Industry-driven production cost intelligent accounting and scheduling coordination system

Abstract

This invention discloses an industry-driven intelligent accounting and scheduling collaborative system for production costs, relating to the field of intelligent manufacturing technology. The system collects input interval objects, including cost tokens and candidate production line time periods, from existing business systems via a data interface, and performs differential encoding to generate a constraint differential sequence. Based on the work order identifier, it extracts constraint feature fields from the constraint differential sequence and constructs a field arrangement pattern sequence. It calculates the first and second conditional entropies using an arrangement transfer entropy algorithm, and obtains the arrangement transfer entropy value by subtracting them. The arrangement transfer entropy values ​​are sorted in descending order, and the first-ranked transmission field is extracted and marked as the transmission code for the written field. An admission edge is constructed, and a bipartite graph is built by combining the work order identifier and the time period identifiers in the candidate production line time periods. Candidate matching results are then filtered, and the candidate matching result with the most work order identifiers is counted to obtain the target matching result. Finally, the work order scheduling status is written.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, specifically to an industry-driven intelligent accounting and scheduling collaborative system for production costs. Background Technology

[0002] In the field of intelligent manufacturing technology, manufacturing enterprises typically need to organize production around work orders, materials, processes, shifts, changeovers, and delivery cycles. As order sources, production line capacity, material prices, and delivery rhythms constantly change under industry-driven models, relying solely on manual experience or fixed scheduling rules makes it difficult to simultaneously address cost accounting and production line time slot arrangements. Especially in production scenarios involving multiple work orders, multiple production lines, and multiple time slots operating in parallel, intelligent cost accounting and scheduling collaboration are gradually becoming important technical directions for manufacturing enterprises in production planning management. The core of this approach lies in first analyzing work order cost constraints and candidate production line time slot characteristics, and then obtaining suitable work order time slot matching solutions for inclusion in the scheduling results.

[0003] In existing intelligent production cost accounting and scheduling collaboration methods, cost accounting typically focuses on processing individual data items such as material prices, process time, or labor costs, while scheduling relies heavily on production line idle periods, shift plans, or delivery deadlines. These two processes are disconnected in their data processing. For constraints such as material price lock-in ranges, process occupancy ranges, shift permission vectors, changeover entry strings, and delivery control ranges, existing methods often lack a unified differential expression method and struggle to analyze the transmission direction of different constraint fields in time slot recommendation. Consequently, when faced with a large number of candidate production line time slots and intertwined and changing work order constraints, problems easily arise such as unclear work order admission criteria, lack of data support for candidate time slot ranking, and difficulty in verifying scheduling results. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an industry-driven intelligent accounting and scheduling collaborative system for production costs, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an industry-driven intelligent accounting and scheduling collaborative system for production costs, comprising a counterfactual coding module, a transmission code generation module, an access edge generation module, and an access decision module; The counterfact coding module collects input interval objects, including cost tokens and candidate production line time periods, from the existing business system through a data interface, and performs differential coding to generate a constrained differential sequence. The transmission code generation module extracts constraint feature fields from the constraint difference sequence based on the work order identifier in the cost token, constructs a field arrangement pattern sequence, calculates the first conditional entropy and the second conditional entropy through the arrangement transfer entropy algorithm, and obtains the arrangement transfer entropy value by subtracting the first conditional entropy from the second conditional entropy. The admission edge generation module sorts the propagation entropy values ​​in descending order, extracts the first-ranked propagation field pair and writes it into the source field code, the guided field code, the field propagation order and the time period recommendation table to form a field propagation code for marking. The admission decision module constructs admission edges and combines them with work order identifiers and time period identifiers in candidate production line time periods to build a bipartite graph and filter candidate matching results. It then counts the candidate matching results with the most work order identifiers to obtain the target matching result and executes work order scheduling status writing based on the target matching result and the field transmission code marking result.

[0006] Preferably, the counterfactual encoding module includes an object reading unit and a differential encoding unit; The object reading unit is used to collect input range objects from existing business systems through a data interface; The existing business systems include a work order cost data system and a production planning data system; The input interval object includes cost tokens and candidate production line time periods, and the input interval object adopts a left-closed and right-open interval; wherein, the left-closed and right-open interval means that the interval includes the start point but does not include the end point; The cost token includes a work order identifier, a material price lock range, a process occupancy range, a shift permission vector, a changeover entry string, and a delivery control range. The candidate production line time period includes time period identifier, time period interval, time period capacity interval, candidate shift code, and predecessor tail status code.

[0007] Preferably, the differential encoding unit is used to differentially encode the cost token and the candidate production line time period to obtain a production differential value data set; The production differential data set includes price lock-in differential, process differential, shift differential, changeover differential, and delivery differential; The production difference data groups under the same work order identifier are arranged according to the order of candidate production line time periods to obtain the constrained difference sequence; The constraint difference sequence includes a price lock field difference sequence, a process field difference sequence, a shift field difference sequence, a changeover field difference sequence, and a delivery field difference sequence.

[0008] Preferably, the conduction code generation module includes a sequence conversion unit and an entropy transfer orientation unit; The sequence conversion unit is used to group the constraint difference sequence according to the work order identifier to obtain the constraint difference sequence record; The constraint feature fields are extracted from the constraint difference sequence records, specifically: the text field of the material lock-up price range is extracted to obtain the material lock-up price field; the text field of the process occupancy range is extracted to obtain the process occupancy field; the text field of the shift permission vector is extracted to obtain the shift adaptation field; the text field of the change-up entry string is extracted to obtain the change-up acceptance field; and the text field of the delivery control range is extracted to obtain the delivery occupancy field. The constraint feature fields include any one of the following: material lock-in price field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field.

[0009] Preferably, the entropy transmission orientation unit includes a mapping sampling unit and an entropy value writing unit; The mapping sampling unit is used to perform field mapping between the constraint difference sequence and the constraint feature field to obtain the field arrangement pattern sequence; The field mapping is as follows: Read the permutation pattern sequences Ri and Rj of any two constraint feature fields Fi and Fj, where Fi and Fj represent any field among the material lock-in field, process occupation field, shift adaptation field, changeover acceptance field, or delivery occupation field, and Fi≠Fj. Ri represents the field permutation pattern sequence corresponding to constraint feature field Fi, and Rj represents the field permutation pattern sequence corresponding to constraint feature field Fj, and the number of field permutation patterns of the two is the same. The r-th field arrangement pattern in the field arrangement pattern sequence Rj is taken as the target current sample, the r-th field arrangement pattern in the field arrangement pattern sequence Ri is taken as the source current sample, and the r+1-th field arrangement pattern in the field arrangement pattern sequence Rj is taken as the target successor sample to form a transfer entropy sample group. This process is repeated until r is the number of field arrangement patterns minus 1, to obtain all transfer entropy sample groups from constraint feature field Fi to constraint feature field Fj.

[0010] Preferably, the entropy value writing unit is used to sequentially form a binary sample joint term by the target current sample and the target successor sample in the same transfer entropy sample group, and count the number of occurrences of each target current sample, target successor sample and binary sample joint term in all transfer entropy sample groups, and calculate the first conditional entropy according to the permutation transfer entropy algorithm. In the same transfer entropy sample group, the source current sample, the target current sample, and the target successor sample are arranged in order to form a ternary sample joint term, and the occurrence frequency of the source current sample, the target current sample, the target successor sample, and the ternary sample joint term in all transfer entropy sample groups is counted. The second conditional entropy is calculated according to the permutation transfer entropy algorithm. Subtract the second conditional entropy from the first conditional entropy to obtain the initial permutation transfer entropy value of constraint feature fields Fi to Fj. If the initial permutation transfer entropy value is negative, write the permutation transfer entropy value of constraint feature fields Fi to Fj to 0; otherwise, determine the initial permutation transfer entropy value as the permutation transfer entropy value of constraint feature fields Fi to Fj.

[0011] Preferably, the admitted edge generation module includes a transcribing unit and an admitted edge writing unit; The transmission coding unit is used to sort the permutation transmission entropy values ​​of all constraint feature fields Fi to constraint feature fields Fj in descending order. The constraint feature field Fi corresponding to each permutation transmission entropy value is recorded as the preceding field, and the constraint feature field Fj is recorded as the following field. A candidate transmission field pair is formed by the preceding field, the following field, and the current permutation transmission entropy value. All candidate transmission field pairs are arranged according to the descending order of the permutation transmission entropy values ​​to obtain a transmission field pair sequence. Each transmission field pair in the transmission field pair sequence includes a preceding field, a following field, a permutation transmission entropy value, and a sorting position. Extract the first field transmission pair in the field transmission pair sequence, read the preceding field, following field, and sorting position of the current field transmission pair, and establish field transmission code elements. In the field transmission code elements, fill the preceding field into the source field code position, fill the following field into the guided field code position, and fill the sorting position into the field transmission order position. The admission edge writing unit is used to form a tuple of the source field difference value and the guided field difference value under the same time period identifier when the source field code, the guided field code and the field transmission order have been written, and to arrange all time period identifiers in ascending order according to the lexicographical order of the tuples, and to set the sorted time period identifier sequence as the time period recommendation table. If the two time period identifiers correspond to the same tuple, then they are arranged in lexicographical order according to the time period identifiers; If the source field code, the guided field code, the field transmission order, and the time period recommendation table all exist, the field transmission code is formed and marked as generating an admission edge; otherwise, it is marked as a review.

[0012] Preferably, the access decision module includes a transmission code screening unit and an access decision unit; The transmission code filtering unit is used to read the field transmission code corresponding to each work order identifier, and read each time period identifier in the time period recommendation table from the current field transmission code, generate the left node code based on the work order identifier, and generate the right node code based on the time period identifier; The left node code, right node code, source field code, guided field code, field propagation order, and time period order are concatenated to form an edge code string. A hash operation is performed on the edge code string to obtain the edge primary key code. The edge primary key code is then used as an index to write the edge attribute to form an admission edge. The edge attributes include left node code, right node code, source field code, guided field code, field propagation order, and time period order; When the work order identifier and the time period identifier are the same in the primary key codes of two edges, it is determined that the two edge identifiers correspond to the same admission edge. When the work order identifier or time period identifier in the primary key code of the two edges is different, it is determined that the two edge identifiers correspond to different access edges; By summing up all the different admission edges, we obtain the admission edge set.

[0013] Preferably, the admission decision unit includes an edge matching unit and a scheduling coordination unit; The edge order matching unit is used to traverse all the admission edges in the admission edge set, and read the work order identifier and time period identifier in each admission edge. The work order identifier is used as the left node identifier, the time period identifier is used as the right node identifier, and the admission edge is used as the candidate matching edge to build a bipartite graph. From all candidate matching edges in the bipartite graph, select a candidate matching edge that simultaneously satisfies that there are no identical work order identifiers and no identical time period identifiers among the candidate matching edges, and record it as a candidate matching result. The number of matched work order identifiers in each candidate matching result is counted using statistical methods, and the candidate matching result with the largest number of matched work order identifiers is selected as the first candidate matching result. When there are more than two first candidate matching results, read all the selected inbound edges in each first candidate matching result, sort them from front to back according to the order of the inbound edges, and select the matching result corresponding to the first selected edge sequence as the target matching result.

[0014] Preferably, the scheduling coordination unit is used to mark the cost scheduling table of the work order identifier that has been matched in the target matching result as an admission, that is: allowing the work order to be scheduled corresponding to the current work order identifier to be arranged to the candidate production line time period corresponding to the current time period identifier; Read the field passcode corresponding to the work order identifier that was not matched by the target matching result; When the field transmission code is marked as generating an admission edge, the cost scheduling table is marked as re-analyzed. That is, the current work order identifier has generated a field transmission code and an admission edge, but the current admission edge is not selected in the target matching results. Therefore, the current work order identifier is not written into the formal production scheduling results, and the counterfactual coding module is returned to reselect the candidate production line time period for iterative analysis. When the field transmission code is marked as "reviewed", the cost scheduling table is also marked as "reviewed". This means that the constraint difference sequence corresponding to the current work order identifier has not generated a field transmission code that can be used for the admission edge set. Therefore, the current work order identifier does not participate in this round of bipartite graph matching, and the list of work orders to be reviewed is output. After manual confirmation, the list is returned to the counterfactual coding module to reselect candidate production line time periods for iterative analysis.

[0015] This invention provides an industry-driven intelligent accounting and scheduling collaborative system for production costs. It offers the following advantages: (1) The counterfactual coding module of this system is used to collect input interval objects from the existing business system through the data interface. The input interval objects include cost tokens and candidate production line time periods. Among them, the cost tokens include work order identifiers, material lock-in price intervals, process occupancy intervals, shift permission vectors, changeover entry strings, and delivery control intervals. The candidate production line time periods include time period identifiers, time period intervals, time period capacity intervals, candidate shift codes, and predecessor tail status codes. The counterfactual coding module generates lock-in price difference values ​​based on the material lock-in price intervals and time period intervals, generates process difference values ​​based on the process occupancy intervals and time period capacity intervals, generates shift difference values ​​based on the shift permission vectors and candidate shift codes, generates changeover difference values ​​based on the changeover entry strings and predecessor tail status codes, and generates delivery difference values ​​based on the delivery control intervals and time period intervals. These are then combined into a production difference value data group to finally obtain a constraint difference sequence. Compared with the existing method of relying on manual reading of work order cost data systems and production planning data systems to determine the feasibility of production scheduling, this module transforms multiple types of cost constraints and production line time period constraints into a computable data sequence, so that subsequent work order scheduling judgments have a unified data foundation.

[0016] (2) The system's transmission code generation module is used to extract constraint feature fields from the constraint difference sequence based on the work order identifier. The constraint feature fields include material lock-in price field, process occupancy field, shift adaptation field, type change acceptance field, and delivery occupancy field. It calculates the first conditional entropy and the second conditional entropy through the field arrangement pattern sequence to obtain the permutation transmission entropy value of constraint feature field Fi to constraint feature field Fj. The admission edge generation module then sorts all the permutation transmission entropy values ​​of constraint feature field Fi to constraint feature field Fj in descending order, extracts the first-ranked transmission field pair, writes the preceding field into the source field code, writes the following field into the guided field code, writes the sorting position into the field transmission order, and combines it with the time period recommendation table to form the field transmission code. Compared to existing fixed rule priorities or single cost value sorting methods, this process first analyzes the transmission status of material lock-in price field, process occupation field, shift adaptation field, changeover acceptance field, and delivery occupation field in the current work order data, and then obtains the source field code, guided field code, field transmission order, and time period recommendation table that participate in the admission judgment, so that the selection of candidate production line time periods no longer depends solely on static scheduling rules.

[0017] (3) The system's admission decision module is used to read field transmission codes, generate left-side node codes based on work order identifiers, generate right-side node codes based on time period identifiers, and then concatenate the left-side node codes, right-side node codes, source field codes, guided field codes, field transmission order, and time period order to form an edge code string. After hashing, the edge primary key code is obtained and written into the edge attributes to form an admission edge. Finally, it is summarized into an admission edge set. The admission decision module establishes a bipartite graph based on the admission edge set, filters candidate matching results, and determines the target matching result based on the number of matched work order identifiers and the admission edge order. For work order identifiers that have been matched in the target matching result, the cost scheduling table is marked as admission. For work order identifiers that have generated field transmission codes and admission edges but have not been selected by the target matching result, the cost scheduling table is marked as re-analysis. For work order identifiers that have not generated field transmission codes that can be used for the admission edge set, the cost scheduling table is marked as review. Compared to existing technologies that separate production cost accounting and production planning and scheduling, this system integrates cost token analysis, candidate production line time period analysis, field transmission code generation, admission edge generation, bipartite graph matching, and work order scheduling status writing into the same data processing flow. This allows production cost accounting results to directly participate in work order admission matching and enables pending work orders to obtain clear flow paths in the three states of admission, re-analysis, and review. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the process of the industry-driven intelligent accounting and scheduling collaborative system for production costs of the present invention; Figure 2 This is a logic block diagram of the intelligent accounting and scheduling collaborative system for industry-driven production costs of the present invention. Detailed Implementation

[0019] 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 embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1 Please see Figure 1 This invention provides an industry-driven intelligent accounting and scheduling collaborative system for production costs. To achieve the above objectives, this invention is implemented through the following technical solutions: including a counterfactual encoding module, a transmission code generation module, an admission edge generation module, and an admission decision module. The counterfact coding module collects input interval objects, including cost tokens and candidate production line time periods, from the existing business system through a data interface, and performs differential coding to generate a constrained differential sequence. The transmission code generation module extracts constraint feature fields from the constraint difference sequence based on the work order identifier in the cost token, constructs a field arrangement pattern sequence, calculates the first conditional entropy and the second conditional entropy through the arrangement transfer entropy algorithm, and obtains the arrangement transfer entropy value by subtracting the first conditional entropy from the second conditional entropy. The admission edge generation module sorts the propagation entropy values ​​in descending order, extracts the first-ranked propagation field pair and writes it into the source field code, the guided field code, the field propagation order and the time period recommendation table to form a field propagation code for marking. The admission decision module constructs admission edges and combines them with work order identifiers and time period identifiers in candidate production line time periods to build a bipartite graph and filter candidate matching results. It then counts the candidate matching results with the most work order identifiers to obtain the target matching result and executes work order scheduling status writing based on the target matching result and the field transmission code marking result.

[0021] In this embodiment, the counterfactual coding module first collects input interval objects, including cost tokens and candidate production line time periods, from the existing business system through a data interface, and then performs differential coding to generate a constrained differential sequence. For example, a certain cost token corresponds to the work order identifier "WO-001", its material lock-in price range is "08:00-12:00", its process occupancy range is "09:00-11:00", its shift permission vector indicates that the early shift is allowed, its changeover entry string is "material A-mold B", and its delivery control range is "before 16:00 on the same day"; there are time period identifiers "P-01" and "P-02" in the candidate production line time periods, where the time period identifier "P-01" has a time period range of "08:30-11:30", a time period capacity range of "08:00-12:00", a candidate shift code of early shift, and a predecessor tail state code of "material A-mold B". The counterfactual coding module compares cost tokens and candidate production line time periods, encoding material lock-in price ranges with time period ranges, process occupancy ranges with time period capacity ranges, shift permission vectors with candidate shift codes, changeover entry strings with predecessor tail state codes, and delivery control ranges with time period ranges, ultimately forming a constrained differential sequence. In this way, cost and time period information, originally scattered across existing business systems, is transformed into a data foundation that can be subsequently calculated, sorted, and matched.

[0022] After obtaining the constraint difference sequence, the transmission code generation module extracts constraint feature fields from the constraint difference sequence based on the work order identifier in the cost token, and constructs a field arrangement pattern sequence. It then calculates the first and second conditional entropies using the arrangement transmission entropy algorithm, and obtains the arrangement transmission entropy value by subtracting them. Taking work order identifier "WO-001" as an example, if the constraint difference sequence shows a good match between the material lock-in price range and the candidate production line time period, but the match between the changeover entry string and the predecessor tail state code varies significantly, the system will convert the material lock-in price field, process occupancy field, shift adaptation field, changeover acceptance field, and delivery occupancy field into a field arrangement pattern sequence, and then analyze which constraint feature transmission field has a more dominant role in the subsequent time period selection. Assuming the permutation propagation entropy value shows that the impact of the "Type Change Acceptance Field" on the "Process Occupation Field" is ranked first, the admission edge generation module sorts the permutation propagation entropy values ​​in descending order, extracts the top-ranked propagation field pair, and writes it into the source field code "Type Change Acceptance Field," the guided field code "Process Occupation Field," the field propagation order "1," and the time period recommendation table. This is then used to assemble the field propagation code for marking. Compared to existing technologies that rely on manual experience, fixed priorities, or single-capacity sequence for work order scheduling, this solution does not directly insert work orders into idle time periods. Instead, it first analyzes the constraint difference sequence, then obtains the permutation propagation entropy value, propagation field pair, source field code, guided field code, field propagation order, and time period recommendation table, giving the selection of candidate production line time periods a clear computational path.

[0023] The admission decision module constructs admission edges based on field transmission codes and builds a bipartite graph by combining work order identifiers and time period identifiers in candidate production line time periods. It then filters candidate matching results, counts the candidate matching result with the most work order identifiers to obtain the target matching result, and finally writes the work order scheduling status based on the target matching result and the field transmission code marking results. For example, work order identifier "WO-001" can match time period identifiers "P-01" and "P-02", and work order identifier "WO-002" can also match time period identifier "P-01". In this case, if only sorted by idle capacity, both work orders might be directed to the same candidate production line time period. However, the admission decision module uses the work order identifier as one side node of the bipartite graph, the time period identifier as the other side node, and the admission edges as candidate matching edges, filtering out the target matching result with the most work order identifiers that do not conflict from the candidate matching results. If the target matching result is "WO-001—P-01, WO-002—P-02", the system executes the work order scheduling status writing based on the target matching result, writing the matched work order identifier into the admission status; if a work order identifier already has a field transmission code but has not entered the target matching result, it is written into the re-analysis status; if a work order identifier has not formed a usable field transmission code, it is written into the review status. Thus, this scheme undertakes tasks such as input interval object acquisition, constraint difference sequence generation, constraint feature field extraction, field arrangement pattern sequence construction, arrangement propagation entropy value calculation, field transmission code marking, admission edge construction, bipartite graph matching, and work order scheduling status writing, enabling production cost accounting results to enter the work order scheduling collaboration process, achieving the goal of balancing cost constraints, time period adaptation, and production scheduling conflict handling.

[0024] Example 2 Please refer to Figure 2 Specifically: the counterfactual encoding module includes an object reading unit and a differential encoding unit; The object reading unit is used to collect input range objects from existing business systems through a data interface; The existing business systems include a work order cost data system and a production planning data system; The input interval object includes cost tokens and candidate production line time periods, and the input interval object adopts a left-closed and right-open interval; wherein, the left-closed and right-open interval means that the interval includes the start point but does not include the end point; Read cost tokens from the work order cost data system and candidate production line time periods from the production plan data system; The cost token includes a work order identifier, a material price lock range, a process occupancy range, a shift permission vector, a changeover entry string, and a delivery control range. The candidate production line time period includes time period identifier, time period interval, time period capacity interval, candidate shift code, and predecessor tail status code.

[0025] The differential encoding unit is used to differentially encode the cost token and candidate production line time period to obtain a production differential value data group, as follows; The material price lock-in range is used as the first comparison range, and the time period range is used as the second comparison range. The lock-in price difference score is obtained by encoding according to Allen's interval coding rules. The delivery control interval is used as the first comparison interval, and the time interval is used as the second comparison interval. The delivery difference value is obtained by encoding according to Allen's interval encoding rules. Using the process occupancy interval as the first comparison interval and the time period accommodating interval as the second comparison interval, the interval inclusion code is executed to obtain the process difference value. The interval inclusion code includes: When the start of the process-occupied interval is no earlier than the start of the time period-accommodated interval, and the end of the process-occupied interval is no later than the end of the time period-accommodated interval, the process difference value is written as 0. When the process occupancy interval overlaps with the time period accommodating interval, and the process occupancy interval does not meet the complete accommodating condition, the process difference value is written as 1; When the process occupancy interval and the time period accommodation interval do not overlap, the process difference value is written as 2; Based on the one-hot coding technique, the candidate shift code is converted into a one-hot vector of the same length as the shift permission vector, and the Hamming distance between the one-hot vector and the shift permission vector is calculated by the Hamming distance algorithm and denoted as the shift difference value. The Levenshtein edit distance between the transformation entry string and the predecessor tail state code is calculated based on the minimum edit distance algorithm and recorded as the transformation difference value; The production differential data set includes price lock-in differential, process differential, shift differential, changeover differential, and delivery differential; The production difference data groups under the same work order identifier are arranged according to the order of candidate production line time periods to obtain the constrained difference sequence; Among them, the price lock-in difference value forms a price lock-in field difference sequence according to the arrangement order of the time period identifiers; the process difference value forms a process field difference sequence according to the arrangement order of the time period identifiers; the shift difference value forms a shift field difference sequence according to the arrangement order of the time period identifiers; the changeover difference value forms a changeover field difference sequence according to the arrangement order of the time period identifiers; and the delivery difference value forms a delivery field difference sequence according to the arrangement order of the time period identifiers. The constraint difference sequence includes a price lock field difference sequence, a process field difference sequence, a shift field difference sequence, a changeover field difference sequence, and a delivery field difference sequence.

[0026] In this embodiment, the object reading unit first collects input interval objects from the existing business system through the data interface. The existing business system includes a work order cost data system and a production plan data system. The input interval objects include cost tokens and candidate production line time periods, and the input interval objects adopt a left-closed and right-open interval. The object reading unit reads the cost token from the work order cost data system and the candidate production line time periods from the production plan data system. The cost token includes a work order identifier, material lock-in price range, process occupancy range, shift permission vector, changeover entry string, and delivery control range. The candidate production line time periods include a time period identifier, time period range, time period capacity range, candidate shift code, and predecessor tail state code. Subsequently, the differential coding unit performs differential coding on the cost token and candidate production line time periods: using the material lock-in price range as the first comparison range and the time period range as the second comparison range, coding is performed according to Allen's range coding rules to obtain the lock-in price difference value; using the delivery control range as the first comparison range and the time period range as the second comparison range, coding is performed according to Allen's range coding rules to obtain the delivery difference value; using the process occupancy range as the first comparison range and the time period accommodation range as the second comparison range, range inclusion coding is performed to obtain the process difference value; using one-hot coding technology, the candidate shift code is converted into a one-hot vector of the same length as the shift permission vector, and the Hamming distance between the one-hot vector and the shift permission vector is calculated using the Hamming distance algorithm, which is recorded as the shift difference value; using the minimum edit distance algorithm, the Levenshtein edit distance between the change-in entry string and the predecessor tail state code is calculated, and recorded as the change-in difference value, thereby forming a production difference value data group including lock-in price difference value, process difference value, shift difference value, change-in difference value, and delivery difference value. Subsequently, the system arranges the production difference value data groups under the same work order identifier according to the arrangement order of candidate production line time periods, obtaining a constraint difference sequence. Among them, the price lock difference value forms a price lock field difference sequence according to the arrangement order of time period identifiers, the process difference value forms a process field difference sequence according to the arrangement order of time period identifiers, the shift difference value forms a shift field difference sequence according to the arrangement order of time period identifiers, the changeover difference value forms a changeover field difference sequence according to the arrangement order of time period identifiers, and the delivery difference value forms a delivery field difference sequence according to the arrangement order of time period identifiers. The counterfactual encoding module transforms the material price lock interval, process occupancy interval, shift permission vector, changeover entry string, delivery control interval, time period interval, time period capacity interval, candidate shift code, and predecessor tail state code, which were originally distributed in the work order cost data system and production planning data system, into a unified constraint difference sequence. This enables the subsequent transmission code generation module, admission edge generation module, and admission decision module to perform field transmission analysis, field transmission code generation, admission edge construction, and work order scheduling status writing based on the same set of difference results.Compared to existing technologies that rely solely on manual experience, fixed production scheduling rules, or single capacity availability information for work order arrangement, this implementation method first analyzes the interval differences, shift differences, and changeover differences between cost tokens and candidate production line time periods. Then, it obtains differential sequences of price locking, process, shift, changeover, and delivery fields, enabling production cost accounting data to directly enter the scheduling and coordination process. This achieves the goals of unified cost constraint expression, supporting candidate production line time period screening, and assisting in work order scheduling status determination. Furthermore, it ensures that subsequent target matching results meet on-site constraints such as material price locking, process occupancy, shift permission, changeover acceptance, and delivery control.

[0027] Example 3 Please refer to Figure 2 Specifically: the transduction code generation module includes a sequence conversion unit and an entropy transmission orientation unit; The sequence conversion unit is used to group the constraint difference sequence according to the work order identifier to obtain the constraint difference sequence records arranged in time period order under the same work order identifier; The constraint feature fields are extracted from the constraint difference sequence records, specifically: the text field of the material lock-up price range is extracted to obtain the material lock-up price field; the text field of the process occupancy range is extracted to obtain the process occupancy field; the text field of the shift permission vector is extracted to obtain the shift adaptation field; the text field of the change-up entry string is extracted to obtain the change-up acceptance field; and the text field of the delivery control range is extracted to obtain the delivery occupancy field. The constraint feature fields include any one of the following: material lock-in price field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field.

[0028] The entropy transmission orientation unit includes a mapping sampling unit and an entropy value writing unit; The mapping sampling unit is used to perform field mapping between the constraint difference sequence and the constraint feature field to obtain the field arrangement pattern sequence; The field mapping is as follows: Read the permutation pattern sequences Ri and Rj of any two constraint feature fields Fi and Fj, where Fi and Fj represent any field among the material lock-in field, process occupation field, shift adaptation field, changeover acceptance field, or delivery occupation field, and Fi≠Fj. Ri represents the field permutation pattern sequence corresponding to constraint feature field Fi, and Rj represents the field permutation pattern sequence corresponding to constraint feature field Fj, and the number of field permutation patterns of the two is the same. The r-th field arrangement pattern in the field arrangement pattern sequence Rj is taken as the target current sample, the r-th field arrangement pattern in the field arrangement pattern sequence Ri is taken as the source current sample, and the r+1-th field arrangement pattern in the field arrangement pattern sequence Rj is taken as the target successor sample to form a transfer entropy sample group. This process is repeated until r is the number of field arrangement patterns minus 1, to obtain all transfer entropy sample groups from constraint feature field Fi to constraint feature field Fj. Here, r starts from 1 and does not take the last field in the arrangement pattern.

[0029] The entropy writing unit is used to sequentially combine the current target sample and the subsequent target sample into a binary sample joint term within the same transfer entropy sample group. If either the current target sample or the subsequent target sample is missing, a binary sample joint term cannot be formed. The unit also counts the occurrences of each current target sample, subsequent target sample, and binary sample joint term in all transfer entropy sample groups and calculates the first conditional entropy based on the permutation transfer entropy algorithm. Specifically: Where H1 represents the first conditional entropy, Ns represents the total number of transitive entropy sample groups, y and z represent the sequence of one-field permutation patterns of the current target sample and the subsequent target sample, respectively, log2 represents the logarithmic function with base 2, and N(y) and N(y, z) represent the number of times the joint term of the current target sample and the binary sample appears in the total transitive entropy sample groups, respectively. In the same transitive entropy sample set, the source current sample, target current sample, and target successor sample are sequentially combined into a ternary sample joint term. If any one of the source current sample, target current sample, or target successor sample is missing, a ternary sample joint term cannot be formed. The occurrence counts of the source current sample, target current sample, target successor sample, and ternary sample joint term in all transitive entropy sample sets are counted. The second conditional entropy is calculated according to the permutation transitive entropy algorithm, specifically: Where H2 represents the second conditional entropy, x represents a sequence of field permutations of the current source sample, and N(x, y, z) represents the number of times the ternary sample joint term appears in the entire transitive entropy sample group; Subtract the second conditional entropy from the first conditional entropy to obtain the initial permutation transfer entropy value of constraint feature fields Fi to Fj. If the initial permutation transfer entropy value is negative, write the permutation transfer entropy value of constraint feature fields Fi to Fj to 0; otherwise, determine the initial permutation transfer entropy value as the permutation transfer entropy value of constraint feature fields Fi to Fj.

[0030] In this embodiment, the sequence conversion unit first groups the constraint difference sequence generated by the counterfactual encoding module according to the work order identifier, obtaining constraint difference sequence records arranged in time period order under the same work order identifier, and extracts the material lock-in price field, process occupation field, shift adaptation field, changeover acceptance field, and delivery occupation field from the constraint difference sequence records; subsequently, the mapping sampling unit in the entropy orientation unit performs field mapping between the constraint difference sequence and the constraint feature fields to obtain the field arrangement pattern sequence, and reads the field arrangement pattern sequence Ri and the field arrangement pattern sequence Rj for any two different constraint feature fields Fi and Fj, taking the r-th field arrangement pattern in the field arrangement pattern sequence Rj as the target current sample, the r-th field arrangement pattern in the field arrangement pattern sequence Ri as the source current sample, and the r+1-th field arrangement pattern in the field arrangement pattern sequence Rj as the target successor sample, forming a transfer entropy sample group, and traversing it sequentially to obtain constraint feature fields Fi to constraint feature fields. The algorithm first calculates the first conditional entropy by subtracting the second conditional entropy from the first conditional entropy. Then, it sequentially combines the current target sample and the subsequent target sample into a binary sample joint term within the same transfer entropy sample group. The occurrence counts of the current target sample, the subsequent target sample, and the binary sample joint term are counted in the entire transfer entropy sample group. Simultaneously, the current source sample, the current target sample, and the subsequent target sample are sequentially combined into a ternary sample joint term. The occurrence counts of the current source sample, the current target sample, the subsequent target sample, and the ternary sample joint term are counted in the entire transfer entropy sample group. The second conditional entropy is calculated using the permutation transfer entropy algorithm. Finally, the first conditional entropy is subtracted from the second conditional entropy to obtain the initial permutation transfer entropy value from constraint feature field Fi to constraint feature field Fj. When the initial permutation transfer entropy value is negative, the permutation transfer entropy value from constraint feature field Fi to constraint feature field Fj is written to 0; otherwise, the initial permutation transfer entropy value is determined as the permutation transfer entropy value from constraint feature field Fi to constraint feature field Fj. The purpose of this implementation is to transform the temporal changes in the constraint difference sequence into a computable field propagation direction. This allows the admission edge generation module to no longer rely solely on fixed priority to select fields, but instead to determine the sequential role of constraint feature fields in work order scheduling based on the first conditional entropy, the second conditional entropy, and the permutation propagation entropy value. Compared with existing techniques that rely on manual experience, static rules, or single-field sorting, this method can incorporate material lock-in fields, process occupancy fields, shift adaptation fields, changeover acceptance fields, and delivery occupancy fields into the same analysis process. This makes the time period recommendation table more closely match the actual differences between cost tokens and candidate production line time periods, and lays a data foundation for subsequent admission edge construction, bipartite graph matching, target matching result screening, and work order scheduling status writing. This allows production cost accounting results to enter the work order scheduling collaboration process more smoothly.

[0031] Example 4 Please refer to Figure 2 Specifically: the admitted edge generation module includes a transmission coding unit and an admitted edge writing unit; The transmission coding unit is used to sort the permutation transmission entropy values ​​of all constraint feature fields Fi to constraint feature fields Fj in descending order. The constraint feature field Fi corresponding to each permutation transmission entropy value is recorded as the preceding field, and the constraint feature field Fj is recorded as the following field. A candidate transmission field pair is formed by the preceding field, the following field, and the current permutation transmission entropy value. All candidate transmission field pairs are arranged according to the descending order of the permutation transmission entropy values ​​to obtain a transmission field pair sequence. Each transmission field pair in the transmission field pair sequence includes a preceding field, a following field, a permutation transmission entropy value, and a sorting position. If the entropy values ​​of two candidate transmission field pairs are the same, then the candidate transmission field pairs are arranged according to the fixed order of the preceding field in the material lock-in price field, process occupation field, shift adaptation field, changeover acceptance field, or delivery occupation field. If the preceding fields are the same, they are arranged according to the fixed order of the following fields in each candidate transmission field pair: material lock-in field, process occupation field, shift adaptation field, changeover acceptance field, or delivery occupation field. Extract the first field transmission pair in the sequence of field transmission pairs. Read the preceding field, following field, and sorting position of the current field transmission pair and establish field transmission code elements. In the field transmission code elements, fill the preceding field into the source field code position, fill the following field into the guided field code position, fill the sorting position into the field transmission order position, and write 1 to the field transmission order position. The transmitted field pair is the first transmitted field pair in the sequence of transmitted field pairs, so the field transmission order position is written as 1. The admitted edge writing unit is used to write the source field difference value when the source field code, the guided field code, and the field transmission order have been written: When the source field code is a material lock price field, the source field difference value is the lock price difference value; When the source field code is a field occupied by the process, the source field difference value is the process difference value; When the source field code is a shift matching field, the source field difference value is the shift difference value; When the source field code is a type-transformation acceptor field, the source field difference value is the type-transformation difference value; When the source field code is a delivery occupied field, the source field difference value is the delivery difference value; The difference value of the guided field: When the guided field code is a material lock-in price field, the guided field difference value is the lock-in price difference value; When the guided field code is a field occupied by the process, the guided field difference value is the process difference value; When the guided field code is a shift matching field, the guided field difference value is the shift difference value; When the code of the guided field is a conversion acceptor field, the difference value of the guided field is the conversion difference value; When the guided field code is a delivery occupied field, the guided field difference value is the delivery difference value; Form a tuple from the source field difference value and the guided field difference value under the same time period identifier, sort all time period identifiers in ascending order according to the lexicographical order of the tuples, and set the sorted time period identifier sequence as the time period recommendation table; If the two time period identifiers correspond to the same tuple, then they are arranged in lexicographical order according to the time period identifiers; If the source field code, the guided field code, the field transmission order, and the time period recommendation table all exist, the field transmission code is formed and marked as generating an admission edge; otherwise, it is marked as a review.

[0032] In this embodiment, the transmission coding unit first sorts the permutation transmission entropy values ​​of all constraint feature fields Fi to Fj in descending order. If the permutation transmission entropy values ​​are the same, the constraint feature field Fi is arranged according to the fixed order in the material lock-in field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field. If the constraint feature fields Fi are the same, the constraint feature field Fj is arranged according to the fixed order in the material lock-in field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field. The sorted transmission field pair sequence is recorded as the transmission field pair sequence. Each transmission field pair includes a preceding field, a following field, permutation transmission entropy value, and sorting position. Then, the first-ranked transmission field pair in the transmission field pair sequence is extracted, the preceding field is written into the source field code, the following field is written into the guided field code, the sorting position is written into the field transmission order, and the field transmission order is written to 1. When the source field code, guided field code, and field transmission order have been written, the admission edge writing unit identifies the same process. For each time period under a single identifier, the source field difference value and the guided field difference value are respectively taken as the price lock-in field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field. When the source field code is a material price lock-in field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field, the source field difference value is taken as the price lock-in difference value, process difference value, shift difference value, changeover difference value, or delivery difference value. When the guided field code is a material price lock-in field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field, the guided field difference value is taken as the price lock-in difference value, process difference value, or delivery difference value. The values ​​are: shift difference value, type change difference value, or delivery difference value. Then, the source field difference value and the guided field difference value under the same time period identifier are formed into a tuple. All time period identifiers are sorted in ascending order according to the lexicographical order of the tuples. The sorted time period identifier sequence is set as the time period recommendation table. If the tuples corresponding to two time period identifiers are the same, they are sorted in lexicographical order of the time period identifiers. When the source field code, the guided field code, the field transmission order, and the time period recommendation table all exist, the field transmission code is formed and marked as generating an admission edge; otherwise, it is marked as a review. This module transforms the field action direction expressed by the permutation transfer entropy value into source field code, guided field code, field transmission order, and time period recommendation table. This allows candidate production line time periods to no longer be sorted solely by idle time or manual experience, but rather by cost tokens, constraint difference sequences, constraint feature fields, and permutation transfer entropy values ​​for pre-admission screening. This achieves the purpose of preparing data for the admission decision module to construct admission edges, generate admission edge sets, screen candidate matching results, and obtain target matching results. Material lock-in price fields, process occupancy fields, shift adaptation fields, changeover acceptance fields, and delivery occupancy fields can all enter a unified coding process. The recommended order of each time period identifier under the work order identifier has a clear calculation basis, and the review situation can also be distinguished separately through the field transmission code marking results. This makes the subsequent work order scheduling status writing more in line with the production site needs such as cost control, process arrangement, shift adaptation, changeover acceptance, and delivery control.

[0033] Example 5 Please refer to Figure 2 Specifically: the access decision module includes a transmission code screening unit and an access decision unit; The transmission code filtering unit is used to read the field transmission code corresponding to each work order identifier, and read each time period identifier in the time period recommendation table from the current field transmission code. It generates a left node code based on the work order identifier and a right node code based on the time period identifier, as follows; The work order identifier is converted into a UTF-8 encoded string and its byte length is read. The original work order node string is formed according to the order of WORK|byte length|work order identifier string. Then, the SHA-256 hash operation is performed on the original work order node string to obtain the work order hash string. The left node code is formed according to the order of NW|work order hash string. The time period identifier is converted into a UTF-8 encoded string and its byte length is read. The original string of the time period node is formed according to the order of PERIOD|byte length|time period identifier string. Then, the SHA-256 hash operation is performed on the original string of the time period node to obtain the time period hash string. The right node code is formed according to the order of NP|time period hash string. The left node code, right node code, source field code, guided field code, field propagation order, and time period order are concatenated to form an edge code string. A hash operation is performed on the edge code string to obtain the edge primary key code. The edge primary key code is then used as an index to write the edge attribute to form an admission edge. The edge attributes include left node code, right node code, source field code, guided field code, field propagation order, and time period order; When the work order identifier and the time period identifier are the same in the primary key codes of two edges, it is determined that the two edge identifiers correspond to the same admission edge. When the work order identifier or time period identifier in the primary key code of the two edges is different, it is determined that the two edge identifiers correspond to different access edges; By summing up all the different admission edges, we obtain the admission edge set.

[0034] The admission decision-making unit includes an edge order matching unit and a scheduling coordination unit; The edge order matching unit is used to traverse all the admission edges in the admission edge set, and read the work order identifier and time period identifier in each admission edge. The work order identifier is used as the left node identifier, the time period identifier is used as the right node identifier, and the admission edge is used as the candidate matching edge to build a bipartite graph. From all candidate matching edges in the bipartite graph, select a candidate matching edge that simultaneously satisfies that there are no identical work order identifiers and no identical time period identifiers among the candidate matching edges, and record it as a candidate matching result. The number of matched work order identifiers in each candidate matching result is counted using statistical methods, and the candidate matching result with the largest number of matched work order identifiers is selected as the first candidate matching result. When there are more than two first candidate matching results, read all the selected inbound edges in each first candidate matching result, sort them from front to back according to the order of the inbound edges, and select the matching result corresponding to the first selected edge sequence as the target matching result.

[0035] The scheduling and coordination unit is used to mark the cost scheduling table of the work order identifier that has been matched in the target matching result as admission, that is: allowing the work order to be scheduled corresponding to the current work order identifier to be arranged to the candidate production line time period corresponding to the current time period identifier; Read the field passcode corresponding to the work order identifier that was not matched by the target matching result; When the field transmission code is marked as generating an admission edge, the cost scheduling table is marked as re-analyzed. That is, the current work order identifier has generated a field transmission code and an admission edge, but the current admission edge is not selected in the target matching results. Therefore, the current work order identifier is not written into the formal production scheduling results, and the counterfactual coding module is returned to reselect the candidate production line time period for iterative analysis. When the field transmission code is marked as "reviewed", the cost scheduling table is also marked as "reviewed". This means that the constraint difference sequence corresponding to the current work order identifier has not generated a field transmission code that can be used for the admission edge set. Therefore, the current work order identifier does not participate in this round of bipartite graph matching, and the list of work orders to be reviewed is output. After manual confirmation, the list is returned to the counterfactual coding module to reselect candidate production line time periods for iterative analysis.

[0036] In this embodiment, the transmission code filtering unit first reads the field transmission code corresponding to the work order identifier, and then reads the time period identifiers from the time period recommendation table from the current field transmission code. Then, it generates a left-side node code based on the work order identifier and a right-side node code based on the time period identifier. Specifically, the work order identifier is converted to a UTF-8 encoded string and its byte length is read. The original work order node string is formed according to the order of WORK|byte length|work order identifier string. Then, a SHA-256 hash operation is performed on the original work order node string to obtain a work order hash string, and a left-side node code is formed according to the order of NW|work order hash string. The time period identifier is converted to a UTF-8 encoded string and its byte length is read. The original time period node string is formed according to the order of PERIOD|byte length|time period identifier string, and then a SHA-256 hash operation is performed on the original time period node string. SHA-256 hashing is performed to obtain the time period hash string, and the right node code is formed according to the order of NP|time period hash string. Then, the left node code, right node code, source field code, guided field code, field propagation order, and time period order are concatenated to form the edge encoding string. The edge encoding string is hashed to obtain the edge primary key code, and the edge attribute is written with the edge primary key code as the index to form the admission edge. The edge attribute includes the left node code, right node code, source field code, guided field code, field propagation order, and time period order. When the work order identifier and the time period identifier in the two edge primary keys are the same, it is determined that the two edge identifiers correspond to the same admission edge. When the work order identifier or the time period identifier in the two edge primary keys are different, it is determined that the two edge identifiers correspond to different admission edges. Then, all the different admission edges are summarized to obtain the admission edge set.The edge matching unit in the admission decision unit traverses all admission edges in the admission edge set, reads the work order identifier and time period identifier from each admission edge, and constructs a bipartite graph by using the work order identifier as the left node identifier, the time period identifier as the right node identifier, and the admission edge as the candidate matching edge. From all candidate matching edges in the bipartite graph, it selects candidate matching edges that simultaneously satisfy the condition that no two candidate matching edges have the same work order identifier or the same time period identifier, and records this as a candidate matching result. Then, it uses statistical methods to count the number of matched work order identifiers in each candidate matching result, and selects the candidate matching result with the largest number of matched work order identifiers as the first candidate matching result. When there are more than two first candidate matching results, it reads all selected admission edges in each first candidate matching result, and then, based on the criteria... The incoming edges are sorted from front to back, and the matching result corresponding to the first selected edge sequence is selected as the target matching result. The scheduling and coordination unit then marks the cost scheduling table of the matched work order identifier in the target matching result as admitted, that is, it allows the work order to be scheduled corresponding to the current work order identifier to be arranged to the candidate production line time period corresponding to the current time period identifier. The field transmission code corresponding to the work order identifier that is not matched by the target matching result is read. When the field transmission code is marked as generating admitted edge, the cost scheduling table is marked as re-analyzed, and the system returns to the counterfactual encoding module to reselect the candidate production line time period for iterative analysis. When the field transmission code is marked as reviewed, the cost scheduling table is marked as reviewed, and a list of work orders to be reviewed is output. After manual confirmation, the system returns to the counterfactual encoding module to reselect the candidate production line time period for iterative analysis. This module transforms the time period recommendation table in the field transmission code into a comparable set of admission edges, and uses a bipartite graph to filter out target matching results that do not occupy the same work order identifier and the same time period identifier. The purpose is to integrate cost tokens, candidate production line time periods, constraint differential sequences, permutation transmission entropy values, source field codes, guided field codes, field transmission order, and time period recommendation tables into the work order scheduling status writing process. Compared with existing technologies that rely solely on manual scheduling, fixed priorities, or idle time period order for arrangement, this implementation method can write the three types of statuses—admission, re-analysis, and review—into the cost scheduling table separately. This allows the formal production scheduling results, work order identifiers that need to reselect candidate production line time periods, and the list of work orders awaiting review that require manual confirmation to enter different processing paths. Its beneficial effects are that the work order scheduling judgment basis is more complete, the source of admission edges is clearer, the candidate matching result screening process is easier to trace, and the production scheduling results are more in line with production site constraints such as material price locking, process occupancy, shift adaptation, changeover acceptance, and delivery control.

[0037] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. An industry-driven intelligent accounting and scheduling collaborative system for production costs, characterized by: It includes a counterfactual encoding module, a transit code generation module, an admission edge generation module, and an admission decision module; The counterfact coding module collects input interval objects, including cost tokens and candidate production line time periods, from the existing business system through a data interface, and performs differential coding to generate a constrained differential sequence. The transmission code generation module extracts constraint feature fields from the constraint difference sequence based on the work order identifier in the cost token, constructs a field arrangement pattern sequence, calculates the first conditional entropy and the second conditional entropy through the arrangement transfer entropy algorithm, and obtains the arrangement transfer entropy value by subtracting the first conditional entropy from the second conditional entropy. The admission edge generation module sorts the propagation entropy values ​​in descending order, extracts the first-ranked propagation field pair and writes the field propagation code into the source field code, the guided field code, the field propagation order, and the time period recommendation table in sequence to form a field propagation code for marking. The admission decision module constructs admission edges and combines them with work order identifiers and time period identifiers in candidate production line time periods to build a bipartite graph and filter candidate matching results. It then counts the candidate matching results with the most work order identifiers to obtain the target matching result and executes work order scheduling status writing based on the target matching result and the field transmission code marking result.

2. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 1, characterized in that: The counterfact coding module includes an object reading unit and a differential coding unit; The object reading unit is used to collect input range objects from existing business systems through a data interface; The existing business systems include a work order cost data system and a production planning data system; The input interval object includes cost tokens and candidate production line time periods, and the input interval object adopts a left-closed and right-open interval; wherein, the left-closed and right-open interval means that the interval includes the start point but does not include the end point; The cost token includes a work order identifier, a material price lock range, a process occupancy range, a shift permission vector, a changeover entry string, and a delivery control range. The candidate production line time period includes time period identifier, time period interval, time period capacity interval, candidate shift code, and predecessor tail status code.

3. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 2, characterized in that: The differential encoding unit is used to differentially encode the cost token and candidate production line time period to obtain a production differential value data group; The production differential data set includes price lock-in differential, process differential, shift differential, changeover differential, and delivery differential; The production difference data groups under the same work order identifier are arranged according to the order of candidate production line time periods to obtain the constrained difference sequence; The constraint difference sequence includes a price lock field difference sequence, a process field difference sequence, a shift field difference sequence, a changeover field difference sequence, and a delivery field difference sequence.

4. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 3, characterized in that: The conduction code generation module includes a sequence conversion unit and an entropy transfer orientation unit; The sequence conversion unit is used to group the constraint difference sequence according to the work order identifier to obtain the constraint difference sequence record; The constraint feature fields are extracted from the constraint difference sequence records, specifically: the text field of the material lock-up price range is extracted to obtain the material lock-up price field; the text field of the process occupancy range is extracted to obtain the process occupancy field; the text field of the shift permission vector is extracted to obtain the shift adaptation field; the text field of the change-up entry string is extracted to obtain the change-up acceptance field; and the text field of the delivery control range is extracted to obtain the delivery occupancy field. The constraint feature fields include any one of the following: material lock-in price field, process occupancy field, shift adaptation field, changeover acceptance field, or delivery occupancy field.

5. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 4, characterized in that: The entropy transmission orientation unit includes a mapping sampling unit and an entropy value writing unit; The mapping sampling unit is used to perform field mapping between the constraint difference sequence and the constraint feature field to obtain the field arrangement pattern sequence; The field mapping is as follows: Read the permutation pattern sequences Ri and Rj of any two constraint feature fields Fi and Fj, where Fi and Fj represent any field among the material lock-in field, process occupation field, shift adaptation field, changeover acceptance field, or delivery occupation field, and Fi≠Fj. Ri represents the field permutation pattern sequence corresponding to constraint feature field Fi, and Rj represents the field permutation pattern sequence corresponding to constraint feature field Fj, and the number of field permutation patterns of the two is the same. The r-th field arrangement pattern in the field arrangement pattern sequence Rj is taken as the target current sample, the r-th field arrangement pattern in the field arrangement pattern sequence Ri is taken as the source current sample, and the r+1-th field arrangement pattern in the field arrangement pattern sequence Rj is taken as the target successor sample to form a transfer entropy sample group. This process is repeated until r is the number of field arrangement patterns minus 1, to obtain all transfer entropy sample groups from constraint feature field Fi to constraint feature field Fj.

6. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 5, characterized in that: The entropy value writing unit is used to sequentially form a binary sample joint term by the target current sample and the target successor sample in the same transfer entropy sample group, and to count the number of occurrences of each target current sample, target successor sample and binary sample joint term in all transfer entropy sample groups, and to calculate the first conditional entropy according to the permutation transfer entropy algorithm. In the same transfer entropy sample group, the source current sample, the target current sample, and the target successor sample are arranged in order to form a ternary sample joint term, and the occurrence frequency of the source current sample, the target current sample, the target successor sample, and the ternary sample joint term in all transfer entropy sample groups is counted. The second conditional entropy is calculated according to the permutation transfer entropy algorithm. Subtract the second conditional entropy from the first conditional entropy to obtain the initial permutation transfer entropy value of constraint feature fields Fi to Fj. If the initial permutation transfer entropy value is negative, write the permutation transfer entropy value of constraint feature fields Fi to Fj to 0; otherwise, determine the initial permutation transfer entropy value as the permutation transfer entropy value of constraint feature fields Fi to Fj.

7. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 6, characterized in that: The admitted edge generation module includes a transcribing unit and an admitted edge writing unit; The transmission coding unit is used to sort the permutation transmission entropy values ​​of all constraint feature fields Fi to constraint feature fields Fj in descending order. The constraint feature field Fi corresponding to each permutation transmission entropy value is recorded as the preceding field, and the constraint feature field Fj is recorded as the following field. A candidate transmission field pair is formed by the preceding field, the following field, and the current permutation transmission entropy value. All candidate transmission field pairs are arranged according to the descending order of the permutation transmission entropy values ​​to obtain a transmission field pair sequence. Each transmission field pair in the transmission field pair sequence includes a preceding field, a following field, a permutation transmission entropy value, and a sorting position. Extract the first field transmission pair in the field transmission pair sequence, read the preceding field, following field, and sorting position in the current field transmission pair, and establish field transmission code elements. In the field transmission code elements, write the preceding field into the source field code position, fill the following field into the guided field code position, and fill the sorting position into the field transmission order position. The admission edge writing unit is used to form a tuple of the source field difference value and the guided field difference value under the same time period identifier when the source field code, the guided field code and the field transmission order have been written, and to arrange all time period identifiers in ascending order according to the lexicographical order of the tuples, and to set the sorted time period identifier sequence as the time period recommendation table. If the source field code, the guided field code, the field transmission order, and the time period recommendation table all exist, the field transmission code is formed and marked as generating an admission edge; otherwise, it is marked as a review.

8. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 7, characterized in that: The access decision module includes a transmission code screening unit and an access decision unit; The transmission code filtering unit is used to read the field transmission code corresponding to each work order identifier, and read each time period identifier in the time period recommendation table from the current field transmission code, generate the left node code based on the work order identifier, and generate the right node code based on the time period identifier; The left node code, right node code, source field code, guided field code, field propagation order, and time period order are concatenated to form an edge code string. A hash operation is performed on the edge code string to obtain the edge primary key code. The edge primary key code is then used as an index to write the edge attribute to form an admission edge. The edge attributes include left node code, right node code, source field code, guided field code, field propagation order, and time period order; When the work order identifier and the time period identifier are the same in the primary key codes of two edges, it is determined that the two edge identifiers correspond to the same admission edge. When the work order identifier or time period identifier in the primary key code of the two edges is different, it is determined that the two edge identifiers correspond to different access edges; By summing up all the different admission edges, we obtain the admission edge set.

9. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 8, characterized in that: The admission decision-making unit includes an edge order matching unit and a scheduling coordination unit; The edge order matching unit is used to traverse all the admission edges in the admission edge set, and read the work order identifier and time period identifier in each admission edge. The work order identifier is used as the left node identifier, the time period identifier is used as the right node identifier, and the admission edge is used as the candidate matching edge to build a bipartite graph. From all candidate matching edges in the bipartite graph, select a candidate matching edge that simultaneously satisfies that there are no identical work order identifiers and no identical time period identifiers among the candidate matching edges, and record it as a candidate matching result. The number of matched work order identifiers in each candidate matching result is counted using statistical methods, and the candidate matching result with the largest number of matched work order identifiers is selected as the first candidate matching result. When there are more than two first candidate matching results, read all the selected inbound edges in each first candidate matching result, sort them from front to back according to the order of the inbound edges, and select the matching result corresponding to the first selected edge sequence as the target matching result.

10. The industry-driven intelligent accounting and scheduling collaborative system for production costs according to claim 9, characterized in that: The scheduling and coordination unit is used to mark the cost scheduling table of the work order identifier that has been matched in the target matching result as admission, that is: allowing the work order to be scheduled corresponding to the current work order identifier to be arranged to the candidate production line time period corresponding to the current time period identifier; Read the field passcode corresponding to the work order identifier that was not matched by the target matching result; When the field transmission code is marked as generating an admission edge, the cost scheduling table is marked as re-analyzed. That is, the current work order identifier has generated a field transmission code and an admission edge, but the current admission edge is not selected in the target matching results. Therefore, the current work order identifier is not written into the formal production scheduling results, and the counterfactual coding module is returned to reselect the candidate production line time period for iterative analysis. When the field transmission code is marked as "reviewed", the cost scheduling table is also marked as "reviewed". This means that the constraint difference sequence corresponding to the current work order identifier has not generated a field transmission code that can be used for the admission edge set. Therefore, the current work order identifier does not participate in this round of bipartite graph matching, and the list of work orders to be reviewed is output. After manual confirmation, the list is returned to the counterfactual coding module to reselect candidate production line time periods for iterative analysis.

Images