Ceramic capacitor manufacturing performance monitoring system based on integrated time series prediction algorithm

Abstract

The application discloses a ceramic capacitor manufacturing performance monitoring system based on an integrated time sequence prediction algorithm, relates to the technical field of intelligent monitoring and quality control, and comprises a batch chain establishing module, a residual error analysis module, a margin monitoring module, a double-track isolation module and a pulse replacement module.The batch chain establishing module is based on a feeding time, a stock bin switching signal and a process beat to construct a batch switching time chain, maps discrete time nodes corresponding to batch switching into continuous time intervals, forms a transition corridor covering the batch switching process, and determines a batch transition anchor at the tail of the transition corridor.The application realizes continuous manufacturing monitoring through time chain construction and batch anchoring, realizes early intervention in combination with a residual error density band and an inhibition path, keeps data flow stable through double-track isolation and breathing replacement, and guarantees the quality consistency and performance stability of ceramic capacitor manufacturing.

Description

Technical Field

[0001] This invention relates to the field of intelligent monitoring and quality control technology, specifically to a ceramic capacitor manufacturing performance monitoring system based on an integrated time-series prediction algorithm. Background Technology

[0002] The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithms is an intelligent monitoring technology solution for the entire ceramic capacitor production process. It introduces a supervisory control and data acquisition system to uniformly collect and time-series organize multi-dimensional continuous data generated during manufacturing, constructing a time-series data system that reflects the operating status and performance change patterns of key processes. Based on this, an integrated time-series prediction algorithm is introduced to comprehensively analyze and proactively predict the evolution trends of equipment operating parameters, process condition fluctuations, and product performance indicators. This enables continuous monitoring and early identification of manufacturing performance stability, deviation risks, and potential anomalies. Relying on the supervisory control and data acquisition system, the system dynamically correlates the prediction results with the actual operating status, providing data support for manufacturing process status assessment, quality control, and production adjustment. This allows the ceramic capacitor manufacturing process to gradually shift from post-event detection to process perception and trend prediction.

[0003] The existing technology has the following shortcomings: Under current technological conditions, when multiple batches of raw materials alternately enter the production line according to the production rhythm during the manufacturing process of ceramic capacitors, the subtle differences in particle distribution, physical property response, and process adaptability characteristics between different batches of raw materials will have a superimposed impact on the multidimensional time-series data of the manufacturing process in a short period of time. This causes the deviation between the monitoring results based on time-series prediction and the actual operating state to rapidly concentrate and amplify, manifested as the prediction residual suddenly jumping from a dispersed, low-density state to a high-density clustered state on the time axis. Under the condition of this dense residual occurrence, the residual used in the existing technology to suppress short-term fluctuations and maintain the smoothness of the prediction output becomes ineffective. Suppression mechanisms are prone to entering a unlocked state due to continuous triggering, causing discontinuous jumps in the predicted output between adjacent time slices. This results in manufacturing performance monitoring results losing temporal continuity and trend readability at critical stages. After the continuity of monitoring results is disrupted, the quality control process struggles to accurately distinguish between structural anomalies caused by batch switching and normal process fluctuations. Consequently, it becomes impossible to implement effective interventions in the early stages of raw material alternation. Ultimately, this leads to the equivalent treatment and implicit mixing of different batches of raw materials in critical manufacturing processes. This problem then manifests as systemic failure risks such as decreased dielectric performance consistency and reliability degradation at the finished product stage.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a ceramic capacitor manufacturing performance monitoring system based on an integrated timing prediction algorithm to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a ceramic capacitor manufacturing performance monitoring system based on an integrated timing prediction algorithm, comprising a batch chain establishment module, a residual analysis module, a margin monitoring module, a dual-track isolation module, and a pulse replacement module: The batch chain building module constructs a batch switching time chain based on the feeding time, silo switching signal and process cycle time, maps the discrete time nodes corresponding to batch switching to continuous time intervals, forms a transition corridor covering the batch switching process, and determines the batch transition anchor at the end of the transition corridor. The residual analysis module uses the batch transition anchor as the time alignment benchmark, guides the predicted residual sequence output by the integrated time series prediction algorithm, performs residual density analysis on the time slices covered by the batch transition anchor, extracts the time segments of the predicted residual surge and forms the residual density band. The margin monitoring module constructs a suppression margin reduction path based on the residual density band, and monitors the consumption status of the suppression margin by the predicted residual along the suppression margin reduction path. When the suppression margin reduction path approaches the limit section, it locks the time position corresponding to the suppression margin returning to zero, forming the unlock trigger point. The dual-track isolation module establishes a dual-track isolation structure around the unlock trigger point, migrates the time slice corresponding to the unlock trigger point to the shadow material ledger, and generates an isolation bypass queue based on the shadow material ledger to drive the process cycle time, so that the data stream corresponding to the shadow material ledger is separated from the main link of manufacturing performance monitoring. The pulse replacement module initiates a breathing-style batch pulse replacement mechanism based on the isolation bypass queue. It adjusts the replacement window according to the changing rhythm of the residual density band and simultaneously shifts the batch transition anchor to maintain the continuity of the ceramic capacitor manufacturing performance monitoring results on the time axis.

[0007] Preferably, the steps for determining the batch transition anchor are as follows: Time calibration is performed for each raw material feeding action, and the feeding time is used as the reference point for the start time of the batch entering the manufacturing process. At the same time, the silo switching signal associated with the feeding action and the running rhythm of the manufacturing process are acquired. After aligning the feeding time, silo switching signal, and process cycle time, the feeding time and silo switching signal corresponding to adjacent batches are correlated and processed. Combined with the process cycle time, the time interval before and after batch switching is extended to form a transition corridor covering the batch switching process. After the transition corridor is formed, the time position at the end of the transition corridor is locked based on the completion status of the silo switching signal and the time node when the process cycle time returns to stability, and the batch transition anchor used to characterize the completion of batch switching is determined. The batch transition anchor is integrated with the feeding time, silo switching signal and process cycle to construct a batch switching time chain for continuously characterizing the batch switching process.

[0008] Preferably, the steps for predicting the formation of the residual density band are as follows: Introducing batch transition anchors into the time organization process of manufacturing performance monitoring, the time position corresponding to the batch transition anchor is used as the time alignment benchmark, and the manufacturing process is covered before and after the time position. After completing the time alignment, the predicted residual sequence is expanded according to continuous time slices around the time range covered by the batch transition anchor, so that the predicted residual forms a time structure with the batch transition anchor as the center. After the prediction residual completion time is expanded, the distribution of the prediction residual in adjacent time slices is clustered to identify the time segments in which the prediction residual continuously rises and appears in a concentrated manner. Adjacent predicted residual concentration segments on the time axis are aggregated to form a residual density band that characterizes the concentrated distribution of predicted residuals.

[0009] Preferably, the residual density band is constructed with the time position corresponding to the batch transition anchor as the center. The time range covered by the residual density band maintains a time correspondence with the transition corridor in the batch switching time chain, so that the residual density band only represents the concentrated distribution state of the predicted residuals in the batch switching stage and is used to limit the time range of subsequent manufacturing performance monitoring.

[0010] Preferably, the steps for forming the unlock trigger point are as follows: After the residual density band is formed, the extension range of the residual density band on the time axis is used as the effective range to construct a suppression margin reduction path to characterize the suppression capacity consumption process. After the suppression margin decreases along the path, monitoring points are set up continuously along the time axis to carry the consumption status of the suppression margin by the prediction residual at the corresponding time position. After the continuous monitoring sites are deployed, the state of suppression margin consumption of the prediction residual at each monitoring site is characterized over time along the suppression margin reduction path, so that the suppression margin shows a gradual decreasing trend over time. When the suppression margin decreases and approaches the limit section, the time position corresponding to the suppression margin returning to zero is locked, and this time position is determined as the unlock trigger point.

[0011] Preferably, the continuous monitoring sites are distributed continuously along the path of decreasing suppression margin in chronological order, and the time position corresponding to the unlocking trigger point is used as the time boundary for subsequent manufacturing process data organization to distinguish between the suppression capacity consumption stage and the suppression capacity depletion stage.

[0012] Preferably, the steps for establishing the dual-track isolation structure are as follows: After the lockout trigger point is formed by suppressing the decrease of the margin, the time slice in the manufacturing performance monitoring is defined by the time position corresponding to the lockout trigger point, and the time slice is extracted from the continuous time structure to form a time branch. After the time slice is removed, the corresponding time slice is migrated as a whole to form a shadow material ledger, with the unlock trigger point as the boundary, and the shadow material ledger maintains a time index relationship consistent with the manufacturing process. After the shadow material ledger is formed, an isolated bypass queue is generated in parallel with the manufacturing process based on the time index and process cycle time relationship in the shadow material ledger. After the isolated bypass queue is generated, the data stream corresponding to the shadow material ledger is separated from the main manufacturing performance monitoring link and transmitted independently along the isolated bypass queue.

[0013] Preferably, after the isolated bypass queue is formed, the time synchronization between the isolated bypass queue and the manufacturing process is maintained according to the process cycle, so that the time slice corresponding to the shadow material account continues to advance according to the manufacturing cycle and maintains the time phase consistency with the main link of manufacturing performance monitoring during the advancement process.

[0014] Preferably, the steps for initiating a breathing-type batch pulse replacement mechanism based on an isolated bypass queue, adjusting the replacement window according to changes in the residual density band, and simultaneously shifting the batch transition anchor to maintain the continuity of manufacturing performance monitoring results over time are as follows: After the isolated bypass queue is formed, the range of the permutation window used for data exchange between the main link and the isolated bypass is determined by the variation rhythm of the residual density band on the time axis; After the replacement window range is determined, alternating contraction and expansion operations are performed on the replacement window according to the change state of the residual density band in order to adjust the data flow rhythm between the main link and the isolation bypass. During the alternating adjustment of the replacement window, the batch transition anchor is synchronously shifted by combining the time structure of the isolation bypass queue, so that the batch transition anchor is time-aligned with the current manufacturing state. By periodically adjusting the replacement window and synchronously shifting the batch transition anchor, the continuity of manufacturing performance monitoring results on the time axis is maintained.

[0015] Preferably, the shrinkage operation of the displacement window occurs during the stage when the residual density band is concentrated and enhanced, and the expansion operation of the displacement window occurs during the stage when the residual density band is diffused and flat. Furthermore, the lateral movement direction of the batch transition anchor is consistent with the change direction of the displacement window to ensure continuous alignment of the manufacturing performance monitoring time reference.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention constructs a batch switching time chain and batch transition anchors to achieve a continuous representation of batch switching events in the manufacturing process over time, enabling manufacturing performance monitoring to maintain a stable time reference during the alternating use of raw materials. Through the synergistic effect of residual density bands and suppression margin reduction paths, predicted residuals are structurally identified and intervened in advance during the concentration phase, avoiding discontinuous jumps in monitoring output at critical time slices, thereby maintaining the trend readability and temporal continuity of manufacturing performance monitoring results.

[0017] This invention establishes a dual-track isolation structure around the unlock trigger point and combines it with a breathing-style batch pulse replacement mechanism to achieve dynamic separation and flexible connection between the main manufacturing monitoring link and the shadow data link. This allows abnormal batch data to be independently isolated and systematically returned without affecting the overall monitoring rhythm. This method ensures that the monitoring output remains stable and the data flow is consistent even under multiple batch alternation conditions during the manufacturing process, providing continuous and reliable monitoring support for quality control and performance stability in ceramic capacitor manufacturing. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a schematic diagram of the ceramic capacitor manufacturing performance monitoring system based on the integrated timing prediction algorithm of the present invention. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] This invention provides, for example Figure 1The ceramic capacitor manufacturing performance monitoring system shown includes a batch chain establishment module, a residual analysis module, a margin monitoring module, a dual-track isolation module, and a pulse replacement module. The batch chain building module constructs a batch switching time chain based on the feeding time, silo switching signal and process cycle time, maps the discrete time nodes corresponding to batch switching to continuous time intervals, forms a transition corridor covering the batch switching process, and determines the batch transition anchor at the end of the transition corridor. To continuously perceive and organize the process changes resulting from multiple batches of raw materials entering the production line at alternating intervals, this step focuses on the feeding behavior and process rhythm, structurally unfolding the batch changeover process in the time dimension to construct a batch changeover time chain that reflects the continuous characteristics of batch changeover. The specific implementation process is as follows: During production, each raw material feeding action is time-calibrated, with the feeding moment serving as the initial time reference point for the batch entering the manufacturing process. Simultaneously, the silo switching signal associated with the feeding action and the current manufacturing process's operating rhythm information are collected. The silo switching signal is used to characterize the switching status of different batches of raw materials in the physical storage and transportation path, while the process rhythm is used to characterize the time progression rate of each process unit in the manufacturing process. By aligning the feeding moment, silo switching signal, and process rhythm in the same time coordinate system, the entry process of each batch of raw materials forms a clear starting point on the time axis, providing a basic time reference for the continuous description of subsequent batch switching processes.

[0022] After aligning the feeding time with the relevant operating signals, the feeding times of adjacent batches of raw materials and their corresponding silo switching signals are correlated and analyzed. The batch switching nodes, which originally existed as single points in time, are extended in time towards the cycle time of the preceding and following processes. By combining the manufacturing process advancement rhythm reflected by the cycle time, a certain range of time segments before and after batch switching is uniformly marked, so that batch switching is no longer just a discrete event that occurs instantaneously, but is mapped to a continuous time interval covering feeding behavior, silo switching process, and process cycle time transition. This forms a transition corridor on the time axis that can completely cover the entire batch switching process. This transition corridor is used to carry the continuous changes in the manufacturing process state during batch alternation.

[0023] After the transition corridor is formed, the time position inside the transition corridor is further refined. Based on the completion status of the material switching signal and the time node when the process cycle returns to stability, the tail boundary of the transition corridor is locked. This tail boundary is used to characterize the time position when the influence of the previous batch of raw materials in the manufacturing process gradually decreases and the manufacturing process is dominated by the next batch of raw materials. By extracting this time position from the continuous time interval, it forms an anchor position with a clear time direction in the batch switching time chain, thereby providing a stable time anchor point for distinguishing the manufacturing state before and after the batch switching.

[0024] After determining the time position at the end of the transition corridor, this time position, along with the aforementioned material feeding time, silo switching signal, and process cycle time, is incorporated into the batch switching time chain for unified organization. This ensures that the batch switching time chain includes not only the start time information reflecting the batch entry process, but also continuous time interval information describing the batch alternation transition process, and anchor time positions used to calibrate the batch completion transition. The batch switching time chain constructed in this way can continuously reflect the dynamic evolution of the manufacturing process during the alternation of different batches of raw materials on the time axis, and provides a stable and consistent time structure foundation for subsequent manufacturing performance monitoring and status analysis around batch switching.

[0025] The residual analysis module uses the batch transition anchor as the time alignment benchmark, guides the predicted residual sequence output by the integrated time series prediction algorithm, performs residual density expansion on the time slices covered by the batch transition anchor, extracts the time segments of the predicted residual surge, and aggregates them to form a residual density band. To continuously characterize and structure the changes in manufacturing performance under batch switching conditions, this step organizes and expands the distribution of the predicted residuals along the time dimension around the batch transition anchor, forming a time structure that reflects the concentrated segments of manufacturing state fluctuations. The specific implementation steps are as follows: Based on the established batch transition anchor in the batch switching time chain, the batch transition anchor is introduced into the time organization process of manufacturing performance monitoring as a unified time reference for aligning prediction results with the actual manufacturing process status. By using the time position corresponding to the batch transition anchor as the center, a fixed range of time coverage is extended over the manufacturing process before and after that time position, so that changes in manufacturing status directly related to batch switching are fully incorporated into the same time reference frame. This provides a clear and stable time alignment basis for the continuous observation of prediction residuals during the batch switching stage, avoiding misalignment of data from different time sources on the time axis.

[0026] After time alignment is completed, the predicted residual sequence is expanded at the time slice level around the time range covered by the batch transition anchor. This expansion process uses continuous time slices in the manufacturing process as the basic unit, reorganizing the predicted residuals that were originally arranged in chronological order into a time structure centered on the batch transition anchor. This allows the distribution of the predicted residuals in different time slices before and after batch switching to be observed side by side. In this way, the changes in the predicted residuals under the influence of batch switching are no longer reflected as single-point numerical changes, but are transformed into an overall distribution state within a continuous time slice, thus providing a basis for subsequent identification of concentrated changes in the predicted residuals in the time dimension.

[0027] After the predicted residuals are unfolded over time with the batch transition anchor as the center, the distribution of the unfolded predicted residuals in the time slice dimension is clustered and identified, with a focus on the segments where the predicted residuals rise continuously and appear in clusters in adjacent time slices. By continuously tracking the trend of the predicted residuals in these time slices, the time segments in which the predicted residuals change from a relatively dispersed state to a concentrated state are extracted from the overall time structure. This allows the time segments to reflect the phased characteristics of the rapid accumulation of prediction deviations in the manufacturing process under the influence of batch switching, thus forming a continuous segment on the time axis to characterize the concentrated changes of the predicted residuals.

[0028] After extracting the concentrated segments of the predicted residuals, multiple interconnected and adjacent concentrated segments of the predicted residuals on the time axis are converged to form a continuous and extended band-shaped region in the time structure. This band-shaped region is used to uniformly represent the concentrated distribution characteristics of the predicted residuals during the batch switching stage. By expressing the concentrated changes of the predicted residuals on the time axis in a band-shaped structure, the evolution of the predicted deviation generated before and after the batch transition in the manufacturing process can be presented in a continuous and traceable manner, thereby providing a clear time basis for subsequent continuous adjustment and state differentiation of manufacturing performance monitoring around this band-shaped region.

[0029] The margin monitoring module constructs a suppression margin reduction path based on the residual density band, sets continuous monitoring points on the suppression margin reduction path, monitors the consumption status of the suppression margin by the predicted residual along the suppression margin reduction path, and locks the time position corresponding to the suppression margin returning to zero when the suppression margin reduction path approaches the limit section, forming the unlock trigger point. To avoid the concentrated accumulation of predicted residuals during batch changeover phases affecting the continuity of manufacturing performance monitoring, this step, based on the established residual density bands, structures the suppression capability used to balance the fluctuations of predicted residuals during manufacturing performance monitoring in a time dimension. By constructing a decreasing suppression margin path, the consumption process of suppression capability is continuously characterized, and a clear time marker is formed before the suppression capability is about to be exhausted. The specific implementation process is as follows: Given the established residual density band, the extension of the residual density band along the time axis is taken as the main range of suppression capability consumption, and a suppression margin reduction path is constructed based on this time range. The suppression margin is used to characterize the ability to buffer and balance continuous fluctuations in predicted residuals during manufacturing performance monitoring, and the suppression margin reduction path is used to describe the evolution of suppression capability gradually consumed over time under the continuous action of the residual density band. By associating the residual density band and the suppression margin in the same time dimension, the suppression margin reduction path can be continuously expanded along the time direction of the residual density band, thus forming a time path that reflects the trend of suppression capability consumption.

[0030] After the suppression margin reduction path is formed, multiple continuously distributed monitoring points are set along the suppression margin reduction path on the time axis. Each monitoring point corresponds to a specific time position on the suppression margin reduction path and is used to carry the consumption state of the suppression margin by the prediction residual at that time position. By keeping the monitoring points continuously distributed on the time axis, the change of the suppression margin is no longer only reflected as an overall change result, but can be gradually presented within the time range covered by the residual density band. This enables a detailed characterization of the suppression margin consumption process of the prediction residual under continuous time action, and gives the suppression margin reduction path continuous and traceable time characteristics.

[0031] With continuous monitoring sites already deployed along the suppression margin reduction path, the consumption status of the suppression margin by the predicted residuals at each continuous monitoring site is continuously observed along the suppression margin reduction path. By comparing the changes in the suppression margin at adjacent monitoring sites, the trend of the suppression margin gradually decreasing over time is depicted. In this process, the continuous concentrated appearance of the predicted residuals within the residual density band will drive the suppression margin to move towards the low-value segment along the suppression margin reduction path, so that the suppression margin reduction path shows an evolution characteristic of gradually approaching the depletion state from a sufficient state on the time axis, thus providing a clear time clue for identifying the arrival of the critical state of suppression capability.

[0032] When the suppression margin reduction path advances to near the limit section on the time axis, a comprehensive judgment is made on the changing trend of the suppression margin at continuous monitoring points to lock the specific time position corresponding to the suppression margin returning to zero, and this time position is marked as the unlock trigger point. The unlock trigger point is used to indicate that after this time position, the suppression capability used to balance the prediction residuals in the manufacturing performance monitoring process has been continuously exhausted. If the prediction residuals continue to appear in a concentrated manner, it may cause discontinuous changes in the manufacturing performance monitoring results in the time dimension. By forming the unlock trigger point in advance at the time position corresponding to the suppression margin returning to zero, the subsequent manufacturing process organization and monitoring structure adjustment around this time position have a clear and unified time basis.

[0033] The dual-track isolation module establishes a dual-track isolation structure around the unlock trigger point, migrates the time slice corresponding to the unlock trigger point to the shadow material ledger as a whole, and generates an isolation bypass queue based on the shadow material ledger to drive the process cycle, so that the data flow corresponding to the shadow material ledger is separated from the main manufacturing performance monitoring link. When the suppression margin decreases and progresses to the unlock trigger point, to prevent the manufacturing performance monitoring from being interrupted due to the exhaustion of suppression capacity after this time point, this step dynamically reconstructs the organization structure of the manufacturing process data around the unlock trigger point. By establishing a dual-track isolation structure in the time dimension, manufacturing performance monitoring can complete the isolation and transition of abnormal time slices without interrupting the main monitoring link. The specific implementation steps are as follows: After suppressing the lockout trigger point of the path locking mechanism to reduce residual capacity, the time position corresponding to this lockout trigger point is used as the starting point for reconstructing the manufacturing process data structure. The time slice covering this time position in manufacturing performance monitoring is then defined. This time slice is used to cover the critical period when predicted residuals occur in the batch changeover phase. Its data content includes multi-dimensional information such as manufacturing equipment operating parameters, process operation status, and product performance measurement results. To prevent residual anomalies within this time slice from propagating to subsequent time slices, this time slice needs to be extracted from the continuous time structure and logically branched on the time axis. In this way, an independent path foundation is reserved for subsequent data flow in the time structure, enabling manufacturing performance monitoring to achieve local isolation without disrupting the overall time sequence, laying the foundation for a time-layered structure for the construction of a dual-track isolation structure.

[0034] After the time slice extraction is completed, the entire time slice is migrated to an independent data recording channel to form a shadow material ledger, using the unlock trigger point as the boundary. This shadow material ledger is specifically used to hold time slice data where the manufacturing status fluctuates abnormally due to batch switching, and maintains the same time indexing system as the main manufacturing data ledger to ensure consistency between subsequent analysis and process cycle time. During the migration, the data content originally belonging to the main manufacturing monitoring link is logically copied and reassigned to the shadow material ledger, so that the manufacturing information in the abnormal time slice is separated from the main link and becomes an independent data entity that can be observed and analyzed separately. Through this migration operation, the manufacturing process forms a two-layer data structure in the time dimension where the main ledger and the shadow ledger coexist. The main ledger continues to undertake routine monitoring tasks, while the shadow ledger is used to hold data from the abnormal phase, thereby spatially isolating rather than directly eliminating the prediction deviation during the batch switching phase.

[0035] After the shadow material ledger is formed, to ensure the manufacturing process maintains continuous progress in terms of timing, an isolated bypass queue, parallel to the main manufacturing process, needs to be generated based on the time index and process takt relationship in the shadow material ledger. This isolated bypass queue is used to take over the data flow tasks of the corresponding time slices in the shadow material ledger during the progress of the manufacturing process, ensuring that the manufacturing process can maintain a consistent takt state even after the lockout trigger point. By using the process takt as the core driving parameter, the time slices in the shadow material ledger are sequentially connected to form a bypass sequence synchronized with the main monitoring link, thus creating a dual-track operation pattern in terms of time structure where the main link and the bypass link coexist. This dual-track structure ensures that during batch switching, the main link can continue to receive manufacturing data from subsequent batches, while the bypass link carries the transition data of abnormal batches, with both progressing in parallel on the time axis without mutual interference.

[0036] After the isolated bypass queue is established, the data stream corresponding to the shadow material ledger is completely decoupled from the main manufacturing performance monitoring link. This ensures that the main link retains only time slices related to normal batch operation, while abnormal batch data represented by the shadow material ledger is independently transmitted and stored through the isolated bypass queue. This decoupling operation is not deletion or overwriting, but rather a structured data path adjustment that ensures the data contained in the shadow material ledger operates logically completely independently, thereby preventing abnormal data during batch switching from interfering with the main monitoring link. Through this dual-track parallel and isolated organization, the manufacturing process can maintain the continuity and stability of the main monitoring link when facing discontinuous disturbances caused by batch switching, while retaining detailed process information of the batch transition phase in the shadow material ledger, providing time-consistent data support for subsequent manufacturing performance evaluation and process adjustments.

[0037] The pulse replacement module initiates a breathing-type batch pulse replacement mechanism based on the isolation bypass queue. It alternately performs replacement window contraction and replacement window expansion operations according to the changing rhythm of the residual density band, and simultaneously shifts the batch transition anchor to maintain the continuity of ceramic capacitor manufacturing performance monitoring results on the time axis. In the manufacturing process of ceramic capacitors, after the dual-track isolation structure is established and the shadow material ledger and the main manufacturing link operate in parallel over time, dynamic coordination of the data transfer rhythm between the main link and the isolation bypass is required to maintain the temporal continuity of manufacturing performance monitoring during batch changeover. This step, based on the formation of the isolation bypass queue, establishes a periodically adjustable data replacement mechanism according to the changing rhythm of the residual density band. By alternately performing the contraction and expansion of the replacement window on the time axis, combined with the synchronous lateral movement of the batch transition anchor, continuous maintenance and smooth transition of the manufacturing performance monitoring process are achieved. The specific implementation process is as follows: After the isolation bypass queue is constructed, the changing trend of the residual density band is used as a time reference for manufacturing status fluctuations. The fluctuation rhythm of the residual density band on the time axis is correlated and mapped with the manufacturing process cycle. Through this mapping, the fluctuation period of the manufacturing process during batch switching phase corresponds to the changing period of the residual density band, thereby enabling the determination of the initial range of the data replacement window based on the expansion or contraction state of the residual density band. This replacement window is used for data exchange between the main link and the isolation bypass, and its time span determines the length of the manufacturing process time slice covered by each replacement operation. By using the residual density band as the driving signal to determine the basic scale of the replacement window, the manufacturing process maintains dynamic coordination with the predicted deviation changes in time organization, laying the rhythmic basis for subsequent breathing-style replacement operations.

[0038] After the initial range of the replacement window is established, alternating contraction and expansion operations are performed on the replacement window based on the periodic variation rhythm of the residual density band. When the residual density band shows a concentrated contraction trend on the time axis, the fluctuation of prediction deviation in the manufacturing process tends to intensify. To prevent the spread of anomalies, the replacement window range should be narrowed to reduce the amount of data interaction between the main link and the isolation bypass, thereby achieving local convergence of data flow. When the residual density band shows a spreading or flattening trend on the time axis, the prediction residuals of the manufacturing process gradually return to stability. At this time, the replacement window can be appropriately expanded to allow the data that has returned to normal in the shadow material ledger to be reintegrated into the main manufacturing link. By automatically adjusting the replacement window according to the rhythm of the residual density band, the data replacement process has periodic contraction and expansion characteristics, forming a dynamic reciprocating behavior similar to a breathing rhythm, thereby achieving gradual fusion of abnormal batch data on the time axis.

[0039] While the replacement window is periodically adjusted, the batch transition anchor is synchronously shifted based on the time structure of the isolation bypass queue. The batch transition anchor identifies the turning point in the batch switching process on the time axis. If its position remains static, the replacement window will misalign with the manufacturing cycle time, affecting the time consistency of manufacturing performance monitoring. Therefore, after each contraction or expansion of the replacement window, the batch transition anchor is slightly shifted in the corresponding direction based on the direction and magnitude of the window change, ensuring that the batch transition anchor remains dynamically aligned with the current manufacturing state on the time axis. This synchronous shifting operation ensures that the time reference point for manufacturing performance monitoring is coordinated with the data replacement behavior, avoiding drift in the monitoring benchmark caused by changes in the batch switching rhythm, thus ensuring the continuity of manufacturing performance monitoring results in the time dimension.

[0040] During the continuous operation of the breathing-style batch pulse replacement mechanism, the data flow between the main link and the isolation bypass exhibits a periodic alternation. Each time the replacement window shrinks, the data flow in the shadow inventory is locally compressed, limiting abnormal information to a smaller time frame and preventing the propagation of anomalies. Each time the replacement window expands, the data in the shadow inventory gradually returns to the main link, causing the manufacturing performance monitoring results to revert to overall consistency. Simultaneously, the synchronous lateral shift of the batch transition anchor ensures phase consistency between the main link and the bypass on the time axis, enabling the manufacturing performance monitoring process to maintain temporal continuity and information integrity at different stages of batch switching. This breathing-style dynamic replacement method allows for flexible connection and continuous updating of monitoring data during batch switching, avoiding information gaps caused by monitoring interruptions and ensuring the natural temporal fusion of data from different batches, thereby achieving a continuous and smooth output of manufacturing performance monitoring results on the time axis.

[0041] This invention constructs a batch switching time chain and batch transition anchors to achieve a continuous representation of batch switching events in the manufacturing process over time, enabling manufacturing performance monitoring to maintain a stable time reference during the alternating use of raw materials. Through the synergistic effect of residual density bands and suppression margin reduction paths, predicted residuals are structurally identified and intervened in advance during the concentration phase, avoiding discontinuous jumps in monitoring output at critical time slices, thereby maintaining the trend readability and temporal continuity of manufacturing performance monitoring results.

[0042] This invention establishes a dual-track isolation structure around the unlock trigger point and combines it with a breathing-style batch pulse replacement mechanism to achieve dynamic separation and flexible connection between the main manufacturing monitoring link and the shadow data link. This allows abnormal batch data to be independently isolated and systematically returned without affecting the overall monitoring rhythm. This method ensures that the monitoring output remains stable and the data flow is consistent even under multiple batch alternation conditions during the manufacturing process, providing continuous and reliable monitoring support for quality control and performance stability in ceramic capacitor manufacturing.

[0043] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A ceramic capacitor manufacturing performance monitoring system based on an integrated time-series prediction algorithm, characterized in that, This includes a batch link building module, a residual analysis module, a margin monitoring module, a dual-track isolation module, and a pulse replacement module. The batch chain building module constructs a batch switching time chain based on the feeding time, silo switching signal and process cycle time, maps the discrete time nodes corresponding to batch switching to continuous time intervals, forms a transition corridor covering the batch switching process, and determines the batch transition anchor at the end of the transition corridor. The residual analysis module uses the batch transition anchor as the time alignment benchmark, guides the predicted residual sequence output by the integrated time series prediction algorithm, performs residual density analysis on the time slices covered by the batch transition anchor, extracts the time segments of the predicted residual surge and forms the residual density band. The margin monitoring module constructs a suppression margin reduction path based on the residual density band, and monitors the consumption status of the suppression margin by the predicted residual along the suppression margin reduction path. When the suppression margin reduction path approaches the limit section, it locks the time position corresponding to the suppression margin returning to zero, forming the unlock trigger point. The dual-track isolation module establishes a dual-track isolation structure around the unlock trigger point, migrates the time slice corresponding to the unlock trigger point to the shadow material ledger, and generates an isolation bypass queue based on the shadow material ledger to drive the process cycle time, so that the data stream corresponding to the shadow material ledger is separated from the main link of manufacturing performance monitoring. The pulse replacement module initiates a breathing-type batch pulse replacement mechanism based on the isolation bypass queue. It adjusts the replacement window according to the changing rhythm of the residual density band and simultaneously shifts the batch transition anchor to maintain the continuity of the ceramic capacitor manufacturing performance monitoring results on the time axis. The steps for forming the unlock trigger point are as follows: After the residual density band is formed, the extension range of the residual density band on the time axis is used as the effective range to construct a suppression margin reduction path to characterize the suppression capacity consumption process. After the suppression margin decreases along the path, monitoring points are set up continuously along the time axis to carry the consumption status of the suppression margin by the prediction residual at the corresponding time position. After the continuous monitoring sites are deployed, the state of suppression margin consumption of the prediction residual at each monitoring site is characterized over time along the suppression margin reduction path, so that the suppression margin shows a gradual decreasing trend over time. When the suppression margin decreases and approaches the limit section, lock the time position corresponding to the suppression margin returning to zero, and determine this time position as the unlock trigger point; The steps for establishing a dual-track isolation structure are as follows: After the lockout trigger point is formed by suppressing the decrease of the margin, the time slice in the manufacturing performance monitoring is defined by the time position corresponding to the lockout trigger point, and the time slice is extracted from the continuous time structure to form a time branch. After the time slice is removed, the corresponding time slice is migrated as a whole to form a shadow material ledger, with the unlock trigger point as the boundary, and the shadow material ledger maintains a time index relationship consistent with the manufacturing process. After the shadow material ledger is formed, an isolated bypass queue is generated in parallel with the manufacturing process based on the time index and process cycle time relationship in the shadow material ledger. After the isolated bypass queue is generated, the data stream corresponding to the shadow material ledger is separated from the main manufacturing performance monitoring link and transmitted independently along the isolated bypass queue.

2. The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithm according to claim 1, characterized in that, The steps for determining the batch transition anchor are as follows: Time calibration is performed for each raw material feeding action, and the feeding time is used as the reference point for the start time of the batch entering the manufacturing process. At the same time, the silo switching signal associated with the feeding action and the running rhythm of the manufacturing process are acquired. After aligning the feeding time, silo switching signal, and process cycle time, the feeding time and silo switching signal corresponding to adjacent batches are correlated and processed. Combined with the process cycle time, the time interval before and after batch switching is extended to form a transition corridor covering the batch switching process. After the transition corridor is formed, the time position at the end of the transition corridor is locked based on the completion status of the silo switching signal and the time node when the process cycle time returns to stability, and the batch transition anchor used to characterize the completion of batch switching is determined. The batch transition anchor is integrated with the feeding time, silo switching signal and process cycle to construct a batch switching time chain.

3. The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithm according to claim 2, characterized in that, The steps for predicting the formation of residual density bands are as follows: Introducing batch transition anchors into the time organization process of manufacturing performance monitoring, the time position corresponding to the batch transition anchor is used as the time alignment benchmark, and the manufacturing process is covered before and after the time position. After completing the time alignment, the predicted residual sequence is expanded according to continuous time slices around the time range covered by the batch transition anchor, so that the predicted residual forms a time structure with the batch transition anchor as the center. After the prediction residual completion time is expanded, the distribution of the prediction residual in adjacent time slices is clustered to identify the time segments in which the prediction residual continuously rises and appears in a concentrated manner. Adjacent predicted residual concentration segments on the time axis are aggregated to form a residual density band that characterizes the concentrated distribution of predicted residuals.

4. The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithm according to claim 3, characterized in that, The residual density band is constructed with the time position corresponding to the batch transition anchor as the center. The time range covered by the residual density band maintains a time correspondence with the transition corridor in the batch switching time chain, so that the residual density band only represents the concentrated distribution state of the predicted residuals in the batch switching stage.

5. The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithm according to claim 1, characterized in that, The continuous monitoring points are distributed sequentially along the path of decreasing suppression margin. The time position corresponding to the unlock trigger point is used as the time boundary for subsequent manufacturing process data organization, which is used to distinguish between the suppression capacity consumption stage and the suppression capacity depletion stage.

6. The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithm according to claim 1, characterized in that, After the isolated bypass queue is formed, the time synchronization between the isolated bypass queue and the manufacturing process is maintained according to the process rhythm, so that the time slice corresponding to the shadow material account continues to advance according to the manufacturing rhythm, and maintains the time phase consistency with the main link of manufacturing performance monitoring during the advancement process.

7. The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithm according to claim 1, characterized in that, The following steps are taken to initiate a breathing-type batch pulse replacement mechanism based on an isolated bypass queue, adjusting the replacement window according to changes in the residual density band and simultaneously shifting the batch transition anchor to maintain the continuity of manufacturing performance monitoring results over time: After the isolated bypass queue is formed, the range of the permutation window used for data exchange between the main link and the isolated bypass is determined by the variation rhythm of the residual density band on the time axis; After the replacement window range is determined, alternating contraction and expansion operations are performed on the replacement window according to the change state of the residual density band in order to adjust the data flow rhythm between the main link and the isolation bypass. During the alternating adjustment of the replacement window, the batch transition anchor is synchronously shifted by combining the time structure of the isolation bypass queue, so that the batch transition anchor is time-aligned with the current manufacturing state. By periodically adjusting the replacement window and synchronously shifting the batch transition anchor, the continuity of manufacturing performance monitoring results on the time axis is maintained.

8. The ceramic capacitor manufacturing performance monitoring system based on integrated time-series prediction algorithm according to claim 7, characterized in that, The shrinkage operation of the displacement window occurs during the stage when the residual density band is concentrated and enhanced, while the expansion operation of the displacement window occurs during the stage when the residual density band is diffused and flat. Furthermore, the lateral movement direction of the batch transition anchor is consistent with the change direction of the displacement window to ensure continuous alignment of the manufacturing performance monitoring time reference.

Images