Self-adaptive high-precision lunar soft landing trajectory intelligent optimization control method and system

Abstract

The invention discloses a self-adaptive high-precision lunar soft landing trajectory intelligent optimization control method and system, and when a lunar lander is softly landed on the lunar surface, the control method comprises the steps: employing a speed sensor and a height sensor, and measuring the descending speed of the current lander and the distance between the lander and the lunar surface in real time; a fuel consumption rate control strategy for safe landing of the lunar lander and minimum fuel consumption is obtained by adopting an intelligent optimal control method of self-adaptive orthogonal configuration; and controlling the fuel consumption of the lunar lander by adopting the fuel consumption rate control strategy obtained in the step 2. Functional modules integrated by a lunar lander MCU in the control system comprise an information acquisition module, an initialization module, a time scale transformation module, an orthogonal configuration module, an NLP problem solving module, an adaptive processing module and a control instruction output module. And each module correspondingly completes the control method. And finally, the fuel consumption is reduced to the greatest extent during safe soft landing of the lander.

Description

Technical Field

[0001] The present invention relates to the field of lunar soft landing control, and particularly to an intelligent optimization control system and method for an adaptive high-precision lunar soft landing trajectory. When the lunar lander is about to land, the system can calculate the optimal soft landing strategy to ensure the safe landing of the lander and minimize fuel consumption to the greatest extent. Background Art

[0002] Since the 1990s, a new wave of enthusiasm has emerged. Among them, the soft landing of the lunar lander is a very important technical means. Soft landing is to reduce the vertical speed by certain means before landing so that it lands at an acceptable speed to protect the aircraft and astronauts. Since there is no air on the moon, similar to a vacuum, it is impossible to use a parachute, and it is difficult to control the force of an air cushion. The current technology relies on the reaction force of the lunar lander itself to achieve soft landing. Nowadays, many experts and scholars at home and abroad have carried out in-depth research on this. Summary of the Invention

[0003] The present invention discloses an intelligent optimization control system and method for an adaptive high-precision lunar soft landing trajectory, which is used to control the landing trajectory of a lunar lander. The control system consists of a lunar lander height sensor, a lunar lander speed sensor, a lunar lander MCU, a fuel consumption system parameter input, and a fuel consumption system. When the lunar lander is ready for soft landing, the lunar lander speed sensor and height sensor measure the current descent speed of the lander and its distance from the lunar surface, and transmit the measurement data to the MCU. The MCU immediately executes an intelligent optimal control method based on adaptive orthogonal collocation to calculate a control strategy that enables the lunar lander to land safely and consume the least amount of fuel, and converts it into an operation instruction to be transmitted to the fuel consumption system. The present invention can ensure the safe soft landing of the lunar lander and minimize fuel consumption to the greatest extent.

[0004] Specifically: An intelligent optimization control method for an adaptive high-precision lunar soft landing trajectory, characterized in that the steps when the lunar lander soft lands on the lunar surface include:

[0005] Step 1: Use a speed sensor and a height sensor to measure the current descent speed of the lander and its distance from the lunar surface in real time;

[0006] Step 2: Based on the measurement data in Step 1, use an intelligent optimal control method based on adaptive orthogonal collocation to obtain a control strategy (fuel consumption rate) that enables the lunar lander to land safely and consume the least amount of fuel;

[0007] Step 3: Use the fuel consumption rate control strategy obtained in Step 2 to control the fuel consumption of the lunar lander;

[0008] The derivation process of the mathematical model of the fuel consumption rate control strategy is as follows:

[0009] Describe the soft landing process of the lunar lander as:

[0010]

[0011] Where:

[0012] t represents time, t 0 represents the start time of the soft landing process, and t f represents the end time of the soft landing process, and t f is not fixed;

[0013] is the state variable of the lunar lander during the landing process (including physical parameters such as the speed and acceleration of the lunar lander during the landing process), and the superscript T represents the transpose operation; x 0 is the initial value of the state variable, is its first derivative;

[0014] u(t) represents the fuel consumption rate, and u l , u u are its lower limit value and upper limit value respectively;

[0015] is a system of differential equations established based on the law of conservation of energy and mechanical principles, and is a function of the time, state variables, and fuel consumption rate of the lunar lander during the landing process;

[0016]

[0017] is a constraint condition established for the physical parameters at the end of the soft landing of the lunar lander. n x and n g represent the dimension of the state vector and the dimension of the constraint condition respectively.

[0018] Assume that Φ[x(t f )] represents the total fuel consumption, then the mathematical model for minimizing fuel consumption is expressed as:

[0019]

[0020] Where J[u(t)] represents the control objective, which is determined by the fuel consumption rate u(t); Equation (2) is an optimal control problem;

[0021] In the second step:

[0022] 2.1) Collect measurement data:

[0023] Current descent speed of the lunar lander, altitude reached by the lunar lander, and performance parameters of the fuel consumption system;

[0024] 2.2) Time scale transformation:

[0025] Introduce a new time variable υ such that

[0026] t = (t f - t 0 )υ + t 0 (3)

[0027] Thus, convert the mathematical model (2) with an unfixed end time t f into a mathematical model with a control time domain of [0,1], as follows:

[0028]

[0029] where

[0030]

[0031] 2.3) Orthogonal collocation, discretize the ordinary differential equation system over the time axis [t 0 , t f to form an NLP problem. The steps include:

[0032] Step 2.3.1) Represent the control variable u(t) and the state trajectory x(t) as a linear combination of Lagrange interpolation basis functions, i.e.:

[0033]

[0034] where N is the number of discrete segments of the time axis [t 0 , t f , is the Lagrange interpolation basis function, and the linear combination coefficients u i,j and s i,j are the values of u(t) and x(t) at the Gauss collocation points t i,j respectively;

[0035] Step 2.3.2) Since the derivative function expressions of all basis functions are known, the differential equation system of the state trajectory is expressed in the following form:

[0036]

[0037] Step 2.3.3): Replace the original differential equation system with an approximate differential equation system and discretize it at the Gauss collocation points to obtain approximate algebraic equality equations; Express the objective function and constraints in terms of u i,j and s i,jPerform discrete processing to obtain a non-linear programming NLP problem;

[0038] 2.4) Solve the NLP problem to obtain the control strategy and the corresponding state trajectory:

[0039] 2.5) Perform adaptive processing, analyze the results of the NLP problem solving module. If the convergence condition is met, output the results of step 2.4) as the control strategy; if the convergence condition is not met, return to step 2.3).

[0040] An intelligent optimization control system for an adaptive high-precision lunar soft landing trajectory, characterized by comprising: a speed sensor, an altitude sensor, a lunar lander MCU, a fuel consumption system parameter input, and a fuel consumption system;

[0041] During the soft landing process of the lunar lander on the lunar surface:

[0042] Step 1: When the lunar lander prepares for soft landing, turn on the speed sensor and the altitude sensor to measure the current descent speed of the lander and the distance from the lunar surface in real time, and transmit the measurement data to the lunar lander MCU;

[0043] Step 2: The lunar lander MCU executes an intelligent optimal control method based on adaptive orthogonal collocation internally to calculate the control strategy that enables the lunar lander to land safely and consume the least fuel;

[0044] Step 3: The lunar lander MCU converts the obtained fuel consumption rate control strategy into a fuel consumption system operation instruction and transmits it to the fuel consumption system.

[0045] The beneficial effects of the present invention are mainly manifested in that: the intelligent optimization control system for an adaptive high-precision lunar soft landing trajectory controls the lunar soft landing trajectory by calculating the optimal fuel consumption rate of the lunar soft lander. The method introduced in the present invention has the characteristics of high calculation efficiency and high calculation accuracy. In addition, since in the NLP problem solving module, an intelligent optimization method based on particle swarm is adopted, it has the advantage of converging to the global optimal solution. Description of the Drawings

[0046] Figure 1 is a schematic structural diagram of an intelligent optimization control system for an adaptive high-precision lunar soft landing trajectory;

[0047] Figure 2 is a structural diagram of the internal functional modules of the MCU of an intelligent optimization control system for an adaptive high-precision lunar soft landing trajectory. Detailed Embodiments

[0048] In order to enable the lunar lander to land safely and minimize fuel consumption, the present invention provides an intelligent optimization control system for the lunar soft landing trajectory with high calculation speed and high calculation accuracy, which can meet the constraints and achieve the optimal adaptive high-precision. The control system uses an MCU as the implementation carrier of the optimal control method.

[0049] The soft landing process of the lunar lander can be described as follows:

[0050]

[0051] where t represents time, t 0 represents the start time of the soft landing process, and t f represents the end time of the soft landing process, and t f is not fixed; is called the state variable, representing physical parameters such as the speed and acceleration of the lunar lander. The superscript T represents the transpose of a vector or matrix. x 0 is its initial value, is its first derivative; u(t) represents the fuel consumption rate of the lunar lander, and u l , u u are its lower limit value and upper limit value respectively; is a system of differential equations established based on the law of conservation of energy and the principles of mechanics;

[0052] is a constraint condition established for the physical parameters at the end of the soft landing of the lunar lander. n x and n g represent the dimensions of the state vector and the constraint condition respectively.

[0053] Assume that Φ[x(t f )] represents the total fuel consumption. Then the mathematical model for minimizing fuel consumption can be expressed as:

[0054]

[0055] where J[u(t)] represents the control objective, which is determined by the fuel consumption rate u(t). This problem is essentially an optimal control problem.

[0056] An intelligent optimization method of adaptive orthogonal collocation is integrated in the MCU, and a control system is constructed based on this. The structure of the control system is as Figure 2 shown, and its functional modules include a lunar lander speed sensor, a lunar lander height sensor, a lunar lander MCU, a fuel consumption system, and a fuel consumption system parameter input.

[0057] The operation process of the control system is as follows:

[0058] Step 1: When the lunar lander prepares for a soft landing, turn on the speed sensor and the altitude sensor to measure the current descent speed of the lander and the distance from the lunar surface in real time, and transmit the measurement data to the MCU;

[0059] Step 2: The MCU executes the intelligent optimal control method based on adaptive orthogonal collocation internally to calculate the fuel consumption rate control strategy that enables the lunar lander to land safely and consumes the least fuel;

[0060] Step 3: The MCU converts the obtained fuel consumption rate control strategy into an operation instruction for the fuel consumption system and transmits it to the fuel consumption system.

[0061] The present invention integrates the intelligent optimal control method based on adaptive orthogonal collocation in the MCU. As Figure 2 shown, it internally includes an information acquisition module 21, an initialization module 22, a time scale transformation module 23, an orthogonal collocation module 24, a nonlinear programming (NLP) problem solving module 25, an adaptive processing module 26, and a control instruction output module 27. Corresponding to the control method, each module is used to complete the corresponding steps.

[0062] 1. The information acquisition module includes three sub-modules: the acquisition of the current descent speed of the lunar lander, the acquisition of the altitude reached by the lunar lander, and the acquisition of the performance parameters of the fuel consumption system.

[0063] 2. The time scale transformation module is implemented by the following steps:

[0064] Introduce a new time variable υ such that

[0065] t = (t f - t 0 )υ + t 0 (3)

[0066] Thus, convert the mathematical model (2) with an unfixed end time t f into a mathematical model with a control time domain of [0,1], as follows:

[0067]

[0068] Where,

[0069]

[0070] 3. The orthogonal collocation module is implemented by the following steps:

[0071] Step 1): Represent the control quantity u(t) and the state trajectory x(t) as a linear combination of Lagrange interpolation basis functions, i.e.:

[0072]

[0073] where N is the number of discrete segments of the time axis [t 0 , t f . is the Lagrange interpolation basis function, and the linear combination coefficients u i,j and s i,j are the values of u(t) and x(t) at the Gauss collocation points t i,j respectively.

[0074] Step 2): Since the derivative function expressions of all basis functions are known, the differential equation system of the state trajectory can be expressed in the following form:

[0075]

[0076] Step 3): Replace the original differential equation system with an approximate differential equation system and discretize it at the Gauss collocation points to obtain an approximate algebraic equation. Discretize the objective function and the constraint conditions with u i,j and s i,j to obtain an NLP problem.

[0077] 4. The NLP problem solving module is implemented as follows:

[0078] The vector u to be optimized is evenly divided into NE sub-vectors, that is, In each iteration, only one sub-vector of the vector u to be optimized is updated Each sub-vector is optimized by a sub-population. The update formulas for the velocity and position of the i-th particle in the j-th sub-population in the q-th cycle are:

[0079]

[0080] where, and are the velocities of the i-th particle in the j-th sub-population in the q-th and q - 1-th cycles respectively, and are the positions of the i-th particle in the j-th sub-population in the q - 1-th and q-th cycles respectively, is the individual historical best position of the i-th particle in the j-th sub-population in the q - 1-th cycle, is the population historical best position of the j-th sub-group in the q - 1-th cycle, w (q-1) is the inertia coefficient in the q - 1-th cycle, and are the acceleration coefficients in the q - 1-th cycle, r 1 and r 2is a random number uniformly distributed between 0 and 1. When the particle in the j-th sub-population of the q-th cycle is the main body, the actual number of iterations is q×NE + j generations, which is also denoted as the k-th generation.

[0081] To calculate the objective function value, an environmental vector across sub-populations at the q-th cycle is defined This environmental vector is composed of the population historical optimal positions of each sub-population spliced together, that is

[0082]

[0083] When it is necessary to calculate the objective function of particle i in sub-population j at the q-th cycle, the j-th component of the environmental vector is replaced by the position vector of particle i in sub-population j to obtain an extended position vector at the q-th cycle That is

[0084]

[0085] Similarly, the individual historical optimal position of particle i in sub-population j at the q-th cycle can be extended to

[0086]

[0087] The population historical optimal position of sub-population j at the q-th cycle can be extended to

[0088]

[0089] The position of the i-th particle in the k-th generation is equivalent to The individual historical optimal position of the i-th particle in the k-th generation is equivalent to The population historical optimal position g in the k-th generation (k),best is equivalent to where q×NE + j = k. Therefore, the method for updating the individual historical optimal position of the i-th particle in the q×NE + j (k)-th generation is: if then Otherwise, The method for updating the population historical optimal position in the q×NE + j generation is: select If then Otherwise, where J represents the objective function to be maximized.

[0090] 5. The NLP problem-solving module is implemented as follows:

[0091] Step 1): Analyze the result of the NLP problem-solving module and calculate the slope information σ at time node t k kj

[0092]

[0093] where N T represents the total number of configuration points. u (k+1)j = u j (t k+1 ) represents the value of the jth control component at time t k+1 , and r represents the number of control components. If |σ kj | ≥ σ 0 , then segment the time interval at t k , and approximate each segment using the Lagrange interpolation basis function, where σ 0 is a pre-specified threshold.

[0094] Step 2): Evaluate the violation degree of the optimized solution at the midpoint of the configuration points against the original differential equation system, where the midpoint of the configuration points is defined as:

[0095]

[0096] where τ i and τ i+1 are orthogonal collocation points. Define the error matrix

[0097]

[0098] where X(t) and U(t) are the obtained optimal solutions, and n x is the number of state variables. Define the error comprehensive index v j

[0099]

[0100] where e ij is the element in the i-th row and j-th column of the error matrix E. N s represents the number of collocation points in the s-th segment.

[0101] If v j , (j = 1, …, n x ) is greater than the pre-given threshold ε, then increase the number of configuration points by N add , N add ​is a pre-specified number. If step 1) does not increase the number of segments and step 2) does not increase the number of collocation points, output the solution of the NLP problem solving module; otherwise, recalculate the distribution of collocation points to prepare for the next iteration.

[0102] The present invention will be further described below in conjunction with embodiments.

[0103] The mathematical model of a certain lunar lander's soft landing is as follows:

[0104] min J[u(t)] = x 3 (t f )

[0105]

[0106] x(0) = [10, -2, 0] T

[0107] x 1 (t f ) = 0

[0108] x 2 (t f ) = 0

[0109] 0 ≤ u(t) ≤ 3

[0110] 0 ≤ t ≤ t f

[0111] where x 1 (t), x 2 (t), x 3 (t) respectively represent the distance (m) between the lander and the lunar surface, the descent speed (m / s), and the fuel consumption (kg), and u(t) represents the fuel consumption rate (kg / s).

[0112] The MCU runs an intelligent optimal control method based on adaptive orthogonal collocation, and its running process is as Figure 2 shown:

[0113] Step 1: Initialize the operation of module 22, set the initial grid number N = 1, the initial guess value of the fuel consumption rate control strategy Set the constant value N add = 3, ε = 0.5, set the maximum number of iterations l max = 2 and the termination error tol J = 10 -6 , and let the iteration count l = 0;

[0114] Step 2: Run the time scale transformation module 23 to convert the mathematical model (2) into the form of the mathematical model (4);

[0115] Step 3: Discretize the system of ordinary differential equations over the time axis [t 0 , t f through the orthogonal collocation module 24 to form an NLP problem;

[0116] Step 4: Obtain the required control strategy and corresponding state trajectory through the NLP problem solving module 25, which includes multiple internal iterations. For the angle of attack control quantity u (k) (t) obtained in a certain iteration, if its corresponding objective function value J[u (k) (t)] and the objective function value J[u (k-1) (t)] of the previous iteration differ by less than the precision requirement 10 -6 , then the convergence condition is satisfied, and the result is output to module 27.

[0117] 1. Information acquisition module 21: When the lander starts soft landing, the speed sensor and altitude sensor measure the current descent speed of the lander as -2 m / s and the distance from the lunar surface as 10 m, and transmit these measurement data to the information acquisition module 21 of the MCU.

[0118] 2. Time scale transformation module 23: Introduce a new time variable υ such that

[0119] t = (t f - t 0 )υ + t 0

[0120] Thus, convert the mathematical model with an unfixed end time t f into a mathematical model with a control time domain of [0, 1] as follows:

[0121]

[0122] 0 ≤ υ ≤ 1

[0123] where

[0124]

[0125] 3. Orthogonal collocation module 24: Implement it using the following steps:

[0126] Step 1): Represent the control quantity u(t) and the state trajectory x(t) as a linear combination of Lagrange interpolation basis functions, i.e.:

[0127]

[0128] where N is the number of discrete segments of the time axis [t 0 , t f , is the Lagrange interpolation basis function, and the linear combination coefficient ui,j and s i,j are the values of u(t) and x(t) at the Gauss collocation points t i,j respectively.

[0129] Step 2): Since the derivative function expressions of all basis functions are known, the differential equation system of the state trajectory can be expressed in the following form:

[0130]

[0131] Step 3): Replace the original differential equation system with an approximate differential equation system and discretize it at the Gauss collocation points to obtain an approximate algebraic equation. Discretize the objective function and the constraint conditions with u i,j and s i,j to obtain an NLP problem.

[0132] 4. The NLP problem solving module 25 is implemented by the following steps:

[0133] The vector u to be optimized is evenly divided into NE sub-vectors, that is In each iteration, only one sub-vector of the vector u to be optimized is updated Each sub-vector is optimized by a sub-population. The update formulas for the velocity and position of the i-th particle in the j-th sub-population in the q-th cycle are:

[0134]

[0135] where and are the velocities of the i-th particle in the j-th sub-population in the q-th and q - 1-th cycles respectively, and are the positions of the i-th particle in the j-th sub-population in the q - 1-th and q-th cycles respectively, is the individual historical optimal position of the i-th particle in the j-th sub-population in the q - 1-th cycle, is the population historical optimal position of the j-th sub-sub-population in the q - 1-th cycle, w (q-1) is the inertia coefficient in the q - 1-th cycle, and are the acceleration coefficients in the q - 1-th cycle, r 1 and r 2 are random numbers uniformly distributed between 0 and 1. When the particles in the j-th sub-population in the q-th cycle are the main body, the actual number of iterations is q×NE + j generations, which is also denoted as the k-th generation.

[0136] To calculate the objective function value, define the environmental vector across sub-populations in the q-th cycle The environmental vector is composed of the population historical optimal positions of each sub-population, that is

[0137]

[0138] When it is necessary to calculate the objective function of particle i in sub-population j at the q-th iteration, the j-th component of the environmental vector is replaced by the position vector of particle i in sub-population j to obtain an extended position vector at the q-th iteration, that is

[0139]

[0140] Similarly, the individual historical optimal position of particle i in sub-population j at the q-th iteration can be extended to

[0141]

[0142] The population historical optimal position of sub-population j at the q-th iteration can be extended to

[0143]

[0144] The position of the i-th particle in the k-th generation is equivalent to, and the individual historical optimal position of the i-th particle in the k-th generation is equivalent to. The population historical optimal position g (k),best of the k-th generation is equivalent to, where q×NE + j = k. Therefore, the method for updating the individual historical optimal position of the i-th particle in the q×NE + j (k)-th generation is: if then Otherwise, The method for updating the population historical optimal position of the q×NE + j-th generation is: select If then Otherwise, where J represents the objective function to be maximized.

[0145] 4. The adaptive processing module 26 is implemented by the following steps:

[0146] Step 1): Analyze the result solved by the NLP problem solving module 25 and calculate the slope information σ k at the time node t kj

[0147]

[0148] Among them, N T represents the total number of configuration points. u (k+1)j = u j (t k+1 ) represents the value of the j-th control component at time t k+1 . r represents the number of control components. If |σ kj | ≥ σ 0 , then the time interval is segmented at t k , and each segment is approximated using Lagrange interpolation basis functions, where σ 0 is a pre-specified threshold.

[0149] Step 2): Evaluate the violation degree of the optimized solution at the midpoint of the configuration points for the original differential equation system, where the midpoint of the configuration points is defined as:

[0150]

[0151] Among them, τ i and τ i+1 are orthogonal collocation points. Define the error matrix

[0152]

[0153] where X(t), U(t) are the obtained optimal solutions, and n x is the number of state variables. Define the error comprehensive index v j

[0154]

[0155] where e ij is the element in the i-th row and j-th column of the error matrix E. N s represents the number of configuration points in the s-th segment.

[0156] If v j , (j = 1, …, n x ) is greater than the pre-given threshold ε, then increase N add configuration points, and N add is a pre-specified number. If step 1) does not add new segments and step 2) does not add new configuration points, then output the solution of the NLP problem solving module; otherwise, recalculate the distribution of the configuration points to prepare for the next iteration.

[0157] Finally, the MCU converts the obtained fuel consumption rate control strategy into a fuel consumption system operation instruction and transmits it to the fuel consumption system.

[0158] The above content is a further detailed description of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited only to these descriptions. For those of ordinary skill in the technical field to which the present invention pertains, without departing from the inventive concept, several simple deductions or substitutions can still be made, and all should be regarded as falling within the protection scope of the present invention.

Claims

1. An adaptive high-precision intelligent optimization control method for lunar soft landing trajectory, characterized in that The steps for a lunar lander to softly land on the lunar surface include: Step 1: Use speed sensors and altitude sensors to measure the current descent speed of the lander and the distance from the lunar surface in real time; Step 2: Based on the measured data of step 1, an intelligent optimal control method of adaptive orthogonal configuration is used to obtain a fuel consumption rate control strategy for the lunar lander to land safely and consume the least fuel; Step 3: Use the fuel consumption rate control strategy obtained in step 2 to control the fuel consumption of the lunar lander; The mathematical model derivation process of the fuel consumption rate control strategy is: Describe the soft landing process of the lunar lander as: in: t represents time, t0 represents the start time of the soft landing process, t f represents the end time of the soft landing process, and t f Not fixed; is the state variable of the lunar lander landing process, the superscript T represents the transposition operation; x0 is the initial value of the state variable, is its first-order derivative; u(t) represents the fuel consumption rate, u l 、u u are their lower and upper limits respectively; It is a set of differential equations based on the conservation of energy and the principles of mechanics; is the constraint condition for the physical parameters at the end of the soft landing of the lunar lander, n x and n g They represent the dimensions of the state vector and the dimensions of the constraints respectively. Assume that Φ[x(t f )] represents the total fuel consumption, then the mathematical model that minimizes fuel consumption is expressed as: Where J[u(t)] represents the control target, which is determined by the fuel consumption rate u(t); Formula (2) is an optimal control problem; In step 2: 2.1) Collecting measurement data: The lunar lander's current descent speed, lunar lander's arrival altitude, and fuel consumption system performance parameters; 2.2) Time scale transformation: Introduce a new time variable υ, so that t=(t f -t0)υ+t0 (3) So the end time t f The non-fixed mathematical model (2) is converted into a mathematical model with a control time domain of [0,1] as follows: in, 2.3) Orthogonal configuration, the ordinary differential equations are placed on the time axis [t0,t f ] are all discretized to form an NLP problem, the steps include: Step 2.3.1) The control quantity u(t) and the state trajectory x(t) are expressed as a linear combination of Lagrange interpolation basis functions, that is: Where N is the time axis [t0,t f ] is the number of discrete segments, is the Lagrange interpolation basis function, the linear combination coefficient u i,j and i,j They are u(t) and x(t) at Gauss collocation point t. i,j The value on Step 2.3.2) Since the derivative expressions of all basis functions are known, the differential equations of the state trajectory are expressed as follows: Step 2.3.3): Replace the original differential equations with the approximate differential equations and discretize them at the Gauss collocation points to obtain the approximate algebraic equations; replace the objective function and constraints with u i,j and i,j Perform discrete processing to obtain the nonlinear programming NLP problem; 2.4) Solve the NLP problem and obtain the control strategy and corresponding state trajectory: 2.5) Adaptive processing: Analyze the results of the NLP problem solving module. If the convergence condition is met, the result of step 2.4) is used as the control strategy output; if the convergence condition is not met, return to step 2.3).

2. The adaptive high-precision intelligent optimization control method for lunar soft landing trajectory according to claim 1 is characterized by: In step 2.4): The vector u to be optimized is divided into NE sub-vectors, namely In each iteration, only a subvector of the vector u to be optimized is updated Each subvector is optimized by a subpopulation; The update formulas for the velocity and position of the i-th particle in the j-th subpopulation in the q-th cycle are: in, and are the speed of the i-th particle in the j-th subpopulation in the q-th and q-1-th cycles, respectively. and are the positions of the i-th particle in the j-th subpopulation at the q-1th and qth cycles, respectively. is the individual historical optimal position of the i-th particle in the j-th subpopulation in the q-1th cycle, is the historical optimal position of the jth sub-group in the q-1th cycle, w (q-1) is the inertia coefficient of the q-1th cycle, and is the acceleration coefficient of q-1 cycles, r1 and r2 are random numbers uniformly distributed between 0 and 1; When the particles in the jth subpopulation of the qth cycle are taken as the main body, the number of iterations is q×NE+j generations, also recorded as the kth generation; Define the environment vector across subpopulations in the qth cycle It is composed of the historical optimal positions of each sub-population, that is, When the objective function of particle i in subpopulation j needs to be calculated in the qth cycle, the environment vector The jth component of Replace, and get an extended position vector for the qth cycle Right now Similarly, the individual historical optimal position of particle i in subpopulation j in the qth cycle is Can be expanded to The historical optimal position of subpopulation j in the qth cycle Expand to The position of the i-th particle in the k-th generation and Equivalently, the individual historical optimal position of the i-th particle in the k-th generation is and Equivalently, the historical optimal position g of the population in the kth generation (k),best and Equivalent, where q×NE+j=k; therefore, The method for updating the individual historical optimal position of the i-th particle of the q×NE+j(k) generation is: like but otherwise, The method to update the historical optimal position of the population in the q×NE+j generation is: choose like but otherwise, Here, J represents the objective function to be maximized.

3. The adaptive high-precision lunar soft landing trajectory intelligent optimization control method according to claim 2 is characterized by: In step 2.5): 2.5.1) Analyze the results of the NLP problem solving module and calculate the time node t k The slope information σ kj Among them, N T Indicates the total number of configuration points; u (k+1)j =u j (t k+1 ) indicates that the jth control component is at t k+1 The value at the moment, r represents the number of control components; If |σ kj |≥σ0, then the time interval is t k Disposition segments, each segment is approximated using Lagrange interpolation basis functions, where σ0 is a pre-specified threshold; 2.5.2) Evaluate the degree to which the optimized solution violates the original differential equation system at the midpoint of the configuration point, where: Midpoint of collocation point Defined as: where τ i and τ i+1 are orthogonal collocation points; Defining the error matrix Where X(t) and U(t) are the optimal solutions obtained, respectively. x is the number of state variables; Define the error comprehensive index v j where e ij is the element in the i-th row and j-th column of the error matrix E, N s Indicates the number of collocation points in segment s; If v j ,(j=1,…,n x ) is greater than a pre-given threshold ε, then N is increased add The number of configuration points, N add is a pre-specified number. If step 1) does not add a new number of segments and step 2) does not add a new number of collocation points, then output the solution of the NLP problem solving module; otherwise, recalculate the distribution of collocation points to prepare for the next iteration.

4. An adaptive and high-precision intelligent optimization control system for the lunar soft landing trajectory, characterized by: include: Speed ​​sensor, altitude sensor, lunar lander MCU, fuel consumption system parameter input and fuel consumption system; The lunar lander is softly landing on the lunar surface: Step 1: When the lunar lander is preparing for soft landing, the speed sensor and altitude sensor are turned on to measure the current descent speed of the lander and the distance from the lunar surface in real time, and transmit the measurement data to the lunar lander MCU; Step 2: The lunar lander MCU executes the internal intelligent optimal control method based on adaptive orthogonal configuration to calculate the control strategy that enables the lunar lander to land safely and consume the least fuel; Step 3: The lunar lander MCU converts the obtained fuel consumption rate control strategy into fuel consumption system operation instructions and transmits them to the fuel consumption system.

5. The adaptive high-precision lunar soft landing trajectory intelligent optimization control system according to claim 4 is characterized by: The functional modules of the lunar lander MCU include: information acquisition module, initialization module, time scale transformation module, orthogonal configuration module, nonlinear programming NLP problem solving module, adaptive processing module and control instruction output module; each module corresponds to the step of step 2 of the intelligent optimization control method for the lunar soft landing trajectory.

Images