This paper presents a method of multicopter intercep-tion control based on visual servo and virtual tube in a cluttered environment.The proposed hybrid heuristic function improves the efficiency of the A*algorithm.The...This paper presents a method of multicopter intercep-tion control based on visual servo and virtual tube in a cluttered environment.The proposed hybrid heuristic function improves the efficiency of the A*algorithm.The revised objective function makes the virtual tube generating curve not only smooth but also close to the path points generated by the A*algorithm.In six dif-ferent simulation scenarios,the efficiency of the modified A*algorithm is 6.2%higher than that of the traditional A*algorithm.The efficiency of path planning and virtual tube planning is veri-fied by simulations.The effectiveness of interception control is verified by a software-in-loop(SIL)simulation.展开更多
In this paper,an online midcourse guidance method for intercepting high-speed maneuvering targets is proposed.Firstly,the affine system is used to build a dynamic model and analyze the state constraints.The midcourse ...In this paper,an online midcourse guidance method for intercepting high-speed maneuvering targets is proposed.Firstly,the affine system is used to build a dynamic model and analyze the state constraints.The midcourse guidance problem is transformed into a continuous time optimization problem.Secondly,the problem is transformed into a discrete convex programming problem by affine control variable relaxation,Gaussian pseudospectral discretization and constraints linearization.Then,the off-line midcourse guidance trajectory is generated before midcourse guidance.It is used as the initial reference trajectory for online correction of midcourse guidance.An online guidance framework is used to eliminate the error caused by calculation of guidance instruction time.And the design of discrete points decreases with flight time to improve the solving efficiency.In addition,it is proposed that the terminal guidance capture is used innovatively space to judge the success of midcourse guidance.Numerical simulation shows the feasibility and effectiveness of the proposed method.展开更多
A situation maintenance-based cooperative guidance strategy is proposed to intercept a high-speed and high-maneuverability target via inferior missiles.Reachability and relative motion analyses are conducted to develo...A situation maintenance-based cooperative guidance strategy is proposed to intercept a high-speed and high-maneuverability target via inferior missiles.Reachability and relative motion analyses are conducted to develop and pursue virtual targets,respectively.A two-stage guidance strategy under nonlinear kinematics is developed on the basis of virtual targets.The first stage optimizes the coverage and collision situation by pursuing virtual targets under specific angular constraints.The second stage subsequently intercepts the superior target based on the handover condition optimized by the first stage.Numerical simulation results are provided to compare the effectiveness and superiority of the proposed strategy with those of the reachability-based cooperative strategy(RCS),coverage-based cooperative guidance(CBCG)and augmented proportional navigation(APN)under various maneuvering modes.展开更多
A new proportional navigation(PN) guidance law,called combined proportional navigation(CPN),is proposed.The guidance law is designed to intercept high-speed targets,which is a common case for ballistic targets.The ran...A new proportional navigation(PN) guidance law,called combined proportional navigation(CPN),is proposed.The guidance law is designed to intercept high-speed targets,which is a common case for ballistic targets.The range of target-to-interceptor speed ratio during target interception is derived when guidance laws are applied in high-speed targets interception,and the effectiveness of negative navigation ratio in the PN-based guidance law is proven analytically in some lemmas.Based on the lemmas,the lateral acceleration command of CPN is defined,and the solution to the appearance of singularity in time-varying navigation ratio is given.The simulation results show that CPN can determine headon engagement(as PN) or tail-chase engagement(as RPN) through initial path angle compared with PN and retro proportional navigation(RPN),and can adjust the value of navigation ratio for head-on engagement or tail-chase engagement.Therefore,the capture region of CPN is larger than that of other guidance laws using PN-based methods.展开更多
This paper is mainly on the problem of radiation interception risk control in sensor network for target tracking.Firstly,the sensor radiation interception risk is defined as the product of the interception probability...This paper is mainly on the problem of radiation interception risk control in sensor network for target tracking.Firstly,the sensor radiation interception risk is defined as the product of the interception probability and the cost caused by the interception.Secondly,the radiation interception probability model and cost model are established,based on which the calculation method of interception risk can be obtained.Thirdly,a sensor scheduling model of radiation risk control is established,taking the minimum interception risk as the objective function.Then the Hungarian algorithm is proposed to obtain sensor scheduling scheme.Finally,simulation experiments are mad to prove the effectiveness of the methods proposed in this paper,which shows that compared with the sensor radiation interception probability control method,the interception risk control method can keep the sensor scheduling scheme in low risk as well as protect sensors of importance in the sensor network.展开更多
Background: Within-canopy interception of photosynthetically active radiation(PAR) impacts yield and other agronomic traits in cotton(Gossypium hirsutum L.). Field experiments were conducted to investigate the influen...Background: Within-canopy interception of photosynthetically active radiation(PAR) impacts yield and other agronomic traits in cotton(Gossypium hirsutum L.). Field experiments were conducted to investigate the influence of 6 cotton varieties(they belong to 3 different plant types) on yield, yield distribution, light interception(LI), LI distribution and the relationship between yield formation and LI in Anyang, Henan, in 2014 and 2015.Result: The results showed that cotton cultivars with long branches(loose-type) intercepted more LI than did cultivars with short branches(compact-type), due to increased LI in the middle and upper canopy. Although loose-type varieties had greater LI, they did not yield significantly higher than compact-type varieties, due to decreased harvest index. Therefore, improving the harvest index by adjusting the source-to-sink relationship may further increase cotton yield for loose-type cotton. In addition, there was a positive relationship between reproductive organ biomass accumulation and canopy-accumulated LI, indicating that enhancing LI is important for yield improvement for each cultivar. Furthermore, yield distribution within the canopy was significantly linearly related to vertical LI distribution.Conclusion: Therefore, optimizing canopy structure of different plant type and subsequently optimizing LI distribution within the cotton canopy can effectively enhance the yield.展开更多
Electromagnetic coil launch is an important branch of electromagnetic launch(EML)technology,which is suitable for launching anti-torpedo torpedo(ATT).This paper focuses on the EML parameters and the interception proba...Electromagnetic coil launch is an important branch of electromagnetic launch(EML)technology,which is suitable for launching anti-torpedo torpedo(ATT).This paper focuses on the EML parameters and the interception probability of the EML two ATTs salvo.Based on the launching model of a multi-stage coil launcher,the trajectory model of the ATT and the attacking torpedo,a calculation method for the EML two ATTs salvo parameters is proposed,with the conditions of capture and interception given reasonably.An adaptive particle swarm optimization(APSO)algorithm is proposed to calculate the optimal launching parameters,by designing the adaptive inertia weight and time-varying study factors.According to the analysis of the simulation with Monte Carlo method,EML will improve the interception probability effectively,and the interception probability is affected by the launching range.The results demonstrate good performance of the proposed APSO in calculating EML parameters for the two ATTs salvo in certain combat situation.Implications of these results are particular regarding the command and decision in the anti-torpedo combat.展开更多
This paper investigates the guidance method based on reinforcement learning(RL)for the coplanar orbital interception in a continuous low-thrust scenario.The problem is formulated into a Markov decision process(MDP)mod...This paper investigates the guidance method based on reinforcement learning(RL)for the coplanar orbital interception in a continuous low-thrust scenario.The problem is formulated into a Markov decision process(MDP)model,then a welldesigned RL algorithm,experience based deep deterministic policy gradient(EBDDPG),is proposed to solve it.By taking the advantage of prior information generated through the optimal control model,the proposed algorithm not only resolves the convergence problem of the common RL algorithm,but also successfully trains an efficient deep neural network(DNN)controller for the chaser spacecraft to generate the control sequence.Numerical simulation results show that the proposed algorithm is feasible and the trained DNN controller significantly improves the efficiency over traditional optimization methods by roughly two orders of magnitude.展开更多
The interception probability of a single missile is the basis for combat plan design and weapon performance evaluation,while its influencing factors are complex and mutually coupled.Existing calculation methods have v...The interception probability of a single missile is the basis for combat plan design and weapon performance evaluation,while its influencing factors are complex and mutually coupled.Existing calculation methods have very limited analysis of the influence mechanism of influencing factors,and none of them has analyzed the influence of the guidance law.This paper considers the influencing factors of both the interceptor and the target more comprehensively.Interceptor parameters include speed,guidance law,guidance error,fuze error,and fragment killing ability,while target performance includes speed,maneuverability,and vulnerability.In this paper,an interception model is established,Monte Carlo simulation is carried out,and the influence mechanism of each factor is analyzed based on the model and simulation results.Finally,this paper proposes a classification-regression neural network to quickly estimate the interception probability based on the value of influencing factors.The proposed method reduces the interference of invalid interception data to valid data,so its prediction accuracy is significantly better than that of pure regression neural networks.展开更多
Light attenuation within a row of crops such as cotton is influenced by canopy architecture,which is defined by size,shape and orientation of shoot components.Level of light interception causes an array of morpho-anat...Light attenuation within a row of crops such as cotton is influenced by canopy architecture,which is defined by size,shape and orientation of shoot components.Level of light interception causes an array of morpho-anatomical,physiological and biochemical changes.Physiological determinants of growth include light interception,light use efficiency,dry matter accumulation,duration of growth and dry matter partitioning.Maximum light utilization in cotton production can be attained by adopting cultural practices that yields optimum plant populations as they affect canopy arrangement by modifying the plant canopy components.This paper highlights the extent to which spatial arrangement and density affect light interception in cotton crops.The cotton crop branches tend to grow into the inter-row space to avoid shade.The modification of canopy components suggests a shade avoidance and competition for light.Maximum leaf area index is obtained especially at flowering stage with higher populations which depicts better yields in cotton production.展开更多
To study location problems with congestion demand, a bi-objective model based on flow intercep-tion problem is proposed.The model is formulated from the view of M/M/m queuing systems.Service quantity and quality are s...To study location problems with congestion demand, a bi-objective model based on flow intercep-tion problem is proposed.The model is formulated from the view of M/M/m queuing systems.Service quantity and quality are simultaneously considered as objectives, with constraint on total cost.Service quality includes deviation distance from preplanned trips and customers＇ waiting time.Service quantity is the number of intercep-tion customers.A multi-objective evolutionary algorithm combined with greedy heuristic is proposed.Finally a computational experiment is given, and the algorithm is proved to be efficient.展开更多
To study simultaneous location of different kinds of facilities, a new model of flow interception problem with multi-type of flows is proposed, with the consideration of multi-purpose flows and the influence of facili...To study simultaneous location of different kinds of facilities, a new model of flow interception problem with multi-type of flows is proposed, with the consideration of multi-purpose flows and the influence of facilities on each other. To be more practical, the objective is to maximize the benefits from flow-by customers instead of maximizing flows in the past. Since this problem is NP-hard and there is no optimal solution for large network, greedy and improved greedy heuristic algorithms are proposed. A computational example is presented to show that the optimal location decisions are proved to be strongly dependent on the influence coefficients and average profits by one customer and different types of facilities are almost co-located. Finally the results of sensitivity analysis are reported.展开更多
The traditional tangent impulse interception problem does not consider the influence of actual deviation.However,by taking the actual state deviation of the interceptor into the orbit design process,an interception or...The traditional tangent impulse interception problem does not consider the influence of actual deviation.However,by taking the actual state deviation of the interceptor into the orbit design process,an interception orbit that is more robust than the nominal orbit can be obtained.Therefore,we study the minimum time interception problem and the minimum terminal interception error problem under tangent impulse conditions and give an orbit optimization method that considers the interception time and the interception uncertainty.First,we express the interceptor's transfer time equation as a form of flight path angle,establish a global optimization model for solving the minimum time tangent impulse interception and give a hybrid optimization algorithm based on Augmented Lagrange Genetic Algorithm-Sequential Quadratic Programming(ALGA-SQP).Secondly,we use the universal time equation and Bootstrap resampling technology to calculate the interceptor's terminal error distribution and establish the relevant global optimization model by using the circumscribed cuboid volume of the interceptor's terminal position error ellipsoid as the optimization index.Finally,we combined the above two singleobjective optimization models to establish a global multi-objective optimization model that considers interception time and interception uncertainty and gave a hybrid multi-objective optimization algorithm based on Non-dominated Sorting Genetic Algorithm Ⅱ-Goal Achievement Method(NSGA2-GAM).The simulation example verifies the effectiveness of this method.展开更多
This work proposes a recorded recurrent twin delayed deep deterministic(RRTD3)policy gradient algorithm to solve the challenge of constructing guidance laws for intercepting endoatmospheric maneuvering missiles with u...This work proposes a recorded recurrent twin delayed deep deterministic(RRTD3)policy gradient algorithm to solve the challenge of constructing guidance laws for intercepting endoatmospheric maneuvering missiles with uncertainties and observation noise.The attack-defense engagement scenario is modeled as a partially observable Markov decision process(POMDP).Given the benefits of recurrent neural networks(RNNs)in processing sequence information,an RNN layer is incorporated into the agent’s policy network to alleviate the bottleneck of traditional deep reinforcement learning methods while dealing with POMDPs.The measurements from the interceptor’s seeker during each guidance cycle are combined into one sequence as the input to the policy network since the detection frequency of an interceptor is usually higher than its guidance frequency.During training,the hidden states of the RNN layer in the policy network are recorded to overcome the partially observable problem that this RNN layer causes inside the agent.The training curves show that the proposed RRTD3 successfully enhances data efficiency,training speed,and training stability.The test results confirm the advantages of the RRTD3-based guidance laws over some conventional guidance laws.展开更多
基金supported by the National Natural Science Foundation of China(62303350).
文摘This paper presents a method of multicopter intercep-tion control based on visual servo and virtual tube in a cluttered environment.The proposed hybrid heuristic function improves the efficiency of the A*algorithm.The revised objective function makes the virtual tube generating curve not only smooth but also close to the path points generated by the A*algorithm.In six dif-ferent simulation scenarios,the efficiency of the modified A*algorithm is 6.2%higher than that of the traditional A*algorithm.The efficiency of path planning and virtual tube planning is veri-fied by simulations.The effectiveness of interception control is verified by a software-in-loop(SIL)simulation.
文摘In this paper,an online midcourse guidance method for intercepting high-speed maneuvering targets is proposed.Firstly,the affine system is used to build a dynamic model and analyze the state constraints.The midcourse guidance problem is transformed into a continuous time optimization problem.Secondly,the problem is transformed into a discrete convex programming problem by affine control variable relaxation,Gaussian pseudospectral discretization and constraints linearization.Then,the off-line midcourse guidance trajectory is generated before midcourse guidance.It is used as the initial reference trajectory for online correction of midcourse guidance.An online guidance framework is used to eliminate the error caused by calculation of guidance instruction time.And the design of discrete points decreases with flight time to improve the solving efficiency.In addition,it is proposed that the terminal guidance capture is used innovatively space to judge the success of midcourse guidance.Numerical simulation shows the feasibility and effectiveness of the proposed method.
基金supported by the National Natural Science Foundation of China(Grant No.62203362)the Natural Science Basic Research Program of Shaanxi(Grant No.2023-JC-QN-0569)。
文摘A situation maintenance-based cooperative guidance strategy is proposed to intercept a high-speed and high-maneuverability target via inferior missiles.Reachability and relative motion analyses are conducted to develop and pursue virtual targets,respectively.A two-stage guidance strategy under nonlinear kinematics is developed on the basis of virtual targets.The first stage optimizes the coverage and collision situation by pursuing virtual targets under specific angular constraints.The second stage subsequently intercepts the superior target based on the handover condition optimized by the first stage.Numerical simulation results are provided to compare the effectiveness and superiority of the proposed strategy with those of the reachability-based cooperative strategy(RCS),coverage-based cooperative guidance(CBCG)and augmented proportional navigation(APN)under various maneuvering modes.
文摘A new proportional navigation(PN) guidance law,called combined proportional navigation(CPN),is proposed.The guidance law is designed to intercept high-speed targets,which is a common case for ballistic targets.The range of target-to-interceptor speed ratio during target interception is derived when guidance laws are applied in high-speed targets interception,and the effectiveness of negative navigation ratio in the PN-based guidance law is proven analytically in some lemmas.Based on the lemmas,the lateral acceleration command of CPN is defined,and the solution to the appearance of singularity in time-varying navigation ratio is given.The simulation results show that CPN can determine headon engagement(as PN) or tail-chase engagement(as RPN) through initial path angle compared with PN and retro proportional navigation(RPN),and can adjust the value of navigation ratio for head-on engagement or tail-chase engagement.Therefore,the capture region of CPN is larger than that of other guidance laws using PN-based methods.
基金This article is funded by Chinese national natural science foundation(61573374).
文摘This paper is mainly on the problem of radiation interception risk control in sensor network for target tracking.Firstly,the sensor radiation interception risk is defined as the product of the interception probability and the cost caused by the interception.Secondly,the radiation interception probability model and cost model are established,based on which the calculation method of interception risk can be obtained.Thirdly,a sensor scheduling model of radiation risk control is established,taking the minimum interception risk as the objective function.Then the Hungarian algorithm is proposed to obtain sensor scheduling scheme.Finally,simulation experiments are mad to prove the effectiveness of the methods proposed in this paper,which shows that compared with the sensor radiation interception probability control method,the interception risk control method can keep the sensor scheduling scheme in low risk as well as protect sensors of importance in the sensor network.
基金funded by the National Natural Science Foundation of China(31371561)
文摘Background: Within-canopy interception of photosynthetically active radiation(PAR) impacts yield and other agronomic traits in cotton(Gossypium hirsutum L.). Field experiments were conducted to investigate the influence of 6 cotton varieties(they belong to 3 different plant types) on yield, yield distribution, light interception(LI), LI distribution and the relationship between yield formation and LI in Anyang, Henan, in 2014 and 2015.Result: The results showed that cotton cultivars with long branches(loose-type) intercepted more LI than did cultivars with short branches(compact-type), due to increased LI in the middle and upper canopy. Although loose-type varieties had greater LI, they did not yield significantly higher than compact-type varieties, due to decreased harvest index. Therefore, improving the harvest index by adjusting the source-to-sink relationship may further increase cotton yield for loose-type cotton. In addition, there was a positive relationship between reproductive organ biomass accumulation and canopy-accumulated LI, indicating that enhancing LI is important for yield improvement for each cultivar. Furthermore, yield distribution within the canopy was significantly linearly related to vertical LI distribution.Conclusion: Therefore, optimizing canopy structure of different plant type and subsequently optimizing LI distribution within the cotton canopy can effectively enhance the yield.
基金National Natural Science Foundation of China (Grant No. 51777212)
文摘Electromagnetic coil launch is an important branch of electromagnetic launch(EML)technology,which is suitable for launching anti-torpedo torpedo(ATT).This paper focuses on the EML parameters and the interception probability of the EML two ATTs salvo.Based on the launching model of a multi-stage coil launcher,the trajectory model of the ATT and the attacking torpedo,a calculation method for the EML two ATTs salvo parameters is proposed,with the conditions of capture and interception given reasonably.An adaptive particle swarm optimization(APSO)algorithm is proposed to calculate the optimal launching parameters,by designing the adaptive inertia weight and time-varying study factors.According to the analysis of the simulation with Monte Carlo method,EML will improve the interception probability effectively,and the interception probability is affected by the launching range.The results demonstrate good performance of the proposed APSO in calculating EML parameters for the two ATTs salvo in certain combat situation.Implications of these results are particular regarding the command and decision in the anti-torpedo combat.
基金supported by the National Defense Science and Technology Innovation(18-163-15-LZ-001-004-13).
文摘This paper investigates the guidance method based on reinforcement learning(RL)for the coplanar orbital interception in a continuous low-thrust scenario.The problem is formulated into a Markov decision process(MDP)model,then a welldesigned RL algorithm,experience based deep deterministic policy gradient(EBDDPG),is proposed to solve it.By taking the advantage of prior information generated through the optimal control model,the proposed algorithm not only resolves the convergence problem of the common RL algorithm,but also successfully trains an efficient deep neural network(DNN)controller for the chaser spacecraft to generate the control sequence.Numerical simulation results show that the proposed algorithm is feasible and the trained DNN controller significantly improves the efficiency over traditional optimization methods by roughly two orders of magnitude.
基金supported by the Foundation Strengthening Program Technology Field Foundation(2020-JCJQ-JJ-132)。
文摘The interception probability of a single missile is the basis for combat plan design and weapon performance evaluation,while its influencing factors are complex and mutually coupled.Existing calculation methods have very limited analysis of the influence mechanism of influencing factors,and none of them has analyzed the influence of the guidance law.This paper considers the influencing factors of both the interceptor and the target more comprehensively.Interceptor parameters include speed,guidance law,guidance error,fuze error,and fragment killing ability,while target performance includes speed,maneuverability,and vulnerability.In this paper,an interception model is established,Monte Carlo simulation is carried out,and the influence mechanism of each factor is analyzed based on the model and simulation results.Finally,this paper proposes a classification-regression neural network to quickly estimate the interception probability based on the value of influencing factors.The proposed method reduces the interference of invalid interception data to valid data,so its prediction accuracy is significantly better than that of pure regression neural networks.
基金Source of funding for compiling this review paper is the Department of Research and Specialist Services through the Cotton Research Institute,Zimbabwe.
文摘Light attenuation within a row of crops such as cotton is influenced by canopy architecture,which is defined by size,shape and orientation of shoot components.Level of light interception causes an array of morpho-anatomical,physiological and biochemical changes.Physiological determinants of growth include light interception,light use efficiency,dry matter accumulation,duration of growth and dry matter partitioning.Maximum light utilization in cotton production can be attained by adopting cultural practices that yields optimum plant populations as they affect canopy arrangement by modifying the plant canopy components.This paper highlights the extent to which spatial arrangement and density affect light interception in cotton crops.The cotton crop branches tend to grow into the inter-row space to avoid shade.The modification of canopy components suggests a shade avoidance and competition for light.Maximum leaf area index is obtained especially at flowering stage with higher populations which depicts better yields in cotton production.
基金supported by the National Natural Science Foundation of China (70601011, 70871044)Trans-Century Training Program Foundation for the Talents by the State Education Commission (NCET-06-0653)
文摘To study location problems with congestion demand, a bi-objective model based on flow intercep-tion problem is proposed.The model is formulated from the view of M/M/m queuing systems.Service quantity and quality are simultaneously considered as objectives, with constraint on total cost.Service quality includes deviation distance from preplanned trips and customers＇ waiting time.Service quantity is the number of intercep-tion customers.A multi-objective evolutionary algorithm combined with greedy heuristic is proposed.Finally a computational experiment is given, and the algorithm is proved to be efficient.
基金supported by the National Natural Science Foundation of China(7060101170871044)Trans-Century Traning Program Foundation for the Talents by the State Education Commission(NECT-06-0653).
文摘To study simultaneous location of different kinds of facilities, a new model of flow interception problem with multi-type of flows is proposed, with the consideration of multi-purpose flows and the influence of facilities on each other. To be more practical, the objective is to maximize the benefits from flow-by customers instead of maximizing flows in the past. Since this problem is NP-hard and there is no optimal solution for large network, greedy and improved greedy heuristic algorithms are proposed. A computational example is presented to show that the optimal location decisions are proved to be strongly dependent on the influence coefficients and average profits by one customer and different types of facilities are almost co-located. Finally the results of sensitivity analysis are reported.
文摘The traditional tangent impulse interception problem does not consider the influence of actual deviation.However,by taking the actual state deviation of the interceptor into the orbit design process,an interception orbit that is more robust than the nominal orbit can be obtained.Therefore,we study the minimum time interception problem and the minimum terminal interception error problem under tangent impulse conditions and give an orbit optimization method that considers the interception time and the interception uncertainty.First,we express the interceptor's transfer time equation as a form of flight path angle,establish a global optimization model for solving the minimum time tangent impulse interception and give a hybrid optimization algorithm based on Augmented Lagrange Genetic Algorithm-Sequential Quadratic Programming(ALGA-SQP).Secondly,we use the universal time equation and Bootstrap resampling technology to calculate the interceptor's terminal error distribution and establish the relevant global optimization model by using the circumscribed cuboid volume of the interceptor's terminal position error ellipsoid as the optimization index.Finally,we combined the above two singleobjective optimization models to establish a global multi-objective optimization model that considers interception time and interception uncertainty and gave a hybrid multi-objective optimization algorithm based on Non-dominated Sorting Genetic Algorithm Ⅱ-Goal Achievement Method(NSGA2-GAM).The simulation example verifies the effectiveness of this method.
基金supported by the National Natural Science Foundation of China(Grant No.12072090)。
文摘This work proposes a recorded recurrent twin delayed deep deterministic(RRTD3)policy gradient algorithm to solve the challenge of constructing guidance laws for intercepting endoatmospheric maneuvering missiles with uncertainties and observation noise.The attack-defense engagement scenario is modeled as a partially observable Markov decision process(POMDP).Given the benefits of recurrent neural networks(RNNs)in processing sequence information,an RNN layer is incorporated into the agent’s policy network to alleviate the bottleneck of traditional deep reinforcement learning methods while dealing with POMDPs.The measurements from the interceptor’s seeker during each guidance cycle are combined into one sequence as the input to the policy network since the detection frequency of an interceptor is usually higher than its guidance frequency.During training,the hidden states of the RNN layer in the policy network are recorded to overcome the partially observable problem that this RNN layer causes inside the agent.The training curves show that the proposed RRTD3 successfully enhances data efficiency,training speed,and training stability.The test results confirm the advantages of the RRTD3-based guidance laws over some conventional guidance laws.
