A method for FRD fracture attribution determination based on a thimble process chain of evidence

Abstract

The application discloses a kind of FRD fracture attribution determination methods based on ejector pin process evidence chain, it is related to the field of semiconductor packaging technology, including: collecting patch process parameters form process evidence data;Process evidence is encoded according to action section, support section and suction section to generate process signature string;Based on process signature string, ejector pin action information and support response information are extracted, matching relationship is constructed and ejector pin support correlation spectrum is formed, evidence chain closure determination is executed to generate evidence chain state;When evidence chain is closed, needle position projection shell is constructed and chip back trace detection and isotopicity are judged;Based on evidence chain state, trace detection result and isotopicity judgment result, FRD fracture attribution conclusion is generated by evidence fusion rule.The patch process parameters are segmented and encoded, and the correlation spectrum is constructed, the needle position projection shell and the trace isotopicity analysis are combined, the corresponding relationship between the process structure and the contact trace is established, so as to improve the accuracy and consistency of FRD fracture attribution.

Description

Technical Field

[0001] This invention relates to the field of semiconductor packaging technology, specifically to a method for determining the cause of FRD breakage based on the evidence chain of the ejector pin process. Background Technology

[0002] In semiconductor packaging and die-mount manufacturing, chips typically require a push-pin mechanism for support during mounting or transfer operations, and are then picked up and transported by a pick-up nozzle. During this process, the push-pin assembly, push pins, push pin array, and pick-up nozzle must work collaboratively according to a predetermined process configuration to ensure uniform stress and positional stability of the chip during lifting, support, and pick-up. Abnormalities in the push-pin support, pick-up position, or process configuration can lead to localized stress concentration on the back of the chip, potentially causing FRD (Fracture Related Defect) fracture problems in subsequent processes or during use.

[0003] Existing production management and quality traceability systems typically record some equipment parameters, production batch information, and process execution data for quality management and problem tracing during production. However, this data is mostly in discrete form, lacking a structured organization of the relationships between different process execution stages (such as lifting actions, ejector pin support, and suction operations), making it difficult to form a unified expression that can characterize the complete process configuration relationships. Meanwhile, in some failure analysis scenarios, indentations, microcracks, or contact marks can be identified by inspecting images of the back of the chip to help determine potential failure sources.

[0004] However, in practical applications, the aforementioned information is often scattered across different equipment or management systems, and the relationships between various process parameters are fragmented, making it difficult to form a complete structured correlation analysis. Especially in the absence of a unified coding mechanism to organize multi-stage process parameters, the matching relationships between different process execution components are difficult to express and trace effectively, resulting in the inability to construct a structured correlation model reflecting the correspondence between the ejector pin's action and the support response. When FRD breakage occurs, it is usually necessary to rely on experience or manual analysis of various aspects such as equipment configuration, ejector pin support status, and chip surface marks. The analysis process is complex, and the degree of correlation between different pieces of information is not easily visualized, leading to limitations in the efficiency and consistency of determining the cause of breakage.

[0005] Furthermore, existing technologies lack an analytical mechanism for determining the integrity of the evidence chain based on the relationship between process execution components, making it impossible to make a structured judgment on whether there are breaks or mismatches in the process configuration. At the same time, there is a lack of a unified isotopic analysis method for the spatial correspondence between the trace information on the back of the chip and the theoretical position of the ejector pin, making it difficult to effectively integrate evidence from multiple sources.

[0006] Therefore, in the chip mounting process, there is still room for improvement in how to segment, encode, and structure the relevant process parameters of the ejector pin support, construct an association structure that reflects the relationship between the ejector pin action and the support response, and on this basis, determine the closure of the process evidence chain. At the same time, by combining the chip back trace information and its spatial co-position relationship with the ejector pin action position to perform multi-source evidence fusion analysis, in order to achieve effective attribution determination of the cause of FRD fracture, further improvements are needed. Summary of the Invention

[0007] Based on the shortcomings of the prior art described above, the purpose of this invention is to provide a method for determining the cause of FRD breakage based on the evidence chain of the ejector pin process, so as to solve the above-mentioned technical problems.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process, comprising: S1: Collect process parameters during the chip mounting process to form a set of process evidence data. The process parameters include: top-mounting action parameters, pin array structure, suction execution parameters, and production batch and product traceability information. S2: Encode the process evidence according to the three-segment structure of action segment, support segment and absorption segment to generate a process signature string; S3: Based on the process signature string, extract the top action information corresponding to each action segment and the support response information corresponding to the support segment, construct the matching relationship between the top action and the support response, and form a top support correlation spectrum based on the matching relationship. Then, perform process evidence chain closure reasoning based on the top support correlation spectrum to generate evidence chain states, including: broken state and closed state. S4: When the evidence chain is closed, the spatial action mapping relationship of the pin array is determined based on the support correlation spectrum, and the pin position projection shell is constructed according to the pin array structure. Trace detection is performed on the back image of the chip within the pin position projection shell area to generate trace detection results. The isotopicity between the trace detection results and the corresponding theoretical position of the pin is determined to generate isotopicity judgment results. S5: Based on the status of the evidence chain, trace detection results, and isotope judgment results, generate attribution conclusions according to the preset evidence fusion rules, and finally output the FRD breakage attribution conclusion.

[0009] The present invention is further configured such that S1 includes: Read the top-mounting action parameters during the chip mounting process from the equipment control system. The top-mounting action parameters include: top-mounting height, top-mounting device model, and pin type. Obtain the ejector pin array structure and suction execution parameters from the equipment process formula library, wherein the suction execution parameters include: the crystal suction nozzle specification and the station number; Obtain production batch and product traceability information from the production management system. The production batch and product traceability information includes: chip ID, production batch, process batch, equipment number, and execution time. The parameters of the top-mounted action, the structure of the ejector pin array, the suction execution parameters, and the production batch and product traceability information are correlated and integrated to form a set of process evidence.

[0010] The present invention is further configured such that S2 includes: Based on the set of process evidence, the top action parameters are classified into the action segment, the ejector pin array structure into the support segment, and the suction execution parameters into the suction segment. The parameters in the action segment, support segment, and suction segment are organized in a structured manner according to a preset coding rule, so that the parameters of each segment are arranged in a predetermined order, forming a segmented coding structure to express the process configuration relationship between the top action, the ejector pin support, and the suction execution. The encoding results of the action segment, support segment, and absorb segment are sequentially concatenated to form a process signature string that characterizes the chip mounting process configuration relationship.

[0011] The present invention is further configured such that the lifting support correlation spectrum includes four nodes: the upper lifting device, the lifting pin, the lifting pin array, and the crystal suction nozzle, as well as the mapping relationship between the nodes established based on the matching relationship between the lifting action and the support response.

[0012] The present invention is further configured such that S3 includes: a step of extracting function information and constructing matching relationships, and a step of constructing an association spectrum and determining the closure of the evidence chain.

[0013] The present invention is further configured such that the step of extracting the function information and constructing the matching relationship includes: Based on the process signature string, the codes in the action segment, support segment and absorption segment are parsed to recover the corresponding process parameter values; The upper lifting height, upper lifting device model and ejector pin type in the process parameter values ​​of the action segment are combined according to the preset parameter combination rules, and the lifting action information used to characterize the upper lifting execution state is generated through parameter mapping. The structural parameters of the ejector array in the process parameter values ​​of the support section are analyzed. The ejector arrangement, pin spacing and array size information are extracted by the structural parameter extraction rules. Support response information for characterizing the support state is generated by the structural feature induction method. The jacking action information and support response information are aligned according to time sequence and process segment correspondence, and the two types of information are compared item by item based on preset parameter matching rules to establish the matching relationship between jacking action and support response.

[0014] The present invention is further configured such that the association spectrum construction and evidence chain closure determination steps include: Based on the matching relationship, the corresponding process execution components in the action segment, support segment and suction segment are associated and mapped to determine the connection relationship between the upper ejector, ejector pin, ejector pin array and crystal suction nozzle; Organize each process execution component and its connection relationship according to the preset data structure to generate a jacking support correlation spectrum containing nodes and mapping relationships between nodes; Iterate through the connection relationships of each node in the top support association spectrum, call the pre-stored process rule library, and verify each mapping relationship item by item to determine whether the process matching requirements are met between each node. When any mapping relationship does not meet the process matching requirements, the evidence chain is marked as broken, and the corresponding mismatch node information is recorded. When all mapping relationships meet the process matching requirements, the evidence chain is marked as closed.

[0015] The present invention is further configured such that the pin projection shell is a detection area mask on the back of the chip corresponding to the theoretical contact position of each pin.

[0016] The present invention is further configured such that S4 includes: When the evidence chain is closed, the theoretical contact position of each pin on the back of the chip is determined according to the pin array structure parameters configured in the chip mounting process, and a pin position projection shell corresponding to the pin array is constructed based on the theoretical contact position through a region projection mapping method. Based on the area defined by the pin projection shell, the acquired chip back image is segmented, and the indentation features, microcrack features or contact imprint features within the area are detected by image feature recognition to obtain the corresponding trace detection results. The acquired trace detection results are spatially matched with the theoretical contact positions of each ejector pin, and the isotope between the trace information and the corresponding theoretical position of the ejector pin is determined by the position correspondence determination method, generating the isotope determination result corresponding to the ejector pin contact trace.

[0017] The present invention is further configured such that S5 includes: Obtain the status of the chain of evidence, trace detection results, and isotope determination results; The evidence fusion rules include: When the chain of evidence is broken, an attribution conclusion for abnormal process configuration is generated based on the mapping relationship that does not meet the process rules. When the chain of evidence is closed, attribution is determined by combining trace detection results and isotope assessment results: If no effective traces are detected in the area of ​​the pin projection shell, an attribution conclusion of low risk of cracking at the patching station is generated. If a trace is detected in the area of ​​the pin projection shell and the trace is in the same position as the theoretical position of the corresponding ejector pin, then the attribution conclusion of the crack caused by the patching station is generated and the corresponding pin position information is output. If a trace is detected in the area of ​​the pin projection shell but the trace is not in the same position as the theoretical position of the corresponding pin, then the attribution conclusion that the fracture originates from other workstations is generated. The attribution conclusions are output and combined with the preset output template to form the FRD fracture attribution conclusion.

[0018] This invention provides a method for attributing FRD (Fault Defect) breakage based on a pin process evidence chain. The method involves: S1: collecting process parameters during the chip mounting process to form a process evidence data set, including: pin mounting action parameters, pin array structure, suction execution parameters, and production batch and product traceability information; S2: encoding the process evidence into three segments: action segment, support segment, and suction segment, generating a process signature string; S3: based on the process signature string, extracting the pin mounting action information corresponding to each action segment and the support response information corresponding to the support segment, constructing a matching relationship between the pin mounting action and the support response, and forming a pin-support correlation spectrum based on the matching relationship, thereby executing the process based on the pin-support correlation spectrum. The evidence chain closure reasoning generates evidence chain states, including broken and closed states; S4: When the evidence chain state is closed, the spatial action mapping relationship of the pin array is determined based on the jacking support correlation spectrum, and a pin position projection shell is constructed according to the pin array structure. Trace detection is performed on the chip back image within the pin position projection shell area to generate trace detection results. The isotope between the trace detection results and the corresponding theoretical pin positions is determined to generate isotope judgment results; S5: Based on the evidence chain state, trace detection results, and isotope judgment results, an attribution conclusion is generated according to the preset evidence fusion rules, and finally, the FRD broken attribution conclusion is output. The beneficial effects include: By encoding the chip mounting process parameters into a three-segment structure and generating a process signature string, process information such as the top-mounting action, pin support, and pick-up execution can be expressed in a unified structure. This enables the structured organization and association of configuration relationships between different process execution stages, which is beneficial for the structured recording and traceability of the mounting process configuration relationships and improves the analyzability of process data.

[0019] By constructing a jacking support correlation spectrum that includes nodes of the top jack, pin, pin array, and crystal suction nozzle, and performing a link closure check on the mapping relationship between each node, a structured model of the matching relationship between the jacking action and the support response is formed. Based on the correlation spectrum, the closure determination of the process evidence chain is realized, so that the equipment configuration relationship can be verified in the form of a correlation structure, thereby helping to identify process configuration anomalies and improve the accuracy of anomaly location.

[0020] By constructing a pin projection shell based on the pin array structure and performing isotopic analysis by combining the trace detection results in the chip back image with the theoretical position of the pin, the spatial correspondence between the process execution structure and the actual contact trace is established, and the correspondence between the pin support state and the contact trace on the back of the chip is established, which helps to make a more evidence-based attribution judgment on the cause of FRD fracture.

[0021] 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

[0022] 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 The flowchart illustrates an exemplary embodiment of the present invention of a method for determining the cause of FRD breakage based on the evidence chain of the pin process. Detailed Implementation

[0023] 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.

[0024] 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.

[0025] 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.

[0026] Example: A method for determining the attribution of FRD breakage based on the evidence chain of the pin manufacturing process, such as Figure 1 As shown, it includes: S1: Collect process parameters during the chip mounting process to form a set of process evidence data. The process parameters include: top-mounting action parameters, pin array structure, suction execution parameters, and production batch and product traceability information. S2: Encode the process evidence according to the three-segment structure of action segment, support segment and absorption segment to generate a process signature string; S3: Based on the process signature string, extract the top action information corresponding to each action segment and the support response information corresponding to the support segment, construct the matching relationship between the top action and the support response, and form a top support correlation spectrum based on the matching relationship. Then, perform process evidence chain closure reasoning based on the top support correlation spectrum to generate evidence chain states, including: broken state and closed state. S4: When the evidence chain is closed, the spatial action mapping relationship of the pin array is determined based on the support correlation spectrum, and the pin position projection shell is constructed according to the pin array structure. Trace detection is performed on the back image of the chip within the pin position projection shell area to generate trace detection results. The isotopicity between the trace detection results and the corresponding theoretical position of the pin is determined to generate isotopicity judgment results. S5: Based on the status of the evidence chain, trace detection results, and isotope judgment results, generate attribution conclusions according to the preset evidence fusion rules, and finally output the FRD breakage attribution conclusion.

[0027] The present invention is further configured such that S1 includes: Read the top-mounting action parameters during the chip mounting process from the equipment control system. The top-mounting action parameters include: top-mounting height, top-mounting device model, and pin type. Obtain the ejector pin array structure and suction execution parameters from the equipment process formula library, wherein the suction execution parameters include: the crystal suction nozzle specification and the station number; Obtain production batch and product traceability information from the production management system. The production batch and product traceability information includes: chip ID, production batch, process batch, equipment number, and execution time. The process involves linking and integrating parameters such as the lifting action, pin array structure, pick-up execution parameters, and production batch and product traceability information to form a set of process evidence. Specifically, in the chip mounting process, firstly, process parameters are collected to form a set of process evidence data. The lifting action parameters include: lifting height, lifting device model, and pin type. The lifting height refers to the vertical displacement distance of the pin from its initial position to its lifted position, measured in micrometers (e.g., 150 micrometers). The lifting device model is the identifier of the mechanism that drives the pin to perform the lifting action (e.g., ET-2000). The pin type describes the pin's geometry, categorized as pointed, flat, or ball-shaped (e.g., flat). The pin array structure includes the number of array rows, array columns, pin spacing, and pin diameter (e.g., 4 rows, 4 columns, 0.5 mm pin spacing, 0.1 mm pin diameter). The pick-up execution parameters include the pick-up nozzle specifications and station number. The pick-up nozzle specifications include the nozzle diameter and material (e.g., a 1.0 mm diameter rubber pick-up nozzle), and the station number is, for example, STATION-02. Production batch and product traceability information includes chip ID, production batch, process batch, equipment number, and execution time. The chip ID is a unique identifier for a single chip, such as FRD100-001. The production batch is, for example, BATCH-20250310-01. The process batch is usually the same as the production batch. The equipment number is, for example, DIE-BONDER-03. The execution time is accurate to the second, such as 2025-03-10 09:23:15. The above parameters are automatically collected by constructing three data interfaces: an equipment control system interface, an equipment process recipe library interface, and a production management system interface. The equipment control system interacts with the upper-level system through the equipment communication interface, recording and reporting the top height, top device model, and ejector pin type in real time. The top height is obtained by the equipment measurement unit, the top device model is read from the equipment configuration parameters, and the ejector pin type is identified through the equipment tooling configuration parameters. The equipment process formula library interface queries the formula record corresponding to the current product model from the manufacturing execution system or the equipment's local formula library to obtain the ejector pin array structure, crystal picker specifications, and workstation number; the production management system interface reads the batch identification code on the chip carrier through a barcode scanner, queries the manufacturing execution system for the chip ID, production batch, process batch, and equipment number, and the system generates the execution time.The data acquisition process is executed in the following steps: First, when the placement equipment performs a lifting action, the equipment control system reports event data through the equipment communication interface, including the lifting height value, the lifting device model code, and the ejector pin type identifier. The data acquisition system receives and parses these parameters. Second, before the start of production for the current batch, the system initiates a query request to the process formula library based on the product model to obtain the formula record matching the product. From this, it extracts the ejector pin array structure parameters (including the number of rows, columns, pin spacing, and ejector pin diameter), the pick-up nozzle specifications (nozzle diameter and material), and the station number. These parameters remain unchanged throughout the batch. Third, when the chip carrier arrives at the placement station, a fixed barcode scanner reads the batch identification code on the carrier, and the system sends the data to the manufacturing execution system. The system queries the chip ID list, production batch number, process batch number, and other information contained in the batch, obtains the current execution equipment number, and records the execution time information. The fourth step involves associating and integrating parameters from different sources, creating data records for each chip's processing event, and writing the top height, top mount model, ejector pin type, ejector pin array structure, pick-up nozzle specification, workstation number, chip ID, production batch, process batch, equipment number, and execution time into the record. To ensure the validity of the collected data, the system can perform integrity verification and value range checks on key parameters, mark abnormal data, and store the records in the process evidence database after verification for subsequent process signature string generation and top support correlation spectrum construction.Here is a complete implementation example: A semiconductor packaging and testing plant performs a surface mount process on FRD-100 chips, with a production batch of BATCH-20250310-01, totaling 100 chips. The equipment engineer has configured the FRD100-RECIPE-V2 recipe for the FRD-100 product in the process recipe library. This recipe includes a 4-row x 4-column pin array, a pin pitch of 0.5 mm, a pin diameter of 0.1 mm, and a 1.0 mm diameter rubber pick. The workstation number is STATION-02. The currently installed pin fixture on the surface mount equipment is a flat-headed pin, and the ejector model is ET-2000. When the batch starts, the operator scans the batch code on the carrier. The system identified batch BATCH-20250310-01 and queried the Manufacturing Execution System to confirm that this batch contained chip IDs from FRD100-001 to FRD100-100. Based on the product model FRD-100, the system loaded recipe FRD100-RECIPE-V2 from the process recipe library, obtaining the ejector pin array structure, pick-up nozzle specifications, and workstation number. When processing the first chip, FRD100-001, the equipment performed a lifting action, measuring an upward lifting height of 150 micrometers. The equipment control system reported the upward lifting height of 150 micrometers, the ejector model ET-2000, and the flat-head ejector pin information. The system recorded the execution time as 2025-03-10. At 09:23:15, the system integrates all information into a single complete record: chip ID is FRD100-001, production batch is BATCH-20250310-01, process batch is BATCH-20250310-01, equipment number is DIE-BONDER-03, execution time is 2025-03-10 09:23:15, top height is 150 micrometers, top mount model is ET-2000, pin type is flat, pin array structure is 4 rows × 4 columns, spacing is 0.5 mm, pin diameter is 0.1 mm, pick-up nozzle specification is 1.0 mm diameter rubber nozzle, station number is STATION-02. The above process is then repeated for the remaining 99 chips in the batch, generating an independent process evidence record for each chip. All records are stored in the process evidence database, providing a complete data foundation for subsequent generation of process signature strings and construction of the top support correlation spectrum.

[0028] The present invention is further configured such that S2 includes: Based on the set of process evidence, the top action parameters are classified into the action segment, the ejector pin array structure into the support segment, and the suction execution parameters into the suction segment. The parameters in the action segment, support segment, and suction segment are organized in a structured manner according to a preset coding rule, so that the parameters of each segment are arranged in a predetermined order, forming a segmented coding structure to express the process configuration relationship between the top action, the ejector pin support, and the suction execution. The encoding results of the action segment, support segment, and pick-up segment are sequentially concatenated to form a process signature string that characterizes the chip mounting process configuration relationship. Specifically, to generate the process signature string, the system predefines a set of encoding rules, which specify the division method of the three-segment structure, the arrangement order of parameters within each segment, the unified expression method of parameter values, and the separation method between segments. The action segment includes two parameters: the ejector type and the ejector height; the support segment includes two parameters: the ejector pin type and the ejector pin array structure; the pick-up segment includes two parameters: the pick-up nozzle specification and the station number. The parameter arrangement order within the action segment is: ejector type first, then ejector height; the parameter arrangement order within the support segment is: ejector pin type first, then ejector pin array structure; and the parameter arrangement order within the pick-up segment is: pick-up nozzle specification first, then station number. Before encoding, parameter values ​​are standardized according to uniform rules: the top ejector model directly uses the original string, the top ejector height is converted into a numerical string, the ejector pin type directly uses the original string, the ejector pin array structure is combined in the order of row number, column number, pin spacing and ejector pin diameter to form structural expression information, the crystal suction nozzle specification is expressed by combining the suction nozzle diameter and suction nozzle material, and the workstation number directly uses the original string; parameters within each segment are connected by preset connectors, and the action segment, support segment and suction segment are distinguished by preset separators. The process signature string generation process is executed according to the following steps: Step 1: Extract the process parameters corresponding to the current chip from the process evidence set. The system queries the complete record of the chip from the process evidence database based on the chip ID to obtain the top mount model, top mount height, ejector pin type, ejector pin array structure, pick-up nozzle specification, and station number. Step 2: Classify the parameters according to the division rules of action segment, support segment, and pick-up segment. The top mount model and top mount height are classified into the action segment, the ejector pin type and ejector pin array structure into the support segment, and the pick-up nozzle specification and station number into the pick-up segment. Step 3: Arrange the parameters within each segment in a predetermined order and perform normalization processing. The action segment first processes the top mount model, then processes the top mount height, and combines the two according to a preset connection method to form the action segment encoding result. The support segment first processes the ejector pin type, then performs a structured representation of the ejector pin array structure, combining the array row number, column number, pin spacing, and ejector pin diameter according to a preset structure. The ejector pin type is then connected to the array structure representation to form the support segment encoding result. The pick-up segment first combines the pick-up nozzle specifications, then processes the workstation number, and forms the pick-up segment encoding result through a preset connection method. The fourth step involves concatenating the action segment encoding result, support segment encoding result, and pick-up segment encoding result in the order of action segment, support segment, and pick-up segment to form a complete process signature string. The fifth step involves storing the generated process signature string in the process evidence database, associated with the current chip ID, for subsequent use by the support-up association spectrum analysis module to support the association analysis and processing of chip mounting process configuration relationships.

[0029] The present invention is further configured such that the lifting support correlation spectrum includes four nodes: an upper ejector, ejector pins, ejector pin array, and pick-up nozzle, as well as mapping relationships between nodes established based on the matching relationship between the lifting action and the support response. Specifically, the lifting support correlation spectrum is a structured correlation model constructed based on the configuration relationship between key execution components in the chip mounting process, used to describe the process coordination relationship between the lifting action, ejector pin support, and pick-up execution; the lifting support correlation spectrum includes four nodes: an upper ejector, ejector pins, ejector pin array, and pick-up nozzle, as well as mapping relationships between nodes, wherein the upper ejector node represents the device unit that performs the chip lifting action, the ejector pin node represents the ejector pin element participating in chip lifting support, the ejector pin array node represents the spatial arrangement structure information of the ejector pins, and the pick-up nozzle node represents the pick-up execution unit that performs the chip pick-up operation; by establishing mapping relationships between the above nodes, the action parameters of the upper ejector can be associated with the ejector pin support structure and the pick-up behavior of the pick-up nozzle, thereby forming a complete lifting-supporting-pick-up process relationship structure.

[0030] The present invention is further configured such that S3 includes: a step of extracting function information and constructing matching relationships, and a step of constructing an association spectrum and determining the closure of the evidence chain. Specifically, the action information extraction and matching relationship construction step is used to analyze and process the process parameters of the action segment and the support segment, and establish the correspondence between the upper ejector execution behavior and the support response. By combining and encoding the upper ejector height, upper ejector model and ejector pin type in the action segment and matching them with preset parameter mapping rules, and by parsing and classifying the field parameters of the ejector pin array structure in the support segment, the action information of the upper ejector and the support response information of the support are obtained, and the action information of the upper ejector, the support response information and the matching relationship between them are output. The matching relationship is used to construct the upper ejector-support correlation spectrum. The correlation spectrum construction and evidence chain closure determination step is used to construct the upper ejector-support correlation spectrum based on the matching relationship and determine the integrity of the process action chain. By establishing the node connection structure between the upper ejector, ejector pin, ejector pin array and crystal suction nozzle according to the equipment connection relationship and process path sequence, and performing support response matching checks node by node along the node path based on the action information of the upper ejector, the evidence chain closure determination result is obtained, and the evidence chain state parameters are output. The evidence chain state parameters are used to perform FRD fracture attribution determination in conjunction with spatial isotope analysis.

[0031] The present invention is further configured such that the step of extracting the function information and constructing the matching relationship includes: Based on the process signature string, the codes in the action segment, support segment and absorption segment are parsed to recover the corresponding process parameter values; The upper lifting height, upper lifting device model and ejector pin type in the process parameter values ​​of the action segment are combined according to the preset parameter combination rules, and the lifting action information used to characterize the upper lifting execution state is generated through parameter mapping. The structural parameters of the ejector array in the process parameter values ​​of the support section are analyzed. The ejector arrangement, pin spacing and array size information are extracted by the structural parameter extraction rules. Support response information for characterizing the support state is generated by the structural feature induction method. The action information and support response information are aligned according to time sequence and process segment correspondence, and the two types of information are compared item by item based on preset parameter matching rules to establish a matching relationship between the action and support responses. Specifically, the process signature string is a serialized data structure formed by uniformly encoding the original data of each process segment during chip placement. It contains field identifiers and field value codes corresponding to the action segment, support segment, and pick-up segment. The code is parsed field by field through a preset field mapping table to restore each field code to the corresponding physical process parameter value. Among them, the top height is collected by the equipment displacement sensor, the top mount model is read from the equipment configuration parameter table, and the pin type is obtained by looking up the pin type number in the process configuration file. The pin array structure parameters are provided by the equipment structure database, including the pin arrangement method (such as matrix or ring), pin spacing (center distance between adjacent pins), and array size (total number of pins or array dimension). After parameter recovery is completed, the top height, top device model, and ejector pin type in the action segment are combined according to preset parameter combination rules. The combination rules are to construct a multi-field combination key in the order of "top device model - ejector pin type - top height range", and to search and match the combination key based on a pre-established parameter mapping table. The parameter mapping table is constructed by statistical analysis of historical process debugging data and calibration normalization of equipment calibration data. It records the top action mode corresponding to different combination parameters. The top action category label corresponding to the current combination parameter is determined by looking up the table, thereby forming top action information. For the support segment data, the structural parameters of the ejector pin array are first processed by field decomposition. Structural features are extracted in the order of "arrangement method - pin spacing - array size". The extracted results are then classified based on preset structural classification rules. The structural classification rules divide the array into levels according to the ejector pin coverage density and force distribution characteristics. The ejector pin coverage density is calculated based on the number of ejector pins per unit area in the ejector pin array. The force distribution characteristics are determined by rule mapping based on the ejector pin arrangement method and pin spacing parameters. On this basis, the classification results are labeled to generate support response information, which is used to characterize the support capacity and force distribution characteristics of the ejector pin array under corresponding process conditions.After obtaining the jacking action information and support response information, the two types of information are synchronously sorted according to timestamps, and the correspondence between the action segment and the support segment is established based on the process segment division identifier. Subsequently, the two types of information are compared item by item based on preset parameter matching rules. The parameter matching rules include type consistency judgment and range consistency judgment. Type consistency is used to determine whether the jacking action category and the support response category belong to a preset matching set. Range consistency is used to determine whether the action area corresponding to the jacking action is covered by the support response. The action area corresponding to the jacking action is determined according to the contact range model corresponding to the jacking height and jacking pin type. The contact range model is obtained by calibrating and testing the actual contact area under different jacking pin types and jacking height conditions and establishing a correspondence. The coverage area of ​​the support response is calculated based on the jacking pin array structure parameters. When both conditions are met, it is determined that the match is successful, and the matching relationship between the jacking action and the support response is established accordingly for the subsequent construction of the jacking support correlation spectrum.

[0032] The present invention is further configured such that the association spectrum construction and evidence chain closure determination steps include: Based on the matching relationship, the corresponding process execution components in the action segment, support segment and suction segment are associated and mapped to determine the connection relationship between the upper ejector, ejector pin, ejector pin array and crystal suction nozzle; Organize each process execution component and its connection relationship according to the preset data structure to generate a jacking support correlation spectrum containing nodes and mapping relationships between nodes; Iterate through the connection relationships of each node in the top support association spectrum, call the pre-stored process rule library, and verify each mapping relationship item by item to determine whether the process matching requirements are met between each node. When any mapping relationship does not meet the process matching requirements, the evidence chain is marked as broken, and the corresponding mismatch node information is recorded. When all mapping relationships meet the process matching requirements, the evidence chain is marked as closed. Specifically, the matching relationship serves as the input constraint for the associated mapping, originating from the item-by-item comparison results of the top action information and support response information in the preceding steps. The matching relationship includes the action segment parameter combination identifier and support segment structural feature identifier with established correspondences, and is further associated with the suction nozzle operation parameters in the suction segment. The suction nozzle operation parameters include suction position coordinates, suction pressure, and suction timing information. The suction position coordinates are obtained by the patch device's visual positioning system, the suction pressure is collected in real time by a negative pressure sensor, and the suction timing information is obtained from the execution log of the device control system. When performing process execution component association mapping, the data of the action section, support section and suction section are first grouped according to the process section identifier. Then, based on the component correspondence table predefined in the equipment configuration file, the upper ejector, ejector pin, ejector pin array and crystal suction nozzle are physically mapped. The correspondence between the upper ejector and ejector pin is determined by the equipment structure assembly information, the affiliation between ejector pin and ejector pin array is determined by the array structure parameters, and the spatial association between ejector pin array and crystal suction nozzle is established through the position mapping relationship in the equipment coordinate system, thus forming the initial set of component connection relationships. When generating the lifting support correlation spectrum, the above-mentioned components and their connection relationships are organized according to a preset data structure. The data structure adopts the form of node-connection edge, wherein the upper ejector, ejector pin, ejector pin array and crystal suction nozzle are respectively used as nodes, and the action path relationship between nodes is used as connection edge. Each node is accompanied by corresponding process parameter attribute information, including the upper ejector node's lifting height and model attribute, the ejector pin node's type attribute, the ejector pin array node's arrangement and pin spacing attribute, and the crystal suction nozzle node's suction position and pressure attribute. The lifting support correlation spectrum containing node attributes and connection relationships is constructed in this way. When performing the evidence chain closure determination, the connection edges in the lifting support correlation spectrum are traversed according to the process path sequence, and a pre-stored process rule library is called to verify each connection relationship. The process rule library is constructed from equipment process specifications and historical stable process window data, which includes the allowed matching relationships between various lifting actions and support responses, as well as the constraints of the suction operation. During the verification process, for the connection relationship from the upper ejector to the ejector pin, it is verified whether the lifting action category matches the ejector pin type; for the connection relationship from the ejector pin to the ejector pin array, it is verified whether the ejector pin distribution conforms to the array structure constraints; for the connection relationship from the ejector pin array to the suction nozzle, it is verified whether there is a spatial coverage relationship between the ejector pin action area and the suction position.When a non-matching condition in the process rule base is detected during the verification of any connection relationship, the subsequent traversal of the current path is immediately terminated, and the node in the corresponding connection relationship is marked as a mismatch node. At the same time, the process parameter information corresponding to the node is recorded as abnormal evidence and output, and the evidence chain status is marked as broken. When all connection relationships pass the verification in sequence, it is determined that the entire process action link meets the continuity and consistency requirements, the evidence chain status is marked as closed, and the closed status identifier is output for subsequent attribution analysis.

[0033] The invention is further configured such that the pin projection shell is a detection area mask on the back of the chip corresponding to the theoretical contact position of each pin. Specifically, its construction is based on the number of rows, columns, pin spacing, and pin diameter of the pin array. By calculating the theoretical contact position of each pin on the back of the chip as the center, and using the pin diameter plus a preset mechanical deviation or image registration error as the radius, a series of circular areas are generated. The set of these circular areas constitutes the pin projection shell. The function of the pin projection shell is to strictly limit the trace detection range on the back of the chip to the theoretical area that the pins may contact, thereby eliminating the influence of interference factors such as scratches, stains, random textures, or reflections in non-contact areas on the detection results. At the same time, when a crack or indentation is detected in the projection shell area, the corresponding pin can be accurately located based on the circular area where the trace is located, establishing a spatial correspondence between the trace and the specific pin. In overall attribution determination, the pin projection shell is a key tool for matching process parameters with physical traces. By transforming the abstract pin array structure into a quantifiable image analysis area, it provides direct physical evidence for determining whether the fracture is caused by the chip mounting station.

[0034] The present invention is further configured such that S4 includes: When the evidence chain is closed, the theoretical contact position of each pin on the back of the chip is determined according to the pin array structure parameters configured in the chip mounting process, and a pin position projection shell corresponding to the pin array is constructed based on the theoretical contact position through a region projection mapping method. Based on the area defined by the pin projection shell, the acquired chip back image is segmented, and the indentation features, microcrack features or contact imprint features within the area are detected by image feature recognition to obtain the corresponding trace detection results. The acquired trace detection results are spatially matched with the theoretical contact positions of each pin, and the isotopicity between the trace information and the corresponding theoretical pin position is determined by a position correspondence determination method, generating an isotopicity judgment result for the pin contact trace. Specifically, when the evidence chain is closed, the theoretical contact position of each pin on the back of the chip is determined based on the pin array structure parameters configured in the chip mounting process. The pin array structure parameters include the number of rows, the number of columns, the pin pitch, and the pin diameter, for example, 4 rows, 4 columns, a pin pitch of 0.5 mm, and a pin diameter of 0.1 mm. The relative position of each pin on the back of the chip is determined based on the row and column size of the pin array and the pin pitch parameters, thus forming a coordinate set consisting of all the theoretical contact positions of the pins. Based on the theoretical contact positions, a pin position projection shell corresponding to the pin array is constructed through a region projection mapping method. The pin position projection shell is a set of detection areas established around each theoretical contact position of the pin. Each detection area is centered on the theoretical contact position of the corresponding pin, and its range is determined by the pin size parameters and the system's allowable mechanical deviation and image registration error. For example, in some implementations, the radius of the detection area can be determined based on the pin diameter and allowable deviation range. For instance, if the pin diameter is 0.1 mm, the detection area radius can be set to approximately 0.07 mm to cover the pin contact area and any potential error range. In actual image processing, the physical coordinates of the theoretical contact position need to be converted to image coordinates. This conversion is determined based on the spatial resolution parameters of the image acquisition system. For example, when the image resolution is 100 pixels per millimeter, the physical coordinates can be mapped to image pixel coordinates according to the corresponding ratio, and the detection area range is converted to the corresponding pixel size. Based on the area defined by the pin projection shell, the acquired chip back image is segmented, retaining only the image data within the projection shell. Within the projection shell area, indentation features, microcrack features, or contact mark features are detected using image feature recognition. Image feature recognition can be achieved through image denoising, edge extraction, and region connectivity analysis. The detected image features are then filtered according to preset feature judgment rules to identify trace information that may be generated by pin contact. The image coordinates of the center position of the detected trace are recorded and converted to the corresponding physical spatial position based on the image resolution. The acquired trace information is spatially matched with the theoretical contact positions of each ejector pin. The closest ejector pin position is determined by calculating the spatial distance between the center of the trace and each theoretical contact position. The isotopicity between the trace information and the corresponding theoretical ejector pin position is determined using a position correspondence determination method. The isotopicity determination threshold can be set according to the ejector pin size parameters and the system's allowable deviation range. For example, in some embodiments, when the ejector pin diameter is 0.1 mm, the isotopicity determination threshold can be set to approximately 0.15 mm.When the distance between the center of a trace and the theoretical position of the corresponding pin is less than or equal to the threshold, the trace is considered to be in situ with the theoretical position of the corresponding pin, and the array number of the corresponding pin is recorded. When the distance is greater than the threshold, it is considered to be in situ. If no trace is detected within the projection shell area, it is considered to be without a trace. Finally, the situation judgment result corresponding to the pin contact trace is generated, which includes whether a trace exists, the number of traces, and the situation judgment information of the pin position corresponding to each trace. For example, for an array of 4 rows and 4 columns, a pin spacing of 0.5 mm, and a pin diameter of 0.1 mm, with an image resolution of 100 pixels / mm, a trace is detected within the projection shell area. The physical position corresponding to its image coordinates is approximately (0.52, 0.07) mm, which is close to the theoretical contact position of the pin in the 1st row and 2nd column. After spatial distance comparison, it is within the situation judgment threshold range, and therefore is judged to be a situational trace. In another example, if the distance between the detected trace location and the nearest theoretical contact location of the ejector pin is significantly greater than the set isotope determination threshold, it is determined to be a non-isotope trace. If no trace information is detected within the projection shell area, a trace-free detection result is generated, and this result is output as the isotope determination result corresponding to the ejector pin contact trace.

[0035] The present invention is further configured such that S5 includes: Obtain the status of the chain of evidence, trace detection results, and isotope determination results; The evidence fusion rules include: When the chain of evidence is broken, an attribution conclusion for abnormal process configuration is generated based on the mapping relationship that does not meet the process rules. When the chain of evidence is closed, attribution is determined by combining trace detection results and isotope assessment results: If no effective traces are detected in the area of ​​the pin projection shell, an attribution conclusion of low risk of cracking at the patching station is generated. If a trace is detected in the area of ​​the pin projection shell and the trace is in the same position as the theoretical position of the corresponding ejector pin, then the attribution conclusion of the crack caused by the patching station is generated and the corresponding pin position information is output. If a trace is detected in the area of ​​the pin projection shell but the trace is not in the same position as the theoretical position of the corresponding pin, then the attribution conclusion that the fracture originates from other workstations is generated. The attribution conclusions are combined with a preset output template to form the FRD fracture attribution conclusion. Specifically, the evidence chain state originates from the preceding correlation spectrum construction and closure determination steps. It is the result of determining whether the top-level action path is continuous and effective, specifically including "closed state" and "fractured state". A fractured state is determined when the mapping relationship between any process execution component does not meet the constraints of the process rule base; otherwise, a closed state is determined. The evidence chain state is output by the evidence chain verification module and serves as the first constraint condition for attribution determination. The trace detection results originate from the defect detection processing of the chip bottom image. The chip bottom image is acquired by an online visual inspection device. An industrial camera images the bottom area after chip mounting, and after denoising, grayscale normalization, and edge enhancement processing, it is input into the defect detection algorithm for feature recognition to extract trace information of suspected FRD fracture areas. The trace detection results include trace position coordinates, trace morphological features, and trace area information. The trace position coordinates are the two-dimensional position of the trace area in the chip coordinate system, obtained by image coordinate transformation. The trace morphological features are used to characterize the shape characteristics of cracks or indentations, and the trace area is calculated statistically from the pixel area. The isotope determination result is obtained by comparing the spatial projection model of the pin's action area with the trace detection result. The theoretical pin position is calculated from the device structural parameters and pin array configuration, and mapped to the chip image plane through the calibration relationship between the device coordinate system and the image coordinate system, forming a pin position projection shell area. Based on this, it is determined whether the trace position coordinates fall into the corresponding pin position projection shell area, and the spatial offset between the trace center and the corresponding theoretical pin position is further calculated. If the offset is less than a preset tolerance range, it is determined to be an isotope relationship; otherwise, it is determined to be a non-isotope relationship, thus obtaining the isotope determination result. During attribution determination, branching is first performed based on the evidence chain state: when the evidence chain state is broken, the mismatch node information and corresponding process parameters recorded during the correlation spectrum verification process are directly read, mapped to the specific process execution component, and an attribution conclusion for process configuration anomalies is generated according to preset anomaly type mapping rules. These anomaly type mapping rules are used to map mismatches to specific device configuration anomaly types, such as pin type mismatch or unreasonable array structure.When the evidence chain is closed, the attribution is further refined by combining the trace detection results and the isotope judgment results: First, it is determined whether there are valid traces in the pin projection shell area. The valid traces are determined by area threshold and morphological feature screening rules. When no traces that meet the conditions are detected, it is determined that the cracking risk of the patching station is low. When valid traces are detected, the cracking is further classified according to the isotope judgment results. If the trace and the theoretical position of the ejector pin meet the isotope relationship, the corresponding ejector pin node is marked as the cracking source, and the position number of the ejector pin in the array is extracted as the pin position information output. If the trace and the theoretical position of the ejector pin do not meet the isotope relationship, it is determined that the fracture source does not belong to the current ejector pin action path, but is attributed to other stations or other process links. After the attribution determination is completed, the attribution conclusions and corresponding key parameter information are structured and formatted according to a preset output template. The output template predefines the attribution category, corresponding process section, abnormal component, and key parameter fields. By filling in the corresponding fields, a standardized FRD fracture attribution conclusion is generated and output to the host computer system or quality traceability system for subsequent analysis and decision-making.

[0036] 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 determining the attribution of FRD breakage based on the evidence chain of the pin manufacturing process, characterized in that, include: S1: Collect process parameters during the chip mounting process to form a set of process evidence data. The process parameters include: top-mounting action parameters, pin array structure, suction execution parameters, and production batch and product traceability information. S2: Encode the process evidence according to the three-segment structure of action segment, support segment and absorption segment to generate a process signature string; S3: Based on the process signature string, extract the top action information corresponding to each action segment and the support response information corresponding to the support segment, construct the matching relationship between the top action and the support response, and form a top support correlation spectrum based on the matching relationship. Then, perform process evidence chain closure reasoning based on the top support correlation spectrum to generate evidence chain states, including: broken state and closed state. S4: When the evidence chain is closed, the spatial action mapping relationship of the pin array is determined based on the support correlation spectrum, and the pin position projection shell is constructed according to the pin array structure. Trace detection is performed on the back image of the chip within the pin position projection shell area to generate trace detection results. The isotopicity between the trace detection results and the corresponding theoretical position of the pin is determined to generate isotopicity judgment results. S5: Based on the status of the evidence chain, trace detection results, and isotope judgment results, generate attribution conclusions according to the preset evidence fusion rules, and finally output the FRD breakage attribution conclusion.

2. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 1, characterized in that, S1 includes: Read the top-mounting action parameters during the chip mounting process from the equipment control system. The top-mounting action parameters include: top-mounting height, top-mounting device model, and pin type. Obtain the ejector pin array structure and suction execution parameters from the equipment process formula library, wherein the suction execution parameters include: the crystal suction nozzle specification and the station number; Obtain production batch and product traceability information from the production management system. The production batch and product traceability information includes: chip ID, production batch, process batch, equipment number, and execution time. The parameters of the top-mounted action, the structure of the ejector pin array, the suction execution parameters, and the production batch and product traceability information are correlated and integrated to form a set of process evidence.

3. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 1, characterized in that, S2 includes: Based on the set of process evidence, the top action parameters are classified into the action segment, the ejector pin array structure into the support segment, and the suction execution parameters into the suction segment. The parameters in the action segment, support segment, and suction segment are organized in a structured manner according to a preset coding rule, so that the parameters of each segment are arranged in a predetermined order, forming a segmented coding structure to express the process configuration relationship between the top action, the ejector pin support, and the suction execution. The encoding results of the action segment, support segment, and absorb segment are sequentially concatenated to form a process signature string that characterizes the chip mounting process configuration relationship.

4. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 1, characterized in that, The jacking support correlation spectrum includes four nodes: the upper jack, the jacking pin, the jacking pin array, and the crystal suction nozzle, as well as the mapping relationship between the nodes established based on the matching relationship between the jacking action and the support response.

5. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 4, characterized in that, The S3 includes: the step of extracting function information and constructing matching relationships, and the step of constructing association spectrum and determining the closure of evidence chain.

6. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 5, characterized in that, The steps for extracting functional information and constructing matching relationships include: Based on the process signature string, the codes in the action segment, support segment and absorption segment are parsed to recover the corresponding process parameter values; The upper lifting height, upper lifting device model and ejector pin type in the process parameter values ​​of the action segment are combined according to the preset parameter combination rules, and the lifting action information used to characterize the upper lifting execution state is generated through parameter mapping. The structural parameters of the ejector array in the process parameter values ​​of the support section are analyzed. The ejector arrangement, pin spacing and array size information are extracted by the structural parameter extraction rules. Support response information for characterizing the support state is generated by the structural feature induction method. The jacking action information and support response information are aligned according to time sequence and process segment correspondence, and the two types of information are compared item by item based on preset parameter matching rules to establish the matching relationship between jacking action and support response.

7. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 6, characterized in that, The steps for constructing the correlation spectrum and determining the closure of the evidence chain include: Based on the matching relationship, the corresponding process execution components in the action segment, support segment and suction segment are associated and mapped to determine the connection relationship between the upper ejector, ejector pin, ejector pin array and crystal suction nozzle; Organize each process execution component and its connection relationship according to the preset data structure to generate a jacking support correlation spectrum containing nodes and mapping relationships between nodes; Iterate through the connection relationships of each node in the top support association spectrum, call the pre-stored process rule library, and verify each mapping relationship item by item to determine whether the process matching requirements are met between each node. When any mapping relationship does not meet the process matching requirements, the evidence chain is marked as broken, and the corresponding mismatch node information is recorded. When all mapping relationships meet the process matching requirements, the evidence chain is marked as closed.

8. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 1, characterized in that, The pin projection shell is a mask on the back of the chip corresponding to the detection area where each pin is theoretically in contact with the chip.

9. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 8, characterized in that, S4 includes: When the evidence chain is closed, the theoretical contact position of each pin on the back of the chip is determined according to the pin array structure parameters configured in the chip mounting process, and a pin position projection shell corresponding to the pin array is constructed based on the theoretical contact position through a region projection mapping method. Based on the area defined by the pin projection shell, the acquired chip back image is segmented, and the indentation features, microcrack features or contact imprint features within the area are detected by image feature recognition to obtain the corresponding trace detection results. The acquired trace detection results are spatially matched with the theoretical contact positions of each ejector pin, and the isotope between the trace information and the corresponding theoretical position of the ejector pin is determined by the position correspondence determination method, generating the isotope determination result corresponding to the ejector pin contact trace.

10. The method for determining the cause of FRD breakage based on the evidence chain of the pin manufacturing process according to claim 1, characterized in that, S5 includes: Obtain the status of the chain of evidence, trace detection results, and isotope determination results; The evidence fusion rules include: When the chain of evidence is broken, an attribution conclusion for abnormal process configuration is generated based on the mapping relationship that does not meet the process rules. When the chain of evidence is closed, attribution is determined by combining trace detection results and isotope assessment results: If no effective traces are detected in the area of ​​the pin projection shell, an attribution conclusion of low risk of cracking at the patching station is generated. If a trace is detected in the area of ​​the pin projection shell and the trace is in the same position as the theoretical position of the corresponding ejector pin, then the attribution conclusion of the crack caused by the patching station is generated and the corresponding pin position information is output. If a trace is detected in the area of ​​the pin projection shell but the trace is not in the same position as the theoretical position of the corresponding pin, then the attribution conclusion that the fracture originates from other workstations is generated. The attribution conclusions are output and combined with the preset output template to form the FRD fracture attribution conclusion.

Images