CN118426792A - Automatic switching method, device, equipment and storage medium for chip burning - Google Patents

Abstract

The application relates to the technical field of chip burning, and discloses an automatic switching method, device and equipment for chip burning and a storage medium. The method comprises the following steps: acquiring information of a plurality of chips to be burned and constructing a chip information database; identifying and obtaining the chip model and obtaining the specification parameters of the target burning chip; matching a burner and a burning program from a chip information database; establishing physical connection and checking electrical connection state information; the programming program is used for controlling the burner to execute erasing, programming and checking operations, and current parameter data and voltage parameter data are monitored in real time; performing abnormality detection in the burning process through a naive Bayes model to obtain an abnormality detection result; if the abnormality detection result is abnormal, generating an abnormality processing scheme, if the abnormality detection result is normal, continuing the burning operation and automatically switching the next chip to be burned to execute the burning operation.

Description

Automatic switching method, device, equipment and storage medium for chip burning

Technical Field

The present application relates to the field of chip recording technologies, and in particular, to a method, an apparatus, a device, and a storage medium for automatically switching chip recording.

Background

With the increasing complexity of functions and the increasing integration of electronic devices, the types and models of chips are increased, and the demands for writing chips are becoming urgent. The conventional chip burning method relies on manual operation, and an operator needs to manually identify the type of the chip, select corresponding burning equipment and programs, and physically connect the chip to a burner. The method is low in efficiency, and is easy to cause artificial errors, so that the burning of the chip fails or the quality is unstable. In addition, the specifications and the burning demands of different chips are different, the manual selection and the connection process are complicated, and the automation degree and the production efficiency of the production line are greatly restricted.

The prior art has significant shortcomings in terms of automation and intelligence. The prior art lacks effective processing of chip information, the identification and information verification process of the chip mostly depends on manual intervention, the automatic identification means is limited, the information acquisition is incomplete, the real-time monitoring and exception handling mechanism is lacking, once an exception condition occurs, the exception condition cannot be found and handled in time, and the chip and equipment are easy to damage.

Disclosure of Invention

The application provides an automatic switching method, device and equipment for chip burning and a storage medium, which are used for improving the accuracy of chip burning and realizing the automatic switching of the chip burning process.

In a first aspect, the present application provides an automatic switching method for chip burning, where the automatic switching method for chip burning includes:

acquiring a plurality of pieces of chip information to be burned, classifying and structuring to obtain a plurality of pieces of standard chip information, and constructing a chip information database according to the plurality of pieces of standard chip information;

Performing surface optical character recognition on a target burning chip to obtain a chip model, and performing information verification through a radio frequency identification technology to obtain specification parameters of the target burning chip;

according to the chip model and the specification parameters, matching a burner and a burning program of the target burning chip from the chip information database;

Establishing physical connection between the target burning chip and the burner, and performing electric connection state inspection based on chip pin definition to obtain electric connection state information;

Based on the electrical connection state information, programming instructions and data are sent through the programming program, the burner is controlled to execute erasing, programming and checking operations, and meanwhile, current parameter data and voltage parameter data of the target burning chip are monitored in real time;

inputting the current parameter data and the voltage parameter data into a preset naive Bayesian model for abnormality detection in the burning process, and obtaining an abnormality detection result;

if the abnormality detection result is abnormal, performing abnormality investigation on the burner and generating an abnormality processing scheme, if the abnormality detection result is normal, controlling the burner to continue the burning operation, and when the target burning chip finishes burning, automatically switching the next chip to be burnt to execute the burning operation.

In a second aspect, the present application provides an automatic switching device for chip burning, where the automatic switching device for chip burning includes:

the acquisition module is used for acquiring a plurality of pieces of chip information to be burned, classifying and structuring the chip information to obtain a plurality of pieces of standard chip information, and constructing a chip information database according to the plurality of pieces of standard chip information;

The verification module is used for carrying out surface optical character recognition on the target burning chip to obtain a chip model, and carrying out information verification through a radio frequency identification technology to obtain specification parameters of the target burning chip;

The matching module is used for matching the burner and the burning program of the target burning chip from the chip information database according to the chip model and the specification parameters;

The checking module is used for establishing physical connection between the target burning chip and the burner and checking the electric connection state based on the definition of chip pins to obtain electric connection state information;

The processing module is used for sending programming instructions and data through the programming program based on the electrical connection state information, controlling the writer to execute erasing, programming and checking operations, and simultaneously monitoring current parameter data and voltage parameter data of the target programming chip in real time;

the detection module is used for inputting the current parameter data and the voltage parameter data into a preset naive Bayesian model to perform abnormality detection in the burning process, so as to obtain an abnormality detection result;

and the switching module is used for carrying out abnormality investigation on the burner and generating an abnormality processing scheme if the abnormality detection result is abnormal, controlling the burner to continue the burning operation if the abnormality detection result is normal, and automatically switching the next chip to be burned to execute the burning operation when the target burning chip finishes the burning.

The third aspect of the present application provides an automatic switching device for chip burning, comprising: a memory and at least one processor, the memory having instructions stored therein; and the at least one processor calls the instruction in the memory so that the automatic switching equipment for the chip burning executes the automatic switching method for the chip burning.

A fourth aspect of the present application provides a computer-readable storage medium having instructions stored therein, which when run on a computer, cause the computer to perform the above-described automatic switching method of chip burn-in.

According to the technical scheme provided by the application, through classification and structuring processing, a comprehensive chip information database is constructed, data query and update are efficiently performed, and the accuracy and the comprehensiveness of information in the burning process are ensured. By utilizing the optical character recognition and wireless radio frequency recognition technology, the model and specification parameters of the chip are automatically acquired, the error of manual operation is reduced, and the recognition precision and efficiency are improved. Based on the data in the chip information database, the most suitable burning equipment and program are automatically matched according to the chip type and specification parameters, so that the technical index of the burning equipment meets or exceeds the chip requirement, and the manual matching error is avoided. The automatic connection of the chip and the burning equipment is realized through the mechanical arm and the alignment device, and the electric connection state inspection is carried out according to the definition of the pins of the chip, so that the reliability of physical connection and the accuracy of electric connection are ensured, and the risk of manual connection errors is reduced. In the burning process, current and voltage parameters of the chip are monitored in real time through the equipment interface, and abnormality detection in the burning process is carried out by using a preset naive Bayesian model, so that abnormal conditions can be found and processed in time, and the burning quality and equipment safety are ensured. The automatic control burning device executes erasing, programming and checking operations, and consistency and efficiency of burning operations are improved. After the burning is finished, the chip to be burned is automatically switched to the next chip to be burned, so that the waiting time is reduced, and the equipment utilization rate and the production rhythm are improved. When an abnormality is detected in the burning process, abnormality troubleshooting is automatically performed and an abnormality processing scheme is generated, so that normal operation can be quickly recovered even under abnormal conditions, and the influence on a subsequent burning task is avoided.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained based on these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic diagram of an embodiment of an automatic switching method for chip burning in an embodiment of the present application;

fig. 2 is a schematic diagram of an embodiment of an automatic switching device for chip burning in an embodiment of the application.

Detailed Description

The embodiment of the application provides an automatic switching method, device and equipment for chip burning and a storage medium. The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.

For easy understanding, the following describes a specific flow of an embodiment of the present application, referring to fig. 1, and an embodiment of an automatic switching method for chip burning in an embodiment of the present application includes:

Step S101, obtaining a plurality of pieces of chip information to be burned, classifying and structuring to obtain a plurality of pieces of standard chip information, and constructing a chip information database according to the plurality of pieces of standard chip information;

It can be understood that the execution body of the present application may be an automatic switching device for chip burning, and may also be a terminal or a server, which is not limited herein. The embodiment of the application is described by taking a server as an execution main body as an example.

Specifically, basic information of the chip to be burned is obtained through chip information inquiry, and the basic information may come from manufacturer data, inventory system or other sources of the chip. The information is classified according to a certain standard, the classification basis can be the type, model or other relevant parameters of the chip, and different types of chips can be better managed and distinguished through classification. And carrying out structural processing on the classified chip information, and arranging the information according to a predefined format to ensure normalization and consistency of the information. Such structured information includes the type, model number, specification parameters, number of chips, and unique identification codes for each chip. And generating a unique identification code and associating the unique identification code with the corresponding structured chip information to obtain associated chip information. On the basis, the related chip information is subjected to standardized processing, so that all the chip information accords with unified standards and formats, and subsequent database construction and information management are facilitated. And constructing a chip information database according to the standardized chip information. The database contains standard information of all chips to be burnt, can be used for subsequent chip burning operation, quickly obtains detailed information of each chip by inquiring the database, and is matched with a proper burner and a burning program, so that the high efficiency and the accuracy of the burning process are ensured.

Step S102, carrying out surface optical character recognition on a target burning chip to obtain a chip model, and carrying out information verification through a radio frequency identification technology to obtain specification parameters of the target burning chip;

Specifically, the surface gray level image acquisition is carried out on the target burning chip through the high-resolution industrial camera to obtain a surface gray level image, and the surface gray level image is subjected to image preprocessing to obtain a preprocessed surface image. Through the image preprocessing steps including denoising, contrast enhancement, edge detection and other operations, the accuracy of subsequent character positioning and segmentation is improved. And according to the preprocessed surface image, positioning and segmenting the surface characters of the target burning chip to obtain segmented character images. In the character positioning process, the position of the character on the surface of the chip is accurately found out through a positioning algorithm, and the character is separated from the background through a segmentation algorithm, so that a clear segmentation character image is formed. And carrying out character recognition on the segmented character image to obtain the chip model. Character recognition processes typically utilize optical character recognition techniques to convert segmented character images into readable chip models through a trained recognition model. The identified chip model is used for subsequent checksum comparison. Meanwhile, a radio frequency identification tag of the target burning chip is obtained. And reading the RFID tag on the chip by using a wireless radio frequency identification technology to obtain a wireless radio frequency identification result containing chip information. In order to ensure the accuracy of identification, the radio frequency identification result is compared with the chip model to obtain a comparison result. If the comparison result is inconsistent or not matched, the surface optical character recognition and the radio frequency recognition are needed to be carried out on the target burning chip again until the comparison result is consistent and matched. Through cyclic detection and verification, accuracy and reliability of chip model identification are guaranteed, and burning failure or other problems caused by identification errors are avoided. And when the comparison result is consistent and matched, acquiring specification parameters of the target burning chip from a chip information database according to the chip model. These specification parameters include chip package type, chip pin count, chip operating voltage, and chip memory capacity.

Step S103, according to the model and specification parameters of the chip, matching a burner and a burning program of a target burning chip from a chip information database;

Specifically, a chip information database is established, and the database contains model numbers and specification parameters of various chips, such as chip packaging types, chip pin numbers, chip working voltages, chip storage capacities and the like. When the chip is required to be burned, the system can search in the database by reading the model and specification parameters of the target burned chip so as to find the record corresponding to the target burned chip. The query process ensures that the retrieved record corresponds exactly to the target chip by matching the key field of the chip model. And extracting detailed specification parameters of the chip according to the inquired records. These parameters are the basis for selecting the appropriate burner and programming. Each chip typically corresponds to one or more specific models of writers that are capable of precise operation based on the physical and electrical characteristics of the chip. The specification parameters of the chip, such as pin number and arrangement, operating voltage range, storage capacity, etc., directly affect the selection of the burner. Therefore, the system will screen out compatible burner models from the database based on these parameters. At the same time, each chip also requires a specific programming procedure. The programming program not only comprises basic programming instructions, but also comprises optimization parameters and a verification algorithm aiming at a specific chip, so that the accuracy and the integrity of programming data are ensured. Thus, the system extracts the burn program matching the target chip from the database while selecting the burner. Each programming program in the database is recorded in detail and contains information such as the type of the applicable chip, the required programming environment, specific operation steps and the like. In the matching process, the compatibility and the optimality of the selected burner and the burning program are ensured through a plurality of algorithms and rules. These algorithms will take into account all specification parameters of the chip in combination, preferably selecting those combinations that perform best in practice.

Step S104, establishing physical connection between the target burning chip and the burner, and performing electric connection state inspection based on the definition of the chip pins to obtain electric connection state information;

Specifically, according to the specification parameters, a chip pin definition is obtained from a chip information database, wherein the chip pin definition comprises a power supply pin, a ground pin, a clock pin, a data pin and a control pin. Based on the pin definitions, a pin connection configuration file of the target burning chip is generated, wherein the configuration file lists the number, the name, the type and the connection mode of each pin, and the accuracy and the consistency in the physical connection are ensured. And placing the target burning chip on a chip clamp of the burner in the generation of the pin connection configuration file, and physically connecting an adapter pin of the burner with a pin of the target burning chip through a mechanical arm according to the pin connection configuration file to establish reliable physical connection between the target burning chip and the burner. And after the physical connection is established, carrying out power-on test on the target burning chip and the burner to obtain a power-on test result. The power-on test comprises the steps of applying voltage to the power pin and the ground pin, and detecting whether current is in a normal range, so as to judge whether the connection between the power pin and the ground pin is normal. By preliminary testing the connection of the power and ground pins, some basic connection problems can be found and eliminated. And according to the power-on test result and the pin connection configuration file, checking the electrical connection state of each pin of the target burning chip to obtain the electrical connection state of each pin. The electrical connection state check not only includes detecting whether the pins are connected, but also includes detecting whether the pin voltage is normal and whether the pin resistance is within a reasonable range. By applying a specific voltage and current, the response of each pin is detected, ensuring that it meets the expected electrical characteristics. And summarizing the electrical connection state information of each pin to obtain the whole electrical connection state information. This information records the connection of each pin, including connectivity, voltage level, and resistance.

Step 105, based on the electrical connection state information, performing programming instruction and data transmission through a programming program, controlling a writer to perform erasing, programming and checking operations, and simultaneously monitoring current parameter data and voltage parameter data of a target writing chip in real time;

Specifically, based on the electrical connection state information, programming instructions and data are sent through a programming program, and the burner is controlled to execute erasing, programming and checking operations. By analyzing the electrical connection status information, the programming program can determine the specific model and requirements of the target chip and then send the appropriate programming instructions and data. The burning program interacts with the burner through the communication interface and sends an erasing instruction. The erasing operation is to clear the original data in the target chip, so as to ensure that the writing of new data is not interfered. After the erasing process is completed, the programming program sends programming instructions and new data is written into the target chip. This process involves transmitting blocks of data, each of which needs to be checked to ensure accuracy and integrity of the writing. During programming, current parameter data and voltage parameter data are monitored in real time. The monitoring module built in the burner can continuously collect current and voltage data of the target chip and transmit the data to the control system in real time. By analyzing these real-time data, the control system detects any anomalies, such as excessive current or abnormal voltage fluctuations, which may be predictive of potential problems with the chip or connection. If abnormality is found, the system immediately stops programming operation, and performs fault checking and processing to prevent chip damage or data writing errors. After the programming operation is completed, the programming program performs a verification operation. The verification is to read the written chip data and compare the written chip data with the original data to confirm that the data is accurate. Any mismatch is recorded and reported during the verification process, and the system decides whether to reprogram or make further diagnostics and repairs as the case may be.

Step S106, inputting the current parameter data and the voltage parameter data into a preset naive Bayesian model for abnormality detection in the burning process, and obtaining an abnormality detection result;

Specifically, wavelet decomposition is performed on the current parameter data to obtain a multi-scale current component sequence, and according to the multi-scale current component sequence, the time domain statistical characteristic and the frequency domain energy characteristic of the current are extracted to obtain a current characteristic vector set. The complex current signal is decomposed into components with different scales through wavelet decomposition, so that characteristics in different frequency ranges are captured, and the time domain statistical characteristics and the frequency domain energy characteristics can comprehensively reflect the change of the current signal. And meanwhile, performing short-time Fourier transform on the voltage parameter data to obtain a time-frequency energy distribution image, and performing image segmentation and mode extraction on the time-frequency energy distribution image to obtain a voltage characteristic image set. The short-time Fourier transform can provide joint distribution of voltage signals in time and frequency, and key voltage characteristic information is effectively extracted from time-frequency energy distribution images through image segmentation and mode extraction technology. And carrying out data fusion on the current characteristic vector set and the voltage characteristic image set to obtain a burning parameter fusion characteristic set. Based on the burning parameter fusion feature set, a feature space and a class prior probability of a naive Bayes model are constructed. The naive bayes model calculates the prior probability of each category by using the sample data in the feature space by assuming independence among features, and the prior probabilities reflect the distribution condition of each category in the data. And carrying out Gaussian kernel function mapping on the feature space to obtain the Gaussian kernel feature space. The Gaussian kernel function mapping can convert the original characteristic space into a high-dimensional space, so that the data distribution is easier to be divided by the linear classifier in the high-dimensional space, and the classification accuracy is improved. And carrying out parameter estimation on the naive Bayesian model according to the Gaussian kernel feature space and the class prior probability to obtain a classifier parameter set. By maximizing the likelihood estimation method, the training data is utilized to optimize the model parameters, so that the model can reflect the statistical characteristics of the data more accurately. And inputting the burning parameter fusion feature set into a naive Bayes model to obtain a classification probability vector. The classification probability vector represents the probability that each feature sample belongs to a different class, and by comparing these probabilities, the most likely class can be determined. And determining the class label of the abnormal detection result according to the classification probability vector. The class label is used for identifying whether an abnormality exists in the burning process, if the classification probability vector indicates that the sample belongs to the abnormality class, the system can send out an alarm and take corresponding measures to ensure the safety and the reliability of the burning process.

And performing kernel matrix calculation on the Gaussian kernel feature space to obtain a kernel matrix, and constructing an objective function and constraint conditions of the semi-positive planning problem according to the kernel matrix. The kernel matrix is obtained by calculating the gaussian kernel function value, reflecting the similarity between sample points in the feature space. The nonlinear problem is converted into linear problem processing through a kernel matrix. And solving the semi-positive definite programming problem by using a sequence minimum optimization algorithm to obtain a support vector set. The sequence minimum optimization algorithm is an efficient optimization method, and can quickly find the optimal solution of the problem. The support vector set includes sample points that play a critical role in the classification decision process. From these support vector sets, weight vectors of gaussian kernel functions of the naive bayes model are calculated. And carrying out Laplacian correction on the class prior probability to obtain the smoothed class prior probability. Laplace correction is a common smoothing technique, and can avoid the problem that the prior probability is zero, so that the model is more robust. And calculating a category weight vector of the naive Bayes model according to the smoothed category prior probability. The class weight vector reflects the prior probability distribution for each class. And performing matrix splicing on the weight vector of the Gaussian kernel function and the category weight vector to obtain a weight parameter matrix of the naive Bayesian model. By stitching the two weight vectors together, a comprehensive parameter matrix is constructed, containing all the key parameters of the model. And carrying out principal component analysis on the weight parameter matrix to obtain a parameter set after dimension reduction. The principal component analysis is a dimension reduction technology, and by extracting the principal components of data, the redundancy of the features can be reduced, and the calculation efficiency of the model can be improved. And constructing a decision function of the naive Bayes model according to the parameter set after the dimension reduction. The decision function is the core of classifying the model, and the classifying result can be output by calculating the input data. And outputting the coefficients of the decision function as a classifier parameter set.

Feature normalization processing is carried out on the burning parameter fusion feature set to obtain a normalized feature set, and a normalized feature vector of a naive Bayes model is constructed according to the normalized feature set. The feature normalization processing ensures that all features are in the same range by normalizing feature data of different scales, and avoids unstable model training or poor classification effect caused by overlarge difference of feature values. And carrying out Gaussian kernel function mapping on the normalized feature vector to obtain a kernel feature vector. Gaussian kernel function mapping is a common kernel method that maps the original feature space to a high-dimensional space through nonlinear transformation, making the data distribution easier to classify in the high-dimensional space. And calculating a class discrimination function value vector according to the kernel feature vector and the weight parameter matrix in the classifier parameter set. The class discrimination function value vector is obtained by carrying out linear combination calculation on the kernel feature vector and the weight parameter, and reflects the possibility that each sample belongs to different classes. And carrying out softmax normalization on the class discrimination function value vector to obtain an unnormalized probability vector. The softmax function converts the discrimination function value vector into probability distribution so that the probability sum of all categories is 1, and subsequent probability calculation and correction are convenient. And carrying out probability correction according to the unnormalized probability vector and the category weight vector in the classifier parameter set to obtain a normalized probability vector. Probability correction adjusts the probability value of each category by combining the category weights, so that the classification result of the model is more accurate. And carrying out logarithmic operation on the normalized probability vector to obtain a log-likelihood probability vector. The logarithmic operation can convert multiplication operation into addition operation, simplify the calculation process, and improve the numerical stability. And constructing a maximum expected estimation problem according to the log likelihood probability vector. The maximum expected estimation problem finds the optimal probability distribution parameters by maximizing the log likelihood probability, and ensures that the prediction result of the model has the highest reliability. And carrying out quasi-Newton iterative optimization on the maximum expected estimation problem to obtain a probability correction vector. The quasi-Newton method is an efficient optimization algorithm, and the optimal solution is continuously approximated through iterative calculation. The probability correction vector reflects the direction and magnitude of the correction of the initial probability during the iteration. And adding the probability correction vector with the normalized probability vector to obtain a classification probability vector. The classification probability vector is the final classification result and represents the probability that each sample belongs to a different class.

And S107, if the abnormality detection result is abnormal, performing abnormality investigation on the burner and generating an abnormality processing scheme, if the abnormality detection result is normal, controlling the burner to continue the burning operation, and when the target burning chip finishes burning, automatically switching the next chip to be burnt to execute the burning operation.

Specifically, when the system detects that the current and voltage parameters in the burning process are abnormal through a naive Bayesian model, the current burning operation is immediately paused so as to prevent further errors or damage. In this state, the abnormality detection program is automatically started, and detailed inspection is performed on the burner and the related components. This process includes various checks such as checking the hardware connection status of the burner, the power supply status, the integrity of the data transmission lines, and the feedback signals from the various sensors. By a comprehensive analysis of these factors, the specific cause of the abnormality can be determined. And generating an exception handling scheme according to the detection result. This approach includes fault diagnosis reporting and suggested repair steps, which may involve replacement of damaged components, readjustment of connection or calibration equipment parameters, etc. And after the abnormal situation is solved, the electric connection state inspection is carried out again, so that all connection points are ensured to be recovered to be normal. If the checking result shows that all parameters are in the normal range, restarting the burning process, and continuously executing the unfinished burning task. If the abnormal detection result is normal, the system controls the burner to continue the burning operation. In the process, the burner performs erasing, programming and checking operations according to preset programming instructions and data streams, so that the target chip is ensured to accurately write the required data. In the whole burning process, current and voltage parameters are continuously monitored, and any abnormal situation possibly occurring is detected in real time. When the target writing chip finishes writing, final verification is automatically performed to confirm that all data are correctly written, and the chip functions normally. After the verification is completed, the burning state and parameter information of the current chip are recorded, and data support is provided for quality tracing. And automatically switching to the next chip to be burned. The new chip is placed in the burner fixture by a robotic arm or other automated means and the physical connection and electrical inspection is repeated. After the next chip is ready, the system repeats the previous burning flow, from the electric connection state checking to the parameter detection, to the actual burning operation, until the burning is completed.

In the embodiment of the application, through classification and structuring processing, a comprehensive chip information database is constructed, data query and update are efficiently carried out, and the accuracy and the comprehensiveness of information in the burning process are ensured. By utilizing the optical character recognition and wireless radio frequency recognition technology, the model and specification parameters of the chip are automatically acquired, the error of manual operation is reduced, and the recognition precision and efficiency are improved. Based on the data in the chip information database, the most suitable burning equipment and program are automatically matched according to the chip type and specification parameters, so that the technical index of the burning equipment meets or exceeds the chip requirement, and the manual matching error is avoided. The automatic connection of the chip and the burning equipment is realized through the mechanical arm and the alignment device, and the electric connection state inspection is carried out according to the definition of the pins of the chip, so that the reliability of physical connection and the accuracy of electric connection are ensured, and the risk of manual connection errors is reduced. In the burning process, current and voltage parameters of the chip are monitored in real time through the equipment interface, and abnormality detection in the burning process is carried out by using a preset naive Bayesian model, so that abnormal conditions can be found and processed in time, and the burning quality and equipment safety are ensured. The automatic control burning device executes erasing, programming and checking operations, and consistency and efficiency of burning operations are improved. After the burning is finished, the chip to be burned is automatically switched to the next chip to be burned, so that the waiting time is reduced, and the equipment utilization rate and the production rhythm are improved. When an abnormality is detected in the burning process, abnormality troubleshooting is automatically performed and an abnormality processing scheme is generated, so that normal operation can be quickly recovered even under abnormal conditions, and the influence on a subsequent burning task is avoided.

In a specific embodiment, the process of executing step S101 may specifically include the following steps:

(1) Chip information inquiry is carried out on a plurality of chips to be burned to obtain information of the plurality of chips to be burned;

(2) Classifying the chips to be burned according to the information of the chips to be burned to obtain information of a plurality of classified chips;

(3) Carrying out structuring treatment on the plurality of classified chip information to obtain a plurality of structured chip information, wherein the structured chip information comprises chip types, chip models, specification parameters, chip numbers and chip IDs;

(4) Generating a unique identification code of each chip to be burned according to the plurality of structured chip information, and associating the unique identification code with the corresponding structured chip information to obtain a plurality of associated chip information;

(5) And carrying out standardization processing on the plurality of associated chip information to obtain a plurality of standard chip information, and constructing a chip information database according to the plurality of standard chip information.

Specifically, the information of the chip to be burned is collected from various sources including the data of the chip manufacturer, the warehouse management system, the supply chain management system and the like through the information inquiry of the chip. By integrating and analyzing these data, the detailed information of all the chips to be burned is comprehensively known, including their production date, manufacturing lot and technical specifications. And classifying the plurality of chips to be burned according to the information of the plurality of chips to be burned. Chips with similar features are grouped into classes for subsequent processing and management. The classification depends on the type of chip (e.g., microcontroller, memory, sensor, etc.), the model of the chip (e.g., specific model of different series and specifications), and the specification parameters of the chip (e.g., voltage, pin count, package type, etc.). By means of the classification, different types of chips can be managed and processed more effectively, and each type of chip can be burned in an optimal mode. And carrying out structuring treatment on the classified chip information. The raw data is converted into data having a uniform format and standard for storage, retrieval and analysis. In this process, the information of each chip is sorted according to a predefined format to form structured chip information. The structured chip information includes chip type, chip model number, specification parameters, chip number, chip ID, and the like. And generating a unique identification code of each chip to be burned according to the structured chip information. The unique identification code is a unique string or code that can uniquely identify each chip. In generating the UID, a unique code is generated using various methods, for example, in combination with the model number, date of manufacture, and serial number of the chip. And associating the UID with the corresponding structured chip information to form associated chip information. And carrying out standardization processing on the associated chip information. The data formats are unified to facilitate subsequent data management and analysis. Redundancy and inconsistencies in the data can be eliminated by the normalization process, ensuring that all data meets predefined criteria. In the standardization process, data cleaning, format conversion, consistency check and other operations are performed. For example, all specification parameters are uniformly converted into the same unit, so that the consistency and comparability of data are ensured. After the normalization process is completed, a plurality of pieces of standard chip information are obtained. The standardized information has high normalization and consistency, and is convenient to store and retrieve. Based on these standard chip information, a chip information database is constructed. The chip information database is a system for storing and managing all chip information in a centralized way, and can provide efficient data query and management functions. In constructing the database, a relational database (e.g., mySQL, postgreSQL) or a NoSQL database (e.g., mongoDB) may be used, and the appropriate database type may be selected according to specific requirements.

In a specific embodiment, the process of executing step S102 may specifically include the following steps:

(1) Acquiring a surface gray level image of a target burning chip through a high-resolution industrial camera to obtain a surface gray level image, and performing image preprocessing on the surface gray level image to obtain a preprocessed surface image;

(2) According to the preprocessed surface image, positioning and segmenting the surface characters of the target burning chip to obtain segmented character images;

(3) Performing character recognition on the segmented character image to obtain a chip model, and acquiring a radio frequency identification tag of a target burning chip;

(4) Performing radio frequency identification on the radio frequency identification tag to obtain a radio frequency identification result, and comparing the radio frequency identification result with the chip model to obtain a comparison result;

(5) If the comparison result is inconsistent or not matched, carrying out surface optical character recognition and wireless radio frequency recognition on the target burning chip again until the comparison result is consistent and matched;

(6) And according to the chip type, acquiring specification parameters of the target burning chip from a chip information database, wherein the specification parameters comprise the chip packaging type, the chip pin number, the chip working voltage and the chip storage capacity.

Specifically, the surface gray level image acquisition is carried out on the target burning chip through the high-resolution industrial camera to obtain a surface gray level image, and the surface gray level image is subjected to image preprocessing to obtain a preprocessed surface image. The image preprocessing comprises the steps of denoising, contrast enhancement, edge detection and the like, and the image quality can be improved through the processing, so that the target characters are more clearly visible. For example, the denoising process may use a gaussian filter, and the contrast enhancement may be achieved by histogram equalization. And according to the preprocessed surface image, positioning and segmenting the surface characters of the target burning chip to obtain segmented character images. Character positioning is achieved by detecting character areas in the image, and techniques such as contour detection or morphological operations can be used. After positioning, each character area is segmented, so that each character is ensured to be separated independently, and a segmented character image is formed. Character segmentation is typically achieved through morphological processing and connected domain analysis. And carrying out character recognition on the segmented character image to obtain the chip model. Character recognition may employ optical character recognition techniques to convert the segmented character image into corresponding text via a trained neural network model, such as a convolutional neural network. Meanwhile, a radio frequency identification tag (RFID) of the target burning chip is obtained. The RFID tag stores the relevant information of the chip, and the data in the tag can be read through a radio frequency identification technology. And carrying out radio frequency identification on the radio frequency identification tag to obtain a radio frequency identification result. An RFID reader is used for communicating with an RFID tag through electromagnetic signals, and information in the tag, such as a serial number, a batch number and the like of a chip, is read. And comparing the radio frequency identification result with the chip model to obtain a comparison result. If the comparison result is inconsistent or not matched, carrying out surface optical character recognition and radio frequency recognition on the target burning chip again until the comparison result is consistent and matched. In the comparison process, if inconsistent is found, image acquisition and recognition can be carried out again so as to ensure the accuracy of a final result. And according to the chip type, acquiring the specification parameters of the target burning chip from a chip information database. The chip information database contains detailed specification parameters of all chips, including package types, pin numbers, working voltages, storage capacities and the like. All necessary information of the target chip is rapidly acquired by querying the database.

In a specific embodiment, the process of executing step S104 may specifically include the following steps:

(1) Acquiring chip pin definitions from a chip information database according to specification parameters, wherein the chip pin definitions comprise a power supply pin, a ground pin, a clock pin, a data pin and a control pin;

(2) Generating a pin connection configuration file of the target burning chip according to the definition of the pins of the chip, wherein the pin connection configuration file comprises the number, the name, the type and the connection mode of each pin;

(3) Placing a target writing chip on a chip clamp of a writer, and according to a pin connection configuration file, physically connecting an adapter pin of the writer with a pin of the target writing chip through a mechanical arm to establish physical connection between the target writing chip and the writer;

(4) Carrying out an electrifying test on the target burning chip and the burner after the physical connection is established to obtain an electrifying test result, wherein the electrifying test comprises the steps of applying voltage to a power pin and a ground pin and detecting whether current is in a normal range or not;

(5) According to the power-on test result and the pin connection configuration file, checking the electrical connection state of each pin of the target burning chip to obtain the electrical connection state of each pin, wherein the electrical connection state comprises whether the pins are communicated, whether the pin voltage is normal or not and whether the pin resistance is in a reasonable range or not;

(6) Summarizing the electric connection state of each pin to obtain electric connection state information

Specifically, the specification parameters of the target chip are analyzed, and the pin definition of the chip is obtained by inquiring the chip information database. Chip pin definition is an important component of chip design, defining the function and connection requirements of each pin. For example, a record in a database may include the following fields: pin number, pin name, pin type, and connection mode. And generating a pin connection configuration file of the target burning chip according to the chip pin definition. The pin connection configuration file records the number, name, type and connection mode of each pin. For example, assume that the chip has the following key pins: the power supply pin, the ground pin, the clock pin, the data pin and the control pin. And placing the target burning chip on a chip clamp of the burner, and physically connecting an adapter pin of the burner with a pin of the target burning chip through a mechanical arm according to the pin connection configuration file. The high precision control of the robotic arm can ensure that each adapter pin is accurately docked to the corresponding chip pin without error, establishing a reliable physical connection. After the physical connection is established, the target burning chip and the burner are subjected to power-on test, and a power-on test result is obtained. Power-on testing is an important step in verifying connection correctness and initially checking chip functionality. In the power-on test process, voltages are applied to the power pin and the ground pin, and whether the connection is normal or not is judged by measuring the current. Assuming that the applied voltage isThe measured current isThe resistance can be calculated by ohm's law：

；

If the resistance value is within the expected range, the power and ground pins are normally connected. And according to the power-on test result and the pin connection configuration file, checking the electrical connection state of each pin of the target burning chip to obtain the electrical connection state of each pin. The electrical connection status check includes detecting whether the pins are connected, whether the pin voltages are normal, and whether the pin resistances are within a reasonable range. For example, connectivity and functionality of pins may be verified by applying a specific test signal to each pin and measuring its response. Assume that a certain pin should display a voltage ofThe actual measured voltage isIf the difference between the two is within the allowable error range, the pin state is normal:

；

wherein, Is the allowable error range. And summarizing the electrical connection state of each pin to obtain the electrical connection state information. The electrical connection state information integrates the detection results of all pins to form a complete report. The report records the status of each pin, including possible problems and suggested solutions. For example, if the voltage of a pin is abnormal, the problem will be described in detail in the report and it is recommended to check the relevant connection or replace the chip.

In a specific embodiment, the process of executing step S106 may specifically include the following steps:

(1) Carrying out wavelet decomposition on the current parameter data to obtain a multi-scale current component sequence, and extracting time domain statistical characteristics and frequency domain energy characteristics of the current according to the multi-scale current component sequence to obtain a current characteristic vector set;

(2) Performing short-time Fourier transform on the voltage parameter data to obtain a time-frequency energy distribution image, and performing image segmentation and mode extraction on the time-frequency energy distribution image to obtain a voltage characteristic image set;

(3) Carrying out data fusion on the current characteristic vector set and the voltage characteristic image set to obtain a burning parameter fusion characteristic set;

(4) According to the burning parameters, fusing the feature sets, and constructing a feature space and a class prior probability of a naive Bayes model;

(5) Carrying out Gaussian kernel function mapping on the feature space to obtain Gaussian kernel feature space;

(6) According to the Gaussian kernel feature space and the class prior probability, carrying out parameter estimation on the naive Bayesian model to obtain a classifier parameter set;

(7) Inputting the burning parameter fusion feature set into a naive Bayes model to obtain a classification probability vector;

(8) And determining the class label of the abnormal detection result according to the classification probability vector.

Specifically, the current parameter data is subjected to wavelet decomposition. Wavelet decomposition is an effective signal processing method, and can decompose complex current signals into components with different scales and capture characteristics in different frequency ranges. The current signal is decomposed by wavelet transformation into a series of multi-scale component sequences, each component corresponding to a particular frequency range. By analyzing these component sequences, time domain statistical features and frequency domain energy features, such as mean, variance, peak and frequency domain energy, etc., of the current signal are extracted. And carrying out short-time Fourier transform on the voltage parameter data to obtain a time-frequency energy distribution image. The short-time fourier transform can provide a joint distribution of the voltage signal in time and frequency, and the frequency spectrum information in each time period is obtained by performing a segmented fourier transform on the signal. And extracting characteristic modes of the voltage signals at different times and frequencies by carrying out image segmentation and mode extraction on the time-frequency energy distribution image to form a voltage characteristic image set. And carrying out data fusion on the current characteristic vector set and the voltage characteristic image set to obtain a burning parameter fusion characteristic set. Features from different signal sources are combined to form a comprehensive feature set to more fully describe the state of the system. And (3) according to the burning parameters, fusing the feature sets, and constructing a feature space and a class prior probability of the naive Bayes model. The naive bayes model calculates the prior probability of each category by assuming independence between features, using sample data in feature space. Let the feature space beClass prior probability ofWhereinRepresenting the category. And (3) calculating the distribution of each category in the feature space by analyzing the historical data, and constructing the category prior probability. And carrying out Gaussian kernel function mapping on the feature space to obtain the Gaussian kernel feature space. Gaussian kernel mapping is a method of non-linearly mapping the original feature space to a high-dimensional space, such that in the high-dimensional space the data distribution is more linearly separable. Let the gaussian kernel function be:

；

Wherein the method comprises the steps of AndThe feature vector is represented by a vector of features,Representing the core parameters. Through Gaussian kernel function mapping, gaussian kernel feature space is obtained. And carrying out parameter estimation on the naive Bayesian model according to the Gaussian kernel feature space and the class prior probability to obtain a classifier parameter set. The parameter estimation is to optimize the model parameters by using sample data in the Gaussian kernel feature space through a maximum likelihood estimation method, so that the model can reflect the statistical characteristics of the data more accurately. Set the model parameter set asWhereinRepresenting categoriesAnd (3) optimizing the parameters through a log-likelihood function to obtain an optimal parameter set. And inputting the burning parameter fusion feature set into a naive Bayes model to obtain a classification probability vector. The probability that each sample belongs to a different class is calculated by inputting the fused feature set into a naive bayes model. Let the classification probability vectorRepresenting that the sample belongs to the categoryIs a probability of (2). Calculating the classification probability by a Bayes formula:

；

wherein, Expressed in categoryThe conditional probability of the lower feature set,Representing a class prior probability that is based on the class prior probability,Representing the marginal probability of the feature set. And determining the class label of the abnormal detection result according to the classification probability vector. By comparing the classification probabilities of each category, the category with the highest probability is selected as the final classification result. Let maximum classification probability beThe class label of the abnormality detection result is. By the method, abnormal conditions in the burning process are accurately detected, and corresponding processing is performed.

In a specific embodiment, the performing step performs parameter estimation on the naive bayes model according to the gaussian kernel feature space and the class prior probability, and the process of obtaining the classifier parameter set may specifically include the following steps:

(1) Performing kernel matrix calculation on the Gaussian kernel feature space to obtain a kernel matrix, and constructing an objective function and constraint conditions of the semi-positive planning problem according to the kernel matrix;

(2) Solving the semi-positive rule problem by using a sequence minimum optimization algorithm to obtain a support vector set, and calculating a weight vector of a Gaussian kernel function of a naive Bayes model according to the support vector set;

(3) Carrying out Laplacian correction on the class prior probability to obtain smoothed class prior probability, and calculating a class weight vector of the naive Bayes model according to the smoothed class prior probability;

(4) Performing matrix splicing on the weight vector of the Gaussian kernel function and the category weight vector to obtain a weight parameter matrix of the naive Bayesian model;

(5) Performing principal component analysis on the weight parameter matrix to obtain a parameter set after dimension reduction, and constructing a decision function of a naive Bayes model according to the parameter set after dimension reduction;

(6) And outputting the coefficients of the decision function as a classifier parameter set.

Specifically, a kernel matrix is calculated on the Gaussian kernel feature space to obtain a kernel matrix, and an objective function and constraint conditions of the semi-positive planning problem are constructed according to the kernel matrix. Gaussian kernel feature space makes data more linearly separable in high-dimensional space by non-linearly mapping the original feature space to the high-dimensional space. Nuclear matrixIs obtained by calculating the Gaussian kernel function value, and the formula is as follows:

；

wherein, AndRespectively, the feature vectors are represented by the feature vectors,Is a nuclear parameter. Nuclear matrixEach element of (2) represents the similarity of two feature vectors in a high-dimensional space. And constructing an objective function and constraint conditions of the semi-positive planning problem, and finding out an optimal support vector and a classification hyperplane through optimizing and solving. The objective function can be expressed as:

；

wherein, Is the lagrange multiplier and is a function of the lagrange,Is a sampleCategory labels of (c). The constraint conditions are as follows:

；

wherein, Is a regularization parameter that controls the complexity of the model. And solving a semi-positive planning problem by a Sequence Minimum Optimization (SMO) algorithm to obtain a support vector set. The SMO algorithm is an efficient optimization method, by decomposing the original problem into a series of small problems, gradually optimizing each small problem, and finally converging to a globally optimal solution. The support vector sets refer to those corresponding Lagrangian multipliersSamples above zero, they play a key role in classification decisions. And calculating the weight vector of the Gaussian kernel function of the naive Bayesian model according to the support vector set. Weight vectorThe support vector and its Lagrangian multiplier are used to determine the formula:

；

wherein, Is the mapping of the support vector in high-dimensional space. And carrying out Laplacian correction on the class prior probability to obtain the smoothed class prior probability. The laplace correction avoids the problem of zero prior probability by adding a small smoothing parameter to each class count. Corrected class prior probabilityThe calculation formula is as follows:

；

wherein, Is a category ofIs used for the number of samples of (a),Is the total number of samples that are to be taken,Is the total number of categories. And calculating a category weight vector of the naive Bayes model according to the smoothed category prior probability. The class weight vector reflects the weight of the prior probability of each class in the classification decision, and the calculation formula is as follows:

；

wherein, Is a category ofIs a weight of (2). And performing matrix splicing on the weight vector of the Gaussian kernel function and the category weight vector to obtain a weight parameter matrix of the naive Bayesian model. And carrying out principal component analysis on the weight parameter matrix to obtain a parameter set after dimension reduction. The redundancy of the features is reduced by extracting the main components of the data, and the calculation efficiency of the model is improved. The principal component analysis step includes calculating a covariance matrix of the weight parameter matrixThen, the feature value decomposition is performed, and feature vectors corresponding to the first several largest feature values are selected as main components. The reduced parameter set is expressed as:

；

wherein, Is a parameter set after the dimension reduction,Is a principal component feature vector matrix. And constructing a decision function of the naive Bayes model according to the parameter set after the dimension reduction. The decision function is used for classifying the new sample, and the posterior probability of each class is calculated, wherein the formula is as follows:

；

wherein, Is a sampleBelongs to the category ofIs used to determine the posterior probability of (1),Is a class-conditional probability of being a class,Is a class a priori probability of being,Is the marginal probability of the feature. And outputting the coefficients of the decision function as a classifier parameter set. The classifier parameter set contains all key parameters for classification decision including Gaussian kernel weight vector, class weight vector and smoothed class prior probability.

In a specific embodiment, the performing step inputs the burning parameter fusion feature set into a naive bayes model, and the process of obtaining the classification probability vector may specifically include the following steps:

(1) Feature normalization processing is carried out on the burning parameter fusion feature set to obtain a normalized feature set, and a normalized feature vector of a naive Bayes model is constructed according to the normalized feature set;

(2) Carrying out Gaussian kernel function mapping on the normalized feature vector to obtain a kernel feature vector, and calculating a class discrimination function value vector according to the kernel feature vector and a weight parameter matrix in a classifier parameter set;

(3) Carrying out softmax normalization on the class discrimination function value vector to obtain an unnormalized probability vector, and carrying out probability correction according to the unnormalized probability vector and a class weight vector in the classifier parameter set to obtain a normalized probability vector;

(4) Carrying out logarithmic operation on the normalized probability vector to obtain a log-likelihood probability vector, and constructing a maximum expected estimation problem according to the log-likelihood probability vector;

(5) And carrying out quasi-Newton method iterative optimization on the maximum expected estimation problem to obtain a probability correction vector, and adding the probability correction vector and the normalized probability vector to obtain a classification probability vector.

Specifically, feature normalization processing is performed on the burning parameter fusion feature set to obtain a normalized feature set, and a normalized feature vector of a naive Bayes model is constructed according to the normalized feature set. The feature normalization process converts feature values of different scales into a common range, typically between 0 and 1, so as to avoid excessive impact on model training due to different scales of certain features. And carrying out Gaussian kernel function mapping on the normalized feature vector to obtain a kernel feature vector. Gaussian kernel mapping is a nonlinear transformation method that can map the original feature space to a high-dimensional space, making the data more linearly separable in the high-dimensional space. The formula of the gaussian kernel function is:

；

wherein, AndIs the normalized feature vector of the object to be processed,Is a nuclear parameter. By calculating the gaussian kernel function values, kernel feature vectors are obtained, which will be used in the subsequent classification process. And calculating a class discrimination function value vector according to the kernel feature vector and the weight parameter matrix in the classifier parameter set. The class discrimination function value vector is obtained by linearly combining the kernel feature vector and the weight parameter matrix, and the formula is as follows:

；

wherein, Is a category ofIs used for judging the function value of the (a),Is a category ofIs used for the weight vector of (a),Is a kernel feature vector. By calculating the discrimination function value of each class, class discrimination function value vectors are obtained, and the vectors represent the possibility that the sample belongs to different classes. And carrying out softmax normalization on the class discrimination function value vector to obtain an unnormalized probability vector. The softtmax function converts the class discrimination function value into a probability distribution such that the sum of probabilities for all classes is 1, the formula is:

；

wherein, Is a category ofIs used to determine the non-normalized probability of (1),Is a category ofIs a discrimination function value of (a). By softmax normalization, unnormalized probability vectors are obtained, which represent the probabilities that samples belong to different classes. And carrying out probability correction according to the unnormalized probability vector and the category weight vector in the classifier parameter set to obtain a normalized probability vector. The probability correction is to adjust the probability value of each category by combining the category weights, so that the classification result of the model is more accurate. The modified normalized probability vector is expressed as: . Wherein, Is a category ofIs used for the normalization of the probability of (1),Is a probability of being non-normalized,Is a category ofIs used for the weight vector of (a). And carrying out logarithmic operation on the normalized probability vector to obtain a log-likelihood probability vector. The logarithmic operation can convert multiplication operation into addition operation, so that the calculation process is simplified, and meanwhile, the numerical stability can be improved. The log likelihood probability vector is expressed as: . By means of a logarithmic operation, log-likelihood probability vectors are obtained, which will be used to construct the maximum expectation estimation problem. And constructing a maximum expected estimation problem according to the log likelihood probability vector. The maximum expected estimation problem finds the optimal probability distribution parameters by maximizing the log likelihood probability, and ensures that the prediction result of the model has the highest reliability. The objective function of the maximum expectation estimation problem can be expressed as:

；

By optimizing the objective function, an optimal probability distribution parameter can be obtained. And carrying out quasi-Newton iterative optimization on the maximum expected estimation problem to obtain a probability correction vector. The quasi-Newton method is an efficient optimization algorithm, and the optimal solution is continuously approximated through iterative calculation. In the optimization process, calculating the gradient of each step and the approximate value of the hessian matrix, and adjusting probability distribution parameters until the optimal solution is converged. The probability correction vector is expressed as:

；

Iterative optimization by quasi-newton method results in probability correction vectors that will be used to correct normalized probabilities. And adding the probability correction vector and the normalized probability vector to obtain a classification probability vector. The classification probability vector represents the final probability that each sample belongs to a different class, the formula is:

；

And adding the probability correction vector and the normalized probability vector to obtain a classification probability vector, and determining a classification result of the sample.

The above describes the method for automatically switching the chip burn in the embodiment of the present application, and the following describes the device for automatically switching the chip burn in the embodiment of the present application, referring to fig. 2, one embodiment of the device for automatically switching the chip burn in the embodiment of the present application includes:

The acquisition module 201 is configured to acquire a plurality of pieces of chip information to be burned, perform classification and structuring processing to obtain a plurality of pieces of standard chip information, and construct a chip information database according to the plurality of pieces of standard chip information;

the verification module 202 is used for carrying out surface optical character recognition on the target burning chip to obtain a chip model, and carrying out information verification through a radio frequency identification technology to obtain specification parameters of the target burning chip;

The matching module 203 is configured to match a burner and a burning program of the target burning chip from the chip information database according to the chip model and the specification parameter;

The checking module 204 is configured to establish a physical connection between the target writing chip and the writer and perform an electrical connection status check based on the definition of the chip pins, so as to obtain electrical connection status information;

The processing module 205 is configured to perform programming instruction and data transmission through a programming program based on the electrical connection status information, control the writer to perform erasing, programming and verification operations, and monitor current parameter data and voltage parameter data of the target writing chip in real time;

The detection module 206 is configured to input the current parameter data and the voltage parameter data into a preset naive bayes model to perform abnormality detection in the burning process, so as to obtain an abnormality detection result;

And the switching module 207 is configured to perform abnormality investigation on the writer and generate an abnormality processing scheme if the abnormality detection result is abnormal, and if the abnormality detection result is normal, control the writer to continue the writing operation, and automatically switch the next chip to be written to perform the writing operation when the target writing chip completes the writing.

Through the cooperation of the components, a comprehensive chip information database is constructed through classification and structuring, data query and updating are efficiently carried out, and the accuracy and the comprehensiveness of information in the burning process are ensured. By utilizing the optical character recognition and wireless radio frequency recognition technology, the model and specification parameters of the chip are automatically acquired, the error of manual operation is reduced, and the recognition precision and efficiency are improved. Based on the data in the chip information database, the most suitable burning equipment and program are automatically matched according to the chip type and specification parameters, so that the technical index of the burning equipment meets or exceeds the chip requirement, and the manual matching error is avoided. The automatic connection of the chip and the burning equipment is realized through the mechanical arm and the alignment device, and the electric connection state inspection is carried out according to the definition of the pins of the chip, so that the reliability of physical connection and the accuracy of electric connection are ensured, and the risk of manual connection errors is reduced. In the burning process, current and voltage parameters of the chip are monitored in real time through the equipment interface, and abnormality detection in the burning process is carried out by using a preset naive Bayesian model, so that abnormal conditions can be found and processed in time, and the burning quality and equipment safety are ensured. The automatic control burning device executes erasing, programming and checking operations, and consistency and efficiency of burning operations are improved. After the burning is finished, the chip to be burned is automatically switched to the next chip to be burned, so that the waiting time is reduced, and the equipment utilization rate and the production rhythm are improved. When an abnormality is detected in the burning process, abnormality troubleshooting is automatically performed and an abnormality processing scheme is generated, so that normal operation can be quickly recovered even under abnormal conditions, and the influence on a subsequent burning task is avoided.

The application also provides an automatic switching device for chip burning, which comprises a memory and a processor, wherein the memory stores computer readable instructions, and when the computer readable instructions are executed by the processor, the processor executes the steps of the automatic switching method for chip burning in the above embodiments.

The present application also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, or may be a volatile computer readable storage medium, where instructions are stored in the computer readable storage medium, where the instructions when executed on a computer cause the computer to perform the steps of the method for automatically switching between chip burn.

It will be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, systems and units may refer to the corresponding processes in the foregoing method embodiments, which are not repeated herein.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (random acceS memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims (10)

1. An automatic switching method for chip burning is characterized by comprising the following steps:

acquiring a plurality of pieces of chip information to be burned, classifying and structuring to obtain a plurality of pieces of standard chip information, and constructing a chip information database according to the plurality of pieces of standard chip information;

Performing surface optical character recognition on a target burning chip to obtain a chip model, and performing information verification through a radio frequency identification technology to obtain specification parameters of the target burning chip;

according to the chip model and the specification parameters, matching a burner and a burning program of the target burning chip from the chip information database;

Establishing physical connection between the target burning chip and the burner, and performing electric connection state inspection based on chip pin definition to obtain electric connection state information;

Based on the electrical connection state information, programming instructions and data are sent through the programming program, the burner is controlled to execute erasing, programming and checking operations, and meanwhile, current parameter data and voltage parameter data of the target burning chip are monitored in real time;

inputting the current parameter data and the voltage parameter data into a preset naive Bayesian model for abnormality detection in the burning process, and obtaining an abnormality detection result;

if the abnormality detection result is abnormal, performing abnormality investigation on the burner and generating an abnormality processing scheme, if the abnormality detection result is normal, controlling the burner to continue the burning operation, and when the target burning chip finishes burning, automatically switching the next chip to be burnt to execute the burning operation.

2. The method for automatically switching chip burn-in according to claim 1, wherein the steps of obtaining a plurality of pieces of chip information to be burned-in, classifying and structuring the pieces of chip information to obtain a plurality of pieces of standard chip information, and constructing a chip information database according to the pieces of standard chip information, comprise:

Chip information inquiry is carried out on a plurality of chips to be burned to obtain information of the plurality of chips to be burned;

classifying the chips to be burned according to the information of the chips to be burned to obtain information of a plurality of classified chips;

Carrying out structuring treatment on the plurality of classified chip information to obtain a plurality of structured chip information, wherein the structured chip information comprises chip types, chip models, specification parameters, chip numbers and chip IDs;

Generating a unique identification code of each chip to be burned according to the plurality of structured chip information, and associating the unique identification code with the corresponding structured chip information to obtain a plurality of associated chip information;

And carrying out standardization processing on the plurality of associated chip information to obtain a plurality of standard chip information, and constructing a chip information database according to the plurality of standard chip information.

3. The automatic switching method of chip burning according to claim 1, wherein the performing surface optical character recognition on the target burning chip to obtain a chip model, and performing information verification by using a radio frequency identification technology to obtain specification parameters of the target burning chip comprises:

acquiring a surface gray level image of a target burning chip through a high-resolution industrial camera to obtain a surface gray level image, and performing image preprocessing on the surface gray level image to obtain a preprocessed surface image;

according to the preprocessed surface image, positioning and segmenting the surface characters of the target burning chip to obtain segmented character images;

performing character recognition on the segmented character image to obtain a chip model, and acquiring a radio frequency identification tag of the target burning chip;

performing radio frequency identification on the radio frequency identification tag to obtain a radio frequency identification result, and comparing the radio frequency identification result with the chip model to obtain a comparison result;

if the comparison result is inconsistent or not matched, carrying out surface optical character recognition and wireless radio frequency recognition on the target burning chip again until the comparison result is consistent and matched;

And according to the chip model, acquiring specification parameters of the target burning chip from a chip information database, wherein the specification parameters comprise the chip packaging type, the chip pin number, the chip working voltage and the chip storage capacity.

4. The method for automatically switching between chip burn-in according to claim 1, wherein said establishing a physical connection between the target burn-in chip and the burner and performing an electrical connection status check based on a chip pin definition to obtain electrical connection status information comprises:

Acquiring chip pin definitions from a chip information database according to the specification parameters, wherein the chip pin definitions comprise a power supply pin, a ground pin, a clock pin, a data pin and a control pin;

generating a pin connection configuration file of the target burning chip according to the chip pin definition, wherein the pin connection configuration file comprises the number, the name, the type and the connection mode of each pin;

placing the target writing chip on a chip clamp of the writer, and according to the pin connection configuration file, physically connecting an adapter pin of the writer with a pin of the target writing chip through a mechanical arm to establish physical connection between the target writing chip and the writer;

Carrying out an electrifying test on the target burning chip and the burner after the physical connection is established to obtain an electrifying test result, wherein the electrifying test comprises the steps of applying voltage to a power pin and a ground pin and detecting whether current is in a normal range or not;

according to the power-on test result and the pin connection configuration file, checking the electrical connection state of each pin of the target burning chip to obtain the electrical connection state of each pin, wherein the electrical connection state comprises whether the pins are communicated, whether the pin voltage is normal and whether the pin resistance is in a reasonable range;

and summarizing the electrical connection state of each pin to obtain the electrical connection state information.

5. The automatic switching method for chip programming according to claim 1, wherein the inputting the current parameter data and the voltage parameter data into a preset naive bayes model to perform programming process anomaly detection, to obtain an anomaly detection result, includes:

Performing wavelet decomposition on the current parameter data to obtain a multi-scale current component sequence, and extracting time domain statistical characteristics and frequency domain energy characteristics of current according to the multi-scale current component sequence to obtain a current characteristic vector set;

Performing short-time Fourier transform on the voltage parameter data to obtain a time-frequency energy distribution image, and performing image segmentation and mode extraction on the time-frequency energy distribution image to obtain a voltage characteristic image set;

Carrying out data fusion on the current characteristic vector set and the voltage characteristic image set to obtain a burning parameter fusion characteristic set;

according to the burning parameters, fusing the feature sets, and constructing a feature space and a class prior probability of a naive Bayes model;

carrying out Gaussian kernel function mapping on the feature space to obtain Gaussian kernel feature space;

according to the Gaussian kernel feature space and the class prior probability, carrying out parameter estimation on the naive Bayesian model to obtain a classifier parameter set;

Inputting the burning parameter fusion feature set into the naive Bayes model to obtain a classification probability vector;

And determining the class label of the abnormal detection result according to the classification probability vector.

6. The automatic switching method of chip burning according to claim 5, wherein the performing parameter estimation on the naive bayes model according to the gaussian kernel feature space and the class prior probability to obtain a classifier parameter set includes:

Performing kernel matrix calculation on the Gaussian kernel feature space to obtain a kernel matrix, and constructing an objective function and constraint conditions of a semi-positive planning problem according to the kernel matrix;

Solving the semi-positive programming problem by using a sequence minimum optimization algorithm to obtain a support vector set, and calculating a weight vector of a Gaussian kernel function of the naive Bayes model according to the support vector set;

Carrying out Laplacian correction on the class prior probability to obtain smoothed class prior probability, and calculating a class weight vector of the naive Bayesian model according to the smoothed class prior probability;

performing matrix splicing on the weight vector of the Gaussian kernel function and the category weight vector to obtain a weight parameter matrix of a naive Bayesian model;

Performing principal component analysis on the weight parameter matrix to obtain a parameter set after dimension reduction, and constructing a decision function of the naive Bayes model according to the parameter set after dimension reduction;

And outputting the coefficients of the decision function as a classifier parameter set.

7. The automatic switching method of chip programming according to claim 6, wherein the inputting the programming parameter fusion feature set into the naive bayes model to obtain the classification probability vector includes:

feature normalization processing is carried out on the burning parameter fusion feature set to obtain a normalized feature set, and a normalized feature vector of the naive Bayes model is constructed according to the normalized feature set;

Carrying out Gaussian kernel function mapping on the normalized feature vector to obtain a kernel feature vector, and calculating a class discrimination function value vector according to the kernel feature vector and a weight parameter matrix in the classifier parameter set;

Carrying out softmax normalization on the class discrimination function value vector to obtain an unnormalized probability vector, and carrying out probability correction according to the unnormalized probability vector and a class weight vector in the classifier parameter set to obtain a normalized probability vector;

carrying out logarithmic operation on the normalized probability vector to obtain a log-likelihood probability vector, and constructing a maximum expected estimation problem according to the log-likelihood probability vector;

and carrying out quasi-Newton iterative optimization on the maximum expected estimation problem to obtain a probability correction vector, and adding the probability correction vector and the normalized probability vector to obtain a classification probability vector.

8. An automatic switching device for chip burning, for executing the automatic switching method for chip burning according to any one of claims 1 to 7, characterized in that the device comprises:

the acquisition module is used for acquiring a plurality of pieces of chip information to be burned, classifying and structuring the chip information to obtain a plurality of pieces of standard chip information, and constructing a chip information database according to the plurality of pieces of standard chip information;

The verification module is used for carrying out surface optical character recognition on the target burning chip to obtain a chip model, and carrying out information verification through a radio frequency identification technology to obtain specification parameters of the target burning chip;

The matching module is used for matching the burner and the burning program of the target burning chip from the chip information database according to the chip model and the specification parameters;

The checking module is used for establishing physical connection between the target burning chip and the burner and checking the electric connection state based on the definition of chip pins to obtain electric connection state information;

The processing module is used for sending programming instructions and data through the programming program based on the electrical connection state information, controlling the writer to execute erasing, programming and checking operations, and simultaneously monitoring current parameter data and voltage parameter data of the target programming chip in real time;

the detection module is used for inputting the current parameter data and the voltage parameter data into a preset naive Bayesian model to perform abnormality detection in the burning process, so as to obtain an abnormality detection result;

and the switching module is used for carrying out abnormality investigation on the burner and generating an abnormality processing scheme if the abnormality detection result is abnormal, controlling the burner to continue the burning operation if the abnormality detection result is normal, and automatically switching the next chip to be burned to execute the burning operation when the target burning chip finishes the burning.

9. An automatic switching device for chip burning, which is characterized by comprising: a memory and at least one processor, the memory having instructions stored therein;

The at least one processor invokes the instructions in the memory to cause the automatic switching device of chip burning to perform the automatic switching method of chip burning of any of claims 1-7.

10. A computer readable storage medium having instructions stored thereon, wherein the instructions when executed by a processor implement the method for automatically switching chip burn according to any of claims 1-7.