High-speed train optimal traction coordination method based on dynamic average consistency
Through the distributed optimization algorithm based on elastic fixed time dynamic average consistency, the problem of optimal coordination distribution of traction power in distributed coordination control of high-speed trains is solved, and the safe and fast distribution of traction power for high-speed trains without a control center is realized, improving the safety and rapid distribution of traction power for high-speed trains System reliability and network security are used to meet the fast and high-precision control needs of high-speed and high-density operations of high-speed trains.
Patent Information
- Application Number
- CN202411323686.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-23
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-09-23
AI Technical Summary
The existing distributed coordination control scheme for high-speed trains has failed to effectively solve the problem of optimal coordinated distribution of traction power, making it difficult to achieve low-cost operation, and affects the safe operation of the train in the case of cyber attacks.
Using a distributed optimization algorithm based on elastic fixed time dynamic average consistency, the traction power coordination optimization problem is converted into a dynamic average consistency problem, and the optimal traction force of each traction power unit is distributedly to achieve the situation without a control center. The traction power of the high-speed train is safe, fast and optimally allocated.
It realizes distributed control with optimized and coordinated traction power of high-speed trains, avoids single-node failure of the control center, improves system reliability and network security, and meets the fast and high-precision control needs of high-speed and high-density operations of high-speed trains.
Smart Images

Figure CN119037482A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of high-speed train operation control and relates to an optimal traction coordination method for high-speed trains based on dynamic average consistency. Background Art
[0002] In order to meet the higher safety, comfort, energy saving, punctuality and other performance requirements of railway transportation, high-speed trains must be equipped with advanced operation control systems to control the traction and braking of the trains. The traction power of the train is mainly divided into two modes: centralized traction power mode and decentralized traction power mode. Since the high-speed operation of the train requires a powerful power drive, the high-speed train generally adopts a decentralized traction power mode, that is, multiple traction power units are distributed in multiple carriages. The existing train operation control system uses the control center to coordinate, optimize and control the decentralized traction power. This centralized coordination method has the disadvantages of single-node failure, heavy communication burden on the control center, and poor scalability. It is difficult to guarantee the future high-speed, high-density, and public transportation-like high-speed train operation needs.
[0003] In order to overcome the drawbacks of centralized coordination optimization and control of high-speed trains, academia and industry have proposed distributed coordination control methods for high-speed trains based on multi-agent system consistency in recent years. For example, the patent application with application number 201910588591.1 proposed a distributed control method for multiple power units of high-speed trains; the patent application with application number 201810557997.9 proposed a parking consistency control method for EMUs based on multi-agents; the paper Distributed adaptive fault-tolerant control for high-speed trains using multi-agent system model published in the IEEE Transactions on Vehicular Technology journal proposed a distributed adaptive fault-tolerant control method for high-speed trains.
[0004] The above-mentioned distributed coordinated control scheme for high-speed trains mainly solves the control problems of train position and speed, but ignores the optimal coordinated allocation of traction power, which makes it impossible for high-speed trains to operate at low cost under distributed control; secondly, the existing distributed coordinated control scheme for high-speed trains only achieves asymptotic convergence, which makes it difficult to ensure fast and high-precision control of the train; in addition, the existing distributed coordinated control scheme for high-speed trains considers ideal communication conditions, while the actual communication network is extremely vulnerable to network attacks, thus affecting the safe operation of the train. Summary of the invention
[0005] In view of this, the purpose of the present invention is to provide an optimal traction coordination method for high-speed trains based on dynamic average consistency, and to achieve safe, fast and optimal allocation of decentralized traction power for high-speed trains under hostile network attack control input based on a distributed optimization algorithm with elastic fixed-time dynamic average consistency in the absence of a control center.
[0006] In order to achieve the above object, the present invention provides the following technical solutions:
[0007] A high-speed train optimal traction coordination method based on dynamic average consistency, the method comprising: considering a high-speed train with no less than two traction power units, establishing a traction power coordination optimization problem of the high-speed train according to the operating status of each traction power unit; using the optimal solution characteristics, converting the traction power coordination optimization problem into a dynamic average consistency problem; and obtaining the optimal traction force of each traction power unit of the high-speed train in a distributed manner based on a coordination optimization algorithm of elastic fixed-time dynamic average consistency.
[0008] Furthermore, the coordinated optimization problem of traction power of high-speed trains includes:
[0009] The cost function that needs to be optimized to establish high-speed train traction power coordination is:
[0010]
[0011] Where C(f) represents the total cost function of traction, f = [f1,f2,…,f n ] T , f i represents the traction force provided by the i-th traction power unit, C i (f i ) represents the cost function of the i-th traction power unit, and n represents the total number of traction power units.
[0012] The constraints that need to be satisfied to minimize the cost function are established, including the traction-drag balance constraint, which is expressed as:
[0013]
[0014] Where R(t) represents the time-varying total resistance of the high-speed train, and the upper and lower limits of traction are expressed as:
[0015] f i m ≤f i ≤f i M
[0016] In the formula, f i mrepresents the lower limit of the traction force of the i-th traction power unit, f i M Represents the upper limit of the traction force of the i-th traction power unit.
[0017] The smooth penalty function is used to transform the above traction power coordination optimization problem with equality and inequality constraints into a traction power coordination optimization problem with only equality constraints:
[0018]
[0019] In the formula, C σ (f) represents the total cost function with penalty cost, represents the cost function with penalty cost of the i-th traction power unit;
[0020]
[0021] In the formula, σ i >0 is the penalty parameter for the coordination optimization problem of the i-th traction power unit.
[0022] Furthermore, based on the traction power coordination optimization problem with only equality constraints, the traction power coordination optimization problem is transformed into a dynamic average consistency problem according to the optimal solution characteristics of the coordination optimization problem. is twice continuously differentiable and strongly convex, then the optimal solution of the coordinated optimization problem is the stationary point of the Lagrangian function; the Lagrangian function is expressed as:
[0023]
[0024] Where λ represents the Lagrange multiplier of the equality constraint;
[0025] According to the Newton-Raphson method, the stationary point of the Lagrangian function L can be obtained by the following equation:
[0026]
[0027] Where η=[f1,f2,…,f n ,λ] T , Δη=[Δf1,Δf2,…,Δf n ,Δ] T , and They represent the gradient and Hessian matrix of the Lagrangian function L at η respectively; then according to the above formula, the expression of the centralized optimal solution can be obtained as follows:
[0028]
[0029] In the formula, fi * represents the optimal traction power value of the i-th traction power unit. According to the optimal solution expression, the key to the distributed solution of the traction power coordination optimization problem is how to obtain the time-varying λ-Δλ in a distributed manner through the dynamic average consistency algorithm. Specifically, a coordinated optimization algorithm based on elastic fixed-time dynamic average consistency can be used to obtain the time-varying λ-Δλ and the optimal traction power value f in a distributed manner. i * .
[0030] Furthermore, the optimal traction power value f of each traction power unit is estimated in a distributed manner based on the coordinated optimization algorithm of elastic fixed-time dynamic average consistency. i * :
[0031]
[0032] In the formula, represents the estimated optimal traction power value of the i-th traction power unit; and represents the auxiliary variable of the coordinated optimization algorithm of the i-th traction power unit, which is used to estimate λ-Δλ.
[0033] Update by the following formulas respectively and
[0034]
[0035] Where N i represents the communication neighbor set of the i-th traction power unit, a ij represents the communication weight coefficient. If the jth and ith traction power units can obtain each other's information through communication connection, then j∈N i And a ij =a ji >0, otherwise And a ij =a ji = 0, the communication topology between the traction power units meets the conditions of being undirected and connected; and represents the state variable of the coordinated optimization algorithm of the ith traction power unit, and is the state variable of the jth traction power unit coordination optimization algorithm; R vi (t) represents the virtual time-varying resistance assigned to the i-th traction power unit coordination optimization algorithm, which is continuously differentiable and satisfies
[0036] Update the state variables by the following formulas respectively and
[0037]
[0038] In the formula, and The initial value of is any bounded real value; and represents the control input of the coordinated optimization algorithm of the ith traction power unit, and They represent the control input of the network attacker to the coordination optimization algorithm of the i-th traction power unit. and The multiplicative attack component applied, and are the control inputs of the network attacker to the coordinated optimization algorithm of the i-th traction power unit. and The additive attack component imposed. Among them, the multiplicative attack component is time-varying and unknown, satisfying and and is a known bounded positive constant; the additive attack component is also time-varying and unknown, satisfying and and is a known bounded positive constant.
[0039] Among them, the control input of the coordination optimization algorithm of the i-th traction power unit is and Respectively expressed as:
[0040]
[0041] In the formula, represents a positive control gain, 0<β<0.5 represents the control parameter, T1>0 is the preset fixed convergence time parameter of the controller, and represents the auxiliary variable consistency error of the coordinated optimization algorithm of the ith traction power unit, sig(·) 1-2β It is represented by sign(·)|·| 1-2β , where |·| is the absolute value function.
[0042] Through the above-mentioned coordinated optimization algorithm based on elastic fixed-time dynamic average consistency, the estimated optimal traction power value can be After a preset fixed convergence time T1, the time-varying optimal traction power f is accurately tracked. i * .
[0043] The beneficial effects of the present invention are:
[0044] (1) The high-speed train traction power optimization coordination method proposed in the present invention is distributed and does not require global information collection and processing by the control center, thus avoiding single-node failure of the control center, reducing the communication and computing burden of the controller, and being conducive to improving the reliability of the train control operation system.
[0045] (2) The high-speed train traction power optimization coordination method proposed in the present invention tracks the optimal solution within a preset fixed time, and its fixed time convergence rate can be directly adjusted by controller parameters. The preset fixed time convergence characteristics are conducive to meeting the fast and high-precision control performance requirements required for high-speed train high-speed, high-density, and public transportation operations.
[0046] (3) The high-speed train traction power optimization and coordination method proposed in the present invention can operate normally even under the control input of the network attack controller, which is beneficial to ensuring the network safe and stable operation of the high-speed train.
[0047] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:
[0049] Figure 1 It is a flow chart of the high-speed train optimal traction coordination method based on dynamic average consistency according to the present invention;
[0050] Figure 2 This is a topology diagram of the communication connection between traction power units. DETAILED DESCRIPTION
[0051] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.
[0052] Aiming at the optimal coordinated allocation of traction power of high-speed trains under control input of hostile network attacks, the present invention proposes a distributed optimization algorithm based on elastic fixed-time dynamic average consistency, which can realize safe, fast and optimal allocation of decentralized traction power of high-speed trains in the absence of a control center.
[0053] like Figure 1 As shown, the method proposed in the present invention mainly includes: considering a high-speed train with n (n is an integer greater than or equal to 2) traction power units. Firstly, according to the operating state of each traction power unit, the traction power coordination optimization problem of the high-speed train is established; secondly, the traction power coordination optimization problem is transformed into a dynamic average consistency problem by using the optimal solution characteristics; finally, the optimal traction force of each traction power unit is obtained based on the elastic fixed time dynamic average consistency algorithm.
[0054] In one embodiment of the present invention, a high-speed train including six traction power units is taken as an example to explain the method proposed by the present invention in detail. Figure 2 shown.
[0055] Specifically, the method includes:
[0056] 1. First, according to the operating status of the six traction power units, the traction power coordination optimization problem of the high-speed train is established. Specifically, the cost function that needs to be optimized for the traction power coordination of the high-speed train is established:
[0057]
[0058] Where C(f) represents the total cost function for traction, f = [f1, f2, ..., f6] T , f i represents the traction force provided by the i-th traction power unit, represents the cost function of the i-th traction power unit, which is twice continuously differentiable and strongly convex. i1 , α i2 , α i3 , α i4 And the upper and lower limit parameters of traction force f i M 、f i m As shown in Table 1:
[0059] Table 1
[0060]
[0061]
[0062] Establish the constraints that need to be satisfied to minimize the cost function, including:
[0063] 1) Traction-resistance balance constraint:
[0064]
[0065] Among them, R(t)=50000+30sin(t)+200tanh(t) represents the time-varying total resistance of the high-speed train.
[0066] 2) Upper and lower limit constraints of traction force of traction power unit:
[0067] f i m ≤f i ≤f i M
[0068] Then, a smooth penalty function is used to transform the proposed traction power coordination optimization problem with equality and inequality constraints into a traction power coordination optimization problem with only equality constraints, as shown in the following formula:
[0069]
[0070] Among them, C σ (f) represents the total cost function with penalty cost, The cost function representing the i-th traction power unit with penalty cost is expressed as:
[0071]
[0072] Among them, σ i =100 is the penalty parameter for the i-th traction power unit coordination optimization problem; g i,1 (f i ) = f i m -f i , g i,2 (f i ) = f i -f i M .
[0073] 2. Secondly, the traction power coordination optimization problem is transformed into a dynamic average consistency problem by utilizing the optimal solution characteristics.
[0074] In the above traction power coordination optimization problem with only equality constraints, due to is strongly convex, so the optimal solution to the coordination optimization problem is the Lagrangian function The stationary point of , where λ is the Lagrange multiplier of the equality constraint.
[0075] According to the Newton-Raphson method, the stationary point of the Lagrangian function L can be obtained by the following formula:
[0076]
[0077] η=[f1,f2,…,f6,λ] T =[8000,8000,8000,8000,10000,8000,10] T
[0078] Δη=[Δf1,Δf2,…,Δf6,Δλ] T
[0079] in, and are the gradient and Hessian matrix of the Lagrangian function L at η respectively. From the above formula, we can get the following centralized optimal solution expression:
[0080]
[0081] Among them, f i * It indicates the optimal traction power value of the i-th traction power unit. From the above formula, we can see that the key to the distributed solution of the traction power coordination optimization problem is how to use the dynamic average consistency algorithm to obtain the time-varying λ-Δλ in a distributed manner.
[0082] 3. Based on the elastic fixed-time dynamic average consistency algorithm, the optimal traction force of the six traction power units is obtained in a distributed manner.
[0083] The optimal traction power value of each traction power unit is estimated in a distributed manner based on a coordinated optimization algorithm based on elastic fixed-time dynamic average consistency:
[0084]
[0085] in, is the estimated optimal traction power value of the i-th traction power unit; and It is the auxiliary variable of the coordinated optimization algorithm of the i-th traction power unit and is used to estimate λ-Δλ.
[0086] and Updated by the following formulas respectively:
[0087]
[0088] Among them, N i is the communication neighbor set of the i-th traction power unit, a ij is the communication weight coefficient (its value is shown in Table 2), and is the state variable of the coordination optimization algorithm of the i-th traction power unit, and is the state variable of the coordination optimization algorithm for the jth traction power unit.
[0089] Table 2
[0090] <![CDATA[α 11 =0]]> <![CDATA[α 12 =1]]> <![CDATA[α 13 =0]]> <![CDATA[α 14 =0]]> <![CDATA[α 15 =0]]> <![CDATA[α 16 =1]]> <![CDATA[α 21 =1]]> <![CDATA[α 22 =0]]> <![CDATA[α 23 =1]]> <![CDATA[α 24 =0]]> <![CDATA[α 25 =0]]> <![CDATA[α 26 =0]]> <![CDATA[α 31 =0]]> <![CDATA[α 32 =1]]> <![CDATA[α 33 =0]]> <![CDATA[α 34 =1]]> <![CDATA[α 35 =0]]> <![CDATA[α 36 =0]]> <![CDATA[α 41 =0]]> <![CDATA[α 42 =0]]> <![CDATA[α 43 =1]]> <![CDATA[α 44 =0]]> <![CDATA[α 45 =1]]> <![CDATA[α 46 =0]]> <![CDATA[α 51 =0]]> <![CDATA[α 52 =0]]> <![CDATA[α 53 =0]]> <![CDATA[α 54 =1]]> <![CDATA[α 55 =0]]> <![CDATA[α 56 =1]]> <![CDATA[α 61 =1]]> <![CDATA[α 62 =0]]> <![CDATA[α 63 =0]]> <![CDATA[α 64 =0]]> <![CDATA[α 65 =1]]> <![CDATA[α 66 =0]]>
[0091] R vi (t) is the virtual time-varying resistance assigned to the coordinated optimization algorithm of the i-th traction power unit, and its value is:
[0092] R v1 (t) = 8000 + 10sin(t)
[0093] R v2 (t) = 9000 + 20sin(t)
[0094] R v3 (t) = 8000 + 60tanh (t)
[0095] R v4 (t) = 8000
[0096] R v5 (t) = 9000 + 60tanh (t)
[0097] R v6 (t) = 8000 + 80tanh (t)
[0098] State variables and Update them respectively by the following formulas:
[0099]
[0100] in, and The initial value of is 0. and is the control input of the coordination optimization algorithm for the i-th traction power unit, are the control inputs of the network attacker to the coordinated optimization algorithm of the i-th traction power unit. The multiplicative attack component applied, are the control inputs of the network attacker to the coordinated optimization algorithm of the i-th traction power unit. The additive attack component applied.
[0101] The attacker controls the input and The multiplicative attack component applied and its lower bound and As shown in Table 3, the control input and Additive attack component applied and its upper bound and As shown in Table 4.
[0102] Table 3
[0103]
[0104] Table 4
[0105]
[0106] Control input of the coordinated optimization algorithm for the ith traction power unit and It is expressed as:
[0107]
[0108] in, is a positive control gain, β = 0.3 is a control parameter, T1 = 2 is a preset fixed convergence time parameter of the controller, and is the auxiliary variable consistency error of the coordinated optimization algorithm of the i-th traction power unit, sig(·) 1-2β It is represented by sign(·)|·| 1-2β , where |·| is the absolute value function.
[0109] The proposed traction power unit coordination optimization algorithm based on elastic fixed-time dynamic average consistency can make the estimated optimal traction power value After the preset time T1 = 2s, accurately track the time-varying optimal solution f i * .
[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.
Claims
1. A high-speed train optimal traction coordination method based on dynamic average consistency, Features: Considering a high-speed train with no less than two traction power units, the traction power coordination optimization problem of the high-speed train is established according to the operating status of each traction power unit; By using the optimal solution characteristics, the traction power coordination optimization problem is transformed into a dynamic average consistency problem; The coordinated optimization algorithm based on elastic fixed-time dynamic average consistency obtains the optimal traction force of each traction power unit of the high-speed train in a distributed manner.
2. The method according to claim 1, Features: The problem of establishing coordinated optimization of traction power for high-speed trains includes: The cost function that needs to be optimized to establish high-speed train traction power coordination is: Where C(f) represents the total cost function of traction, f = [f 1 ,f 2 ,…,f n ] T , f i represents the traction force provided by the i-th traction power unit, C i (f i ) represents the cost function of the i-th traction power unit, and n represents the total number of traction power units; Establish the constraints that need to be satisfied to minimize the cost function, including the traction-drag balance constraint and the upper and lower limits of traction force f i m ≤f i ≤f i M ; where R(t) represents the time-varying total resistance of the high-speed train, f i m and f i M They represent the lower limit and upper limit of the traction force of the i-th traction power unit respectively; The smooth penalty function is used to transform the above traction power coordination optimization problem with equality and inequality constraints into a traction power coordination optimization problem with only equality constraints: In the formula, C σ (f) represents the total cost function with penalty cost, represents the cost function with penalty cost of the i-th traction power unit; In the formula, σ i >0 represents the penalty parameter of the i-th traction power unit coordination optimization problem.
3. The method according to claim 2, Features: Based on the traction power coordination optimization problem with only equality constraints, the traction power coordination optimization problem is transformed into a dynamic average consistency problem according to the optimal solution characteristics of the coordination optimization problem; wherein, in the traction power coordination optimization problem with only equality constraints, is twice continuously differentiable and strongly convex, then the optimal solution of the coordinated optimization problem is the stationary point of the Lagrangian function; the Lagrangian function is expressed as: Where λ represents the Lagrange multiplier of the equality constraint; According to the Newton-Raphson method, the stationary point of the Lagrangian function L is obtained by the following formula: In the formula, η=[f 1 ,f 2 ,…,f n ,λ] T , Δη=[Δf 1 ,Δf 2 ,…,Δf n ,Δλ] T , and They represent the gradient and Hessian matrix of the Lagrangian function L at η respectively; then the expression of the centralized optimal solution is obtained according to the above formula: In the formula, f i * represents the optimal traction power value of the i-th traction power unit; according to the optimal traction power value f i * The key to distributed solution of the optimal traction power value is to obtain the time-varying λ-Δλ in a distributed manner. The coordinated optimization algorithm based on elastic fixed-time dynamic average consistency is used to obtain the time-varying λ-Δλ and f in a distributed manner. i * .
4. The method according to claim 3, Features: The coordinated optimization algorithm based on elastic fixed-time dynamic average consistency estimates the optimal traction power value f of each traction power unit in a distributed manner. i * : In the formula, represents the estimated optimal traction power value of the i-th traction power unit; and represents the auxiliary variable of the coordination optimization algorithm of the i-th traction power unit, which is used to estimate λ-Δλ; Update by the following formulas respectively and Where N i represents the communication neighbor set of the i-th traction power unit; a ij represents the communication weight coefficient. If the jth and ith traction power units can obtain each other's information through communication connection, then j∈N i And a ij =a ji >0, otherwise And a ij =a ji = 0, the communication topology between the traction power units meets the conditions of being undirected and connected; and represents the state variable of the coordinated optimization algorithm of the ith traction power unit; R vi (t) represents the virtual time-varying resistance assigned to the i-th traction power unit coordination optimization algorithm, R vi (t) is continuously differentiable and satisfies Among them, the state variable and Updated by the following formulas respectively: In the formula, and represents the control input of the coordination optimization algorithm of the i-th traction power unit; and They represent the control input of the network attacker to the coordination optimization algorithm of the i-th traction power unit. and The multiplicative attack component applied is time-varying and unknown, satisfying and and is a known bounded positive constant; and They represent the control input of the network attacker to the coordination optimization algorithm of the i-th traction power unit. and The additive attack component applied is time-varying and unknown, satisfying and and is a known bounded positive constant; The control input and Respectively expressed as: In the formula, represents a positive control gain, 0<β<0.5 represents a control parameter, T 1 >0 indicates the preset fixed convergence time parameter of the controller. and represents the auxiliary variable consistency error of the coordinated optimization algorithm of the ith traction power unit, sig(·) 1-2β Represents sign(·)|·| 1-2β , where |·| is the absolute value function.
Citation Information
Patent Citations
Distributed control method of multiple power units of high-speed train
CN110333655A
Multi-train distributed cooperative cruise control method and system
CN115071784A
Train safety protection method based on dynamic parameters
CN115709748A
Train power distribution method and device
WO2022088316A1
Cited By
Fault-tolerant distributed preset time high-speed train traction power distribution method
CN120462456A
A fault-tolerant distributed pre-scheduled time high-speed train traction power distribution method
CN120462456B
