Dynamic arc consistent time constraint reasoning method for deep space probe mission planning

Abstract

The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning disclosed by the present invention belongs to the field of aerospace technology. The implementation method of the present invention is as follows: establish a time planning problem model for deep space probes; before performing time constraint reasoning, represent the activity time variables and time constraints as variable points and edges in a simple time network (STN), then initialize the time constraint network, and dynamically add the time variables and time constraints of newly added activities in the plan to this network; judge the constraint propagation through the variable value domain update list, only calculate the new value domain for the variables within the influence range of the new constraints, and limit the constraint propagation by whether the calculated value domain is reduced or not, narrow the scope of constraint propagation, reduce the number of time constraint reasoning times, save the limited computing resources of deep space probes, improve the efficiency of complex time constraint reasoning, improve the efficiency of deep space probe mission planning, and enhance the real-time performance of mission planning for deep space probes in case of emergencies.

Description

Technical Field

[0001] The present invention relates to a dynamic arc-consistent time constraint reasoning method for deep space probe mission planning, belonging to the field of aerospace technology. Background Art

[0002] During the space flight of a deep space probe, the distance to the target is far, the flight time is long, the environment is not fully known, and the dynamic influence is large. The long communication delay caused by the long distance makes it difficult for the traditional ground station control mode to meet the real-time requirements of the mission. The key to solving this problem is that the deep space probe has the ability of autonomous operation, and the autonomous mission planning technology is one of the key technologies to realize the autonomous operation of the deep space probe.

[0003] To realize the autonomous mission planning of the probe, one of the cores is to express the time constraints between the executable activities of the probe system, and to perform constraint reasoning on the newly added time constraints during the planning process to determine whether the constraint consistency requirements are met. In terms of time constraint representation, most use the Simple Time Network (STN) to quantitatively represent time variables and time constraints, and perform time constraint reasoning on this basis. Currently, this method is adopted in both the Remote Agent planning software carried by the "Deep Space 1" probe and the EUROPA planning and scheduling platform developed by NASA. Currently, the time constraint reasoning methods mainly include two categories: path consistency and arc consistency. The most typical path consistency method is the Floyd-Warshall algorithm, which calculates the shortest path of the entire network based on the STN, and its time complexity is O(n 3 )). This method has a large number of constraint reasoning operations and low calculation efficiency. Therefore, Bliek proposed the Partial Path Consistency (PPC) algorithm, which effectively reduces the number of constraint reasoning operations. The ΔSTP algorithm combines PPC and the triangulation of the STN graph, and improves the calculation efficiency at the cost of storage resources, with a time complexity of O(t 2 ). P 3 C algorithm enforces partial path consistency by combining the directed path consistency (PPC) on the triangulated STN. Each triangle in the constraint network only needs to perform two constraint reasoning operations, further improving the efficiency of time constraint consistency reasoning, and its time complexity is O(t). The arc consistency method determines whether there is a solution to the problem under the current constraints by tightening the value range of time variables through constraint propagation. The main methods include AC-1 to AC-7, AC-2000, AC-2001, ACSTP, etc.

[0004] However, in the dynamic planning process, for the above methods, every time an activity is added, it is necessary to perform a traversal calculation on all time variables of the current network. As the constraints increase, the computational complexity of time constraints will increase sharply, thus affecting the planning efficiency. To address this issue, some scholars have proposed incremental maintenance methods for time constraint networks. For example, the IFPC algorithm maintains global path consistency by updating part of the network; the IAPSP algorithm is an incremental variant based on the APSP algorithm; the IPPC algorithm maintains the current network by keeping partial path consistency when adding new constraints or tightening existing constraints. Most of the above incremental methods are based on path consistency, and new constraints may be added additionally during the calculation process, thus increasing the computational cost. Summary of the Invention

[0005] The technical problems to be solved by the dynamic arc-consistent time constraint reasoning method for deep space probe mission planning disclosed by the present invention are as follows: (1) Establish a problem model for deep space probe time planning; (2) Equivalently represent the time constraint problem as a simple time network; (3) Initialize the time constraint network and dynamically add the time constraints of newly added activities in the planning to this network; (4) Based on the arc-consistent algorithm, perform dynamic constraint reasoning on the time constraint network after adding new constraints, so as to determine whether the current planning meets the time constraint consistency. The present invention can realize dynamic time constraint reasoning for deep space probes based on arc consistency, reduce the number of time constraint reasoning times, improve the efficiency of complex time constraint reasoning, and further improve the autonomous mission planning efficiency of deep space probes, thereby enhancing the real-time performance of mission planning for deep space probes in case of emergencies.

[0006] The object of the present invention is achieved by the following technical solutions.

[0007] The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning disclosed by the present invention establishes a problem model for deep space probe time planning, including probe states, system executable activity time variables, time constraints of activities, initial states, and target states. Before performing time constraint reasoning, the activity time variables and time constraints are represented as variable points and edges in a simple time network (STN). Then, the time constraint network is initialized, and the time variables and time constraints of newly added activities in the planning are dynamically added to this network. The constraint propagation is judged through a variable value range update list, and the new value range is calculated only for the variables within the influence range of the new constraints. The constraint propagation is restricted by whether the calculated value range is reduced or not, the constraint propagation range is narrowed, the number of time constraint reasoning times is reduced, the limited computing resources of deep space probes are saved, the efficiency of complex time constraint reasoning is improved, and further the mission planning efficiency of deep space probes is improved.

[0008] The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning disclosed by the present invention includes the following steps:

[0009] Step 1. Establish a time planning problem model for deep space probes. During the planning process, a series of executable activities that meet the constraints are searched in the planning space based on the time planning problem model for deep space probes, so that the probe reaches the target state. The time planning problem model for deep space probes includes probe states, system executable activity time variables, time constraints of activities, initial states, and target states.

[0010] Define the time planning problem of deep space probes as a tuple shown in Equation (1)

[0011] Π = [S, O, C, s0, g] (1)

[0012] where S represents the set of probe states, O = {o1, o2,..., o n} represents the set of executable activity time variables of the probe. For any activity o of the probe i has a start time point s i , an end time point e i , and a duration d i , that is, o i = {s i , e i , d i}. C = {c1, c2,…, c m} represents the time constraint relationship between any two activities. s0 represents the initial state of the probe, and g represents the target state. The planning process is to start from the initial state and search for a series of executable activities that meet the constraints in the planning space, so that the probe reaches the target state.

[0013] Step 2. Represent the activity time variables and time constraint relationships in Step 1. Represent the activity time variables and time constraints as vertices and edges in a simple time network STN respectively, which is convenient for subsequent Steps 3 and 4 to use graph-based methods for time constraint reasoning.

[0014] Define the start and end time points of activity x as variables x s , x e .

[0015] For activity variable x s , whose value range is a ≤ x s ≤ b, the self-constraint of the activity variable is represented as I xs = [a, b]. The internal duration constraint of the activity is the constraint between the end time and the start time. The constraint {x s , x e} has a time relationship c ≤ x e - x s ≤ d, then this constraint is represented as interval I xsxe = [c, d], and Among them, constraint I xsxe and I xexs represent the same constraint. Similarly, if variables u and w belong to different activities respectively, then I uw = [e, f] represents the constraint between activities.

[0016] Represent the activity time variables and time constraints as vertices and edges in a simple time network STN, that is, T = <V, E>. V represents the set of all time variables included in the detector. The set of time variables includes the start time point, end time point, and time zero point of the activity. E represents the edge set of the time constraint relationship between variables, and the constraint value is represented as the weight of each edge. Create a list of variables Ne i with constraint relationships for variable i in STN. This list contains all variables in the current network that have constraint relationships with variable i.

[0017] Step 3: Initialize the STN according to the initial state and target state corresponding activities of the detector in Step 1. Establish a list of constraints to be processed Procon based on the constraints of the added activities during the planning process. Add the new constraints in this list to the current time constraint network, and judge the impact of the new constraints on the consistency of the current network according to the subordinate relationship between the variables of the new constraints and the variables of the current constraint network. If the constraints are consistent, continue to take out the constraints in the list of constraints to be processed until the list is empty, and end the time constraint reasoning. If the constraints are inconsistent, return the conclusion of "constraint inconsistent" to the planning search process, end the time constraint reasoning, and realize the dynamic consistency judgment of the time constraints of the newly added activities in the planning based on arc consistency.

[0018] Step 3.1: Initialize the STN according to the initial state and target state corresponding activities of the detector in Step 1. Establish a set of variables V and add the relative time zero variable z. Establish a set of constraints E. Add the activity time variables corresponding to the above initial state and target state to the set V, and add the time constraints of the activities to the set E to complete the initialization of the STN. Establish a list of constraints to be processed Procon based on the constraints of the added activities during the planning process, and add the constraints to be processed during the planning process to the list Procon.

[0019] Step 3.2: Take out the constraints in the list Procon in sequence and add them to the current STN. If the variable u in the new constraint {u, v} is not in the list Ne vIf it is, add it. Perform the same operation for variable v. After adding each new constraint {u, v}, execute Step 4. According to the subordination relationship between the variables of the new constraint and the variables of the current constraint network, judge the impact of the new constraint on the consistency of the current network, and judge the consistency of the new STN. If the constraints are consistent, continue to take out the constraints in the list Procon of constraints to be processed until the list is empty, and end the time constraint reasoning. If the constraints are inconsistent, return the conclusion of "inconsistent constraints" to the planning search process, end the time constraint reasoning, and realize the dynamic consistency judgment based on arc consistency for the time constraints of the newly added activities in the plan.

[0020] Step 4: If the STN before adding the new constraint {u, v} is consistent, it indicates that the current planning activities meet the time constraints. Add the new constraint {u, v} to the set E described in Step 2. Perform constraint dynamic reasoning on the STN after adding the new constraint according to the subordination relationship between the variables and the variables of the current constraint network. Judge the constraint propagation through the variable value range update list. Only calculate the new value range for the variables within the influence range of the new constraint, and limit the constraint propagation by whether the calculated value range is reduced or not, narrow the constraint propagation range, reduce the number of time constraint reasoning times, save the limited computing resources of the deep space probe, and improve the efficiency of complex time constraint reasoning.

[0021] Step 4.1: Judge whether the variables of the new constraint {u, v} are in the set V.

[0022] If neither u nor v is in the set V, execute Step 4.2;

[0023] If one of u and v is in the set V, execute Step 4.3;

[0024] If both u and v are in the set V, execute Step 4.4.

[0025] Step 4.2: If neither u nor v is in the set V, add the constraints I zu =[-∞, +∞] and I zv =[-∞, +∞] to the set E, and add the variables u and v to the set V. The new constraint {u, v} will not be propagated in the original STN and has no impact on the consistency of the original STN. The new STN is consistent, and add the new constraint {u, v} to the set E.

[0026] Step 4.3: If one of u and v is in the set V, define this variable as u, then add the constraint I zv =[-∞, +∞] and the new constraint {u, v} to the set E, and add the variable v to the set V. Calculate as the new value range of the variable v. The symbol means to sum the endpoints of the two intervals respectively. When I1 = [a, b] and I2 = [c, d], then Since I zv = [-∞, +∞], thus I' zv must not be an empty set, indicating that the new STN is consistent.

[0027] Step 4.4: If both u and v are in set V, then create a variable value range update list Q to judge constraint propagation, calculate the variable value ranges within the influence scope of new constraints, judge whether the value ranges of variables are reduced after calculation to limit constraint propagation, narrow the scope of constraint propagation, reduce the number of time constraint reasoning times, save the limited computing resources of the deep space probe, and improve the efficiency of complex time constraint reasoning.

[0028] Step 4.4.1: For variable u, calculate as the new value range of variable u. If I' zu ≠ I zu , then add variable u to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistent". For variable v, calculate as the new value range of variable v. If I' zv ≠ I zv , then add variable v to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistent".

[0029] Step 4.4.2: If the variable value range update list Q is empty, it indicates that the new STN is consistent. If the list is non-empty, then execute Step 4.4.3 until the variable value range update list Q is empty or return the conclusion of "constraint inconsistent".

[0030] Step 4.4.3: For each variable i in the variable value range update list Q, execute Step 4.4.4 until all variables in Q are traversed.

[0031] Step 4.4.4: For variable j in the list Ne i of variable i, calculate as the new value range of variable j. If I' zj ≠ I zj , then add variable j to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistent" until all variables in the list Ne i are traversed. Then remove variable i from the variable value range update list Q.

[0032] Step Five: For the time constraint network that dynamically changes during the planning process, perform dynamic arc-consistent time constraint reasoning based on Step Three and Step Four, reduce the number of time constraint reasoning times, improve the efficiency of complex time constraint reasoning, and further improve the autonomous mission planning efficiency of the deep space probe, thereby enhancing the mission planning real-time performance of the deep space probe in case of emergencies.

[0033] Beneficial effects:

[0034] 1. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning disclosed in the present invention establishes a time planning problem model for deep space probes, including probe states, system executable activity time variables, time constraints of activities, initial states, and target states, etc. Before performing time constraint reasoning, the activity time variables and time constraints are represented as variable points and edges in a simple time network (STN), and then the time constraint network is initialized, and the time constraints of newly added activities in the plan are dynamically added to this network. On this basis, based on the arc-consistent algorithm, dynamic reasoning is performed on the current time constraint network to solve whether there is a value range for the activity time variables after the new constraints are added, and it is judged whether the time constraints are consistent during the planning process according to the reasoning results. Thus, fast reasoning of complex time constraints in the autonomous planning process of multi-node probes is realized, and the planning efficiency is improved.

[0035] 2. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning disclosed in the present invention is based on a simple time network. New constraints are added on the basis of the current consistent network, and the influence of the new constraints on the consistency of the current network is judged according to the subordinate relationship between the variables of the new constraints and the variables of the current constraint network, reducing unnecessary consistency calculations and effectively reducing the number of constraint reasoning times.

[0036] 3. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning disclosed in the present invention uses the arc-consistent method to dynamically maintain the time constraint network. During the constraint reasoning process, the propagation of constraints is judged through a variable value range update list, the propagation range of new constraints is restricted, the number of time constraint reasoning times is reduced, the limited computing resources of deep space probes are saved, and the time constraint reasoning efficiency is improved. Description of the drawings

[0037] Figure 1 is a schematic diagram of a simple time network.

[0038] Figure 2 is a partial constraint relationship diagram of the embodiment adopted by the present invention.

[0039] Figure 3 is a flowchart of the dynamic arc-consistent time constraint reasoning method for deep space probe mission planning. Detailed implementation manners

[0040] In order to better illustrate the purpose and advantages of the present invention, the content of the invention will be further described below in conjunction with the drawings and embodiments.

[0041] In order to verify the feasibility of the method, as Figure 2As shown, the time constraints of some activities of the deep space probe are displayed. There are 500 activities and 1996 constraints in this example. The method of the present invention is used to judge the consistency of the time constraints in the planning process, realizing dynamic and rapid reasoning of time constraints and improving the planning efficiency.

[0042] As Figure 2 shown, the dynamic arc-consistent time constraint reasoning method for the deep space probe mission planning disclosed in this embodiment is specifically implemented as follows:

[0043] Step 1: Establish a time planning problem model for the deep space probe. During the planning process, based on the time planning problem model of the deep space probe, a series of executable activities that meet the constraints are searched in the planning space, so that the probe reaches the target state. The time planning problem model of the deep space probe includes the probe state, the time variables of the system executable activities, the time constraints of the activities, the initial state, and the target state.

[0044] Define the time planning problem of the deep space probe as a tuple as shown in Equation (1)

[0045] Π = [S, O, C, s0, g] (2)

[0046] where S represents the set of probe states, O = {o1, o2,..., o n} represents the set of time variables of the executable activities of the probe. For any activity o of the probe i it has a start time point s i , an end time point e i , and a duration d i , that is, o i = {s i , e i , d i}}. C = {c1, c2,..., c m} represents the time constraint relationship between any two activities. s0 represents the initial state of the probe, and g represents the target state. The planning process is to start from the initial state and search for a series of executable activities that meet the constraints in the planning space, so that the probe reaches the target state.

[0047] Step 2: Represent the activity time variables and time constraint relationships in Step 1. Represent the activity time variables and time constraints as vertices and edges in a simple time network STN respectively, which is convenient for using graph-based methods for time constraint reasoning in subsequent Steps 3 and 4.

[0048] Define the start and end time points of activities x, y, and t as variables x s , x e , y s , y e , ts , t e .

[0049] For the active variable x s , its value range is unconstrained. The constraint {x s , x e} has a temporal relationship of 4 ≤ x e - x s ≤ 183. Then this constraint is represented as the interval I xsxe = [4, 183], and Similarly, for variables x s , y e belonging to different activities respectively, I yexs = [-21, 179] represents the constraint between activities. Some constraints of this embodiment are shown in Table 1.

[0050] Table 1 Some constraints to be processed in the plan

[0051] variable <![CDATA[x s > <![CDATA[x e > <![CDATA[y s > <![CDATA[y e > <![CDATA[t s > <![CDATA[t e > <![CDATA[x s > —— [4,183] —— [-179,21] —— [-245,57] <![CDATA[x e > [-183,-4] —— [6,190] —— —— —— <![CDATA[y s > —— [-190,-6] —— [-187,95] —— [-238,-19] <![CDATA[y e > [-21,179] —— [-95,187] —— [71,179] —— <![CDATA[t s > —— —— —— [-179,-71] —— —— <![CDATA[t e > [-57,245] —— [19,238] —— —— ——

[0052] Represent the activity time variables and time constraints as vertices and edges in a simple temporal network STN, i.e., T = <V, E>. V represents the set of all time variables included in the detector. The set of time variables includes the activity start time point, end time point, and time zero point. E represents the set of edges of the temporal constraint relationships between variables. The constraint values are represented as the weights of each edge. Create a list of variables with constraint relationships Ne i for variable i in STN. This list contains all variables in the current network that have constraint relationships with variable i.

[0053] Step 3: Initialize STN according to the initial state and target state corresponding activities of the detector in Step 1. Establish a list of constraints to be processed based on the constraints of the added activities during the planning process. Add the new constraints in this list to the current temporal constraint network. Judge the impact of the new constraints on the consistency of the current network according to the subordination relationship between the variables of the new constraints and the variables of the current constraint network. If the constraints are consistent, continue to take out the constraints in the list of constraints to be processed until the list is empty, and end the temporal constraint reasoning. If the constraints are inconsistent, return the conclusion of "constraint inconsistent" to the planning search process and end the temporal constraint reasoning to achieve dynamic consistency judgment based on arc consistency for the temporal constraints of the newly added activities in the plan.

[0054] Step 3.1: Initialize the STN according to the initial state of the detector and the activities corresponding to the target state in Step 1. Establish a variable set V, and add the relative time zero variable z. Establish a constraint set E. Add the activity time variables corresponding to the above initial state and target state to the set V, and add the time constraints of the activities to the set E to complete the initialization of the STN. Establish a list of constraints to be processed Procon according to the constraints for adding activities during the planning process, and add the constraints to be processed during the planning process to the list Procon. At this time, there are 1990 constraints in the list Procon except for the activity constraints corresponding to the initial state and the target state.

[0055] Step 3.2: Take out the constraints in the list Procon in sequence and add them to the current STN. If the variable u in the new constraint {u, v} is not in the list Ne of v, add it, and perform the same operation for the variable v. After adding each new constraint {u, v}, perform Step 4. Judge the impact of the new constraint on the consistency of the current network according to the subordinate relationship between the variables of the new constraint and the variables of the current constraint network, and perform a consistency judgment on the new STN. If the constraints are consistent, continue to take out the constraints in the list of constraints to be processed Procon until the list is empty, and end the time constraint reasoning. If the constraints are inconsistent, return the conclusion of "inconsistent constraints" to the planning search process, end the time constraint reasoning, and realize the dynamic consistency judgment based on arc consistency for the time constraints of the newly added activities in the plan. v After calculation, the STN before adding the new constraint I = [-21, 179] is consistent, indicating that the current planned activities meet the time constraints. At this time, add the new constraint I = [-21, 179] to the set E. Perform constraint dynamic reasoning on the STN after adding the new constraint according to the subordinate relationship between the variables and the variables of the current constraint network. Judge the constraint propagation through the variable value range update list. Only calculate the new value range for the variables within the influence range of the new constraint, and limit the constraint propagation by whether the calculated value range is reduced or not, narrow the constraint propagation range, reduce the number of time constraint reasoning times, save the limited computing resources of the deep space detector, and improve the efficiency of complex time constraint reasoning, that is, meet the time constraints.

[0056] Step 4: After calculation, the STN before adding the new constraint I yexs =[-21, 179] is consistent, indicating that the current planned activities meet the time constraints. At this time, add the new constraint I yexs =[-21, 179] to the set E. Perform constraint dynamic reasoning on the STN after adding the new constraint according to the subordinate relationship between the variables and the variables of the current constraint network. Judge the constraint propagation through the variable value range update list. Only calculate the new value range for the variables within the influence range of the new constraint, and limit the constraint propagation by whether the calculated value range is reduced or not, narrow the constraint propagation range, reduce the number of time constraint reasoning times, save the limited computing resources of the deep space detector, and improve the efficiency of complex time constraint reasoning, that is, meet the time constraints.

[0057] Step 4.1: Judge whether the variables of the new constraint I yexs =[-21, 179] are in the set V.

[0058] If neither y e , nor x s is in the set V, perform Step 4.2;

[0059] If y e , nor x sIf one of them is in set V, execute step 4.3;

[0060] If y e , x s are both in set V, execute step 4.4.

[0061] According to the foregoing calculation, y e , x s are both in set V, execute step 4.4.

[0062] Step 4.2: If neither u nor v is in set V, add the constraints I zu = [-∞, +∞] and I zv = [-∞, +∞] to set E, and add the variables u and v to set V. The new constraint {u, v} will not be propagated in the original STN and has no impact on the consistency of the original STN. The new STN is consistent, and add the new constraint {u, v} to set E.

[0063] Step 4.3: If one of u and v is in set V, assume this variable is u, then add the constraint I zv = [-∞, +∞] and the new constraint {u, v} to set E, and add the variable v to set V. Calculate as the new value range of the variable v, and the symbol means summing the endpoints of the two intervals respectively. Example: I1 = [a, b], I2 = [c, d], then Since I zv = [-∞, +∞], so I' zv must not be an empty set, and the new STN is consistent.

[0064] Step 4.4: y e , x s are both in set V, then create a variable value range update list Q to judge the constraint propagation, calculate the variable value ranges within the influence range of the new constraint, judge whether the variable value ranges are reduced after calculation to limit the constraint propagation, narrow the constraint propagation range, reduce the number of time constraint inferences, save the limited computing resources of the deep space probe, and improve the efficiency of complex time constraint inference.

[0065] Step 4.4.1: For the variable y e , calculate as the new value range of the variable y e , if I' zye ≠ I zye , then add the variable y e to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistent". For the variable xs , calculate as the variable x snew value range. If I′ zxs ≠I zxs then add the variable x s to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistency".

[0066] Step 4.4.2: If the variable value range update list Q is empty, it indicates that the new STN is consistent. If the list is non-empty, execute Step 4.4.3 until the variable value range update list Q is empty or return the conclusion of "constraint inconsistency".

[0067] Step 4.4.3: For each variable i in the variable value range update list Q, execute Step 4.4.4 until all variables in Q are traversed.

[0068] Step 4.4.4: For the variable j in the list Ne i of the variable i, calculate as the new value range of the variable j. If I′ zj ≠I zj then add the variable j to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistency" until all variables in the list Ne i are traversed. Then remove the variable i from the variable value range update list Q.

[0069] Step Five: For the time constraint network that dynamically changes during the planning process, based on Step Three and Step Four, perform dynamic arc-consistent time constraint reasoning, reduce the number of time constraint reasoning times, improve the efficiency of complex time constraint reasoning, and further improve the autonomous mission planning efficiency of deep space probes, thereby enhancing the mission planning real-time performance of deep space probes under emergencies.

[0070] According to the above steps, for the time constraint network that dynamically changes during the planning process, adopt a dynamic constraint consistency judgment method, only calculate the new value range for the variables within the influence range of the new constraint, and limit the constraint propagation by whether the calculated value range is reduced or not, narrow the constraint propagation range, reduce the number of detections during a single constraint, and effectively realize the rapid judgment of time constraint consistency during the planning process of deep space probes, improving the planning speed.

[0071] Under the method of the present invention, after performing dynamic constraint reasoning on the adopted embodiments, the number of constraint reasonings and the constraint reasoning time are shown in Table 2.

[0072] Table 2 Number of Constraint Reasonings and Constraint Reasoning Time

[0073] method non-incremental dynamic arc consistency number of constraint inferences 16395644 11152 constraint inference time 3626ms 63ms

[0074] As can be seen from the data in the table, compared with the arc-consistency-based non-incremental method, the dynamic arc-consistency method proposed by the present invention significantly reduces the number of constraint inferences and improves the constraint inference efficiency.

[0075] The above specific description further details the purpose, technical solution and beneficial effects of the invention. It should be understood that the above is only a specific embodiment of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A dynamic arc-consistent time constraint reasoning method for deep space probe mission planning, characterized in that: It includes the following steps: Step 1: Establish a time planning problem model for a deep space probe. During the planning process, based on the time planning problem model of the deep space probe, search for a series of executable activities that meet the constraints in the planning space, so that the probe reaches the target state; The time planning problem model of the deep space probe includes the probe state, the time variables of the system executable activities, the time constraints of the activities, the initial state, and the target state; Step 2: Represent the activity time variables and time constraint relationships in Step 1. Represent the activity time variables and time constraints as vertices and edges in a simple time network (STN) respectively, which is convenient for subsequent Step 3 and Step 4 to use a graph-based method for time constraint reasoning; Step 3: Initialize the STN according to the activities corresponding to the initial state and target state of the probe in Step 1; establish a list of constraints to be processed Procon based on the constraints of the added activities during the planning process. Add the new constraints in this list to the current time constraint network, and judge the impact of the new constraints on the consistency of the current network according to the subordination relationship between the variables of the new constraints and the variables of the current constraint network; if the constraints are consistent, continue to take out the constraints in the list of constraints to be processed until the list is empty, and end the time constraint reasoning. If the constraints are inconsistent, return the conclusion of "constraint inconsistent" to the planning search process and end the time constraint reasoning, so as to realize the dynamic consistency judgment based on arc consistency for the time constraints of the newly added activities in the planning; Step 4: If the STN before adding the new constraint {u, v} is consistent, it indicates that the current planned activity meets the time constraint. Add the new constraint {u, v} to the set E in Step 2, and perform constraint dynamic reasoning on the STN after adding the new constraint according to the subordination relationship between the variables and the variables of the current constraint network. Judge the constraint propagation through the variable value range update list, only calculate the new value range for the variables within the influence range of the new constraint, and limit the constraint propagation by whether the calculated value range is reduced or not, narrow the constraint propagation range, reduce the number of time constraint reasoning times, save the limited computing resources of the deep space probe, and improve the efficiency of complex time constraint reasoning.

2. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning according to claim 1, characterized in that: It also includes Step 5: For the time constraint network that dynamically changes during the planning process, perform dynamic arc consistent time constraint reasoning based on Step 3 and Step 4, reduce the number of time constraint reasoning times, improve the efficiency of the autonomous mission planning of the deep space probe, and thus enhance the real-time performance of the mission planning of the deep space probe under emergencies.

3. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning according to claim 1 or 2, characterized in that: The implementation method of Step 1 is: Define the time planning problem of the deep space probe as a tuple as shown in Equation (1) Π = [S, O, C, s0, g] (1) Among them, S represents the set of detector states, O = {o1, o2,..., o n} represents the set of activity time variables that the detector can execute. For any activity o of the detector i has a start time point s i , an end time point e i , and a duration d i , that is, o i = {s i , e i , d i}; C = {c1, c2,..., c m} represents the time constraint relationship between any two activities, s0 represents the initial state of the detector, and g represents the target state; the planning process is to start from the initial state and search for a series of executable activities that satisfy the constraints in the planning space, so that the detector reaches the target state.

4. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning according to claim 3, characterized in that: The implementation method of Step 2 is: Define the start and end time points of activity x as variables x s and x e ; For the active variable x s , whose value range is a ≤ x s ≤ b, the self - constraint of the active variable is expressed as I xs = [a, b]; the internal - duration constraint of the activity is the constraint between the end time and the start time. For the constraint {x s , x e} with the time relationship c ≤ x e - x s ≤ d, this constraint is expressed as the interval I xsxe = [c, d], and where the constraints I xsxe and I xexs represent the same constraint; similarly, if the variables u and w belong to different activities respectively, then I uw = [e, f] represents the constraint between activities; The activity time variables and time constraints are represented as vertices and edges in a Simple Temporal Network (STN), i.e., T = <V, E>. V represents the set of all time variables included in the detector. The set of time variables includes the start time point, end time point, and time zero point of the activity. E represents the set of edges of the time constraint relationships between variables, and the constraint value is represented as the weight of each edge. Create a list of variables Ne with constraint relationships for variable i in the STN i , which contains all the variables in the current network that have constraint relationships with variable i.

5. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning according to claim 4, wherein: The implementation method of Step 3 is: Step 3.1: Initialize the STN according to the activities corresponding to the initial state and target state of the probe in Step 1; establish a variable set V, and add the relative time zero variable z. Establish a constraint set E, add the activity time variables corresponding to the above initial state and target state to the set V, and add the time constraints of the activities to the set E to complete the initialization of the STN; Establish a list of constraints to be processed Procon based on the constraints of the added activities during the planning process, and add the constraints to be processed during the planning process to the list Procon; Step 3.2: Take out the constraints in the list Procon in order and add them to the current STN. If the variable u in the new constraint {u, v} is not in the list Ne of v, then add it, and perform the same operation on the variable v; v If not, add it, and perform the same operation on variable v; After adding each new constraint {u, v}, step four is executed. According to the subordination relationship between the variables of the new constraint and the variables of the current constraint network, the impact of the new constraint on the consistency of the current network is judged, and the consistency of the new STN is judged. If the constraints are consistent, the constraints in the pending constraint list Procon are continuously taken out until the list is empty, and the time constraint reasoning is ended. If the constraints are inconsistent, the conclusion of "inconsistent constraints" is returned to the planning search process, and the time constraint reasoning is ended, realizing the dynamic consistency judgment based on arc consistency for the time constraints of the newly added activities in the plan.

6. The dynamic arc-consistent time constraint reasoning method for deep space probe mission planning according to claim 5, wherein: The implementation method of step four is as follows: Step 4.1: Judge whether the variables of the new constraint {u, v} are in the set V; If neither u nor v is in the set V, execute step 4.2; If one of u and v is in the set V, execute step 4.3; If both u and v are in the set V, execute step 4.4; Step 4.2: If neither u nor v is in set V, add the constraints I zu = [-∞, +∞] and I zv = [-∞, +∞] to set E, and add variables u and v to set V; the new constraint {u, v} will not be propagated in the original STN and has no impact on the consistency of the original STN. The new STN is consistent, so add the new constraint {u, v} to set E; Step 4.3: If one of u and v is in the set V, define this variable as u, then add the constraint I zv = [-∞, +∞] and the new constraint {u, v} to the set E, and add the variable v to the set V; calculate as the new value range of the variable v, and the symbol means summing the endpoints of the two intervals respectively; when I1 = [a, b] and I2 = [c, d], then Since I zv = [-∞, +∞], so I z ′ v must not be an empty set, indicating that the new STN is consistent; Step 4.4: If both u and v are in the set V, then create a variable value range update list Q to judge the constraint propagation, calculate the variable value ranges within the influence range of the new constraint, judge whether the variable value ranges are reduced after calculation to limit the constraint propagation, narrow the constraint propagation range, reduce the number of time constraint reasoning times, save the limited computing resources of the deep space probe, and improve the efficiency of complex time constraint reasoning; Step 4.4.1: For variable u, calculate as the new value range of variable u. If I z ′ u ≠I zu , add variable u to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistency". For variable v, calculate as the new value range of variable v. If I z ′ v ≠I zv , add variable v to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistency"; Step 4.4.2: If the variable value range update list Q is empty, it indicates that the new STN is consistent. If the list is not empty, execute step 4.4.3 until the variable value range update list Q is empty or the conclusion of "inconsistent constraints" is returned; Step 4.4.3: For each variable i in the variable value range update list Q, execute step 4.4.4 until all variables in Q are traversed; Step 4.4.4: For variable j in the list Ne of variable i, calculate i as the new value range of variable j. If I ′ z ≠ I j zj , then add variable j to the variable value range update list Q; if the value range is an empty set, stop the calculation and return the conclusion of "constraint inconsistency" until all variables in the list Ne i have been traversed; then remove variable i from the variable value range update list Q.​

Images