₀represent that train starts the time point braked, x ₂(t ₀) be t ₀the speed in moment, s _bt ₀the distance of moment train distance dangerous situation spot, t _frepresent the time point of train by dangerous situation spot, at t _fmoment requires that the distance of train driving is no more than s _b.As can be seen from this description, the math modeling of the urgent train braking process of train is one group of differential algebraic equations.

Make the shortest time that train delays, be in fact equivalent to the braking force that braking procedure applies train minimum.Represent time dependent braking force with F (t), then the final expression formula of this problem is:

$\min J\left[F\left(t\right)\right]={\∫}_{t_0}^{t_f}{F}^2\left(t\right)\mathrm{dt}$

$s.t.{\stackrel{\·}{x}}_1\left(t\right)=x_2\left(t\right)$

${\stackrel{\·}{x}}_2\left(t\right)=F\left(t\right)$

x ₁(t ₀)=0

x ₂(t ₀)=x ₂₀

x ₁(t _f)≤s _b

This question essence is optimal control problem.Wherein J [F (t)] is the objective function of problem, is determined by braking force F (t).

The technical solution adopted for the present invention to solve the technical problems is: in high speed high speed train, control in MCU the method for optimally controlling being integrated with current main-stream---control variable parametric method (Control variable parameterization, be called for short CVP), automatically export braking instruction to brake unit when needs emergency braking by described MCU, realize emergency deceleration or parking.Described MCU can be considered as emergency brake signal generator, and its holonomic system as shown in Figure 2, comprises in car speed sensor, dangerous situation distance/processing time input block, high speed train and controls MCU, brake unit, emergency braking warning and status display unit.Described intrasystem each component part connects by data bus in car is unified.

The operational process of described system is as follows:

Steps A 1: high speed train opens car speed sensor in the process of moving, for measuring the moving velocity of this train current in real time;

Steps A 2: at certain moment t ₀, train operator is apprised of front distance s _bhave dangerous situation to occur outward, the time that processing this dangerous situation needs is t _f-t ₀.Train operator is by dangerous situation distance s _band dangerous situation processing time t _f-t ₀input dangerous situation distance/processing time input block;

Steps A 3: control MCU in high speed train and perform inner method for optimally controlling, calculates and train safe can be made by dangerous situation spot, the braking strategy of shortest time that makes again train delay simultaneously;

Steps A 4: control MCU in high speed train and be converted to braking instruction by calculating the braking strategy obtained, issue brake unit, send emergency braking alerting signal simultaneously.

Being integrated with in the high speed train of method for optimally controlling and controlling MCU is core of the present invention, as shown in Figure 3, its inside comprises information acquisition module, initialization module, ordinary differential system (Ordinary Differential Equation, be called for short ODE) more new module, nonlinear programming problem (Non-linear Programming is called for short NLP) solve module, control command output module to solve module, convergence judge module, dynamical parameter.Wherein information acquisition module comprises that dangerous situation distance gathers, the dangerous situation processing time gathers, current vehicle speed gathers three submodules, and NLP solves that module comprises search direction calculating, optimizing step size computation, NLP convergence judge three submodules.

The process that described middle control MCU produces emergency brake signal is as follows:

Step B1: information acquisition module obtains the setting value being input to middle control MCU from dangerous situation distance/processing time input block, and be input to the current vehicle speed value of middle control MCU from car speed sensor.Perform method for optimally controlling---the CVP method from step B2;

Step B2: initialization module brings into operation, arranges the initial guess F of the segments of braking procedure time, braking strategy ^(k)(t), setup algorithm precision tol, by iterations k zero setting;

Step B3: solve the target function value J [F that module obtains current iteration by ODE ^(k)(t)] and constraint functional value.Skip step B4 as k=0 and directly perform step B5;

Step B4: if J is [F ^(k)(t)] with the target function value J [F of last iteration ^(k-1)(t)] the difference of absolute value be less than precision tol, then judge that convergence meets, and the braking strategy of current iteration outputted to brake unit as instruction; If convergence does not meet, then continue to perform step B5;

Step B5: upgrade related dynamic parameters: use F ^(k)t the value of () covers F ^(k-1)the value of (t), and iterations k is increased by 1;

Step B6:NLP solves module and utilizes the target function value and constraint functional value that obtain in step B3, by calculating search direction and optimizing step-length, obtaining and comparing F ^(k-1)t new brake strategy F that () is more excellent ^(k)(t).Again step B3 is jumped to, till convergence judge module meets after this step is complete.

Described ODE solves module, and the method for employing is four step Adams methods, and computing formula is:

$x_1\left(t_{i+1}\right)=x_1\left(t_i\right)+\frac{h}{24}[{55x}_2\left(t_i\right)-{59x}_2\left(t_{i-1}\right)+{37x}_2\left(t_{i-2}\right)-{9x}_2\left(t_{i-3}\right)]$

$x_2\left(t_{i+1}\right)=x_2\left(t_i\right)+\frac{h}{24}[55F\left(t_i\right)-59F\left(t_{i-1}\right)+37F\left(t_{i-2}\right)-9F\left(t_{i-3}\right)]$

Wherein t represents the time, t _ito represent in the braking procedure of Adams method choice point sometime, t _i-1represent the t selected in Adams method _iprevious time point, t _i+1represent the t selected in Adams method _ia rear time point, by that analogy.Integration step h is the difference of any two adjacent time points.X ₁(t _i) represent that train is at t _ithe operating range in moment, x ₂(t _i) represent that train is at t _ithe moving velocity in moment, F (t _i) represent at t _ithe braking force in moment.

Described NLP solves module, adopts following steps to realize:

Step C1: by braking strategy F ^(k-1)t (), as certain point in vector space, is denoted as P ₁, P ₁corresponding target function value is exactly J [F ^(k-1)(t)];

Step C2: from a P ₁set out, according to a search direction d in the NLP algorithm construction vector space selected ^(k-1)with step-length α ^(k-1)

Step C3: through type F ^(k)(t)=F ^(k-1)(t)+α ^(k-1)d ^(k-1)corresponding u in structure vector space ^(k)another one point P ₂, make P ₂corresponding target function value J [F ^(k)(t)] than J [F ^(k-1) (t)] more excellent.

Beneficial effect of the present invention is mainly manifested in: 1, can ensure high speed train safety dangerous situation spot; The shortest time that 2, simultaneously train can be made again to delay.

Accompanying drawing explanation

Fig. 1 is functional schematic of the present invention;

Fig. 2 is structural representation of the present invention;

Fig. 3 controls MCU internal module constructional drawing in the present invention;

Fig. 4 is the emergency braking policy map of embodiment 1.

Detailed description of the invention

Embodiment 1

Suppose that high speed train in the process of moving, driver is apprised of and occurs obstacle suddenly on 1km place, front track, and clearing of obstruction needs 30 seconds.Driver is by these two information input dangerous situation distance/processing time input blocks, and now car speed sensor imports the current vehicle speed of middle control MCU into is 300km/h.Middle control MCU brings into operation inner method for optimally controlling---CVP method immediately, its operational process as shown in Figure 3, for:

Step D1: initialization module 32 brings into operation, the segments arranging the braking procedure time is 20, arranges the initial guess F of braking strategy ^(k)t () is-0.5, setting numerical stability tol is 0.01, by iterations k zero setting;

Step D2: solve the target function value J [F that module 33 obtains current iteration by ODE ^(k)(t)] and constraint functional value.Skip step D3 as k=0 and directly perform step D4;

Step D3: if J is [F ^(k)(t)] with the target function value J [F of last iteration ^(k-1)(t)] the difference of absolute value be less than accuracy requirement 0.01, then judge that convergence meets, and the braking strategy of current iteration outputted to brake unit as instruction; If convergence does not meet, then continue to perform step D4;

Step D4: upgrade related dynamic parameters: use F ^(k)t the value of () covers F ^(k-1)the value of (t), and iterations k is increased by 1;

Step D5:NLP solves module 36 and utilizes the target function value and constraint functional value that obtain in step d 2, by calculating search direction and optimizing step-length, obtaining and comparing F ^(k-1)t new brake strategy F that () is more excellent ^(k)(t).Again step D2 is jumped to, till convergence judge module 04 meets after this step is complete.

Described ODE solves module, and the method for employing is four step Adams methods, and computing formula is:

$x_1\left(t_{i+1}\right)=x_1\left(t_i\right)+\frac{h}{24}[{55x}_2\left(t_i\right)-{59x}_2\left(t_{i-1}\right)+{37x}_2\left(t_{i-2}\right)-{9x}_2\left(t_{i-3}\right)]$

$x_2\left(t_{i+1}\right)=x_2\left(t_i\right)+\frac{0.01}{24}[55F\left(t_i\right)-59F\left(t_{i-1}\right)+37F\left(t_{i-2}\right)-9F\left(t_{i-3}\right)]$

Wherein t represents the time, t _ito represent in the braking procedure of Adams method choice point sometime, t _i-1represent the t selected in Adams method _iprevious time point, t _i+1represent the t selected in Adams method _ia rear time point, by that analogy.Integration step 0.01 is the difference of any two adjacent time points, can meet realistic accuracy demand.X ₁(t _i) represent that train is at t _ithe operating range in moment, x ₂(t _i) represent that train is at t _ithe moving velocity in moment, F (t _i) represent at t _ithe braking force in moment.

Described NLP solves module, adopts following steps to realize:

Step e 1: by braking strategy F ^(k-1)t (), as certain point in vector space, is denoted as P ₁, P ₁corresponding target function value is exactly J [F ^(k-1)(t)];

Step e 2: from a P ₁set out, according to a search direction d in the SQP algorithm construction vector space selected ^(k-1)with step-length α ^(k-1)

Step e 3: through type F ^(k)(t)=F ^(k-1)(t)+α ^(k-1)d ^(k-1)corresponding u in structure vector space ^(k)another one point P ₂, make P ₂corresponding target function value is than J [F ^(k-1)(t)] more excellent.

The result of calculation of method for optimally controlling as shown in Figure 4.Coordinate is through normalized, and ordinate value is-1 expression maximum braking force, and value is 1 expression tractive force limit.The value of whole piece controlling curve F (t) is all no more than 0, shows that this is a control for brake curve.It is 20 that asterisk number on curve represents time slice number.Value on curve is just only 0 at the end of braking procedure, shows train when safety barrier without the need to braking again.

Finally, the braking control strategy of acquisition is outputted to brake unit as instruction by middle control MCU, completes brake operating mechanically, sends emergency braking alerting signal simultaneously.

Above content is in conjunction with concrete preferred implementation further description made for the present invention, can not assert that specific embodiment of the invention is only limited to these explanations.For general technical staff of the technical field of the invention, under the prerequisite not departing from inventive concept, some simple deduction or replace can also be made, all should be considered as belonging to protection scope of the present invention.

Claims (1)

1., based on a high speed train emergency brake signal automatic processing device for optimal control, train of sening as an envoy to can be calculated when there is emergency in train front and to waste time the shortest braking strategy, and it can be used as speed-slackening signal to be implemented; It is characterized in that: form by controlling MCU, brake unit, emergency braking warning and status display apparatus in car speed sensor, dangerous situation distance/processing time input block, high speed train, each component part connects by data bus in car; The operational process of described device comprises:

Steps A 1: open car speed sensor and be used for measuring current vehicle speed in real time;

Steps A 2: train operator is by dangerous situation distance and input dangerous situation processing time, input block dangerous situation distance/processing time;

Steps A 3: control MCU in high speed train and perform inner method for optimally controlling, calculates and train safe can be made by dangerous situation spot, the braking strategy of shortest time that makes again train delay simultaneously;

Steps A 4: control MCU in high speed train and be converted to braking instruction by calculating the braking strategy obtained, issue brake unit, send emergency braking alerting signal simultaneously;

Control MCU in described high speed train, comprise information acquisition module, initialization module, ordinary differential system solve module, convergence judge module, dynamical parameter more new module, nonlinear programming problem solve module, control command output module; Wherein information acquisition module comprises that dangerous situation distance gathers, the dangerous situation processing time gathers, current vehicle speed gathers three submodules, and nonlinear programming problem solves that module comprises search direction calculating, optimizing step size computation, nonlinear programming problem convergence judge three submodules;

Controlling the method for optimally controlling that MCU produces speed-slackening signal automatically in described high speed train is control variable parametric method, and operating procedure is as follows:

Step B1: information acquisition module (31) obtains and is input to high speed train from dangerous situation distance/processing time input block the setting value controlling MCU, and is input to high speed train from car speed sensor the current vehicle speed value controlling MCU; Perform method for optimally controlling---the control variable parametric method from step B2;

Step B2: initialization module (32) brings into operation, arranges the initial guess F of the segments of braking procedure time, braking strategy ^(k)(t), setup algorithm precision tol, by iterations k zero setting;

Step B3: solve the target function value J [F that module (33) obtains current iteration by ordinary differential system ^(k)(t)] and constraint functional value; Wherein J [F (t)] is the objective function of problem, and by representing the decision of time dependent braking force F (t), expression formula is:

$\begin{array}{cc}\min & J\left[F\left(t\right)\right]={\∫}_{t_0}^{t_f}{F}^2\left(t\right)\mathrm{dt}\end{array}$

$\begin{array}{cc}s.t.& {\stackrel{\·}{x}}_1\left(t\right)=x_2\left(t\right)\end{array}$

${\stackrel{\·}{x}}_2\left(t\right)=F\left(t\right)$

x ₁(t ₀)＝0

x ₂(t ₀)＝x ₂₀

x ₁(t _f)≤s _b；

Wherein t represents the time, t ₀represent that train starts the time point braked, t _frepresent the time point of train by dangerous situation spot, x ₁t () represents the distance of train driving, x ₂t () represents the moving velocity of train, s _bt ₀the distance of moment train distance dangerous situation spot;

Skip step B4 as k=0 and directly perform step B5;

Step B4: if J is [F ^(k)(t)] with the target function value J [F of last iteration ^(k-1)(t)] the difference of absolute value be less than precision tol, then judge that convergence meets, and the braking strategy of current iteration outputted to brake unit as instruction; If convergence does not meet, then continue to perform step B5;

Step B5: upgrade related dynamic parameters: use F ^(k)t the value of () covers F ^(k-1)the value of (t), and iterations k is increased by 1;

Step B6: nonlinear programming problem solves module (36) and utilizes the target function value and constraint functional value that obtain in step B3, by calculating search direction and optimizing step-length, obtaining and comparing F ^(k-1)t new brake strategy F that () is more excellent ^(k)(t); Again step B3 is jumped to, till convergence judge module (34) meets after this step is complete;

Described ordinary differential system solves module, and the method for employing is four step Adams methods, and computing formula is:

$x_1\left(t_{i+1}\right)=x_1\left(t_i\right)+\frac{h}{24}[{55x}_2\left(t_i\right)-{59x}_2\left(t_{i-1}\right)+{37x}_2\left(t_{i-2}\right)-{9x}_2\left(t_{i-3}\right)]$

$x_2\left(t_{i+1}\right)=x_2\left(t_i\right)+\frac{h}{24}[55F\left(t_i\right)-59F\left(t_{i-1}\right)+37F\left(t_{i-2}\right)-9F\left(t_{i-3}\right)]$

Wherein t represents the time, t _ito represent in the braking procedure of Adams method choice point sometime, t _i-1represent the t selected in Adams method _iprevious time point, t _i+1represent the t selected in Adams method _ia rear time point, by that analogy; Integration step h is the difference of any two adjacent time points; x ₁(t _i) represent that train is at t _ithe operating range in moment, x ₂(t _i) represent that train is at t _ithe moving velocity in moment, F (t _i) represent at t _ithe braking force in moment;

Described nonlinear programming problem solves module, adopts following steps to realize:

Step C1: by braking strategy F ^(k-1)t (), as certain point in vector space, is denoted as P ₁, P ₁corresponding target function value is exactly J [F ^(k-1)(t)];

Step C2: from a P ₁set out, according to a search direction d in the nonlinear programming problem algorithm construction vector space selected ^(k-1)with step-length α ^(k-1);

Step C3: through type F ^(k)(t)=F ^(k-1)(t)+α ^(k-1)d ^(k-1)corresponding u in structure vector space ^(k)another one point P ₂, make P ₂corresponding target function value J [F ^(k)(t)] than J [F ^(k-1)(t)] more excellent.