CN117104310A - Virtual marshalling control method and system based on data-driven predictive control - Google Patents
Virtual marshalling control method and system based on data-driven predictive control Download PDFInfo
- Publication number
- CN117104310A CN117104310A CN202310680225.5A CN202310680225A CN117104310A CN 117104310 A CN117104310 A CN 117104310A CN 202310680225 A CN202310680225 A CN 202310680225A CN 117104310 A CN117104310 A CN 117104310A
- Authority
- CN
- China
- Prior art keywords
- speed
- vehicle
- periods
- rear vehicle
- short
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000001133 acceleration Effects 0.000 claims description 10
- 230000006870 function Effects 0.000 claims description 10
- 238000013507 mapping Methods 0.000 claims description 10
- 230000000737 periodic effect Effects 0.000 claims description 5
- 239000013598 vector Substances 0.000 claims description 5
- 238000012549 training Methods 0.000 claims description 2
- 238000005457 optimization Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000000903 blocking effect Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000010845 search algorithm Methods 0.000 description 2
- 241000251468 Actinopterygii Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005293 physical law Methods 0.000 description 1
- 238000004886 process control Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 238000002922 simulated annealing Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Landscapes
- Feedback Control In General (AREA)
Abstract
The embodiment of the specification provides a virtual grouping control method and a system based on data-driven predictive control, wherein the method comprises the following steps: predicting the speed values of the last m periods of the front vehicle through a pre-trained short-time dynamics model according to the speed, control quantity and gradient information of the first n periods of the front vehicle, acquiring a predicted speed sequence according to the speed values of the last m periods, and transmitting the predicted speed sequence to the rear vehicle, wherein m and n are natural numbers; the method comprises the steps of obtaining a predicted speed sequence through a rear vehicle, determining a grouping distance of a rear m periods according to the predicted speed sequence, calculating a rear vehicle target speed of the rear m periods according to the grouping distance, solving a control quantity required by realizing the rear vehicle target speed according to the rear vehicle target speed, and controlling the rear vehicle to move according to the control quantity. The embodiment of the specification can more accurately predict the track of the front vehicle, so that the tracking performance of the rear vehicle is optimized.
Description
Technical Field
The present document relates to the field of model predictive control technologies, and in particular, to a virtual grouping control method and system based on data-driven predictive control.
Background
Along with the rapid development of urban rail traffic and the continuous improvement of transportation demands, higher requirements are put forward on the operation capability of a train control system, the operation mode based on mobile blocking becomes a core factor influencing the train operation interval, and the problem that the blocking interval limit is broken through to further improve the operation efficiency is solved. One of the development directions of the method can be considered from the angle of a train operation grouping mode, and an advanced train grouping mode is used for improving the transportation efficiency of rail transit, shortening the departure time interval between trains and improving the transportation capacity of the trains, so that the purpose of meeting traffic demand and national economy development demand is achieved.
For this purpose, the concept of virtual grouping is proposed. As shown in fig. 1, the virtual marshalling technology directly and wirelessly communicates with vehicles, and the rear vehicle obtains the running state of the front vehicle to control the running of the rear vehicle, so that the cooperative running mode of trains with the same speed and very small intervals is realized through wireless communication. In this way, the traditional physical coupler hitch is changed into a wireless communication hitch. The virtual marshalling breaks through the limitation of the original blocking interval, the safety protection distance of the train is further shortened, the train is more flexible to adjust, and the running efficiency is further improved.
The proposal of the virtual marshalling concept also provides new challenges for the train control method: in the virtual formation operation, the running track of the following train needs to be dynamically generated and adjusted in real time, so that a good control effect is difficult to achieve by a traditional control mode such as PID, and many control methods aiming at virtual formation are proposed, wherein one of the common methods is a model predictive control (Model Predict Control, MPC) method.
MPC may also be referred to as rolling time domain control (Moving Horizon Control, MHC) is an advanced process control method for controlling a process that satisfies a range of constraints. Since the 80 s of the 20 th century, it has been used in chemical plants and refinery process industries. MPC has the advantage of being able to display strong constraints on the processing system, making it widely applicable in a variety of industriesIn industrial control. In the prior art, y is as shown in FIG. 2 k To predict the trajectory, u k For optimal control sequences, r (t) is the desired trajectory, and MPC can be divided into three steps:
(1) Model prediction: and measuring the current state of the system, taking the measured state information as the initial state of the system, and predicting the state of the system in the limited time domain through the established system model.
(2) And (3) optimizing and solving: and designing an objective function, and obtaining a control sequence of the current state of the system in an optimization solving mode based on the establishment of an optimization problem of the objective function according to the state obtained by prediction.
(3) Performing: applying the first element of the control sequence of the optimization solution to the control object, and repeating step (1) in the next sampling period.
In the MPC method proposed for virtual marshalling, the algorithm of an optimization solving part is usually focused, and a model part is basically a simple kinematic model or a simplified train dynamics model, so that the prediction accuracy is somewhat deficient, and the model part cannot be adjusted along with the change of the performance of the train in running, thereby influencing the final control effect.
Disclosure of Invention
The present invention is directed to a virtual group control method and system based on data-driven predictive control, and aims to solve the above-mentioned problems in the prior art.
The invention provides a virtual marshalling control method based on data driving predictive control, which is used for double-vehicle virtual marshalling formed by a front vehicle and a rear vehicle, and comprises the following steps:
predicting the speed values of the last m periods of the front vehicle through a pre-trained short-time dynamics model according to the speed, control quantity and gradient information of the first n periods of the front vehicle, acquiring a predicted speed sequence according to the speed values of the last m periods, and transmitting the predicted speed sequence to the rear vehicle, wherein m and n are natural numbers;
the method comprises the steps of obtaining a predicted speed sequence through a rear vehicle, determining a grouping distance of a rear m periods according to the predicted speed sequence, calculating a rear vehicle target speed of the rear m periods according to the grouping distance, solving a control quantity required by realizing the rear vehicle target speed according to the rear vehicle target speed, and controlling the rear vehicle to move according to the control quantity.
The invention provides a virtual marshalling control system based on data driving prediction control, which is used for double-vehicle virtual marshalling formed by a front vehicle and a rear vehicle, and comprises the following components:
the prediction module is positioned on the front vehicle and used for predicting the speed values of the last m periods of the front vehicle through the speed, control quantity and gradient information of the first n periods of the front vehicle and a pre-trained short-time dynamics model, acquiring a predicted speed sequence according to the speed values of the last m periods and transmitting the predicted speed sequence to the rear vehicle, wherein m and n are natural numbers;
the solving module is positioned on the rear vehicle and used for acquiring the predicted speed sequence through the rear vehicle, determining the grouping distance of the rear m periods according to the predicted speed sequence, calculating the target speed of the rear vehicle of the rear m periods according to the grouping distance, solving the control quantity required by realizing the target speed of the rear vehicle according to the target speed of the rear vehicle, and controlling the rear vehicle to move according to the control quantity.
By adopting the embodiment of the invention, the track of the front vehicle can be predicted more accurately, so that the tracking performance of the rear vehicle is optimized.
Drawings
For a clearer description of one or more embodiments of the present description or of the solutions of the prior art, the drawings that are necessary for the description of the embodiments or of the prior art will be briefly described, it being apparent that the drawings in the description that follow are only some of the embodiments described in the description, from which, for a person skilled in the art, other drawings can be obtained without inventive faculty.
FIG. 1 is a schematic diagram of a prior art virtual consist;
FIG. 2 is a flow chart of an MPC process in the prior art;
FIG. 3 is a flow chart of a virtual consist control method based on data-driven predictive control in accordance with an embodiment of the present invention;
FIG. 4 is a flowchart illustrating a detailed process of a virtual consist control method based on data-driven predictive control in accordance with an embodiment of the present invention;
FIG. 5 is a schematic diagram of solving control quantities using a search algorithm according to an embodiment of the present invention;
fig. 6 is a schematic diagram of a virtual consist control system based on data-driven predictive control in accordance with an embodiment of the present invention.
Detailed Description
In order to solve the problems in the prior art, the embodiment of the invention establishes a short-term dynamics model of a virtual marshalling front vehicle in a data driving mode, the front vehicle predicts a short-term speed curve according to the short-term speed curve and sends the short-term speed curve to a marshalling rear vehicle when the marshalling front vehicle runs, and the marshalling rear vehicle calculates a control level according to the predicted speed curve of the front vehicle so as to track the speed and the distance. Compared with the existing model predictive control method, the technical scheme of the method embodiment is based on a machine learning mode, and the model is built through data driving so as to more accurately predict the track of the front vehicle and optimize the tracking performance of the rear vehicle.
In order to enable a person skilled in the art to better understand the technical solutions in one or more embodiments of the present specification, the technical solutions in one or more embodiments of the present specification will be clearly and completely described below with reference to the drawings in one or more embodiments of the present specification, and it is obvious that the described embodiments are only some embodiments of the present specification, not all embodiments. All other embodiments, which can be made by one or more embodiments of the present disclosure without inventive faculty, are intended to be within the scope of the present disclosure.
Method embodiment
According to an embodiment of the present invention, there is provided a virtual group control method based on data-driven predictive control, for a two-vehicle virtual group formed by a front vehicle and a rear vehicle, and fig. 3 is a flowchart of the virtual group control method based on data-driven predictive control according to the embodiment of the present invention, as shown in fig. 3, and the virtual group control method based on data-driven predictive control according to the embodiment of the present invention specifically includes:
step S301, predicting the speed values of m periods of a front vehicle through a pre-trained short-time dynamics model according to the speed, control quantity and gradient information of the front n periods of the front vehicle, acquiring a predicted speed sequence according to the speed values of the m periods of the front vehicle, and transmitting the predicted speed sequence to a rear vehicle, wherein m and n are natural numbers;
in the embodiment of the invention, all periods in a period of a preceding vehicle are grouped by taking n periods as a group, the speed, control quantity and gradient information of the n periods in each group are respectively used as input variables of a short-time dynamics model, the speed values of m periods in the later stage are predicted by the short-time dynamics model through the input variables, the input variables and the speed values of each group are used as one sample, a plurality of samples in the period are obtained, the input variables of the plurality of samples are used as input vectors x of the short-time dynamics model, the speed values of the plurality of samples are used as true values y of the short-time dynamics model, and a mapping relation from x to y is obtained in a supervised learning mode, so that the trained short-time dynamics model is obtained. That is, the embodiment of the invention constructs a train short-term dynamics model with self-learning capability through a machine learning method, and performs model predictive control of virtually grouped trains by using the model.
In the embodiment of the invention, the controller of the front vehicle tracks the target speed of the vehicle, calculates the control quantity according to the target speed of the vehicle, drives the vehicle based on the control quantity, and records the periodic speed of the train;
step S302, the predicted speed sequence is obtained through a rear vehicle, the grouping distance of the rear m periods is determined according to the predicted speed sequence, the rear vehicle target speed of the rear m periods is calculated according to the grouping distance, the control quantity required for realizing the rear vehicle target speed is solved according to the rear vehicle target speed, and the rear vehicle driving is controlled according to the control quantity.
The determining the grouping distance of the m periods according to the predicted speed sequence specifically comprises the following steps: in the case of non-virtual marshalling driving, a minimum distance value of the train is usually determined according to the speed of the front and rear double vehicles by a protection model of an emergency braking 'soft wall', wherein the minimum distance value of the front and rear vehicle is calculated (the minimum distance value of the front and rear vehicle is calculated when the double vehicles respectively perform emergency braking at the self speed, the absolute value of the minimum distance value is taken as the minimum distance value if the minimum distance value is smaller than 0, and the minimum distance value is taken as 0 if the minimum distance value is greater than or equal to 0). In the virtual marshalling operation, due to the characteristic that the front and rear vehicle speeds tend to be consistent, the difference value between the rear vehicle speed and the front vehicle speed can be considered to be smaller than a certain deviation, so that the deviation can be added to the front vehicle speed as the rear vehicle speed to be brought into a model, and the marshalling distance under the certain front vehicle speed is calculated.
The calculating the target speed of the rear vehicle in the m periods according to the grouping distance specifically comprises the following steps:
determining a target speed function of the rear vehicle for the rear m periods according to the formula 1:
equation 1;
wherein,for the ith cycle speed of the preceding vehicle, +.>For the ith cycle speed of the rear vehicle, +.>Representing distance item weight,/->For the distance between the rear vehicle head and the front vehicle tail in the ith period,/for the distance between the front vehicle head>For the ith period target consist distance, +.>As shown in equation 2:
equation 2;
wherein,for controlling the period +.>Is the initial distance;
in one control cycle, the initial distance between the front and rear vehiclesAnd m cycle prediction speeds in front of and behind the vehicle +.>~/>Known, i.e. constant, m cycle speeds after the following vehicle +.>~/>For the solution target, the constraint is equation 3 and equation 4:
equation 3;
equation 4;
wherein,for the minimum acceleration that can be produced by the train, +.>For the maximum acceleration that the train can produce,speed limits for consideration of line, vehicle and signal factors;
and obtaining the target speed of the rear vehicle in the last m periods by solving the formula 1, the formula 3 and the formula 4.
Solving the control measure required to achieve the target speed of the rear vehicle specifically includes:
based on a short-time dynamics model of the rear vehicle, a mapping relation of a control quantity and a short-time speed is obtained, and an objective function which takes the control quantity u as a variable and minimizes the error between the short-time speed and a target speed is obtained, as shown in a formula 5:
equation 5;
wherein,predicting m periodic speed sequences behind a rear vehicle under the condition of applying a control quantity u through a short-time model;
based on the control quantity required by solving the target speed of the rear vehicle in the formula 5, the basic flow is as shown in the figure 5, and specifically comprises the following steps:
step 1, inputting a target speed sequence;
step 2, determining an initial value of a control quantity u, and initializing a minimum error to infinity;
step 3, predicting a speed sequence of the control quantity u through a short-time dynamics model;
step 4, calculating the error between the speed sequence and the target speed sequence;
and 5, judging whether the error is smaller than the current optimal value, if so, taking the current error as the minimum error, recording the corresponding control quantity u, if not, judging whether the continuous search condition is met, if so, selecting the control quantity u of the next search according to the search rule, returning to the step 3, and if not, outputting the control quantity u of the solving result.
The above technical solutions of the embodiments of the present invention are described in detail below with reference to the accompanying drawings.
The embodiment of the invention aims at the design of the virtual double-vehicle grouping with high implementation possibility at present, wherein the front vehicle of the grouping only needs to run according to the mode specified by the operation plan so as to ensure the accuracy of the plan, and the distance control and the like of the virtual grouping are ensured by the rear vehicle of the grouping. As shown in fig. 4, the overall algorithm flow still adopts three steps of prediction-solution-execution of MPC, wherein the prediction part predicts the motion state (short-time speed sequence) of the vehicle before the vehicle is grouped, and the prediction part is performed by the vehicle before the vehicle is grouped; the solving part calculates the control quantity needed to be executed by the rear vehicles and is carried out by the rear vehicles after the marshalling. The design method has two consideration, namely, a large amount of data is needed for predicting the motion state of the train, if the data are sent to the post-train for processing, the burden of train communication is increased, and the post-train needs to be subjected to solving steps, so that the calculation amount is large, and other calculation work of the post-train is reduced as much as possible.
1. Establishing a short-time dynamics model:
establishing an accurate speed prediction model is an important premise for implementing effective control. The method has better effect and smaller calculated amount when the running modes of the train in the section are highly consistent, but the prediction effect cannot be ensured when the running modes of the train are changed, such as temporary speed limit setting, running time adjustment and the like; the other is that by combining the information such as the environment information, the train state, the control quantity sequence and the like, a model simulating the train dynamics is input, so that the train running state in the short time in the future is deduced. The second prediction mode is adopted in real time by the invention in consideration of the flexibility of the grouping operation environment and the low requirement of the follow-up control on the length of the speed sequence.
(1) Variable selection:
the method is close to the actual physical law as much as possible, and considers which factors can influence the running speed of the train in the physical range under the actual condition, so that the influence of artificial definition factors such as a control strategy, line speed limit, driving habit and the like is avoided. Based on the above considerations, the embodiment of the present invention selects the train speed, the control amount (divided into the traction control amount and the braking control amount) and the gradient within a certain time window range (n cycles) as the input variables of the model.
(2) And (3) data acquisition:
the running data of the train (if the dynamic performance difference of the same type of train is within an acceptable range, the train is not limited, and the same type of train is only needed) of the train can be collected as a data set for learning by a model, and the running data of the train can be completely recorded with speed, gradient and control quantity information in a period of time in any interval whether the train is driven automatically or manually. Generally, for an autonomous train, the ATO log can automatically record the above data without additional acquisition, and additional installation of acquisition equipment is required if data acquisition from a non-autonomous train is desired.
(3) And (3) establishing a model:
the short-time dynamics model may be a neural network model, a time series model, a decision tree series model, etc., and embodiments of the present invention are not limited to use with a particular machine learning model. If the model needs to predict the speed of m periods after n periods of variable, grouping all periods in a period of a preceding vehicle by taking n periods as a group, respectively taking the speed, control quantity and gradient information of n periods in each group as input variables of a short-time dynamics model, predicting the speed value of m periods in the later period by the input variables through the short-time dynamics model, taking the input variables and the speed value of each group as one sample, acquiring a plurality of samples in the period, taking the input variables of the plurality of samples as input vectors x of the short-time dynamics model, taking the speed values of the plurality of samples as true values y of the short-time dynamics model, and obtaining a mapping relation from x to y in a supervised learning mode to acquire the trained short-time dynamics model.
In addition, although only the front vehicle and the model need to be used for speed prediction during virtual marshalling control, a speed prediction model needs to be built for the virtual rear vehicle, firstly, the relation between the front vehicle and the rear vehicle after turning back changes, and secondly, the model of the rear vehicle needs to be used for solving the subsequent control quantity.
2. Calculating the target speed of the rear vehicle:
the tracking goal of the virtual consist is to make two trains behave as much as possible as one train, i.e. at each moment: (1) the distance between the front vehicle tail and the rear vehicle head approaches to 0; (2) the speeds of the two vehicles tend to be the same. Considering that under certain scenes, two targets cannot reach the optimal condition (for example, the distance between the front car and the rear car is increased due to interference, and the speed of the rear car is required to exceed that of the front car to compensate the pulled distance), the two targets can be combined into a single optimal target in a weighting mode, and the m-cycle condition is considered, namely:
equation 1;
wherein,for the ith cycle speed of the preceding vehicle, +.>For the ith cycle speed of the rear vehicle, +.>Representing distance item weight,/->For the distance between the rear vehicle head and the front vehicle tail in the ith period,/for the distance between the front vehicle head>For the ith period target consist distance, +.>As shown in equation 2:
equation 2;
for the control period.
In one control cycle, the initial distance between the front and rear vehiclesAnd m cycle prediction speed before and after the vehicle>~/>Known, i.e. constant, m cycle speed after the following car +.>~/>To solve the objective, the constraint is:
wherein the method comprises the steps ofFor the initial speed of the rear vehicle>For the minimum acceleration that can be produced by the train, +.>For maximum acceleration that can be produced by the train, +.>To limit speed in view of the line, vehicle, signal, etc.
It is not difficult to see that the optimization problem is a constrained multi-element quadratic programming problem, and a mature solution algorithm exists and is not repeated here.
3. Control quantity solving
The final control objective is to calculate the control quantity u required to be issued by the grouped rear vehicles, and since the target speed of the rear vehicles is already obtained, various control algorithms can be used for solving, and in order to reduce the extra controller design and tuning work, the mapping relation of the control quantity and the short-time speed can be obtained by means of the short-time dynamics model of the rear vehicles, so that the optimization problem of minimizing the short-time speed and the target speed error by taking the control quantity u as a variable is obtained, and the following formula is adopted:
wherein the method comprises the steps ofThe m-cycle speed sequence after the vehicle is predicted by a short-time model under the condition of applying the control quantity u.
Because the short-time dynamics model is usually a nonlinear model, the optimization problem cannot be solved by means of quadratic programming and the like, and a search algorithm such as simulated annealing, artificial fish shoals and the like can be used for solving.
As shown in fig. 5, the method specifically comprises the following steps: firstly, inputting a target speed sequence, selecting a control quantity initial value u, and initializing a minimum error to infinity; predicting a speed sequence of the u control quantity through a dynamics model, and calculating an error between the speed sequence and a target speed; judging whether the error is smaller than the current optimal value, if so, recording the current error as the minimum error, and recording the corresponding control quantity, otherwise, judging whether the condition of continuous searching is continuously met, if so, selecting the next searched u value according to the searching rule, returning to the step of predicting the speed sequence of the u control quantity through the dynamic model, and if not, outputting the solving result, namely the recorded corresponding control quantity.
In summary, the embodiment of the invention establishes the short-term dynamics model of the virtual marshalling front vehicle in a data driving mode, the front vehicle predicts the short-term speed curve according to the short-term dynamics model and sends the short-term speed curve to the marshalling rear vehicle when the marshalling front vehicle runs, and the marshalling rear vehicle calculates the control level according to the predicted speed curve of the front vehicle so as to track the speed and the distance. Compared with the existing model predictive control method, the technical scheme of the method embodiment is based on a machine learning mode, and the model is built through data driving so as to more accurately predict the track of the front vehicle and optimize the tracking performance of the rear vehicle.
System embodiment
According to an embodiment of the present invention, there is provided a virtual group control system based on data-driven predictive control, for a two-vehicle virtual group formed by a front vehicle and a rear vehicle, and fig. 6 is a schematic diagram of the virtual group control system based on data-driven predictive control according to the embodiment of the present invention, as shown in fig. 6, the virtual group control system based on data-driven predictive control according to the embodiment of the present invention specifically includes:
the prediction module 60 is located on the front vehicle, and is configured to predict, through the speed, the control amount, and the gradient information of the front n periods of the front vehicle, the speed values of the rear m periods of the front vehicle through a pre-trained short-time dynamics model, obtain a predicted speed sequence according to the speed values of the rear m periods, and send the predicted speed sequence to the rear vehicle, where m and n are natural numbers;
the solving module 62 is located on the rear vehicle, and is configured to obtain the predicted speed sequence through the rear vehicle, determine a grouping distance of the rear m periods according to the predicted speed sequence, calculate a target speed of the rear vehicle of the rear m periods according to the grouping distance, solve a control quantity required for realizing the target speed of the rear vehicle according to the target speed of the rear vehicle, and control the rear vehicle according to the control quantity. The solving module 62 is specifically configured to:
determining a target speed function of the rear vehicle for the rear m periods according to the formula 1:
equation 1;
wherein,for the ith cycle speed of the preceding vehicle, +.>For the ith cycle speed of the rear vehicle, +.>Representing distance item weight,/->Is the distance between the rear vehicle head and the front vehicle tail of the ith periodLeave, go up>For the ith period target consist distance, +.>As shown in equation 2:
equation 2;
wherein,for controlling the period +.>Is the initial distance;
in one control cycle, the initial distance between the front and rear vehiclesAnd m cycle prediction speeds in front of and behind the vehicle +.>~/>Known, i.e. constant, m cycle speeds after the following vehicle +.>~/>For the solution target, the constraint is equation 3 and equation 4:
equation 3;
equation 4;
wherein,for the initial speed of the rear vehicle>For the minimum acceleration that can be produced by the train, +.>For maximum acceleration that can be produced by the train, +.>Speed limits for consideration of line, vehicle and signal factors;
and obtaining the target speed of the rear vehicle in the last m periods by solving the formula 1, the formula 3 and the formula 4.
Wherein solving the control amount required to achieve the rear vehicle target speed specifically includes:
based on a short-time dynamics model of the rear vehicle, a mapping relation of a control quantity and a short-time speed is obtained, and an objective function which takes the control quantity u as a variable and minimizes the error between the short-time speed and a target speed is obtained, as shown in a formula 5:
equation 5;
wherein,predicting m periodic speed sequences behind a rear vehicle under the condition of applying a control quantity u through a short-time model;
the control amount required for solving the target speed of the rear vehicle is based on equation 5. The method specifically comprises the following steps:
step 1, inputting a target speed sequence;
step 2, determining an initial value of a control quantity u, and initializing a minimum error to infinity;
step 3, predicting a speed sequence of the control quantity u through a short-time dynamics model;
step 4, calculating the error between the speed sequence and the target speed sequence;
and 5, judging whether the error is smaller than the current optimal value, if so, taking the current error as the minimum error, recording the corresponding control quantity u, if not, judging whether the continuous search condition is met, if so, selecting the control quantity u of the next search according to the search rule, returning to the step 3, and if not, outputting the control quantity u of the solving result.
The system may further include:
the model training module is positioned in a front vehicle and a rear vehicle, and is used for grouping all the periods in a period of the front vehicle by taking n periods as a group, respectively taking the speed, the control quantity and the gradient information of the n periods in each group as input variables of a short-time dynamics model, predicting the speed values of m periods in the rear section by the short-time dynamics model through the input variables, taking the input variables and the speed values of each group as one sample, acquiring a plurality of samples in the period, taking the input variables of the plurality of samples as input vectors x of the short-time dynamics model, taking the speed values of the plurality of samples as true values y of the short-time dynamics model, and obtaining a mapping relation from x to y in a supervised learning mode to acquire the trained short-time dynamics model.
The embodiment of the present invention is a system embodiment corresponding to the above method embodiment, and specific operations of each module may be understood by referring to the description of the method embodiment, which is not repeated herein.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.
Claims (12)
1. The virtual marshalling control method based on data driving prediction control is characterized by being used for double-vehicle virtual marshalling formed by a front vehicle and a rear vehicle, and specifically comprises the following steps:
predicting the speed values of the last m periods of the front vehicle through a pre-trained short-time dynamics model according to the speed, control quantity and gradient information of the first n periods of the front vehicle, acquiring a predicted speed sequence according to the speed values of the last m periods, and transmitting the predicted speed sequence to the rear vehicle, wherein m and n are natural numbers;
the method comprises the steps of obtaining a predicted speed sequence through a rear vehicle, determining a grouping distance of a rear m periods according to the predicted speed sequence, calculating a rear vehicle target speed of the rear m periods according to the grouping distance, solving a control quantity required by realizing the rear vehicle target speed according to the rear vehicle target speed, and controlling the rear vehicle to move according to the control quantity.
2. The method according to claim 1, wherein the method further comprises:
grouping all periods in a period of time of a front vehicle by taking n periods as a group, respectively taking the speeds, control amounts and gradient information of the n periods in each group as input variables of a short-time dynamics model, predicting speed values of m periods in a later stage by the short-time dynamics model through the input variables, taking the input variables and the speed values of each group as one sample, acquiring a plurality of samples in the period of time, taking the input variables of the plurality of samples as input vectors x of the short-time dynamics model, taking the speed values of the plurality of samples as true values y of the short-time dynamics model, and obtaining a mapping relation from x to y in a supervised learning mode to acquire the trained short-time dynamics model.
3. The method according to claim 1, wherein calculating the target speed of the following vehicle for the following m periods according to the grouping distance specifically comprises:
determining a target speed function of the rear vehicle for the rear m periods according to the formula 1:
4. equation 1;
wherein,for the ith cycle speed of the preceding vehicle, +.>For the ith cycle speed of the rear vehicle, +.>Representing distance item weight,/->For the distance between the rear vehicle head and the front vehicle tail in the ith period,/for the distance between the front vehicle head>For the ith period target consist distance, +.>As shown in equation 2:
equation 2;
wherein,for controlling the period +.>Is the initial distance;
in one control cycle, the initial distance between the front and rear vehiclesAnd m cycle prediction speeds in front of and behind the vehicle +.>~/>Known, i.e. constant, m cycle speeds after the following vehicle +.>~/>For the solution target, the constraint is equation 3 and equation 4:
equation 3;
equation 4;
wherein,for the minimum acceleration that can be produced by the train, +.>For maximum acceleration that can be produced by the train, +.>Speed limits for consideration of line, vehicle and signal factors;
and obtaining the target speed of the rear vehicle in the last m periods by solving the formula 1, the formula 3 and the formula 4.
5. A method according to claim 3, wherein said solving for the control amount required to achieve said rear vehicle target speed from said rear vehicle target speed comprises:
based on a short-time dynamics model of the rear vehicle, a mapping relation of a control quantity and a short-time speed is obtained, and an objective function which takes the control quantity u as a variable and minimizes the error between the short-time speed and a target speed is obtained, as shown in a formula 5:
equation 5;
wherein,predicting m periodic speed sequences behind a rear vehicle under the condition of applying a control quantity u through a short-time model;
the control amount required for solving the target speed of the rear vehicle is based on equation 5.
6. The method of claim 4, wherein the control measure required to solve the rear truck target speed based on equation 5 specifically comprises:
step 1, inputting a target speed sequence;
step 2, determining an initial value of a control quantity u, and initializing a minimum error to infinity;
step 3, predicting a speed sequence of the control quantity u through a short-time dynamics model;
step 4, calculating the error between the speed sequence and the target speed sequence;
and 5, judging whether the error is smaller than the current optimal value, if so, taking the current error as the minimum error, recording the corresponding control quantity u, if not, judging whether the continuous search condition is met, if so, selecting the control quantity u of the next search according to the search rule, returning to the step 3, and if not, outputting the control quantity u of the solving result.
7. A virtual consist control system based on data-driven predictive control, characterized by a two-vehicle virtual consist for a front vehicle and a rear vehicle, the system comprising in particular:
the prediction module is positioned on the front vehicle and used for predicting the speed values of the last m periods of the front vehicle through the speed, control quantity and gradient information of the first n periods of the front vehicle and a pre-trained short-time dynamics model, acquiring a predicted speed sequence according to the speed values of the last m periods and transmitting the predicted speed sequence to the rear vehicle, wherein m and n are natural numbers;
the solving module is positioned on the rear vehicle and used for acquiring the predicted speed sequence through the rear vehicle, determining the grouping distance of the rear m periods according to the predicted speed sequence, calculating the target speed of the rear vehicle of the rear m periods according to the grouping distance, solving the control quantity required by realizing the target speed of the rear vehicle according to the target speed of the rear vehicle, and controlling the rear vehicle to move according to the control quantity.
8. The system of claim 6, wherein the system further comprises:
the model training module is positioned in a front vehicle and a rear vehicle, and is used for grouping all the periods in a period of the front vehicle by taking n periods as a group, respectively taking the speed, the control quantity and the gradient information of the n periods in each group as input variables of a short-time dynamics model, predicting the speed values of m periods in the rear section by the short-time dynamics model through the input variables, taking the input variables and the speed values of each group as one sample, acquiring a plurality of samples in the period, taking the input variables of the plurality of samples as input vectors x of the short-time dynamics model, taking the speed values of the plurality of samples as true values y of the short-time dynamics model, and obtaining a mapping relation from x to y in a supervised learning mode to acquire the trained short-time dynamics model.
9. The system of claim 6, wherein the solution module is specifically configured to:
determining a target speed function of the rear vehicle for the rear m periods according to the formula 1:
10. equation 1;
wherein,the ith week of the front vehicleStage speed (I/O)>For the ith cycle speed of the rear vehicle, +.>Representing distance item weight,/->For the distance between the rear vehicle head and the front vehicle tail in the ith period,/for the distance between the front vehicle head>For the ith period target consist distance, +.>As shown in equation 2:
equation 2;
wherein,for controlling the period +.>Is the initial distance;
in one control cycle, the initial distance between the front and rear vehiclesAnd m cycle prediction speeds in front of and behind the vehicle +.>~/>Known, i.e. constant, m cycle speeds after the following vehicle +.>~/>For the solution target, the constraint is equation 3 and equation 4:
equation 3;
equation 4;
wherein,for the minimum acceleration that can be produced by the train, +.>For maximum acceleration that can be produced by the train, +.>Speed limits for consideration of line, vehicle and signal factors;
and obtaining the target speed of the rear vehicle in the last m periods by solving the formula 1, the formula 3 and the formula 4.
11. The system of claim 8, wherein the solution module is specifically configured to:
based on a short-time dynamics model of the rear vehicle, a mapping relation of a control quantity and a short-time speed is obtained, and an objective function which takes the control quantity u as a variable and minimizes the error between the short-time speed and a target speed is obtained, as shown in a formula 5:
equation 5;
wherein,predicting m periodic speed sequences behind a rear vehicle under the condition of applying a control quantity u through a short-time model;
the control amount required for solving the target speed of the rear vehicle is based on equation 5.
12. The system of claim 9, wherein the solution module is specifically configured to:
step 1, inputting a target speed sequence;
step 2, determining an initial value of a control quantity u, and initializing a minimum error to infinity;
step 3, predicting a speed sequence of the control quantity u through a short-time dynamics model;
step 4, calculating the error between the speed sequence and the target speed sequence;
and 5, judging whether the error is smaller than the current optimal value, if so, taking the current error as the minimum error, recording the corresponding control quantity u, if not, judging whether the continuous search condition is met, if so, selecting the control quantity u of the next search according to the search rule, returning to the step 3, and if not, outputting the control quantity u of the solving result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310680225.5A CN117104310A (en) | 2023-06-09 | 2023-06-09 | Virtual marshalling control method and system based on data-driven predictive control |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310680225.5A CN117104310A (en) | 2023-06-09 | 2023-06-09 | Virtual marshalling control method and system based on data-driven predictive control |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117104310A true CN117104310A (en) | 2023-11-24 |
Family
ID=88799011
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310680225.5A Pending CN117104310A (en) | 2023-06-09 | 2023-06-09 | Virtual marshalling control method and system based on data-driven predictive control |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117104310A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117521420A (en) * | 2024-01-04 | 2024-02-06 | 北京交通大学 | Rail transit virtual marshalling train dynamics model construction and application method and system |
-
2023
- 2023-06-09 CN CN202310680225.5A patent/CN117104310A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117521420A (en) * | 2024-01-04 | 2024-02-06 | 北京交通大学 | Rail transit virtual marshalling train dynamics model construction and application method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110371112B (en) | Intelligent obstacle avoidance system and method for automatic driving vehicle | |
| CN111098852B (en) | Parking path planning method based on reinforcement learning | |
| CN111145552B (en) | Planning method for vehicle dynamic lane changing track based on 5G network | |
| Wang et al. | Continuous control for automated lane change behavior based on deep deterministic policy gradient algorithm | |
| CN112307564B (en) | Train ATO target running speed curve optimization method and device | |
| US11794757B2 (en) | Systems and methods for prediction windows for optimal powertrain control | |
| WO2020000192A1 (en) | Method for providing vehicle trajectory prediction | |
| CN111267831A (en) | Hybrid vehicle intelligent time-domain-variable model prediction energy management method | |
| CN107169567A (en) | The generation method and device of a kind of decision networks model for Vehicular automatic driving | |
| Yin et al. | Data-driven models for train control dynamics in high-speed railways: LAG-LSTM for train trajectory prediction | |
| CN117104310A (en) | Virtual marshalling control method and system based on data-driven predictive control | |
| EP3803528A1 (en) | Controller, system and method for vehicle control | |
| CN112686453A (en) | Intelligent prediction method and system for locomotive energy consumption | |
| CN106708044A (en) | Full-hovering hovercraft course control method based on grey prediction hybrid genetic algorithm-PID | |
| KR102256644B1 (en) | Artificial intelligence traffic signal host server using BIM object model and control system comprising it and method of controlling traffic signal | |
| CN113815679B (en) | Implementation method for autonomous driving control of high-speed train | |
| CN116520890B (en) | Unmanned aerial vehicle control platform capable of three-dimensional holographic inspection | |
| CN116176654A (en) | Scene self-adaptive track traffic ATO control system | |
| CN111598311B (en) | Novel intelligent optimization method for train running speed curve | |
| CN112835362B (en) | Automatic lane change planning method and device, electronic equipment and storage medium | |
| CN117444978B (en) | Position control method, system and equipment for pneumatic soft robot | |
| Tesar | Increasing Energy Efficiency and Punctuality of a Light Rail Vehicle with Deep Reinforcement Learning | |
| Zhang et al. | Vehicle Acceleration Prediction Based on Nonlinear Auto Regressive Models with Exogenous Inputs | |
| CN117302293B (en) | Wheel diameter value prediction method and device, electronic equipment and storage medium | |
| CN117208021B (en) | Unmanned vehicle control method for complex road conditions |
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 |