A method for arranging a coke pushing plan of a coke oven random note state based on a bidirectional DIJKSTRA algorithm

Abstract

This invention relates to a method for scheduling coke pushing plans in a disordered state in a coke oven based on a bidirectional Dijkstra algorithm. The method includes constructing initial and target state nodes, constructing forward and backward state transition graphs, calculating the edge weights of each directed edge, initializing the forward and backward search starting nodes, performing a bidirectional Dijkstra search and identifying candidate meeting nodes, calculating the total cumulative penalty value and selecting the final meeting node, and concatenating the forward and backward state sequences to obtain a coke pushing plan that restores the current disordered state to a normal sequence state. This method for scheduling coke pushing plans in a disordered state in a coke oven, based on a bidirectional Dijkstra algorithm, solves the problems of existing scheduling methods being easily affected by differences in operator experience, having long solution times, and being unfavorable for timely handling of disordered conditions on the production floor.

Description

Technical Field

[0001] This invention relates to the field of coking production scheduling optimization and intelligent industrial process control technology, specifically to a method for scheduling coke pushing plans under chaotic coke oven conditions based on the bidirectional DIJKSTRA algorithm. Background Technology

[0002] In coke oven production, coke pushing operations are typically organized and implemented according to a pre-set standard sequence and turnaround time to balance oven operation, heat load distribution, and stable coke quality. When equipment failure, maintenance, or temporary operational disruptions occur at the production site, some carbonization chambers may fail to complete coke pushing at the scheduled time, resulting in a disordered state. This disordered state is not simply a matter of sequence deviation, but rather simultaneously affects the turnaround time of multiple subsequent furnace numbers, transforming the coke pushing plan from a regular cyclical problem into a multi-constraint coupled state evolution problem.

[0003] Existing scheduling methods mainly rely on manual adjustment based on experience or on searching based on the idea of ​​unidirectional shortest path. The former is easily affected by differences in operator experience and is difficult to reliably find a plan with a smaller total loss when there are many candidate solutions; the latter, although it can build a graph model for calculation, can easily lead to an excessively large search range when there are many coke oven holes and complex disordered conditions, resulting in increased solution time and hindering timely handling of disordered conditions on the production site.

[0004] Therefore, existing technologies have shortcomings and need to be improved and developed. Summary of the Invention

[0005] The present invention provides a method for scheduling coke pushing plans under the condition of disordered paper in coke ovens based on the bidirectional DIJKSTRA algorithm. This method addresses the problem that existing scheduling methods mainly rely on manual experience for adjustment or on the idea of ​​searching based on the unidirectional shortest path. The former is easily affected by differences in operator experience and is difficult to reliably find a plan with a small total loss when there are many candidate solutions. The latter, although it can establish a graph model for calculation, tends to have an excessively large search range when there are many coke oven holes and the disordered paper condition is complex, resulting in increased solution time and hindering the timely handling of disordered paper conditions on the production site.

[0006] This invention provides a method for scheduling coke pushing plans in a chaotic state of a coke oven based on the bidirectional DIJKSTRA algorithm, including:

[0007] Obtain the set of random oven numbers under the current operating conditions of the coke oven, the actual coking time of each carbonization chamber in the previous round, the current planned coking time, the current coking sequence, the standard coking sequence, and the standard turnaround time, and construct the initial state node and the target state node.

[0008] Using state nodes as graph nodes, a forward state transition graph is constructed, and a backward state transition graph is constructed from the forward state transition graph.

[0009] Based on the current state node, calculate the edge weight of each directed edge;

[0010] Establish the forward unvisited node set, the forward visited node set, the backward unvisited node set, the backward visited node set, the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table, and initialize them with the forward search starting node and the backward search starting node;

[0011] A bidirectional Dijkstra search is performed by alternating between one round of forward expansion and one round of backward expansion. The forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table are updated respectively. The search stops when there is a common node between the forward visited node set and the backward visited node set, and all common nodes are considered as meeting node candidates.

[0012] Calculate the total cumulative penalty value for each candidate meeting node, and select the candidate meeting node with the smallest total cumulative penalty value as the final meeting node;

[0013] According to the forward successor record table, the forward state sequence is obtained by tracing back from the final meeting node to the initial state node. According to the backward successor record table, the backward state sequence is obtained by tracing back from the final meeting node to the target state node. After deleting duplicate final meeting nodes, the forward state sequence and the backward state sequence are concatenated to obtain the focus push plan that restores the current disordered state to the normal sequence state.

[0014] Furthermore, the acquisition of the randomized set of coke oven numbers under the current operating conditions, the actual coking time of each carbonization chamber in the previous round, the current planned coking time, the current coking sequence, the standard coking sequence, and the standard turnaround time, and the construction of initial state nodes and target state nodes, includes:

[0015] Assume the total number of coke oven carbonization chambers is The standard focus push sequence is a sequence. The standard turnaround time is ;

[0016] For any carbonization chamber Based on the difference between the current planned coking time and the actual coking time of the previous round in the carbonization chamber, the actual turnaround time of the carbonization chamber is calculated. ;

[0017] The current coke pushing sequence, the current randomized furnace number set, and the turnover time vector of each carbonization chamber are used as the initial state nodes. The target state node is defined as the state where the standard coking sequence is restored and the set of randomized furnace numbers is empty. ; where any state node is represented as: ,in, Indicates the current focus sequence. This represents the current set of random letter furnace numbers. This represents the current turnover time vector of each carbonization chamber.

[0018] Furthermore, the step of constructing a forward state transition graph using state nodes as graph nodes, and then constructing a backward state transition graph from the forward state transition graph, includes:

[0019] For the current state node Simulate performing a focus pushing operation to obtain the corresponding new state node. ;in, To execute the new focus push sequence after this focus push, To execute the collection of random furnace numbers after this coking process, This is the current turnover time vector for each carbonization chamber after this coke pushing process;

[0020] Will be arrive The state transition is denoted as a directed edge in the forward state transition graph;

[0021] The state transition is repeated for all reachable states to obtain the forward state transition graph;

[0022] For any directed edge in the forward state transition graph Establish corresponding directed edges in the backward state transition graph. And keep the edge weights consistent.

[0023] Furthermore, the step of calculating the edge weights of each directed edge based on the current state node includes:

[0024] For coke pushing operations of normal furnace numbers, the corresponding edge weights Take zero;

[0025] For the coking operation of the random furnace number, the corresponding edge weight Calculate according to the following formula: ,in, , This refers to the actual turnaround time of the carbonization chamber during the coke pushing operation. Standard turnaround time, This refers to the turnaround time deviation, expressed in hours. The coefficient used to adjust the dimensions; This is a monotonically non-decreasing function pre-calibrated based on production experience. The number of random furnace numbers in the new state after the coke pushing operation is performed.

[0026] Furthermore, the penalty function for: ,in, ;

[0027] penalty function for: .

[0028] Furthermore, the establishment of the forward unvisited node set, the forward visited node set, the backward unvisited node set, the backward visited node set, the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table, and the initialization of the forward search starting node and the backward search starting node, includes:

[0029] Establish a set of forward-unvisited nodes The set of nodes visited in the previous direction The set of unvisited nodes and the set of visited nodes in the backward direction ;in, and Initially, all state nodes are included. and Initially, all are empty;

[0030] Establish a forward cumulative penalty value table Backward cumulative penalty value table And establish separate forward and backward record tables. and successor record table ;

[0031] Settings during initialization: , For state nodes other than the initial state node and the target state node, both the forward cumulative penalty value and the backward cumulative penalty value are initialized to infinity;

[0032] Will Let this be the starting node of the forward search. This is denoted as the starting node for the backward search.

[0033] Furthermore, the forward extension includes:

[0034] In the set of forward-unvisited nodes Select the state node with the smallest cumulative penalty value in the current record. As the starting node for the current forward search, and set the state node... from move in ;

[0035] Traversing nodes All adjacent state nodes in the forward state transition diagram ;

[0036] When node belong And satisfy Updated: And record: ;

[0037] After completing the traversal of all adjacent nodes, in Then select the state node with the smallest cumulative penalty value of the current record as the starting node for the next round of forward expansion.

[0038] Furthermore, the backward extension includes:

[0039] In the set of unvisited nodes Select the state node with the smallest cumulative penalty value in the current record. As the starting node for the current backward search, and set the state node... from move in ;

[0040] Traversing nodes All adjacent state nodes in the backward state transition diagram ;

[0041] When node belong And satisfy Updated: And record: ;

[0042] After completing the traversal of all adjacent nodes, in The state node with the smallest cumulative penalty value in the current record is selected as the starting node for the next round of backward expansion.

[0043] Furthermore, the step of calculating the total cumulative penalty value for each candidate meeting node and selecting the candidate meeting node with the smallest total cumulative penalty value as the final meeting node includes:

[0044] For each candidate meeting node Calculate the total cumulative penalty value: ;

[0045] choose The smallest candidate meeting node is selected as the final meeting node;

[0046] When there are multiple candidate meeting nodes with the same total cumulative penalty value, the meeting node with fewer corresponding path steps is selected first; when the corresponding path steps are the same, the node is selected according to the preset node sorting rules; where the number of steps refers to the total number of state transition steps after splicing.

[0047] Beneficial effects:

[0048] As can be seen from the above technical solutions, the present invention provides a method for arranging coke pushing plans in a chaotic state of a coke oven based on the bidirectional DIJKSTRA algorithm, which has the following beneficial effects:

[0049] 1. This paper expresses the push-focus sequence, scrambled paper information, and turnaround time information through state nodes, enabling the scrambled paper recovery problem to be modeled in a unified state space, thus providing a foundation for automatic solution.

[0050] 2. The quality cost caused by a single focus push is quantified as a side weight, and the total cumulative penalty value is used as the overall target, so that the planning and scheduling results can take into account both the restoration of serial order and the control of quality loss.

[0051] 3. By adopting an alternating approach of forward and backward search, the scope of blind expansion on one side in a large-scale state space is reduced, which is beneficial to improving the solution efficiency of the scrambled paper recovery plan.

[0052] 4. Through the comparison of candidate meeting nodes and the path splicing mechanism, the final plan is not the result of local experience adjustment, but a deterministic result based on the comparison of cumulative penalties across the entire path.

[0053] 5. It can adapt to a single random letter furnace number scenario, and also reserves a unified state expression and penalty modeling framework for multiple random letter scenarios, with room for further expansion.

[0054] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below can be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other.

[0055] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the invention, such as features and / or beneficial effects of exemplary embodiments, will become apparent from the following description or may be learned through practice of specific embodiments according to the teachings of the present invention. Attached Figure Description

[0056] The accompanying drawings are not drawn to scale. In the drawings, each identical or nearly identical component shown in the various figures may be denoted by the same reference numeral. For clarity, not every component is labeled in each figure. Embodiments of various aspects of the invention will now be described by way of example and with reference to the accompanying drawings, wherein:

[0057] Figure 1 This is a flowchart illustrating the overall process of arranging a coke oven push plan based on the bidirectional DIJKSTRA algorithm in an embodiment of this application.

[0058] Figure 2 This is a flowchart of step S102 of a method for arranging a coke oven random state coke pushing plan based on the bidirectional DIJKSTRA algorithm in an embodiment of this application.

[0059] Figure 3 This is a flowchart of step S104 of a method for arranging a coke oven random state coke pushing plan based on the bidirectional DIJKSTRA algorithm in an embodiment of this application.

[0060] Figure 4 This is a flowchart of step S108 of a method for arranging a coke oven random state coke pushing plan based on the bidirectional DIJKSTRA algorithm in an embodiment of this application.

[0061] Figure 5 This is a flowchart of step S112 of a method for arranging a coke oven random state coke pushing plan based on the bidirectional DIJKSTRA algorithm in an embodiment of this application.

[0062] Figure 6 This diagram illustrates the comparison between traditional and bidirectional Dijkstra's algorithm search strategies.

[0063] Figure 7 This is a diagram showing the focusing sequence of the initial disordered paper state S0.

[0064] Figure 8 This is a diagram showing the focus sequence in the disordered state S1.

[0065] Figure 9 This is a diagram showing the focusing sequence in the disordered state S2.

[0066] Figure 10 This is a diagram showing the focus sequence in the disordered state S3.

[0067] Figure 11 This is a diagram showing the focus sequence in the disordered state S4.

[0068] Figure 12 For the target normal state S n The focus pushing sequence diagram.

[0069] Figure 13 This is a schematic diagram of an electronic device according to an embodiment of this application. Detailed Implementation

[0070] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the scope of protection of the present invention. Unless otherwise defined, the technical or scientific terms used herein should have the ordinary meaning understood by those skilled in the art to which this invention pertains.

[0071] The terms "first," "second," and similar words used in the specification and claims of this patent application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Similarly, unless the context clearly indicates otherwise, the singular forms of "an," "a," or "the," etc., do not indicate a quantity limitation, but rather indicate the presence of at least one. Terms such as "comprising" or "including" mean that the element or object preceding "comprising" encompasses the features, integrals, steps, operations, elements, and / or components listed following "comprising" or "including," and do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; these relative positional relationships may change accordingly when the absolute position of the described object changes.

[0072] In coke oven production, to protect the oven body, balance the heat load, and ensure stable production, coke pushing operations must follow a strict sequence, i.e., a coke pushing sequence (such as a 5-2 sequence). Each carbonization chamber has its fixed coke pushing time, and the time interval between two consecutive coke pushing operations in the same carbonization chamber is called the turnaround time. However, in actual production, due to equipment failure, maintenance, or other unforeseen circumstances, a certain carbonization chamber may fail to push coke at the scheduled time, and this carbonization chamber becomes a disordered furnace number. To restore the normal coke pushing sequence, this disordered furnace number must choose an extended or shortened turnaround time for coke pushing in subsequent cycles, which will lead to quality loss of coke in that furnace due to over-firing or under-firing. The process of scheduling coke pushing is essentially about finding a reasonable path from the current disordered state back to the normal state in a complex network diagram composed of the state of the carbonization chamber and the timing of coke pushing.

[0073] Coke ovens typically have numerous carbonization chambers (over 100), resulting in a vast number of nodes in a haphazard state. The traditional Dijkstra's algorithm is a unidirectional search algorithm, expanding outwards from the starting point. In such a large network, the traditional Dijkstra's algorithm needs to traverse a large number of nodes unrelated to the target direction, leading to lengthy computation time. In extreme cases, it may even fail to yield a result due to insufficient computational resources or timeouts.

[0074] In view of this, the embodiments of this application introduce a bidirectional Dijkstra search strategy, which initiates the search simultaneously from the starting point (current scrambled state) and the ending point (normal string order state), and the nodes meet at some point in the middle, thereby greatly reducing the range of nodes that need to be explored and improving computational efficiency.

[0075] From a graphical perspective, "referencing" can be understood as referring to... Figure 6 To calculate the distance from point A to point B, the traditional Dijkstra's algorithm starts from point A and expands outwards until it encounters point B; the entire search area is the area of ​​the largest circle in the graph. In contrast, the bidirectional Dijkstra's algorithm provided in this application expands outwards from both points A and B simultaneously. When the radius of the search range of the unidirectional Dijkstra's algorithm is r, the bidirectional Dijkstra's algorithm starts searching from both the starting and ending points simultaneously, and the search radius on each side of its search range is approximately r / 2. Figure 6 As shown, the optimal path is found when the two small circles meet, and the search area is the area of ​​the two small circles in the diagram. Therefore, the total number of nodes to be explored in a bidirectional search is much smaller than that in a unidirectional search. This allows the present invention to greatly reduce the amount of computation and shorten the time required to solve the optimal coke pushing plan, thereby enabling it to meet the real-time scheduling needs of the production site, quickly respond to disordered conditions, and minimize the coke quality loss caused by disorder through plan optimization.

[0076] Specifically, this invention provides a method for scheduling coke pushing plans in a disordered state of a coke oven based on the bidirectional DIJKSTRA algorithm, referring to... Figure 1 ,include:

[0077] Step S102: Obtain the set of random furnace numbers under the current operating conditions of the coke oven, the actual coking time of each carbonization chamber in the previous round, the current planned coking time, the current coking sequence, the standard coking sequence, and the standard turnaround time, and construct the initial state node and the target state node.

[0078] Step S104: Using state nodes as graph nodes, construct a forward state transition graph, and construct a backward state transition graph from the forward state transition graph.

[0079] Step S106: Based on the current state node, calculate the edge weight of each directed edge.

[0080] Step S108: Establish the forward unvisited node set, the forward visited node set, the backward unvisited node set, the backward visited node set, the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table, and initialize the forward search starting node and the backward search starting node.

[0081] Step S110: Perform a bidirectional Dijkstra search by alternating between one round of forward expansion and one round of backward expansion. Update the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table respectively. Stop the search when there is a common node between the forward visited node set and the backward visited node set, and treat all common nodes as meeting node candidates.

[0082] Step S112: Calculate the total cumulative penalty value for each candidate meeting node, and select the candidate meeting node with the smallest total cumulative penalty value as the final meeting node.

[0083] Step S114: Based on the forward successor record table, backtrack from the final meeting node to the initial state node to obtain the forward state sequence. Based on the backward successor record table, trace from the final meeting node to the target state node to obtain the backward state sequence. After deleting duplicate final meeting nodes, concatenate the forward state sequence and the backward state sequence to obtain the focus push plan that restores the current disordered state to the normal sequence state.

[0084] First, the starting and ending states are constructed based on the set of random furnace numbers, the actual coking time of the previous round, the current string order, and the standard string order. Then, forward and backward graphs are established. Next, the edge weights are calculated and the data structure required for bidirectional search is initialized. Then, forward expansion and backward expansion are executed alternately. After meeting node candidates appear, the total cumulative penalty value is compared. Finally, through backtracking and splicing, the coking plan that restores the current random state to the normal string order state is obtained.

[0085] By abstracting the scrambled paper recovery into a bidirectional shortest path search problem centered on state nodes, and introducing the focus quality loss as an edge weight, the problem of restoring the sequence order and reducing the loss are unified into the same solution framework. It can solve the path between the current scrambled paper state and the target normal state around the unified goal of the total cumulative penalty value. This makes the plan generation have clear optimization goals and executable steps, and provides a basis for the refinement and limitation of subsequent dependent claims.

[0086] In some embodiments, the following are obtained: a randomized set of coke oven numbers under the current operating conditions of the coke oven; the actual coking time of each carbonization chamber in the previous round; the current planned coking time; the current coking sequence; the standard coking sequence; and the standard turnaround time. Initial state nodes and target state nodes are then constructed, referring to... Figure 2 ,include:

[0087] Step S1021: Assume the total number of coke oven carbonization chambers is... The standard focus push sequence is a sequence. The standard turnaround time is .

[0088] Step S1022: For any carbonization chamber Based on the difference between the current planned coking time and the actual coking time of the previous round in the carbonization chamber, the actual turnaround time of the carbonization chamber is calculated. .

[0089] Step S1023: Use the current coke pushing sequence, the current randomized furnace number set, and the turnover time vector of each carbonization chamber as the initial state node. The target state node is defined as the state where the standard coking sequence is restored and the set of randomized furnace numbers is empty. ; where any state node is represented as: ,in, Indicates the current focus sequence. This represents the current set of random letter furnace numbers. This represents the current turnover time vector of each carbonization chamber.

[0090] The initial state nodes are constructed using the current coke push sequence, the current set of randomized furnace numbers, and the turnover time vectors of each carbonization chamber. The target state node is formed by restoring the standard coking sequence and the set of randomized furnace numbers to an empty state. and adopt uniformly Any state node is represented in the form of [formula missing]. The sequence information, scrambling information, and time information in the scrambling problem are simultaneously encoded into a single state node, rather than being simplified to a furnace number order. This ensures that each path in the search process reflects not only changes in furnace number arrangement but also the impact of turnaround time deviations on the penalty value, maintaining consistency between the state representation and subsequent edge weight calculations. This provides a unified and computable state description method for state space search, giving nodes in the graph clear physical meaning and facilitating subsequent construction of state transition edges, calculation of penalty values, and implementation of path recovery.

[0091] In some embodiments, state nodes are used as graph nodes to construct a forward state transition graph, and a backward state transition graph is constructed from the forward state transition graph, referring to... Figure 3 ,include:

[0092] Step S1041: For the current state node Simulate performing a focus pushing operation to obtain the corresponding new state node. ;in, To execute the new focus push sequence after this focus push, To execute the collection of random furnace numbers after this coking process, This is the current turnover time vector for each carbonization chamber after this coke pushing operation.

[0093] Step S1042: The material will be... arrive The state transition is denoted as a directed edge in the forward state transition graph.

[0094] Step S1043: Repeat the state transition for all reachable states to obtain the forward state transition diagram.

[0095] Step S1044: For any directed edge in the forward state transition graph Establish corresponding directed edges in the backward state transition graph. And keep the edge weights consistent.

[0096] Starting from the current state node, sequentially select candidate coke pushing furnace numbers that meet the preset operation constraints, simulate one coke pushing operation, and obtain a new state node. and will Let be a directed edge in the forward graph; for any directed edge in the forward graph... Establish corresponding reverse edges in the backward graph. The discrete operation of coke oven pushing is transformed into a state transition behavior in a graphical model. A dual backward state model is constructed based on the forward state model, allowing the search at the starting and ending points to proceed in parallel under a unified weighting system. This provides an engineering-feasible basis for bidirectional search, enabling it to describe both feasible pushing paths from the current state and reachable paths derived from the target state, thus providing a structural foundation for narrowing the search area.

[0097] In some embodiments, based on the current state node, the edge weights of each directed edge are calculated, including:

[0098] For coke pushing operations of normal furnace numbers, the corresponding edge weights Take zero;

[0099] For the coking operation of the random furnace number, the corresponding edge weight Calculate according to the following formula: ,in, , This refers to the actual turnaround time of the carbonization chamber during the coke pushing operation. Standard turnaround time, This refers to the turnaround time deviation, expressed in hours. The coefficient used to adjust the dimensions; This is a monotonically non-decreasing function pre-calibrated based on production experience. The number of random furnace numbers in the new state after the coke pushing operation is performed.

[0100] The quality loss factors in the recovery of disordered coke sheets are broken down into two parts: time deviation penalty and disordered sheet quantity penalty. This reflects both the loss caused by a single furnace number deviating from the standard turnaround time and the impact of the scale of disordered sheet status on the overall recovery difficulty and planned quality. The edge weights constructed in this way are no longer abstract costs, but rather quantitative indicators of production losses with clear technological implications. This establishes a correspondence between the optimization objective of the search algorithm and the on-site coke quality control objectives, thereby avoiding the situation where coke pushing is arranged solely based on the shortest number of steps while ignoring quality costs.

[0101] In some embodiments, the penalty function for: ,in, Penalty function for: .

[0102] Piecewise functions reflect the asymmetric impact of shortened and extended turnaround times on quality loss, allowing penalty functions to move beyond mere principle descriptions. This improves the consistency and repeatability of edge weight calculations, enabling stable single-step penalty values ​​to be obtained under the same input conditions, thereby enhancing the repeatability of the overall coking plan solution.

[0103] In some embodiments, a set of forward unvisited nodes, a set of forward visited nodes, a set of backward unvisited nodes, a set of backward visited nodes, a forward cumulative penalty value table, a backward cumulative penalty value table, a forward predecessor record table, and a backward successor record table are established, and initialized with the forward search starting node and the backward search starting node, as shown in the following example. Figure 4 ,include:

[0104] Step S1081: Establish the set of forward unvisited nodes The set of nodes visited in the previous direction The set of unvisited nodes and the set of visited nodes in the backward direction ;in, and Initially, all state nodes are included. and Initially, all values ​​are empty.

[0105] Step S1082: Establish a forward cumulative penalty value table Backward cumulative penalty value table And establish separate forward and backward record tables. and successor record table .

[0106] Step S1083: Settings during initialization: , For state nodes other than the initial state node and the target state node, both the forward cumulative penalty value and the backward cumulative penalty value are initialized to infinity.

[0107] Step S1084: ... Let this be the starting node of the forward search. This is denoted as the starting node for the backward search.

[0108] The basic data structure for bidirectional Dijkstra's search was initialized, including the set of unvisited nodes, the set of visited nodes, the cumulative penalty value table, the predecessor record table, and the successor record table, and the starting node for the forward search was specified. and the starting node of the backward search Initialization is performed. The two sets of search states required for bidirectional search are modeled independently and initialized synchronously, so that the search on the starting side and the ending side can be unfolded in a mirror manner in the same state space. This provides a record basis for subsequent meeting node determination and path splicing, establishes a clear data organization framework for bidirectional search, and ensures the logical correspondence between the forward cumulative penalty value and the backward cumulative penalty value, and between the forward path record and the backward path record.

[0109] In some embodiments, forward spreading includes:

[0110] In the set of forward-unvisited nodes Select the state node with the smallest cumulative penalty value in the current record. As the starting node for the current forward search, and set the state node... from move in ;

[0111] Traversing nodes All adjacent state nodes in the forward state transition diagram ;

[0112] When node belong And satisfy Updated: And record: ;

[0113] After completing the traversal of all adjacent nodes, in Then select the state node with the smallest cumulative penalty value of the current record as the starting node for the next round of forward expansion.

[0114] The specific steps of the forward expansion are defined, namely, selecting the node with the smallest cumulative penalty value from the set of unvisited nodes. Transfer it to the set of visited nodes and check the state of all its adjacent nodes. Perform the update; update only if the conditions are met. and By combining state transition rules with penalty cost calculation, a forward cost propagation mechanism for scrambled paper recovery scenarios is formed. This mechanism clarifies how the forward search continuously approaches low-penalty paths in the coke oven state graph, gradually filtering out better state transition paths starting from the current scrambled paper state, and providing the optimal substructure of the forward part for calculating the total cumulative penalty value of the final meeting node.

[0115] In some embodiments, backward extension includes:

[0116] In the set of unvisited nodes Select the state node with the smallest cumulative penalty value in the current record. As the starting node for the current backward search, and set the state node... from move in ;

[0117] Traversing nodes All adjacent state nodes in the backward state transition diagram ;

[0118] When node belong And satisfy Updated: And record: ;

[0119] After completing the traversal of all adjacent nodes, in The state node with the smallest cumulative penalty value in the current record is selected as the starting node for the next round of backward expansion.

[0120] The specific steps of the backward expansion are defined, namely, selecting the node with the smallest cumulative penalty value from the set of unvisited backward nodes. For backward adjacent state nodes Perform the update and update. and By introducing backward cost propagation to the target state in addition to the forward search, information from one side of the target state can be transmitted back to the intermediate region, thus forming the other half of the solution structure corresponding to the forward search. The significance of this setting is that path solving no longer relies entirely on unilateral expansion, but is synchronously approximated from both sides. This can shorten the number of expansion layers required to reach the meeting region, improve the search efficiency in a large state space, and provide backward records from the meeting node to the target state for the final path splicing.

[0121] In some embodiments, a total cumulative penalty value is calculated for each candidate meeting node, and the candidate meeting node with the smallest total cumulative penalty value is selected as the final meeting node, referring to... Figure 5 ,include:

[0122] Step S1121: For each candidate meeting node Calculate the total cumulative penalty value: .

[0123] Step S1122: Select The smallest candidate meeting node is selected as the final meeting node.

[0124] Step S1123: When there are multiple candidate meeting nodes with the same total cumulative penalty value, the meeting node with fewer corresponding path steps is selected first; when the corresponding path steps are the same, the node is selected according to the preset node sorting rules; where the number of steps refers to the total number of state transition steps after splicing.

[0125] For each candidate meeting node Calculate the total cumulative penalty value The node with the smallest total cumulative penalty value is selected as the final meeting node. If the total cumulative penalty values ​​are the same, selection is further based on the number of path steps and a preset node sorting rule. Multiple common candidate nodes formed in the intermediate region by the bidirectional search are further incorporated into a unified comparison framework. Instead of using the first meeting as the sole criterion, the minimum total cumulative penalty value is used as the core criterion. Subsequent judgment rules are set for cases with the same penalty value, thus ensuring the determinism of the final output path, avoiding randomness in meeting node selection, and guaranteeing that the final coking plan has clear comparison standards and repeatable output results. The preset node sorting rule can be specified as, for example, sorting by furnace number in ascending order, by standard string index order, or by state code lexicographical order.

[0126] The following provides another embodiment for illustration:

[0127] A coking plant has a set of coke ovens with a total of 110 carbonization chambers, using a 5-2 series coke pushing method. At a certain moment, there is a disordered oven number (No. 12). A plan to restore the disordered state needs to be formulated. The specific implementation steps are as follows:

[0128] (1) Construct a state transition diagram model, where: during normal furnace number coking... When the value is 0, and the focus is adjusted for the random number 12. In this embodiment, there is only one random furnace number, so the intermediate adjustment state corresponds to... A value of 1 corresponds to the target's normal state. It is 0.

[0129] (2) Initialize bidirectional search, forward search starting point: current scrambled paper state (Boiler No. 12 is not in its normal position), backward search starting point: target in normal state (Boiler No. 12 is in the normal 5-2 series sequence position).

[0130] (3) Perform bidirectional search alternately, with the forward search starting from... Expand and explore the new states after different coke oven numbers are pushed, and perform a backward search from... The reverse expansion deduces which focusing operation can reach the current state. After several rounds of alternating searches, in the forward search state... and backward search state An intersection was found.

[0131] (4) Path splicing forms the optimal plan, forward path: (Cumulative penalty value 12), Backward path: (Cumulative penalty value 8), Focus Push Plan: (Total penalty value: 20).

[0132] Figures 7-12 They respectively show the process from the initial state of disordered paper. , disordered paper state , disordered paper state , disordered paper state , disordered paper state Target normal state The order of focusing. Figure 7 The initial state is a jumbled mess of paper, with a cumulative forward penalty of 0. Figure 12 This node is the starting node for the backward search, and the cumulative backward penalty value is also 0. Figures 8 to 11 Only furnace number 12 in the middle is in a state of disorder. Meanwhile, the actual turnaround time for furnace No. 12 in each of the previous charts is 3 hours longer, indicating a turnaround time deviation. If the value is 3, then Since the actual turnaround time of all furnaces except furnace No. 12 is equal to the specified turnaround time, the penalty value is 0. Figures 7 to 8 , Figures 8 to 9 , Figures 9 to 10 , Figures 10 to 11 , Figures 11 to 12 In the graph, the weight of the edge connecting each node to the previous graph is... The model's planned path contains 5 extensions, and the total penalty value is 20.

[0133] By applying the bidirectional Dijkstra algorithm to the calculation of random paper recovery in a 110-hole coke oven, a large number of invalid searches were avoided, and the computational efficiency was reduced from the initial 30s to about 15s. While ensuring that the global optimal solution is found, the response speed is significantly improved, providing an efficient solution for intelligent scheduling of coking production.

[0134] Another embodiment of the present invention also provides a coking schedule arrangement device for coke oven disordered state coking plan based on the bidirectional DIJKSTRA algorithm, comprising:

[0135] The data acquisition module is used to obtain the randomized set of coke oven numbers under the current operating conditions, the actual coking time of each carbonization chamber in the previous round, the current planned coking time, the current coking sequence, the standard coking sequence, and the standard turnaround time, and to construct the initial state node and the target state node.

[0136] The building module is used to construct a forward state transition graph using state nodes as graph nodes, and to construct a backward state transition graph from the forward state transition graph.

[0137] The first calculation module is used to calculate the edge weights of each directed edge based on the current state node.

[0138] The initialization module is used to establish the set of forward unvisited nodes, the set of forward visited nodes, the set of backward unvisited nodes, the set of backward visited nodes, the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table, and initializes the forward search starting node and the backward search starting node.

[0139] The search module performs a bidirectional Dijkstra search by alternating between forward expansion and backward expansion, updating the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table respectively. The search stops when there is a common node between the forward visited node set and the backward visited node set, and all common nodes are considered as meeting node candidates.

[0140] The second calculation module is used to calculate the total cumulative penalty value for each candidate meeting node and select the candidate meeting node with the smallest total cumulative penalty value as the final meeting node.

[0141] The splicing module is used to backtrack from the final meeting node to the initial state node according to the forward successor record table to obtain the forward state sequence, and to trace from the final meeting node to the target state node according to the backward successor record table to obtain the backward state sequence. After deleting duplicate final meeting nodes, the forward state sequence and the backward state sequence are spliced ​​together to obtain the focus pushing plan that restores the current disordered state to the normal sequence state.

[0142] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0143] Based on the same inventive concept as the above method embodiments, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it enables the electronic device to implement the control method described in the above embodiments.

[0144] In one embodiment, the electronic device may be a server, and in this embodiment, the structure of the electronic device may be as follows: Figure 13 As shown, it includes a memory, a communication module, and one or more processors.

[0145] Memory is used to store computer programs executed by the processor. Memory can be mainly divided into a program storage area and a data storage area. The program storage area can store the operating system and programs required to run instant messaging functions, etc.; the data storage area can store various instant messaging information and operation instruction sets, etc.

[0146] Memory can be volatile memory, such as random access memory (RAM); memory can also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory can be any other medium capable of carrying or storing a desired computer program having the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory can be a combination of the above-mentioned types of memory.

[0147] A processor may include one or more central processing units (CPUs) or digital processing units, etc. The processor is used to implement the aforementioned audio data processing methods when it invokes computer programs stored in memory.

[0148] The communication module is used to communicate with terminal devices and other servers.

[0149] This application embodiment does not limit the specific connection medium between the above-described memory, communication module, and processor. This application embodiment... Figure 13 The memory and processor are connected via a bus, and the bus is in... Figure 13 The connections between other components are illustrated with arrows and are for illustrative purposes only, not as limiting information. Buses can be categorized as address buses, data buses, control buses, etc. For ease of description, Figure 13 The text uses only one arrow to describe it, but does not indicate that there is only one bus or one type of bus.

[0150] Based on the same inventive concept as the above-described method embodiments, embodiments of the present invention also provide a computer-readable storage medium for storing a computer program. When the computer program is run on a computer, it enables an electronic device to implement the control methods described in the above embodiments. The computer-readable storage medium can be a readable signal medium or a readable storage medium. A readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0151] Based on the same inventive concept as the above-described method embodiments, embodiments of the present invention also provide a computer program product. The computer program product includes a computer program that, when run on an electronic device, causes the electronic device to perform the steps of the control methods described above according to various exemplary embodiments of this application. The program product may take the form of any combination of one or more readable media. These computer program commands can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the commands executed by the processor of the computer or other programmable data processing device generate a process for implementing... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0152] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for scheduling coke pushing plans under chaotic conditions in a coke oven based on a bidirectional DIJKSTRA algorithm, characterized in that, include: Obtain the set of random oven numbers under the current operating conditions of the coke oven, the actual coking time of each carbonization chamber in the previous round, the current planned coking time, the current coking sequence, the standard coking sequence, and the standard turnaround time, and construct the initial state node and the target state node. Using state nodes as graph nodes, a forward state transition graph is constructed, and a backward state transition graph is constructed from the forward state transition graph. Based on the current state node, calculate the edge weight of each directed edge; Establish the forward unvisited node set, the forward visited node set, the backward unvisited node set, the backward visited node set, the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table, and initialize them with the forward search starting node and the backward search starting node; A bidirectional Dijkstra search is performed by alternating between one round of forward expansion and one round of backward expansion. The forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table are updated respectively. The search stops when there is a common node between the forward visited node set and the backward visited node set, and all common nodes are considered as meeting node candidates. Calculate the total cumulative penalty value for each candidate meeting node, and select the candidate meeting node with the smallest total cumulative penalty value as the final meeting node; According to the forward successor record table, the forward state sequence is obtained by tracing back from the final meeting node to the initial state node. According to the backward successor record table, the backward state sequence is obtained by tracing back from the final meeting node to the target state node. After deleting duplicate final meeting nodes, the forward state sequence and the backward state sequence are concatenated to obtain the focus push plan that restores the current disordered state to the normal sequence state.

2. The method for scheduling coke pushing plans under chaotic conditions in a coke oven based on the bidirectional DIJKSTRA algorithm according to claim 1, characterized in that, The process of acquiring the randomized set of coke oven numbers under the current operating conditions, the actual coking time of each carbonization chamber in the previous round, the current planned coking time, the current coking sequence, the standard coking sequence, and the standard turnaround time, and constructing initial state nodes and target state nodes, includes: Assume the total number of coke oven carbonization chambers is The standard focus push sequence is a sequence. The standard turnaround time is ; For any carbonization chamber Based on the difference between the current planned coking time and the actual coking time of the previous round in the carbonization chamber, the actual turnaround time of the carbonization chamber is calculated. ; The current coke pushing sequence, the current randomized furnace number set, and the turnover time vector of each carbonization chamber are used as the initial state nodes. The target state node is defined as the state where the standard coking sequence is restored and the set of randomized furnace numbers is empty. ; where any state node is represented as: ,in, Indicates the current focus sequence. This represents the current set of random letter furnace numbers. This represents the current turnover time vector of each carbonization chamber.

3. The method for arranging a coke pushing plan based on a bidirectional DIJKSTRA algorithm in a chaotic state of a coke oven, as described in claim 2, is characterized in that... The step of constructing a forward state transition graph using state nodes as graph nodes, and then constructing a backward state transition graph from the forward state transition graph, includes: For the current state node Simulate performing a focus pushing operation to obtain the corresponding new state node. ;in, To execute the new focus push sequence after this focus push, To execute the collection of random furnace numbers after this coking process, This is the current turnover time vector for each carbonization chamber after this coke pushing process; Will be arrive The state transition is denoted as a directed edge in the forward state transition graph; The state transition is repeated for all reachable states to obtain the forward state transition graph; For any directed edge in the forward state transition graph Establish corresponding directed edges in the backward state transition graph. And keep the edge weights consistent.

4. The method for arranging a coke oven push-out plan based on the bidirectional DIJKSTRA algorithm according to claim 3, characterized in that, The step of calculating the edge weights of each directed edge based on the current state node includes: For coke pushing operations of normal furnace numbers, the corresponding edge weights Take zero; For the coking operation of the random furnace number, the corresponding edge weight Calculate according to the following formula: ,in, , This refers to the actual turnaround time of the carbonization chamber during the coke pushing operation. Standard turnaround time, This refers to the turnaround time deviation, expressed in hours. The coefficient used to adjust the dimensions; This is a monotonically non-decreasing function pre-calibrated based on production experience. The number of random furnace numbers in the new state after the coke pushing operation is performed.

5. The method for arranging a coke oven push-out plan based on the bidirectional DIJKSTRA algorithm according to claim 4, characterized in that, penalty function for: ,in, ; penalty function for: .

6. The method for arranging a coke oven push-out plan based on the bidirectional DIJKSTRA algorithm according to claim 3, characterized in that, The process of establishing the forward unvisited node set, the forward visited node set, the backward unvisited node set, the backward visited node set, the forward cumulative penalty value table, the backward cumulative penalty value table, the forward predecessor record table, and the backward successor record table, and initializing them with the forward search starting node and the backward search starting node, includes: Establish a set of forward-unvisited nodes The set of nodes visited in the previous direction The set of unvisited nodes and the set of visited nodes in the backward direction ;in, and Initially, all state nodes are included. and Initially, all are empty; Establish a forward cumulative penalty value table Backward cumulative penalty value table And establish separate forward and backward record tables. and successor record table ; Settings during initialization: , For state nodes other than the initial state node and the target state node, both the forward cumulative penalty value and the backward cumulative penalty value are initialized to infinity; Will Let this be the starting node of the forward search. This is denoted as the starting node for the backward search.

7. The method for arranging a coke oven pushing plan based on a bidirectional DIJKSTRA algorithm in a disordered state, as described in claim 6, is characterized in that... The forward expansion includes: In the set of forward-unvisited nodes Select the state node with the smallest cumulative penalty value in the current record. As the starting node for the current forward search, and set the state node... from move in ; Traversing nodes All adjacent state nodes in the forward state transition diagram ; When node belong And satisfy Updated: And record: ; After completing the traversal of all adjacent nodes, in Then select the state node with the smallest cumulative penalty value of the current record as the starting node for the next round of forward expansion.

8. The method for arranging a coke oven push-out plan based on the bidirectional DIJKSTRA algorithm according to claim 6, characterized in that, The backward expansion includes: In the set of unvisited nodes Select the state node with the smallest cumulative penalty value in the current record. As the starting node for the current backward search, and set the state node... from move in ; Traversing nodes All adjacent state nodes in the backward state transition diagram ; When node belong And satisfy Updated: And record: ; After completing the traversal of all adjacent nodes, in The state node with the smallest cumulative penalty value in the current record is selected as the starting node for the next round of backward expansion.

9. A method for arranging a coke oven push-out plan based on a bidirectional DIJKSTRA algorithm in a disordered state, as described in claim 6, is characterized in that... The step of calculating the total cumulative penalty value for each candidate meeting node and selecting the candidate meeting node with the smallest total cumulative penalty value as the final meeting node includes: For each candidate meeting node Calculate the total cumulative penalty value: ; choose The smallest candidate meeting node is selected as the final meeting node; When there are multiple candidate meeting nodes with the same total cumulative penalty value, the meeting node with fewer corresponding path steps is selected first; when the corresponding path steps are the same, the node is selected according to the preset node sorting rules; where the number of steps refers to the total number of state transition steps after splicing.

Images