A method and system for automatically generating production scheduling plans based on the photovoltaic crystal pulling industry

Abstract

This invention discloses a method and system for automatically generating production schedules in the photovoltaic crystal pulling industry, relating to the field of industrial manufacturing production scheduling technology. The method includes: receiving order data to be scheduled, real-time furnace status data, and material inventory data; constructing a dynamic rule base; calling structured parameters in the dynamic rule base to perform process constraint pre-checks, generating a candidate set of scheduleable orders and furnaces; dividing future production cycles into execution lock zones and pre-schedule adjustment zones, and constructing a comprehensive production scheduling evaluation index; based on the comprehensive production scheduling evaluation index, using a hierarchical allocation strategy to traverse and match the candidate set of scheduleable orders and furnaces, generating an initial production scheduling sequence; when a preset disturbance trigger condition is identified, performing local rearrangement repair based on neighborhood search on affected related orders, and calling the dynamic rule base for hard constraint verification; adjusting the weight coefficients of the comprehensive production scheduling evaluation index according to the performance of core performance indicators for a consecutive preset number of production cycles.

Description

Technical Field

[0001] This invention relates to the field of industrial manufacturing scheduling technology, specifically to a method and system for automatically generating production schedules in the photovoltaic crystal pulling industry. Background Technology

[0002] With the continuous development of the photovoltaic industry, the specifications and process routes of monocrystalline silicon wafers are becoming increasingly diversified, and the crystal pulling process has become one of the key production links. Current photovoltaic crystal pulling production scheduling typically relies on manual experience for order allocation and furnace scheduling to meet production process requirements and delivery deadlines. In existing technologies, most scheduling methods can complete basic order allocation and resource management, achieving planned production tasks by considering furnace processing capacity, process constraints, and material inventory. These methods, to a certain extent, ensure production continuity and material supply coordination, while supporting the simultaneous management of multiple furnace runs and multiple product specifications. However, with the expansion of production scale and the increase in order structure complexity, traditional scheduling methods have certain limitations in dealing with multi-variable constraints and dynamic production conditions. For example, factors such as furnace thermal compatibility, crucible life decay, doping type continuity, and material inventory status need to be comprehensively considered during the scheduling process to ensure production feasibility and efficiency. Existing methods typically rely on static rules or fixed priority strategies for order allocation, lacking real-time response capabilities to production disturbances and dynamic adjustment mechanisms for scheduling indicators, thus limiting their ability to quickly schedule and handle abnormal production conditions or urgent orders. Achieving automated production scheduling, dynamic operation, and optimization evaluation under high-dimensional process constraints remains a pressing issue for production management.

[0003] This invention proposes an automated production scheduling method and system tailored to the characteristics of the photovoltaic crystal pulling industry. Based on receiving order data, furnace status data, and material inventory data, it constructs a dynamic rule base to achieve automatic verification of multi-dimensional process constraints, generation of candidate sets of schedulable orders, quantification of comprehensive production scheduling evaluation indicators, generation of initial production scheduling sequences, disturbance response repair, and weighted closed-loop optimization. This provides intelligent, automated, and refined production scheduling support for photovoltaic crystal pulling production, helping to improve production efficiency and resource utilization, while ensuring timely order delivery and product quality. Summary of the Invention

[0004] Based on the shortcomings of the prior art described above, the purpose of this invention is to provide a method and system for automatically generating production scheduling plans in the photovoltaic crystal pulling industry, so as to solve the above-mentioned technical problems.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for automatically generating production scheduling plans based on photovoltaic crystal pulling industry production, comprising: S1: Receive pending production order data, furnace real-time status data, and material inventory data, and construct a dynamic rule base. The dynamic rule base includes the order demand structure, furnace process capability profile, doping transformation matrix, and inventory snapshot structure. S2: Call the structured parameters in the dynamic rule base and execute the thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching in sequence to eliminate infeasible production scheduling combinations. S3: Generate a candidate set of schedulable orders and furnace platforms that has undergone process constraint pre-inspection. Each record in the candidate set of schedulable orders and furnace platforms includes a constraint label, an available time window, and a changeover time estimation parameter. S4: Divide the future production scheduling cycle into an execution lock-in zone and a pre-schedule adjustment zone, and construct a comprehensive production scheduling evaluation index that includes delay penalty points, changeover penalty points, idle penalty points, and energy consumption costs; S5: Based on the comprehensive production scheduling evaluation index, use a hierarchical allocation strategy to traverse and match the candidate set of available orders and furnaces to generate an initial production scheduling sequence; S6: Monitor production data flow in real time. When a preset disturbance trigger condition is identified, freeze the production sequence of unaffected areas, perform local rearrangement repair based on neighborhood search only on affected related orders, and call the dynamic rule base for hard constraint verification. S7: Based on the performance of core performance indicators for a consecutive preset number of production scheduling cycles, adjust the weight coefficients of the comprehensive production scheduling evaluation indicators, and use the adjusted weight coefficients as the benchmark input for the next production scheduling cycle.

[0006] The present invention is further configured such that S1 includes: Receive data from the pending production order pool, real-time furnace status data, and material inventory data; Based on the pending production order pool data, geometric parameters, process parameters, resource parameters, and bill of materials parameters are extracted and mapped to generate an order demand structure. The geometric parameters include the target crystal diameter; the process parameters include doping type, maximum process temperature requirement, and maximum power requirement; the resource parameters include product specifications, delivery deadline, and priority; and the bill of materials parameters include polysilicon weight, number of quartz crucibles, and protective gas consumption. Based on real-time furnace status data, a furnace process capability profile is constructed. This profile includes a thermal field compatibility substructure, a crucible lifespan substructure, and historical status parameters. The thermal field compatibility substructure includes the thermal field model, compatible diameter range, extreme temperature threshold, extreme power threshold, set of prohibited doping types, forced replacement marker, and estimated replacement time. The crucible lifespan substructure includes the manufacturer's baseline lifespan, cumulative thermal cycle count, and safety decay ratio. The historical status parameters include the doping type completed in the previous furnace cycle, the earliest available time, and a list of occupied immovable time windows. Based on material inventory data, an inventory snapshot structure is constructed; the inventory snapshot structure records the current available inventory quantity of silicon, crucible and gas and the shortage tolerance threshold parameter. Load the doping type change matrix pre-stored in the dynamic rule base; the doping type change matrix is ​​configured to receive the previous doping type and the current order doping type, and output the cleaning and reset requirements and the corresponding cleaning and reset time window parameters; The dynamic rule base is composed of order demand structure, furnace process capacity profile, inventory snapshot structure, and doping transformation matrix.

[0007] The present invention is further configured such that S2 includes: Iterate through all combinations of order demand structure and furnace process capacity profile, and for each combination, sequentially perform thermal field compatibility verification, crucible life decay calculation, doping continuity deduction, and material kitting comparison: Determine whether the target crystal diameter of the order is within the compatible diameter range of the furnace hot zone; determine whether the maximum process temperature requirement and maximum power requirement of the order do not exceed the limit temperature threshold and limit power threshold of the furnace hot zone; and determine whether the doping type of the order is not included in the prohibited doping type set of the furnace hot zone. If any hot zone compatibility check fails, mark the order-furnace combination as unusable. If there is a forced replacement mark for the furnace hot zone, record the estimated downtime for the corresponding hot zone replacement. The remaining usable number of times is calculated by calling the crucible life substructure. The remaining usable number of times is the manufacturer's benchmark life multiplied by the safety attenuation ratio coefficient, rounded down, and then the cumulative number of thermal cycles is deducted. If the remaining usable number of times is less than 1, the corresponding crucible is marked as unavailable for production, and the furnace to which it belongs is blocked from accepting the order. The doping type change matrix is ​​queried, and the doping type of the previous batch in the furnace's historical status parameters is matched with the doping type in the order demand structure. If the two types are inconsistent, it is determined that cleaning and reset need to be performed, and a cleaning and reset time window defined by the doping type change matrix is ​​inserted into the time axis of the furnace, while generating a type change cost level mark. Compare the bill of materials parameters in the order demand structure with the available inventory quantity in the inventory snapshot structure; if the shortage ratio of any key material exceeds the preset shortage tolerance threshold in the inventory snapshot structure, the corresponding order will be marked as pending completion and removed from the current production scheduling pool.

[0008] The present invention is further configured such that S3 includes: The order-furnace combination that passes the thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching comparison is selected to generate a candidate set of order-furnace that can be scheduled. Each record in the candidate set of available orders-furnaces includes a constraint label, an available time window, and a changeover time estimate parameter. The constraint label is used to identify that the order-furnace combination has passed four checks: thermal field, crucible, doping, and materials. The available time window is determined based on the earliest available start time of the furnace and the process cycle, which is obtained by multiplying the standard working hours by a preset process time coefficient. The changeover time estimate is the sum of the estimated downtime for thermal field replacement and the time for doping and cleaning.

[0009] The present invention is further configured such that S4 includes: The preset future production scheduling cycle is divided into an execution lock zone and a pre-scheduling adjustment zone; the execution lock zone is configured to accommodate recently issued work orders and prohibit modification; the pre-scheduling adjustment zone is configured to accommodate long-term plans and allow sequence rearrangement and time offset. A comprehensive production scheduling evaluation index is constructed, and a quantitative evaluation mechanism combining a penalty point system and normalized weighted summation is adopted. The comprehensive production scheduling evaluation index includes penalty points for delays, penalties for changing models, penalties for idle time, and penalties for energy consumption costs. The delay penalty is determined based on the ratio of the order delay time to the corresponding order delay time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the delay penalty factor. The delay penalty factor is multiplied by the corresponding delay weight to obtain the delay penalty. The replacement penalty is determined based on the ratio of the replacement waiting time to the corresponding replacement waiting time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the replacement penalty factor. The replacement penalty factor is multiplied by the corresponding replacement weight to obtain the replacement penalty. The idle penalty is determined based on the ratio of the furnace idle time to the corresponding furnace idle time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the idle penalty factor. The idle penalty factor is multiplied by the corresponding idle weight to obtain the idle penalty. The energy consumption cost penalty is determined based on the ratio of energy consumption cost to the corresponding energy consumption cost benchmark value. This ratio is normalized and the smaller value between it and 1 is selected as the energy consumption cost penalty factor. The energy consumption cost penalty factor is multiplied by the corresponding energy consumption weight to obtain the energy consumption cost penalty.

[0010] The present invention is further configured such that S5 includes: Orders that have undergone pre-inspection under process constraints are sorted in descending order based on delivery urgency and order priority. Iterate through the sorted available orders in sequence, and use the available orders - furnace candidate set to filter the set of available furnaces corresponding to the current order; For the available furnace sets of the current order, a preliminary screening is performed based on the available time window and the estimated changeover time parameters to eliminate order-furnace combinations with time conflicts; For each candidate order-furnace combination after initial screening, the comprehensive production scheduling evaluation index corresponding to each candidate order-furnace combination is calculated based on its corresponding order delay time, changeover waiting time, furnace idle time and energy consumption cost. The furnace corresponding to the candidate order-furnace combination with the best comprehensive production scheduling evaluation index is selected as the selected furnace. Fill the available time window of the selected furnace with the standard process timeline, and update the occupancy status and corresponding material consumption records of the furnace. Repeat the above sorting, traversing, filtering, calculation and selection steps until all available orders are assigned, generating the initial production schedule.

[0011] The present invention is further configured such that S6 includes: Real-time monitoring of production data streams; when any of the following disturbance trigger conditions are detected, the disturbance response mechanism is activated: furnace malfunction or the duration of deviation of key process parameters from standard values ​​exceeds a preset threshold; the number of consecutive failures of key processes reaches a preset upper limit; order delivery dates are significantly changed or urgent orders with extremely short delivery dates are inserted; key materials fail quality inspection or batch replacement occurs. Starting from the current moment, scan backwards through the locked area, construct the process dependency chain, identify the set of furnaces directly affected and the set of related orders, freeze the production sequence of furnaces that are not affected by the disturbance, and only release the production permissions of furnaces affected by the disturbance and spare idle furnaces. The associated orders are reintegrated into the process constraint pre-inspection logic, and a heuristic neighborhood search is initiated based on the current production sequence. The neighborhood search includes: shifting the order start time as a whole under the premise of satisfying the material arrival and preceding process constraints; exchanging the processing order of adjacent or similar doping type orders within the time window of the same furnace; compressing or stretching the order occupancy window with a preset step size and verifying the feasibility of connection with upstream and downstream processes. After each new production schedule is generated, the hard constraint interface in the dynamic rule base is immediately called to verify and eliminate infeasible solutions. The remaining feasible solutions are then sorted and selected based on the comprehensive production schedule evaluation index. This process is repeated until the maximum number of iterations is reached or the objective function converges. If the time deviation between the optimal solution and the original plan exceeds the preset allowable threshold, it will be marked as requiring manual review; otherwise, it will take effect directly.

[0012] The present invention is further configured such that the key process parameters are limited to the highest process temperature requirement and the highest power requirement in the order demand structure; The key process is limited to the crystal pulling process or the constant diameter growth process in the single crystal pulling process. The key materials are limited to polysilicon, quartz crucibles, and protective gases as specified in the bill of materials parameters of the order demand structure.

[0013] The present invention is further configured such that S7 includes: Based on the performance of core performance indicators for a consecutive preset number of production scheduling cycles, the weighting coefficients in the comprehensive production scheduling evaluation indicators are dynamically adjusted. These weighting coefficients include delay weight, changeover weight, idle weight, and energy consumption weight. If the on-time delivery rate is lower than the preset delivery threshold for a consecutive preset number of production scheduling cycles, the delay weight is increased. If the changeover loss rate is higher than the preset loss threshold for a consecutive preset number of cycles, the changeover weight is increased. If the overall furnace utilization rate is lower than the preset utilization threshold for a consecutive preset number of cycles, the idle weight is increased. The energy consumption weight is a flexible adjustment item and is not triggered for adjustment independently; it is only passively updated during the total weight redistribution. To ensure that the sum of all weight coefficients is 1, when any weight coefficient is increased, the remaining weight coefficients are compressed proportionally according to the original ratio; if any weight coefficient is lower than the preset lower limit after compression, the protection mechanism is triggered and manual confirmation is required. The adjusted weighting coefficients are used as the input benchmark for the next production cycle to achieve closed-loop iteration.

[0014] This invention also provides an automatic production scheduling system for the photovoltaic crystal pulling industry, the system comprising: Rule base construction module: used to receive pending production order data, furnace real-time status data and material inventory data, and build a dynamic rule base. The dynamic rule base includes order demand structure, furnace process capacity profile, doping type matrix and inventory snapshot structure. Constraint Pre-inspection Module: Used to call structured parameters in the dynamic rule base and sequentially perform thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching comparison to eliminate infeasible production scheduling combinations; Candidate set generation module: used to generate a candidate set of schedulable orders-furnaces that has been pre-inspected by process constraints. Each record in the candidate set of schedulable orders-furnaces includes a constraint label, an available time window, and a changeover time estimation parameter. Indicator construction module: used to divide the future production scheduling cycle into execution lock-in zone and pre-schedule adjustment zone, and to construct a comprehensive production scheduling evaluation indicator that includes delay penalty, changeover penalty, idle penalty and energy consumption cost; Production scheduling sequence generation module: Based on the comprehensive production scheduling evaluation index, it uses a hierarchical allocation strategy to traverse and match the candidate set of available orders and furnaces to generate an initial production scheduling sequence; Disturbance Repair Module: Used to monitor production data stream in real time. When a preset disturbance trigger condition is identified, the production sequence of unaffected areas is frozen, and only the affected related orders are subjected to local rearrangement repair based on neighborhood search. The module also calls the dynamic rule base for hard constraint verification. Weighting adjustment module: This module is used to adjust the weighting coefficients of the comprehensive production scheduling evaluation indicators based on the performance of core performance indicators for a preset number of consecutive production scheduling cycles, and to use the adjusted weighting coefficients as the benchmark input for the next production scheduling cycle.

[0015] This invention provides a method and system for automatically generating production schedules in the photovoltaic crystal pulling industry. The method comprises: S1: Receiving order data to be scheduled, furnace real-time status data, and material inventory data; constructing a dynamic rule base, which includes order demand structure, furnace process capability profile, doping changeover matrix, and inventory snapshot structure; S2: Calling structured parameters from the dynamic rule base and sequentially performing thermal field compatibility verification, crucible life decay calculation, doping continuity deduction, and material kitting comparison to eliminate infeasible scheduling combinations; S3: Generating a candidate set of scheduleable orders and furnaces that has undergone process constraint pre-checking, where each record in the candidate set includes constraint tags, available time windows, and estimated changeover time parameters; S4: Dividing future production scheduling cycles. To implement the locking zone and pre-scheduling adjustment zone, a comprehensive production scheduling evaluation index is constructed, including delay penalties, type change penalties, idle penalties, and energy consumption costs; S5: Based on the comprehensive production scheduling evaluation index, a hierarchical allocation strategy is used to traverse and match the candidate set of available orders and furnace platforms to generate an initial production scheduling sequence; S6: Production data flow is monitored in real time. When a preset disturbance trigger condition is identified, the production scheduling sequence in the unaffected area is frozen, and only the affected related orders are subjected to local rearrangement repair based on neighborhood search, and a dynamic rule base is called for hard constraint verification; S7: Based on the performance of core performance indicators for a consecutive preset number of scheduling cycles, the weight coefficients of the comprehensive production scheduling evaluation index are adjusted, and the adjusted weight coefficients are used as the benchmark input for the next scheduling cycle. The beneficial effects include: By sequentially performing thermal field compatibility verification, crucible remaining life calculation, doping continuity deduction, and material matching comparison, infeasible order-furnace combinations with diameter mismatch, excessive temperature / power, prohibited doping, crucible life exhaustion, and shortage of key materials are eliminated in advance during the production planning stage. This pre-inspection mechanism constructs hard filtering rules for the unique process constraints of photovoltaic crystal pulling, effectively avoiding furnace idling, abnormal thermal field loss, and material waste caused by planning inaccuracies, and enhancing the engineering feasibility of the production plan. By dividing future production scheduling cycles into an execution lock-in zone that prohibits modifications and a pre-scheduling adjustment zone that allows dynamic rescheduling, the stability of production plans is balanced with the flexibility to cope with uncertainties. At the same time, a weighted comprehensive evaluation index is constructed that integrates delay penalties, changeover penalties, equipment idle penalties, and energy consumption cost penalties to implement multi-objective quantitative evaluation of delivery timeliness, changeover costs, equipment utilization rate, and energy consumption. Compared with single priority or experience-based production scheduling rules, this mechanism effectively reduces changeover and energy consumption costs and improves the overall utilization rate of furnaces while ensuring order delivery. By monitoring the production line data flow in real time, when furnace malfunctions, critical process failures, urgent order insertions, or material anomalies are identified, the production schedule of furnaces unaffected by the disturbance is frozen. Only affected related orders are subject to local rearrangement repair based on neighborhood search. In each iteration, a dynamic rule base is invoked to perform hard constraint checks on thermal field compatibility, crucible life, doping continuity, and material completeness. This approach avoids plan oscillations caused by global plan overturning and rearrangement, and can generate repair solutions that meet all process constraints in a short time, minimizing the impact of abnormal events on the overall schedule.

[0016] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A flowchart illustrating an exemplary embodiment of the present invention is provided, showing a method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry. Figure 2 This is a schematic diagram illustrating an exemplary embodiment of the present invention of a production scheduling system for the photovoltaic crystal pulling industry that automatically generates production plans. Detailed Implementation

[0018] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0019] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0020] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0021] Example 1: A method for automatically generating production scheduling plans based on the photovoltaic crystal pulling industry, such as Figure 1 As shown, it includes: S1: Receive pending production order data, furnace real-time status data, and material inventory data, and construct a dynamic rule base. The dynamic rule base includes the order demand structure, furnace process capability profile, doping transformation matrix, and inventory snapshot structure. S2: Call the structured parameters in the dynamic rule base and execute the thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching in sequence to eliminate infeasible production scheduling combinations. S3: Generate a candidate set of schedulable orders and furnace platforms that has undergone process constraint pre-inspection. Each record in the candidate set of schedulable orders and furnace platforms includes a constraint label, an available time window, and a changeover time estimation parameter. S4: Divide the future production scheduling cycle into an execution lock-in zone and a pre-schedule adjustment zone, and construct a comprehensive production scheduling evaluation index that includes delay penalty points, changeover penalty points, idle penalty points, and energy consumption costs; S5: Based on the comprehensive production scheduling evaluation index, use a hierarchical allocation strategy to traverse and match the candidate set of available orders and furnaces to generate an initial production scheduling sequence; S6: Monitor production data flow in real time. When a preset disturbance trigger condition is identified, freeze the production sequence of unaffected areas, perform local rearrangement repair based on neighborhood search only on affected related orders, and call the dynamic rule base for hard constraint verification. S7: Based on the performance of core performance indicators for a consecutive preset number of production scheduling cycles, adjust the weight coefficients of the comprehensive production scheduling evaluation indicators, and use the adjusted weight coefficients as the benchmark input for the next production scheduling cycle.

[0022] The present invention is further configured such that S1 includes: Receive data from the pending production order pool, real-time furnace status data, and material inventory data; Based on the pending production order pool data, geometric parameters, process parameters, resource parameters, and bill of materials parameters are extracted and mapped to generate an order demand structure. The geometric parameters include the target crystal diameter; the process parameters include doping type, maximum process temperature requirement, and maximum power requirement; the resource parameters include product specifications, delivery deadline, and priority; and the bill of materials parameters include polysilicon weight, number of quartz crucibles, and protective gas consumption. Based on real-time furnace status data, a furnace process capability profile is constructed. This profile includes a thermal field compatibility substructure, a crucible lifespan substructure, and historical status parameters. The thermal field compatibility substructure includes the thermal field model, compatible diameter range, extreme temperature threshold, extreme power threshold, set of prohibited doping types, forced replacement marker, and estimated replacement time. The crucible lifespan substructure includes the manufacturer's baseline lifespan, cumulative thermal cycle count, and safety decay ratio. The historical status parameters include the doping type completed in the previous furnace cycle, the earliest available time, and a list of occupied immovable time windows. Based on material inventory data, an inventory snapshot structure is constructed; the inventory snapshot structure records the current available inventory quantity of silicon, crucible and gas and the shortage tolerance threshold parameter. Load the doping type change matrix pre-stored in the dynamic rule base; the doping type change matrix is ​​configured to receive the previous doping type and the current order doping type, and output the cleaning and reset requirements and the corresponding cleaning and reset time window parameters; A dynamic rule base is constructed from order demand structure, furnace process capacity profile, inventory snapshot structure, and doping type conversion matrix. Specifically, upon receiving the pending production order pool data, all order information is synchronously collected from the Enterprise Resource Planning (ERP) system or Manufacturing Execution System (MES). Each order data entry includes the customer-specified product specifications, required delivery deadline, and pre-set order priority. Order priority is determined comprehensively based on customer level, contractual delivery urgency, and production line load status. Urgent orders involving default risk or special supply guarantee needs are assigned the highest processing level. The order data explicitly records the target crystal diameter in millimeters, determining the required thermal field specifications for subsequent production. The order data also includes doping... The definition of the miscellaneous type and the maximum process temperature and power requirements that must be achieved during the process are included. The bill of materials required to produce this product are also extracted, specifically including the required weight of polysilicon, the number of quartz crucibles, and the amount of protective gas. These scattered parameters are extracted one by one and mapped to the order requirement structure. The order requirement structure adopts a tree-like hierarchical data model, with the root node being a unique order identifier. The lower-level nodes correspond to standardized fields of geometric parameters, process parameters, resource parameters, and bill of materials parameters, respectively. Field values ​​are stored after data cleaning to remove abnormal formats, thus forming a standardized order data model, providing a unified data format for subsequent constraint verification and calculation calls. At the same time, the real-time status of each furnace is collected. The data is used to construct a profile of the furnace platform's process capability. For each furnace platform, the model of the currently assembled hot zone is read, and the compatible diameter range supported by the hot zone is determined based on this model. The compatible diameter range is defined by the technical specifications provided by the hot zone manufacturer, and is also corrected for the effective processing range caused by the edge effect of the hot zone in actual production. This compatible diameter range defines the minimum and maximum diameter of the crystals that can be processed. The extreme temperature threshold and extreme power threshold that the hot zone can withstand are obtained. These two parameters are used to verify whether the process requirements of the order exceed the physical limits of the equipment. The preset set of incompatible doping types for the hot zone is read. The set of incompatible doping types is generated based on the chemical stability experimental data of the hot zone material, recording the reactions of specific doping elements with the hot zone material. The reaction causes all combinations that lead to performance degradation. This set of prohibited doping types lists the doping types that cannot be processed by the hot zone due to material properties or process limitations. If the hot zone is detected to have reached its design life or has damage marks, a forced replacement indicator is recorded, and the estimated downtime required to replace the hot zone is associated with and stored. The estimated downtime includes the total time for hot zone disassembly, new hot zone installation, vacuum leak detection, and preheating processes, and is determined by statistical analysis of the average time of historical replacement operations. Relevant data of the crucible used in the furnace are retrieved, including the manufacturer's baseline life and the cumulative number of thermal cycles experienced. The remaining usable cycles are calculated by combining the preset safety attenuation ratio coefficient, which is used to reserve the necessary safety margin to prevent the risk of crucible breakage.Record the furnace's historical production information, including the doping type completed in the previous batch, the earliest available start time after the completion of the currently executing task, and a list of locked and unadjustable occupied time windows. The occupied time window list includes the time periods occupied by issued but not yet completed orders and planned equipment maintenance periods. The aforementioned data are respectively categorized into thermal field compatibility substructure, crucible life substructure, and historical state parameters. The thermal field compatibility substructure covers the compatible diameter range, extreme temperature threshold, extreme power threshold, and set of prohibited doping types. The crucible life substructure includes the manufacturer's baseline life and the cumulative number of thermal cycles experienced. The system includes the number of cycles, safety attenuation ratio, and remaining available cycles. Historical status parameters include the doping type completed in the previous cycle, the earliest available start time after the completion of the currently executing task, and a list of locked and unadjustable occupied time windows. The thermal field compatibility substructure, crucible lifetime substructure, and historical status parameters are aggregated to form a furnace process capability profile, which characterizes the furnace's process capability status at the current moment. Simultaneously, current material inventory data is acquired to construct an inventory snapshot structure, which records in real time the total weight of currently available polysilicon, the total number of quartz crucibles, and the total reserve of protective gas in the warehouse. Simultaneously, it presets shortage tolerance thresholds for various materials. These thresholds define the maximum allowable material shortage ratio. Production scheduling is only attempted when the shortage of materials required for an order is within this maximum ratio. If the shortage exceeds the threshold, the material is deemed incomplete. This structure provides a quantitative basis for subsequent judgments on whether an order can meet material supply conditions. It also loads a pre-defined doping type conversion matrix stored in a dynamic rule base. This matrix is ​​a predefined rule lookup table used to specify the processing logic when switching between different doping types. When the doping type of the previous batch on the furnace matches the current order to be scheduled... When there are differences in the doping types of individual components, the doping change matrix outputs a cleaning reset instruction and returns the corresponding cleaning reset time window length. The cleaning reset time window length is determined based on the degree of difference in doping types and the detection standard for residual impurity concentration inside the furnace; the greater the difference, the longer the cleaning reset time window. If the two doping types are the same, it is determined that cleaning is not required. This doping change matrix ensures that the doping switching process complies with process specifications and avoids the risk of residual impurities affecting product quality. Finally, the order demand structure, furnace process capacity profile, inventory snapshot structure, and doping change matrix are integrated to form a dynamic rule base.

[0023] The present invention is further configured such that S2 includes: Iterate through all combinations of order demand structure and furnace process capacity profile, and for each combination, sequentially perform thermal field compatibility verification, crucible life decay calculation, doping continuity deduction, and material kitting comparison: Determine whether the target crystal diameter of the order is within the compatible diameter range of the furnace hot zone; determine whether the maximum process temperature requirement and maximum power requirement of the order do not exceed the limit temperature threshold and limit power threshold of the furnace hot zone; and determine whether the doping type of the order is not included in the prohibited doping type set of the furnace hot zone. If any hot zone compatibility check fails, mark the order-furnace combination as unusable. If there is a forced replacement mark for the furnace hot zone, record the estimated downtime for the corresponding hot zone replacement. The remaining usable number of times is calculated by calling the crucible life substructure. The remaining usable number of times is the manufacturer's benchmark life multiplied by the safety attenuation ratio coefficient, rounded down, and then the cumulative number of thermal cycles is deducted. If the remaining usable number of times is less than 1, the corresponding crucible is marked as unavailable for production, and the furnace to which it belongs is blocked from accepting the order. The doping type change matrix is ​​queried, and the doping type of the previous batch in the furnace's historical status parameters is matched with the doping type in the order demand structure. If the two types are inconsistent, it is determined that cleaning and reset need to be performed, and a cleaning and reset time window defined by the doping type change matrix is ​​inserted into the time axis of the furnace, while generating a type change cost level mark. The process compares the bill of materials parameters in the order demand structure with the available inventory quantities in the inventory snapshot structure. If the shortage ratio of any key material exceeds the preset shortage tolerance threshold in the inventory snapshot structure, the corresponding order is marked as pending completion and removed from the current production scheduling pool. Specifically, it iterates through all combinations of the order demand structure and the furnace process capability profile, and sequentially performs thermal field compatibility verification, crucible life decay calculation, doping continuity deduction, and material completion comparison for each combination. The thermal field compatibility verification first determines whether the target crystal diameter of the order is within the compatible diameter range of the furnace thermal field. The compatible diameter range is defined by the technical specifications provided by the thermal field manufacturer, and also takes into account the effective processing area caused by the edge effect of the thermal field in actual production. The process involves several steps: First, adjustments are made to address the edge effect, which refers to the phenomenon where uneven temperature distribution within the inner wall of the hot zone causes the effective processing dimension of the central area to be smaller than the physical inner diameter. Next, it is determined whether the order's maximum process temperature and maximum power requirements exceed the furnace hot zone's limit temperature and power thresholds. These thresholds are dynamically calibrated using regression analysis of historical operating data, taking into account the aging of the equipment's insulation materials and the performance degradation of the cooling system. The aging level is comprehensively assessed based on equipment operating time and maintenance records, while cooling performance is calculated in real-time based on coolant flow rate and temperature difference parameters. Finally, it is determined whether the order's doping type is included in the furnace hot zone's prohibited doping type set. This prohibited doping type set is generated based on experimental data of the hot zone material's chemical stability, recording specific... All combinations of chemical reactions with the thermal field materials that lead to performance degradation, including oxidation corrosion and lattice penetration; if any thermal field compatibility check fails, the order and furnace combination is marked as unusable; if a furnace thermal field has a mandatory replacement mark, the estimated downtime for the corresponding thermal field replacement is recorded. The estimated downtime includes the total time for thermal field disassembly, new thermal field installation, vacuum leak detection, and preheating processes, determined through historical replacement operation time statistics, excluding abnormal interruptions and rework conditions; the remaining usable cycles are calculated using the crucible lifespan substructure, which is obtained by multiplying the manufacturer's baseline lifespan by a safety attenuation ratio factor, rounding down, and then subtracting the cumulative thermal cycle count. The safety attenuation ratio factor is based on the crucible material purity, tensile strength, etc. The temperature fluctuation range of the crystal process and the statistical data of past breakage accidents are dynamically adjusted. The material purity is graded according to the supplier's quality inspection report, and the temperature fluctuation range is calculated based on the standard deviation of the historical process curve. If the remaining available number of times is less than 1, the corresponding crucible is marked as unschedulable, and the furnace to which it belongs is blocked from accepting the order. The doping type change matrix is ​​queried and matched with the doping type in the order requirement structure based on the doping type of the previous batch in the furnace's historical status parameters. The doping type change matrix is ​​maintained by the process engineer based on the diffusion characteristics of the doping elements and the cleaning process verification results. The diffusion characteristics of the doping elements refer to the migration rate and residual concentration gradient of different dopants in the quartz crucible and the hot zone wall. The cleaning process verification results refer to the relationship between the residual threshold and the cleaning time determined by experiments.If the two types are inconsistent, a cleaning and reset is required. A cleaning and reset time window defined by the doping transformation matrix is ​​inserted into the time axis of the furnace. The length of the cleaning and reset time window is determined based on the degree of difference in doping types and the detection standard for residual impurity concentration inside the furnace. The degree of difference is quantified by the difference in periodic table position and chemical activity, and the residual concentration detection standard is set based on the sampling data of the mass spectrometer. At the same time, a transformation cost level mark is generated. The transformation cost level mark is comprehensively evaluated based on the amount of consumables required for cleaning, labor time loss, and equipment downtime loss. The amount of consumables includes the consumption of acid and high-purity gas, and the labor time loss includes the operator's working time and equipment downtime. Time commitment; compare the bill of materials parameters in the order demand structure with the available inventory quantity in the inventory snapshot structure. Inventory data is synchronized in real time through warehouse management operations. Each material inbound, outbound, and scrap operation updates the snapshot value immediately. The synchronization frequency is tied to inventory change events. If the shortage ratio of any key material exceeds the preset shortage tolerance threshold in the inventory snapshot structure, the shortage tolerance threshold is set differently based on the material procurement cycle, the availability of substitute materials, and the urgency of production. Materials with long procurement cycles are set with a higher tolerance, and materials with sufficient substitute materials are set with a lower tolerance. In this case, the corresponding order is marked as pending completion and removed from the current production scheduling pool.

[0024] The present invention is further configured such that S3 includes: The order-furnace combination that passes the thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching comparison is selected to generate a candidate set of order-furnace that can be scheduled. Each record in the candidate set of available orders and furnaces includes a constraint label, an available time window, and a changeover time estimate parameter. The constraint label indicates that the order-furnace combination has passed four checks: thermal field, crucible, doping, and materials. The available time window is determined based on the earliest available start time of the furnace and the process cycle, which is obtained by multiplying standard working hours by a preset process time coefficient. The changeover time estimate is the sum of the estimated downtime for thermal field replacement and the time for doping and cleaning. Specifically, orders and furnace combinations that pass thermal field compatibility checks, crucible life decay calculations, doping continuity projections, and material kitting comparisons are selected. Entities that pass verification are aggregated into a candidate set of scheduled orders and furnace platforms. Each record in the candidate set contains a constraint label, an available time window, and an estimated changeover time parameter. The constraint label indicates that the combination of the order and the furnace platform has passed four verifications: thermal field, crucible, doping, and materials. The constraint label serves as a binary admission identifier and is generated only if all four verifications are logically true. If any verification fails, the label is not generated, ensuring that all combinations entering the candidate set of scheduled orders and furnace platforms are physically executable. The available time window is defined based on the earliest available start time of the furnace platform and the process cycle. The time is calculated by multiplying the standard working hours by a preset process time coefficient. This process time coefficient compensates for the deviation between theoretical and actual working hours, and its value is dynamically corrected based on deviation analysis of historical production data. Deviation analysis is achieved by calculating the relative error rate between historical actual working hours and standard working hours. The standard working hours are preset by the process specifications based on product specifications and doping types. The setting process is based on the operating time benchmark of the historical optimal process path. The optimal path is determined through statistical analysis of the shortest stable production cycle for products of the same specifications. The shortest stable production cycle refers to the shortest time required to maintain yield standards in continuous batch production. Changeover time is estimated based on the thermal field. The estimated downtime for replacement is the sum of the estimated downtime and the time for doping and cleaning. The estimated downtime for hot zone replacement includes the time for disassembling the old hot zone, installing the new hot zone, and vacuum leak detection. The time for each operation is determined based on the average time of historical replacement operations. The time for doping and cleaning is determined based on the cleaning and reset time window output by the doping replacement matrix. The length of the cleaning and reset time window is jointly determined by the degree of difference in doping type and the detection standard for residual impurity concentration inside the furnace. The sum of the two constitutes the complete replacement cost time. This replacement cost time is included in the order preparation stage and is included in the production sequence calculation as the waiting time before the furnace is ready. It is not included in the calculation of the actual crystal pulling time.Each candidate record's available time window is further linked to the list of immovable time windows already occupied by the furnace. By removing locked maintenance periods and work-in-process periods, the actual available interval of the furnace within that time window is calculated. The removal process employs a time axis overlap detection algorithm. This algorithm converts the furnace's occupied time window and candidate available time window into continuous numerical intervals. It determines whether there is an overlap by comparing the positional relationship between the start and end points of the intervals. If the start point of the candidate interval is earlier than the end point of the occupied interval and the end point is later than the start point of the occupied interval, it is identified as an overlap and removed, ensuring that the production schedule does not conflict with the existing schedule. Plan conflicts; constraint labels, available time windows, and estimated changeover time parameters together constitute the complete attributes of the candidate set of scheduled orders for furnaces. The candidate set of scheduled orders for furnaces is stored using a relational data structure, and each attribute field is linked in real-time with the structured parameters in the dynamic rule base. When the furnace status or material inventory changes, the candidate set of scheduled orders for furnaces automatically triggers an update mechanism. The criteria for determining a failed combination are that the furnace status becomes unavailable, material inventory is insufficient, or process parameters exceed limits. The criteria for adding a feasible combination are that the four checks of the original infeasible combination pass again before it is added to the candidate set of scheduled orders for furnaces.

[0025] The present invention is further configured such that S4 includes: The preset future production scheduling cycle is divided into an execution lock zone and a pre-scheduling adjustment zone; the execution lock zone is configured to accommodate recently issued work orders and prohibit modification; the pre-scheduling adjustment zone is configured to accommodate long-term plans and allow sequence rearrangement and time offset. A comprehensive production scheduling evaluation index is constructed, and a quantitative evaluation mechanism combining a penalty point system and normalized weighted summation is adopted. The comprehensive production scheduling evaluation index includes penalty points for delays, penalties for changing models, penalties for idle time, and penalties for energy consumption costs. The delay penalty is determined based on the ratio of the order delay time to the corresponding order delay time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the delay penalty factor. The delay penalty factor is multiplied by the corresponding delay weight to obtain the delay penalty. The replacement penalty is determined based on the ratio of the replacement waiting time to the corresponding replacement waiting time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the replacement penalty factor. The replacement penalty factor is multiplied by the corresponding replacement weight to obtain the replacement penalty. The idle penalty is determined based on the ratio of the furnace idle time to the corresponding furnace idle time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the idle penalty factor. The idle penalty factor is multiplied by the corresponding idle weight to obtain the idle penalty. The energy consumption cost penalty is determined based on the ratio of energy consumption cost to the corresponding energy consumption cost benchmark value. This ratio is normalized, and the smaller value between it and 1 is selected as the energy consumption cost penalty factor. The energy consumption cost penalty factor is multiplied by the corresponding energy consumption weight to obtain the energy consumption cost penalty. Specifically, the preset future production scheduling cycle is divided into an execution lock zone and a pre-scheduling adjustment zone. The execution lock zone is configured to accommodate recently issued work orders and prohibits modifications. The division is based on the proximity of the current time node to the order delivery deadline. The time period that is close to the current time node and has completed the production preparation process is designated as the execution lock zone. Work orders in this zone have completed material matching and process document issuance. Any unplanned changes will cause production site issues. Disorderly work processes and inefficient resource consumption are demarcated based on a preset time span threshold, determined by a combination of the production preparation cycle length and the response speed to process changes. A pre-scheduling adjustment zone is configured to accommodate long-term plans and allow for sequence rearrangement and time offsets. This zone covers periods far from the current time node where material locking and process preparation have not yet been carried out. It allows for dynamic adjustments to processing sequences and schedules based on equipment load fluctuations and order changes to address market volatility and supply chain uncertainties. Adjustments within this zone do not affect locked production execution plans. A comprehensive production scheduling evaluation index is constructed, employing a quantitative evaluation mechanism combining penalty points and normalized weighted summation. The comprehensive production scheduling evaluation index includes delay penalties and changeover penalties. Penalty points include idle time penalties and energy cost penalties. Delay penalties are determined based on the ratio of the order delay time to the corresponding order delay time baseline. The order delay time baseline is calculated based on the difference between the order delivery deadline and the current time. The ratio represents the relative severity of the order delay; a higher ratio indicates a higher risk of delay. Statutory holidays and production shutdown periods are excluded from the delay time calculation. This ratio is normalized, and the smaller value between it and 1 is selected as the delay penalty factor. Normalization maps the ratio to a numerical range of 0 to 1. A linear transformation method is used to convert the original ratio into a standard score. The logic of selecting a smaller value is to limit the excessive influence of a single order on the overall evaluation. When the ratio exceeds 1, only the maximum value is used. 1. Calculation to avoid excessive penalties: This logic ensures that the upper limit of the evaluation indicators is controllable. The delay penalty factor is multiplied by the corresponding delay weight to obtain the delay penalty. The delay weight reflects the company's emphasis on delivery timeliness. The higher the weight, the greater the negative impact of the delay. The initial value of the weight is set by the company's management according to strategic goals and can be dynamically adjusted according to market feedback. The replacement penalty is determined based on the ratio of the replacement waiting time to the corresponding replacement waiting time benchmark value. The replacement waiting time benchmark value is determined based on the average time consumption of historical replacement operations. The ratio represents the efficiency level of the current replacement plan. The statistical sample selects stable time consumption data of the same type of replacement operation within the historical data statistical period, and calculates the average value after removing outliers.The ratio is normalized, and the smaller value between it and 1 is selected as the changeover penalty factor. The logic of selecting the smaller value is also used to limit the impact of extreme values, ensuring the stability of the evaluation system and preventing distortion of evaluation results due to excessively long single abnormal changeover times. The changeover penalty factor is multiplied by the corresponding changeover weight to obtain the changeover penalty. The changeover weight reflects the company's focus on equipment utilization and changeover costs; a higher weight indicates a higher priority for reducing changeovers. The weight setting needs to balance production efficiency and equipment wear costs. The idle penalty is determined based on the ratio of furnace idle time to the corresponding furnace idle time benchmark value. The furnace idle time benchmark value is calculated based on the difference between the furnace's theoretical working time and planned working time. The ratio represents the degree of idleness of equipment resources. The theoretical working time is determined based on the equipment's rated operating time, and the planned working time is determined based on the production schedule. This ratio is normalized, and the smaller value between it and 1 is selected as the idle penalty factor to limit excessive penalties for idle time, avoid the evaluation system being biased towards overly tight schedules, and prevent equipment overload and quality risks due to excessive pursuit of equipment utilization. The idle penalty... The idle penalty score is obtained by multiplying the factor by the corresponding idle weight. The idle weight reflects the company's emphasis on asset utilization; a higher weight indicates greater benefits from reducing equipment idleness. The weight setting needs to consider equipment maintenance needs and personnel scheduling constraints. The energy cost penalty score is determined based on the ratio of energy cost to the corresponding energy cost benchmark. The energy cost benchmark is determined based on the average energy cost of the same process in history. The ratio represents the energy utilization efficiency of the current solution. Energy cost includes the comprehensive costs of electricity, cooling water, and special gases. The benchmark is updated quarterly to reflect energy market price fluctuations. This ratio is normalized, and the smaller value between it and 1 is selected as the energy cost penalty score factor to control the evaluation scope of energy cost, avoid distortion of evaluation results due to energy price fluctuations, and ensure the stability of the evaluation system under different energy price cycles. The energy cost penalty score is obtained by multiplying the energy cost penalty score factor by the corresponding energy consumption weight. The energy consumption weight reflects the company's requirements for green production and operating cost control; a higher weight indicates a higher priority for energy conservation and consumption reduction. The weight setting needs to comply with national energy conservation and emission reduction policies and the company's carbon emission reduction targets.

[0026] The present invention is further configured such that S5 includes: Orders that have undergone pre-inspection under process constraints are sorted in descending order based on delivery urgency and order priority. Iterate through the sorted available orders in sequence, and use the available orders - furnace candidate set to filter the set of available furnaces corresponding to the current order; For the available furnace sets of the current order, a preliminary screening is performed based on the available time window and the estimated changeover time parameters to eliminate order-furnace combinations with time conflicts; For each candidate order-furnace combination after initial screening, the comprehensive production scheduling evaluation index corresponding to each candidate order-furnace combination is calculated based on its corresponding order delay time, changeover waiting time, furnace idle time and energy consumption cost. The furnace corresponding to the candidate order-furnace combination with the best comprehensive production scheduling evaluation index is selected as the selected furnace. Fill the available time window of the selected furnace with the standard process timeline, and update the occupancy status and corresponding material consumption records of the furnace. The above sorting, traversal, filtering, calculation, and selection steps are repeated until all available orders are allocated, generating an initial production schedule. Specifically, available orders that have undergone process constraint pre-inspection are sorted in descending order based on delivery urgency and order priority. Delivery urgency is calculated by inversely converting the difference between the order's delivery deadline and the current time; the smaller the difference, the higher the urgency. Order priority is directly read from the priority field in the enterprise resource planning platform. Sorting uses a composite key sorting method, with the primary key set as delivery urgency and the secondary key as order priority, to ensure that orders with the closest delivery deadlines receive priority in production resources. The sorted available orders are then traversed sequentially, and the available orders - furnace candidate set is queried to obtain... Retrieve the set of available furnace platforms corresponding to the current order. This set is obtained by querying the candidate furnace platform set for scheduled orders, with the query condition being a unique order identifier match. It returns all furnace platform records with feasible combinations to the current order, each record containing corresponding constraint labels, available time windows, and estimated changeover time parameters. For the available furnace platform set of the current order, a preliminary screening is performed based on the available time windows and estimated changeover time parameters, eliminating order-furnace platform combinations with time conflicts. Time conflict detection uses a time axis overlap verification method, extracting the start and end times of the candidate furnace platform's available time windows and comparing them with the required time interval formed by combining the order's required process cycle and estimated changeover time parameters. When the demand... When an interval overlaps with the list of immovable time windows already occupied by the furnace, it is considered a time conflict and removed, retaining only combinations that are completely feasible in terms of time. For each candidate order-furnace combination after initial screening, a comprehensive production scheduling evaluation index is calculated based on its corresponding order delay time, changeover waiting time, furnace idle time, and energy cost. The comprehensive production scheduling evaluation index is obtained by weighted summation of delay penalty, changeover penalty, idle time penalty, and energy cost penalty; the lower the comprehensive production scheduling evaluation index value, the better the overall benefit. The furnace corresponding to the candidate order-furnace combination with the optimal comprehensive production scheduling evaluation index is selected as the chosen furnace. The optimal criterion is the comprehensive production scheduling evaluation index value. At the very least, if multiple production scheduling comprehensive evaluation index values ​​are the same, the combination with the shortest changeover time is selected first. If they are still the same, the combination with the shortest furnace idle time is selected. Multi-level optimization rules ensure decision certainty. The standard process timeline is filled into the available time window of the selected furnace, and the furnace occupancy status and corresponding material consumption records are updated. The standard process timeline is defined according to the order process specification and includes the standard start and end times of each process. When filling it in, a preparation period for the changeover time estimation parameters must be reserved on the timeline. After filling it in, the time period is marked as occupied, and the quantity of the bill of materials parameter corresponding to the order is deducted from the inventory snapshot structure to ensure that the inventory data is accurate in real time when subsequent orders are allocated.The process of sorting, traversing, filtering, calculating, and selecting is repeated until all available orders are assigned, generating an initial production schedule. The loop terminates when the available order pool is empty or there are no available furnace combinations to assign.

[0027] The present invention is further configured such that S6 includes: Real-time monitoring of production data streams; when any of the following disturbance trigger conditions are detected, the disturbance response mechanism is activated: furnace malfunction or the duration of deviation of key process parameters from standard values ​​exceeds a preset threshold; the number of consecutive failures of key processes reaches a preset upper limit; order delivery dates are significantly changed or urgent orders with extremely short delivery dates are inserted; key materials fail quality inspection or batch replacement occurs. Starting from the current moment, scan backwards through the locked area, construct the process dependency chain, identify the set of furnaces directly affected and the set of related orders, freeze the production sequence of furnaces that are not affected by the disturbance, and only release the production permissions of furnaces affected by the disturbance and spare idle furnaces. The associated orders are reintegrated into the process constraint pre-inspection logic, and a heuristic neighborhood search is initiated based on the current production sequence. The neighborhood search includes: shifting the order start time as a whole under the premise of satisfying the material arrival and preceding process constraints; exchanging the processing order of adjacent or similar doping type orders within the time window of the same furnace; compressing or stretching the order occupancy window with a preset step size and verifying the feasibility of connection with upstream and downstream processes. After each new production schedule is generated, the hard constraint interface in the dynamic rule base is immediately called to verify and eliminate infeasible solutions. The remaining feasible solutions are then sorted and selected based on the comprehensive production schedule evaluation index. This process is repeated until the maximum number of iterations is reached or the objective function converges. If the time deviation between the optimal solution and the original plan exceeds a preset allowable threshold, it is marked as requiring manual review; otherwise, it takes effect directly. The present invention is further configured such that the key process parameters are limited to the highest process temperature requirement and the highest power requirement in the order demand structure. The key process is limited to the crystal pulling process or the constant diameter growth process in the single crystal pulling process. The key materials are limited to polysilicon, quartz crucibles, and protective gases as specified in the bill of materials parameters of the order demand structure. Specifically, the production data stream is monitored in real time, and a disturbance response mechanism is activated when any disturbance trigger condition is detected. Disturbance trigger conditions include furnace malfunction or a critical process parameter deviating from its standard value for a duration exceeding a preset threshold. Key process parameters are limited to the highest process temperature and highest power requirements in the order demand structure. Deviation detection is achieved by comparing real-time furnace sensor data with the process standard values. Duration is counted from the first deviation; an alarm is triggered if the threshold is exceeded. A critical process failure is also triggered when the number of consecutive failures reaches a preset upper limit. Critical processes are limited to the crystal pulling process in single-crystal ingot pulling, or similar processes. For the diameter growth process, failure is judged based on process quality control indicators. A crystal pulling success rate below the standard or diameter fluctuation exceeding the tolerance range is considered a failure. The number of consecutive failures is counted using a sliding window counting method. Orders with significant changes to delivery dates or urgent orders with extremely short delivery times (significant changes refer to orders being delivered earlier than originally planned, urgent orders refer to orders whose delivery time is less than a preset urgency threshold) must skip the regular queuing process and be prioritized. Critical materials failing quality inspection or being subject to mandatory batch replacement (critical materials are limited to polysilicon, quartz crucibles, and protective gases as specified in the order's bill of materials) are also considered failures. Quality inspection failures are judged based on laboratory test reports, and mandatory batch replacement refers to issues caused by supplier... Material switching due to quality issues or expired inventory; starting from the current moment, reverse the execution lock area to construct a process dependency chain; the process dependency chain is generated by analyzing the sequence and resource sharing relationships of each process in the order's process route, identifying logical dependencies, i.e., identifying which downstream processes' material input depends on the semi-finished products produced by the disturbed furnace, or identifying which orders are coupled due to sharing the same key equipment resources; identifying the directly affected furnace set and the associated order set. The affected furnace set includes the disturbed furnace and its physically connected or logically coupled upstream and downstream furnaces, and the associated order set includes orders currently being processed or waiting in the queue on the aforementioned furnaces, as well as orders that depend on the output of the aforementioned orders for subsequent processing. Orders are frozen, and the production sequence of unaffected furnaces is frozen. The freeze operation marks the production sequence of unaffected furnaces as read-only, prohibiting any form of modification, to ensure the stability of the plan in the non-disturbed area. Only the production permissions of the affected furnaces and the standby idle furnaces are released. The release operation removes the production lock on the above resources, allowing the scheduling logic to reallocate the affected orders to the standby idle furnaces. The standby idle furnaces are defined as furnaces that are currently in standby mode, have no task assignment, and have passed the process compatibility verification. The associated orders are re-imported into the process constraint pre-inspection logic, and the thermal field compatibility verification, crucible life decay calculation, doping continuity deduction, and material kitting comparison are re-executed to verify that the repair plan meets all hard constraints.Based on the current production schedule, a heuristic neighborhood search is initiated. This search involves shifting the order start time while ensuring material availability and preceding process constraints are met. The shift operation adjusts the order start time, requiring verification of material availability time and preceding process completion time. Within the same furnace's time window, adjacent or similarly doped orders are swapped. Swapping prioritizes orders with the same doping type to reduce changeover time. After swapping, it's necessary to verify if furnace availability time windows overlap and if changeover time is sufficient. Order occupancy windows are compressed or extended using a preset step size, verifying the feasibility of connection with upstream and downstream processes. Compression shortens order processing time while maintaining process quality standards, while extension extends processing time but assesses the chain reaction impact on subsequent orders. Feasibility verification ensures that the interval between upstream and downstream processes meets process specifications. After each new production schedule is generated, the hard constraint interface in the dynamic rule base is immediately called to verify and eliminate infeasible solutions. The hard constraint interface verifies thermal field compatibility, crucible life decay threshold, doping continuity logic, and material availability. If any constraint fails to meet the standard, the solution is deemed infeasible and eliminated. The remaining feasible solutions are then ranked and selected based on the comprehensive production scheduling evaluation index. This ranking is based on the sum of delay penalties, changeover penalties, idle time penalties, and energy cost penalties, arranged in ascending order, with the solution having the lowest comprehensive production scheduling evaluation index value selected. This process is repeated iteratively until the maximum number of iterations is reached or the objective function converges. The maximum number of iterations is preset as a threshold to prevent infinite loops. The objective function convergence criterion is that the improvement in the comprehensive production scheduling evaluation index value after multiple consecutive iterations is less than a preset threshold, indicating that an approximate optimal solution has been reached. If the time offset between the optimal solution and the original plan exceeds a preset allowable threshold (calculated as the absolute value of the change in order start or end time between the old and new plans, with the preset allowable threshold set based on the production site's adjustment tolerance), the solution is marked as requiring manual review. This triggers a notification sent to production management personnel, requiring manual confirmation before it takes effect. Otherwise, it takes effect directly, meaning the new production schedule is automatically updated to the manufacturing execution module and guides on-site production.

[0028] The present invention is further configured such that S7 includes: Based on the performance of core performance indicators for a consecutive preset number of production scheduling cycles, the weighting coefficients in the comprehensive production scheduling evaluation indicators are dynamically adjusted. These weighting coefficients include delay weight, changeover weight, idle weight, and energy consumption weight. If the on-time delivery rate is lower than the preset delivery threshold for a consecutive preset number of production scheduling cycles, the delay weight is increased. If the changeover loss rate is higher than the preset loss threshold for a consecutive preset number of cycles, the changeover weight is increased. If the overall furnace utilization rate is lower than the preset utilization threshold for a consecutive preset number of cycles, the idle weight is increased. The energy consumption weight is a flexible adjustment item and is not triggered for adjustment independently; it is only passively updated during the total weight redistribution. To ensure that the sum of all weight coefficients is 1, when any weight coefficient is increased, the remaining weight coefficients are compressed proportionally according to the original ratio; if any weight coefficient is lower than the preset lower limit after compression, the protection mechanism is triggered and manual confirmation is required. The adjusted weighting coefficients are used as the input benchmark for the next production scheduling cycle to achieve closed-loop iteration. Specifically, the weighting coefficients in the comprehensive production scheduling evaluation indicators are dynamically adjusted based on the performance of core performance indicators for a consecutive preset number of production scheduling cycles. The weighting coefficients include delay weights, changeover weights, idle weights, and energy consumption weights. If the on-time delivery rate is lower than the preset delivery threshold for a consecutive preset number of production scheduling cycles, the on-time delivery rate is calculated based on the proportion of qualified orders completed on time to the total number of orders to be delivered. The preset delivery threshold is set comprehensively based on the company's historical average delivery level and the performance standards agreed upon in the customer's contract. A continuous drop below the threshold indicates a significant gap in the current production scheduling strategy's guarantee of timely delivery; therefore, the delay weight is increased, and the on-time delivery rate is increased. The operation is implemented by increasing the weight of delays. The increase is positively correlated with the degree of deviation of the on-time delivery rate from the threshold; the greater the deviation, the higher the increase. The purpose is to strengthen the sensitivity and priority of delivery deadlines in subsequent production scheduling. If the changeover loss rate exceeds the preset loss threshold for a preset number of consecutive cycles, the changeover loss rate is calculated based on the ratio of the weight of waste generated during changeover operations to the total weight of raw materials input during the same period. The preset loss threshold is set comprehensively based on process standards and cost control targets. Continuously exceeding the threshold indicates that frequent changeovers lead to resource waste and decreased production efficiency; therefore, the changeover weight is increased. This increase is implemented by increasing the weight of the changeover, and the increase is positively correlated with the degree of deviation of the changeover loss rate from the threshold. The aim is to prioritize reducing unnecessary equipment changes during subsequent production scheduling. If the overall furnace utilization rate falls below a preset threshold for a certain number of consecutive cycles, the overall furnace utilization rate is calculated based on the ratio of the furnace's actual effective production time to its total available time. The preset utilization threshold is set based on the equipment's designed capacity, regular maintenance plans, and reasonable backup redundancy. A continuous drop below the threshold indicates that equipment resources are not being fully utilized. In this case, the idle weight is increased. This increase is achieved by adding a value to the idle weight, and the increase is positively correlated with the degree to which the overall furnace utilization rate deviates from the threshold. The aim is to increase attention to reducing equipment idleness during subsequent production scheduling. Energy consumption weight, as a flexible adjustment item, is not triggered independently. Adjustments to energy consumption weight do not... Instead of relying on fluctuations in a single energy consumption index, it acts as a balancing buffer for changes in other weights, ensuring that the sum of all weight coefficients remains a constant of 1. It is only passively updated during total weight redistribution. When the weight of delay, replacement, or idle time is increased, causing the total weight to exceed 1, the energy consumption weight is reduced proportionally to offset the increase. Conversely, if other weights are reduced, the energy consumption weight is expanded proportionally to maintain a constant sum. To ensure that the sum of all weight coefficients is 1, when any weight coefficient is increased, the remaining weight coefficients are reduced proportionally. Proportional reduction means that all weight coefficients except those of the increased weight are simultaneously multiplied by the same reduction coefficient, which is equal to 1 minus the increase in the weight of the increased weight, ensuring that the sum of all weights after adjustment is still 1.If any weight coefficient after compression falls below a preset lower limit (the preset lower limit is set according to the importance of each weight in the business logic; the delay weight lower limit follows the principle of delivery priority to ensure customer fulfillment; the changeover weight lower limit follows the requirement of process continuity to avoid excessive suppression of necessary process adjustments; and the idle weight lower limit follows the requirement of equipment basic utilization rate to ensure the bottom line of production capacity), a protection mechanism is triggered and manual confirmation is required. The protection mechanism suspends the automatic adjustment process, generates a weight adjustment warning message and pushes it to production management personnel. The adjustment plan must be manually reviewed to ensure it meets the current actual production needs and business priorities before it can take effect. The adjusted weight coefficients are used as the input benchmark for the next production scheduling cycle, replacing the original weight parameters, and are used for calculating the comprehensive evaluation index of the next cycle's production scheduling, forming a closed-loop iteration from performance feedback to strategy optimization.

[0029] Example 2: Please see Figure 2 This exemplary production scheduling system for the photovoltaic crystal pulling industry includes: Rule base construction module: used to receive pending production order data, furnace real-time status data and material inventory data, and build a dynamic rule base. The dynamic rule base includes order demand structure, furnace process capacity profile, doping type matrix and inventory snapshot structure. Constraint Pre-inspection Module: Used to call structured parameters in the dynamic rule base and sequentially perform thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching comparison to eliminate infeasible production scheduling combinations; Candidate set generation module: used to generate a candidate set of schedulable orders-furnaces that has been pre-inspected by process constraints. Each record in the candidate set of schedulable orders-furnaces includes a constraint label, an available time window, and a changeover time estimation parameter. Indicator construction module: used to divide the future production scheduling cycle into execution lock-in zone and pre-schedule adjustment zone, and to construct a comprehensive production scheduling evaluation indicator that includes delay penalty, changeover penalty, idle penalty and energy consumption cost; Production scheduling sequence generation module: Based on the comprehensive production scheduling evaluation index, it uses a hierarchical allocation strategy to traverse and match the candidate set of available orders and furnaces to generate an initial production scheduling sequence; Disturbance Repair Module: Used to monitor production data stream in real time. When a preset disturbance trigger condition is identified, the production sequence of unaffected areas is frozen, and only the affected related orders are subjected to local rearrangement repair based on neighborhood search. The module also calls the dynamic rule base for hard constraint verification. Weighting adjustment module: This module is used to adjust the weighting coefficients of the comprehensive production scheduling evaluation indicators based on the performance of core performance indicators for a preset number of consecutive production scheduling cycles, and to use the adjusted weighting coefficients as the benchmark input for the next production scheduling cycle.

[0030] It should be noted that the automated production scheduling system for the photovoltaic crystal pulling industry provided in the above embodiments and the automated production scheduling method for the photovoltaic crystal pulling industry provided in the above embodiments belong to the same concept. The specific methods of execution of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the automated production scheduling system for the photovoltaic crystal pulling industry provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0031] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry, characterized in that, include: S1: Receive pending production order data, furnace real-time status data, and material inventory data; construct a dynamic rule base, which includes an order demand structure, furnace process capability profile, doping type conversion matrix, and inventory snapshot structure. S1 includes: receiving pending production order pool data, furnace real-time status data, and material inventory data; extracting geometric parameters, process parameters, resource parameters, and bill of materials parameters based on the pending production order pool data, and mapping them to generate an order demand structure; the geometric parameters include the target crystal diameter; the process parameters include doping type, maximum process temperature requirement, and maximum power requirement; the resource parameters include product specifications, delivery deadline, and priority; the bill of materials parameters include polysilicon weight, number of quartz crucibles, and protective gas consumption; constructing a furnace process capability profile based on the furnace real-time status data; the furnace process capability profile includes a polymerization thermal field compatibility substructure, a crucible lifespan substructure, and historical status parameters. The system includes the following parameters: the thermal field compatibility substructure, which includes the thermal field model, compatible diameter range, extreme temperature threshold, extreme power threshold, set of prohibited doping types, forced replacement flag, and estimated replacement time; the crucible lifespan substructure, which includes the manufacturer's baseline lifespan, cumulative thermal cycle count, and safety decay ratio; the historical status parameters, which include the doping type completed in the previous batch, the earliest available time, and a list of occupied immovable time windows; an inventory snapshot structure, which records the available inventory quantity of silicon, crucible, and gas, and the notch tolerance threshold parameters; a doping type change matrix, pre-stored in the dynamic rule base, which is configured to receive the previous doping type and the current order doping type, and output the cleaning and reset requirements and the corresponding cleaning and reset time window parameters; and the dynamic rule base, which is composed of the order requirement structure, furnace process capacity profile, inventory snapshot structure, and doping type change matrix. S2: Call the structured parameters in the dynamic rule base and execute the thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching in sequence to eliminate infeasible production scheduling combinations. S3: Generate a candidate set of schedulable orders and furnace platforms that has undergone process constraint pre-inspection. Each record in the candidate set of schedulable orders and furnace platforms includes a constraint label, an available time window, and a changeover time estimation parameter. S4: Divide the future production scheduling cycle into an execution lock-in zone and a pre-schedule adjustment zone, and construct a comprehensive production scheduling evaluation index that includes delay penalty points, changeover penalty points, idle penalty points, and energy consumption cost penalty points; S5: Based on the comprehensive production scheduling evaluation index, use a hierarchical allocation strategy to traverse and match the candidate set of available orders and furnaces to generate an initial production scheduling sequence; S6: Monitor production data flow in real time. When a preset disturbance trigger condition is identified, freeze the production sequence of unaffected areas, perform local rearrangement repair based on neighborhood search only on affected related orders, and call the dynamic rule base for hard constraint verification. S7: Based on the performance of core performance indicators for a consecutive preset number of production scheduling cycles, adjust the weight coefficients of the comprehensive production scheduling evaluation indicators, and use the adjusted weight coefficients as the benchmark input for the next production scheduling cycle.

2. The method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry according to claim 1, characterized in that, S2 includes: Iterate through all combinations of order demand structure and furnace process capacity profile, and for each combination, sequentially perform thermal field compatibility verification, crucible life decay calculation, doping continuity deduction, and material kitting comparison: Determine whether the target crystal diameter of the order is within the compatible diameter range of the furnace hot zone; determine whether the maximum process temperature requirement and maximum power requirement of the order do not exceed the limit temperature threshold and limit power threshold of the furnace hot zone; and determine whether the doping type of the order is not included in the prohibited doping type set of the furnace hot zone. If any hot zone compatibility check fails, mark the order-furnace combination as unusable. If there is a forced replacement mark for the furnace hot zone, record the estimated downtime for the corresponding hot zone replacement. The remaining usable number of times is calculated by calling the crucible life substructure. The remaining usable number of times is the manufacturer's benchmark life multiplied by the safety attenuation ratio coefficient, rounded down, and then the cumulative number of thermal cycles is deducted. If the remaining usable number of times is less than 1, the corresponding crucible is marked as unavailable for production, and the furnace to which it belongs is blocked from accepting the order. The doping type change matrix is ​​queried, and the doping type of the previous batch in the furnace's historical status parameters is matched with the doping type in the order demand structure. If the two types are inconsistent, it is determined that cleaning and reset need to be performed, and a cleaning and reset time window defined by the doping type change matrix is ​​inserted into the time axis of the furnace, while generating a type change cost level mark. Compare the bill of materials parameters in the order demand structure with the available inventory quantity in the inventory snapshot structure; if the shortage ratio of any key material exceeds the preset shortage tolerance threshold in the inventory snapshot structure, the corresponding order will be marked as pending completion and removed from the current production scheduling pool.

3. The method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry according to claim 2, characterized in that, S3 includes: The order-furnace combination that passes the thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching comparison is selected to generate a candidate set of order-furnace that can be scheduled. Each record in the candidate set of available orders-furnaces includes a constraint label, an available time window, and a changeover time estimate parameter. The constraint label is used to identify that the order-furnace combination has passed four checks: thermal field, crucible, doping, and materials. The available time window is determined based on the earliest available start time of the furnace and the process cycle, which is obtained by multiplying the standard working hours by a preset process time coefficient. The changeover time estimate is the sum of the estimated downtime for thermal field replacement and the time for doping and cleaning.

4. The method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry according to claim 1, characterized in that, S4 includes: The preset future production scheduling cycle is divided into an execution lock zone and a pre-scheduling adjustment zone; the execution lock zone is configured to accommodate recently issued work orders and prohibit modification; the pre-scheduling adjustment zone is configured to accommodate long-term plans and allow sequence rearrangement and time offset. A comprehensive production scheduling evaluation index is constructed, and a quantitative evaluation mechanism combining a penalty point system and normalized weighted summation is adopted. The comprehensive production scheduling evaluation index includes penalty points for delays, penalties for changing models, penalties for idle time, and penalties for energy consumption costs. The delay penalty is determined based on the ratio of the order delay time to the corresponding order delay time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the delay penalty factor. The delay penalty factor is multiplied by the corresponding delay weight to obtain the delay penalty. The replacement penalty is determined based on the ratio of the replacement waiting time to the corresponding replacement waiting time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the replacement penalty factor. The replacement penalty factor is multiplied by the corresponding replacement weight to obtain the replacement penalty. The idle penalty is determined based on the ratio of the furnace idle time to the corresponding furnace idle time benchmark value. The ratio is normalized and the smaller value between it and 1 is selected as the idle penalty factor. The idle penalty factor is multiplied by the corresponding idle weight to obtain the idle penalty. The energy consumption cost penalty is determined based on the ratio of energy consumption cost to the corresponding energy consumption cost benchmark value. This ratio is normalized and the smaller value between it and 1 is selected as the energy consumption cost penalty factor. The energy consumption cost penalty factor is multiplied by the corresponding energy consumption weight to obtain the energy consumption cost penalty.

5. The method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry according to claim 1, characterized in that, S5 includes: Orders that have undergone pre-inspection under process constraints are sorted in descending order based on delivery urgency and order priority. Iterate through the sorted available orders in sequence, and use the available orders - furnace candidate set to filter the set of available furnaces corresponding to the current order; For the available furnace sets of the current order, a preliminary screening is performed based on the available time window and the estimated changeover time parameters to eliminate order-furnace combinations with time conflicts; For each candidate order-furnace combination after initial screening, the comprehensive production scheduling evaluation index corresponding to each candidate order-furnace combination is calculated based on its corresponding order delay time, changeover waiting time, furnace idle time and energy consumption cost. The furnace corresponding to the candidate order-furnace combination with the best comprehensive production scheduling evaluation index is selected as the selected furnace. Fill the available time window of the selected furnace with the standard process timeline, and update the occupancy status and corresponding material consumption records of the furnace. Repeat the above sorting, traversing, filtering, calculation and selection steps until all available orders are assigned, generating the initial production schedule.

6. The method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry according to claim 1, characterized in that, S6 includes: Real-time monitoring of production data streams; when any of the following disturbance trigger conditions are detected, the disturbance response mechanism is activated: furnace malfunction or the duration of deviation of key process parameters from standard values ​​exceeds a preset threshold; the number of consecutive failures of key processes reaches a preset upper limit; order delivery dates are significantly changed or urgent orders with extremely short delivery dates are inserted; key materials fail quality inspection or batch replacement occurs. Starting from the current moment, scan backwards through the locked area, construct the process dependency chain, identify the set of furnaces directly affected and the set of related orders, freeze the production sequence of furnaces that are not affected by the disturbance, and only release the production permissions of furnaces affected by the disturbance and spare idle furnaces. The associated orders are reintegrated into the process constraint pre-inspection logic, and a heuristic neighborhood search is initiated based on the current production sequence. The neighborhood search includes: shifting the order start time as a whole under the premise of satisfying the material arrival and preceding process constraints; exchanging the processing order of adjacent or similar doping type orders within the time window of the same furnace; compressing or stretching the order occupancy window with a preset step size and verifying the feasibility of connection with upstream and downstream processes. After each new production schedule is generated, the hard constraint interface in the dynamic rule base is immediately called to verify and eliminate infeasible solutions. The remaining feasible solutions are then sorted and selected based on the comprehensive production schedule evaluation index. This process is repeated until the maximum number of iterations is reached or the objective function converges. If the time deviation between the optimal solution and the original plan exceeds the preset allowable threshold, it will be marked as requiring manual review; otherwise, it will take effect directly.

7. A method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry according to claim 6, characterized in that, The key process parameters are limited to the highest process temperature requirement and the highest power requirement in the order demand structure. The key process is limited to the crystal pulling process or the constant diameter growth process in the single crystal pulling process. The key materials are limited to polysilicon, quartz crucibles, and protective gases as specified in the bill of materials parameters of the order demand structure.

8. The method for automatically generating production scheduling plans in the photovoltaic crystal pulling industry according to claim 1, characterized in that, S7 includes: Based on the performance of core performance indicators for a consecutive preset number of production scheduling cycles, the weighting coefficients in the comprehensive production scheduling evaluation indicators are dynamically adjusted. These weighting coefficients include delay weight, changeover weight, idle weight, and energy consumption weight. If the on-time delivery rate is lower than the preset delivery threshold for a consecutive preset number of production scheduling cycles, the delay weight is increased. If the changeover loss rate is higher than the preset loss threshold for a consecutive preset number of cycles, the changeover weight is increased. If the overall furnace utilization rate is lower than the preset utilization threshold for a consecutive preset number of cycles, the idle weight is increased. The energy consumption weight is a flexible adjustment item and is not triggered for adjustment independently; it is only passively updated during the total weight redistribution. To ensure that the sum of all weight coefficients is 1, when any weight coefficient is increased, the remaining weight coefficients are compressed proportionally according to the original ratio; if any weight coefficient is lower than the preset lower limit after compression, the protection mechanism is triggered and manual confirmation is required. The adjusted weighting coefficients are used as the input benchmark for the next production cycle to achieve closed-loop iteration.

9. A system for automatically generating production schedules in the photovoltaic crystal pulling industry, used to implement the method for automatically generating production schedules in the photovoltaic crystal pulling industry as described in any one of claims 1-8, characterized in that, include: Rule base construction module: used to receive pending production order data, furnace real-time status data and material inventory data, and build a dynamic rule base. The dynamic rule base includes order demand structure, furnace process capacity profile, doping type matrix and inventory snapshot structure. Constraint Pre-inspection Module: Used to call structured parameters in the dynamic rule base and sequentially perform thermal field compatibility verification, crucible life decay calculation, doping continuity deduction and material matching comparison to eliminate infeasible production scheduling combinations; Candidate set generation module: used to generate a candidate set of schedulable orders-furnaces that has been pre-inspected by process constraints. Each record in the candidate set of schedulable orders-furnaces includes a constraint label, an available time window, and a changeover time estimation parameter. Indicator construction module: used to divide the future production scheduling cycle into execution lock-in zone and pre-schedule adjustment zone, and to construct a comprehensive production scheduling evaluation indicator that includes delay penalty, changeover penalty, idle penalty and energy consumption cost; Production scheduling sequence generation module: Based on the comprehensive production scheduling evaluation index, it uses a hierarchical allocation strategy to traverse and match the candidate set of available orders and furnaces to generate an initial production scheduling sequence; Disturbance Repair Module: Used to monitor production data stream in real time. When a preset disturbance trigger condition is identified, the production sequence of unaffected areas is frozen, and only the affected related orders are subjected to local rearrangement repair based on neighborhood search. The module also calls the dynamic rule base for hard constraint verification. Weighting adjustment module: This module is used to adjust the weighting coefficients of the comprehensive production scheduling evaluation indicators based on the performance of core performance indicators for a preset number of consecutive production scheduling cycles, and to use the adjusted weighting coefficients as the benchmark input for the next production scheduling cycle.

Images