Airborne radar optical multi-load multi-constraint task planning and cooperative scheduling method

Abstract

The invention provides an airborne radar optical multi-load multi-constraint task planning and cooperative scheduling method, which identifies complex constraint conditions of a single sensor and multiple sensors, establishes a resource scheduling evaluation index, designs an optimization algorithm for problem characteristics, and solves the problem of multi-sensor cooperative detection task planning under airborne complex multi-constraint conditions. According to the method, radar load detection task planning, infrared optical load detection task planning, visible light load detection task planning and radar and optical multi-type load cooperative detection task planning can be realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of airborne mission planning and collaborative control, and in particular relates to an airborne radar optical multi-load multi-constraint mission planning and collaborative scheduling method. Background Art

[0002] The planning and collaborative scheduling of airborne tasks are of great significance. For example, the Chinese patent application number is 202111628548.7, entitled An aviation platform with multi-dimensional information detection capabilities and a collaborative operation method, which discloses a planning and management subsystem. Input the parameters of each operating remote sensing payload, design the starting and ending points of the routes of each remote sensing payload according to factors such as the operating area and user requirements, and according to imaging requirements such as overlap, and estimate the flight distance and efficiency to realize a multi-sensor collaborative operation method. However, the planned route requires that the radar can only enter from one direction of the route, and the path planning is not flexible. Moreover, the invention lacks the flight distance and efficiency of the route, and lacks a detailed description of the planning method. Summary of the invention

[0003] The purpose of the present invention is to solve the problems existing in the prior art and to provide an airborne radar optical multi-load multi-constraint task planning and collaborative scheduling method suitable for manned / unmanned rotor and fixed-wing airborne conditions.

[0004] The present invention transforms the problem of airborne radar optical multi-load multi-constraint task planning and coordinated scheduling into an optimization problem, whose optimization goal is to maximize the efficiency of resource utilization. The constraints include sensor constraints, flight platform constraints and environmental constraints, and the design variables are vectors composed of three-dimensional coordinates of multiple imaging positions of the sensor. Through problem transformation, the optimization algorithm is used to simplify the problem-solving process. The constraints of the present invention are comprehensive, and the optimization results are closer to the actual applicable conditions. The optimization algorithm used in the present invention has made a targeted design for convergence performance, improved the convergence speed and ensured the global search capability.

[0005] The present invention is oriented to airborne radar and optical detection payloads, and regards the task planning and collaborative scheduling process of collaborative detection of multiple sensors such as radar and optics as a dynamic optimization process with multiple constraints.

[0006] Furthermore, the constraints of the dynamic optimization process are to meet the constraints of imaging position, observation angle, baseline length and threat area for effective detection of the target, and the optimization goal is the efficiency of resource use, planning the detection track for a given target. Under the premise that the working range of the airborne platform has the temporal and spatial coverage of the target, the use efficiency of each sensor is maximized. According to the optimization results, the existing sensor resources and the flight target status (such as trajectory position information, etc.) are known, and the working plan of each sensor is determined, its tasks are formulated, and the sensor detection resources are reasonably allocated in each time period.

[0007] In the present invention, resource utilization efficiency is defined as a function of the detection track path length. Maximizing the utilization efficiency corresponds to minimizing the detection track path length. The detection path needs to comprehensively consider constraints such as the range and imaging of radar and optical payloads, so it belongs to the collaborative scheduling problem of complex and multi-constrained multiple payloads.

[0008] The design variables of the present invention are mainly the three-dimensional coordinates of each imaging on the entire detection path.

[0009] The constraints of the present invention mainly include sensor constraints, flight platform constraints and environmental constraints, among which sensor constraints include sensor performance constraints and imaging constraints, flight platform constraints mainly include dynamic constraints and range constraints of the flight platform, and environmental constraints mainly include obstacles and threats in the working environment. The performance constraints of the sensor itself mainly include the effective distance (detection range), field of view angle, etc., and the imaging constraints mainly include baseline length, flight strip length, etc.

[0010] The overall process of the multi-sensor task resource scheduling algorithm: First, the design variable vector is formed according to the preset number of imaging times, and then the value range of the design variable is obtained according to the relative position relationship between the starting point and the target area. Then, the constraints of the optimization problem are constructed according to the performance of each payload equipment, the flight platform conditions and the external environmental conditions. Then, the design variables, the variable value range, the constraints, and the initial parameters of the algorithm are input into the optimization algorithm. After the algorithm iterative operation meets the termination conditions, the variable design results are obtained, and finally the flight route is formed according to the variable design results.

[0011] After inputting the design variables, variable value ranges, constraints, and initial algorithm parameters into the optimization algorithm, the algorithm will be initialized first. The initial algorithm parameters mainly include the design variable dimension Dim, the population size Size, the maximum number of iterations Iter, the algorithm exit criterion Cret, the initial value of the inertia weight W0, the maximum value of the inertia weight Wmax, the minimum value of the inertia weight Wmin, the initial value of the local learning factor C10, the maximum value of the local learning factor C1max, the minimum value of the local learning factor C1min, the initial value of the global learning factor C20, the maximum value of the global learning factor C2max, and the minimum value of the global learning factor C2min.

[0012] Further, the optimization algorithm workflow is as follows:

[0013] First, the algorithm is initialized according to its initial parameters, a population is generated according to the population size, and variables are initialized for each individual in the population to ensure that the variable values ​​meet the boundary conditions of the variable values.

[0014] Then, the function is evaluated. Each design variable of the population is substituted into the objective function to calculate its result. Then, it is substituted into the constraint condition and the extent of violation of the constraint condition is calculated. The extent of violation is substituted into the penalty function to calculate its result. The objective function and penalty function calculations are combined and superimposed to obtain the fitness value corresponding to the design variable.

[0015] Each individual in the population updates its historical minimum fitness value and the corresponding design variables. Specifically, each individual in the population vertically compares the fitness value obtained this time with the fitness values ​​obtained in previous generations to find the minimum fitness value and the corresponding design variables.

[0016] The population updates the global minimum fitness value and the corresponding design variables. Specifically, each individual in the population compares the global minimum fitness value and the corresponding design variables of this generation with the global fitness value of the previous generation, and updates the global minimum fitness value and the corresponding design variables.

[0017] Next, the algorithm exit criteria are judged, and the algorithm continues to execute when the exit criteria are not met. The population calculates and updates the inertia weight factor, local learning factor, and global learning factor according to the optimization strategy.

[0018] According to the updated inertia weight factor, local learning factor and global learning factor, each individual calculates its new velocity value (design variable change) and new position value (design variable new value).

[0019] The process returns to the function evaluation step, and the new position value of the population individual (new value of the design variable) is brought into the objective function and constraint conditions for calculation, and finally the new objective function value and the development function value are obtained to form a new fitness value. Further update the individual minimum fitness value and the global minimum fitness value, and determine the exit criteria to confirm whether to update the variable. When the exit criteria are met, the algorithm terminates and outputs the final optimization result.

[0020] The design variables of the present invention are the three-dimensional coordinates of each imaging position. Based on the design results, the starting point is added to form all the key waypoints on the entire route. In order to make the entire route smoother, the Clothoid algorithm is used to smooth the route of the two adjacent key waypoints input to obtain the final route, completing multi-load multi-constraint task planning and collaborative scheduling.

[0021] The beneficial effects of the present invention are as follows:

[0022] 1. The present invention proposes an airborne radar optical multi-load multi-constraint task planning and collaborative scheduling method, which transforms the problem into an optimization problem, whose optimization goal is to maximize the efficiency of resource utilization, and the constraints include sensor constraints, flight platform constraints and environmental constraints. The design variables are vectors composed of the three-dimensional coordinates of multiple imaging positions of the sensor. Through problem transformation, an optimization algorithm is used to simplify the problem solving process.

[0023] 2. The present invention proposes a path planning constraint modeling method, which decomposes the constraints of task planning and collaborative scheduling problems into sensor constraints, flight platform constraints and environmental constraints. Sensor constraints include sensor performance constraints (including range requirements and field of view requirements) and sensor imaging condition constraints (including flight strip length requirements and baseline length requirements), flight platform constraints include dynamic constraints and range constraints, and environmental constraints include obstacle range and threat range. The constraint decomposition is comprehensive, and the optimization design results are closer to actual usage conditions.

[0024] 3. The present invention proposes an improved particle swarm optimization algorithm strategy. Aiming at complex multi-constraint optimization problems and the slow convergence speed, an improved particle swarm optimization algorithm strategy is proposed. The inertia weight factor is updated by a nonlinear change strategy, and the global learning factor is updated by a nonlinear change strategy with the weight. The global learning factor is updated by a nonlinear change strategy with the weight, which improves the convergence speed and ensures good global search capability without falling into the local optimum. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The accompanying drawings constituting a part of the present application are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention:

[0026] Figure 1 It is a multi-sensor task resource scheduling process algorithm flow chart of an airborne radar optical multi-load multi-constraint task planning and collaborative scheduling method according to an embodiment of the present invention;

[0027] Figure 2 It is a constraint condition of the embodiment of the present invention;

[0028] Figure 3 It is an overall workflow diagram of the optimization algorithm of an embodiment of the present invention;

[0029] Figure 4 Schematic diagram of the algorithm initialization process used in the embodiment of the present invention;

[0030] Figure 5 It is a schematic diagram of the algorithm function evaluation and fitness value calculation principle adopted by the present invention;

[0031] Figure 6 It is a schematic diagram of the algorithm variable iterative updating principle adopted by the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0033] This embodiment provides an airborne radar optical multi-payload multi-constraint task planning and coordinated scheduling method, the steps are as follows:

[0034] Step 1: In order to solve the problem of airborne radar optical multi-load and multi-constraint task planning and coordinated scheduling, the present invention transforms the problem into an optimization problem. The optimization goal is to maximize the efficiency of resource utilization. The constraints include sensor constraints, flight platform constraints and environmental constraints. The design variables are vectors composed of the three-dimensional coordinates of multiple imaging positions of the sensor.

[0035] Step 1-1: Determine the design variable X = [x1, y1, z1, x2, y2, z2, ..., xn, yn, zn] according to the required number of imaging times, where n is the number of imaging times.

[0036] Step 1-2: The design constraints of the present invention include sensor constraints, flight platform constraints and environmental constraints. The sensor constraints include sensor performance constraints (including range requirements and field of view requirements) and sensor imaging condition constraints (including flight strip length requirements and baseline length requirements), the flight platform constraints include dynamic constraints and range constraints, and the environmental constraints include obstacle range and threat range. The constraints are formed as follows:

[0037] Rg>Rg,min

[0038] Rg≤ri,min,i=0,1,…,Ns

[0039] B=2Rgsin(α / 2)≥Bmin

[0040] Bmin=max{Bmin,i},i=0,1,…,Ns

[0041] αmin=max{αmin,i},i=0,1,…,Ns

[0042] Li=2Rgtan(β / 2)≥Lmin

[0043] Lmin=max{Lmin,i},i=0,1,…,Ns

[0044] S≤Smax

[0045] θ≤θmax

[0046]

[0047] Where Rg is the ground distance between the flight platform and the target, Rg,min is the minimum ground distance to the target, and ri,min is the operating distance of sensor i. B is the baseline between flight strips, Bmin is the minimum baseline length that must be met, Bmin,i is the minimum baseline length required to be met by sensor i, αmin is the center angle corresponding to the limit, and αmin,i is the center angle corresponding to the baseline required to be met by sensor i. Li is the length of the i-th flight strip, β is the center angle corresponding to the i-th flight strip, Lmin is the minimum flight strip length, and Lmin,i is the minimum flight strip length required by sensor i. S is the total length of the planned path, and Smax is the maximum and range of the flight platform. θ is the angle between two adjacent flight strips, and θmax is the maximum angle between two adjacent flight strips allowed by the dynamic constraints of the flight platform. Ω is the obstacle range and threat range, and the design variables should not fall into this area.

[0048] Step 1-3: The design objective function of the present invention is y=max E, where E is the resource utilization efficiency and E=Smax-S

[0049] Step 2: Input the design variables, variable value ranges, constraints, and initial algorithm parameters into the particle swarm optimization algorithm to initialize the algorithm.

[0050] Step 2-1: Select the initial parameters of the optimization algorithm, mainly including the design variable dimension Dim, population size Size, maximum number of iterations Iter, algorithm exit criterion Cret, inertia weight initial value W0, inertia weight maximum value Wmax, inertia weight minimum value Wmin, local learning factor initial value C10, local learning factor maximum value C1max, local learning factor minimum value C1min, global learning factor initial value C20, global learning factor maximum value C2max, global learning factor minimum value C2min. The parameters are selected as follows:

[0051] The particle swarm algorithm dimension of this embodiment is Dim=3*n;

[0052] The particle swarm algorithm population size in this embodiment is Size=80;

[0053] The maximum number of iterations of the particle swarm algorithm in this embodiment is Iter=100;

[0054] The particle swarm algorithm in this embodiment uses a combination of the maximum number of iterations Iter and the maximum number of function evaluations to make a judgment, where the maximum number of evaluations Emax ≥ 5000;

[0055] The initial value of the inertia weight factor of the particle swarm algorithm in this embodiment is W0, the maximum value is Wmax, and the minimum value is Wmin;

[0056] The initial value of the local learning factor of the particle swarm algorithm in this embodiment is C10, the maximum value is C1max, and the minimum value is C1min;

[0057] The initial value of the global learning factor of the particle swarm algorithm in this embodiment is C20, the maximum value is C2max, and the minimum value is C2min;

[0058] Step 2-2: The penalty function of this embodiment is the magnitude of the violation of the constraint condition in the design variable iteration process multiplied by the penalty coefficient, and then superimposed. Specifically, PEN = ΣKi*CONi, i = 1, 2, ..., Nc, where PEN is the penalty value, Ki is the penalty coefficient of constraint i, CONi is the violation magnitude of constraint i, and Nc is the number of constraints.

[0059] Step 2-3: The fitness value designed in this embodiment is the mathematical sum of the objective function calculation value and the penalty function calculation value, that is, FIT=OBJ+PEN, where FIT is the fitness value, OBJ is the objective function calculation value, and PEN is the penalty function calculation value.

[0060] Step 2-4: Initialize the population according to the initial parameters of the algorithm and generate the population and variables.

[0061] Step 3: Determine whether the algorithm exit conditions are met. If so, exit and output the optimization results; if not, evaluate the current design variables and update them.

[0062] Step 3-1: Determine whether the exit condition is met. This embodiment uses a combination of the maximum number of iterations Iter and the maximum number of function evaluations to make the judgment, where the maximum number of evaluations is Emax.

[0063] Step 3-2: If the exit condition is not met, evaluate the current design variables. Substitute the variables of each individual in the population into the objective function and constraints, calculate the objective function value, the magnitude of violation of the constraints, further obtain the penalty function value, and then find the fitness value.

[0064] Step 3-3: This embodiment adopts a nonlinear change strategy to update the inertia weight factor, and the update method is W(k)=Wmax+(Wmax-Wmin)*exp(-25*(k / Iter)), where k is the current iteration number.

[0065] Step 3-4: This embodiment adopts a strategy that changes nonlinearly with the weight to update the local learning factor, and the updating method is c1(k)=A*W^2+B*W+C, where A, B, and C are constant coefficients, and the values ​​are taken according to the test experience of typical test functions.

[0066] Step 3-5: This embodiment adopts a strategy of nonlinear change with the weight to update the global learning factor, and the updating method is c2(k)=2.5-c1(k).

[0067] Step 3-6: Traverse each individual in the population and update its velocity value. The updating method is vi(k+1)=W(k)*vi(k)+C1(k)*rnd1*(pbesti(k)-xi(k))+C2(k)*rnd2*(gbest(k)-xi(k)), where pbesti is the optimal position in the iterative history of particle i, and gbest is the optimal position in the iterative history of the entire population.

[0068] Step 3-7: Traverse each particle in the algorithm and update its position, xi(k+1)=xi(k)+vi(k+1).

[0069] Step 3-8: Repeat steps 3-1 to 3-7, iterate and update continuously, and when the exit condition is met, proceed to step 4.

[0070] Step 4: When the exit condition is met, the particle swarm algorithm optimization results are output. The coordinates of each imaging position can be known from the optimization results, and all waypoints can be obtained by superimposing the initial points.

[0071] Step 5: Use the Clothoid algorithm to traverse two adjacent points. Input the coordinates of two locations to generate a smooth curve as the route between the two locations. After all iterations are completed, a collection of multiple routes is formed, completing the airborne radar optical multi-load multi-constraint mission planning and collaborative scheduling.

[0072] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. An airborne radar optical multi-load multi-constraint task planning and collaborative scheduling method, characterized in that: Here are the steps: a. Forming a design variable vector according to a preset number of imaging times; b. Obtain the value range of the design variable based on the relative position relationship between the starting point and the target area; c. Construct the constraints of the optimization problem based on the performance of each payload, flight platform conditions and external environmental conditions; d. Input the design variables, variable value ranges, constraints, and initial algorithm parameters into the optimization algorithm, and obtain the variable design results after the algorithm iterative operation meets the termination conditions; e. Form a flight route based on variable design results.

2. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 1 is characterized in that: The initial parameters of the algorithm include the design variable dimension Dim, the population size Size, the maximum number of iterations Iter, the algorithm exit criterion Cret, the initial value of the inertia weight W0, the maximum value of the inertia weight Wmax, the minimum value of the inertia weight Wmin, the initial value of the local learning factor C10, the maximum value of the local learning factor C1max, the minimum value of the local learning factor C1min, the initial value of the global learning factor C20, the maximum value of the global learning factor C2max, and the minimum value of the global learning factor C2min.

3. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 1 is characterized in that: The optimization algorithm steps are as follows: a. First, initialize according to the initial parameters of the algorithm, generate a population according to the population size, initialize variables for each individual in the population, and ensure that the variable values ​​meet the boundary conditions of the variable values; b. Perform function evaluation, bring each design variable of the population into the objective function to calculate its result, then bring it into the constraint condition and calculate the extent of violation of the constraint condition, bring the extent of violation into the penalty function to calculate its result, combine and superimpose the objective function and penalty function calculations to obtain the fitness value corresponding to the design variable; c. Each individual in the population updates its historical minimum fitness value and the corresponding design variable. Specifically, each individual in the population compares the fitness value obtained this time with the fitness values ​​obtained in previous generations to find the minimum fitness value and the corresponding design variable. d. The population updates the global minimum fitness value and the corresponding design variables. Specifically, each individual in the population compares the global minimum fitness value and the corresponding design variables of this generation with the global fitness value of the previous generation, and updates the global minimum fitness value and the corresponding design variables; e. Determine the algorithm exit criteria and continue to execute the algorithm when the exit criteria are not met; f. The population calculates and updates the inertia weight factor, local learning factor and global learning factor according to the optimization strategy; based on the updated inertia weight factor, local learning factor and global learning factor, each individual calculates its new speed value and new position value; g. The process returns to step b, and the new position values ​​of the individuals in the population are brought into the objective function and constraints for calculation, and finally the new objective function value and the emission function value are obtained to form a new fitness value; the individual minimum fitness value and the global minimum fitness value are further updated, and the exit criteria are judged to confirm whether to update the variables; when the exit criteria are met, the algorithm terminates and outputs the final optimization result.

4. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 3 is characterized in that: The penalty function is the amplitude of the violation of the constraint condition in the iterative process of the design variable multiplied by the penalty coefficient, and then superimposed.

5. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 3 is characterized in that: The inertia weight factor is updated using a nonlinear change strategy, the local learning factor is updated using a nonlinear change strategy with the weight, and the global learning factor is updated using a nonlinear change strategy with the weight.

6. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 1, characterized in that: The design variables are the three-dimensional coordinates of each imaging on the entire detection path.

7. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 1 is characterized in that: The constraints include sensor constraints, flight platform constraints and environmental constraints, among which sensor constraints include sensor performance constraints and imaging constraints, flight platform constraints include dynamic constraints and range constraints of the flight platform, and environmental constraints include obstacles and threats in the working environment; sensor performance constraints include effective range and field of view angle, and imaging constraints include baseline length and flight strip length.

8. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 1 is characterized in that: In step e, the Clothoid algorithm is used to smooth the route of two adjacent key waypoints input to obtain the final route.

9. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 1, characterized in that: Transformed into an optimization problem, the optimization goal is to maximize the efficiency of resource utilization.

10. The method for planning and coordinating airborne radar optical multi-load and multi-constraint tasks according to claim 1, characterized in that: It can realize the detection mission planning of radar payload, the detection mission planning of infrared optical payload, the detection mission planning of visible light payload, and the coordinated detection mission planning of radar and optical multi-type payloads.

Images