Intelligent diode production quality monitoring method and system based on multi-source data analysis

Abstract

The invention discloses a diode production quality intelligent monitoring method and system based on multi-source data analysis, and relates to the field of production monitoring, and the method comprises the steps: obtaining the real-time standard data of equipment parameters according to the historical data of production equipment, and building a product qualification rate prediction model; generating a predicted product qualification rate of each piece of production equipment; obtaining performance data and appearance data of a to-be-tested product, comparing the performance data and the appearance data with the standard performance threshold value and the standard shape data respectively, and generating unqualified product data and the qualified rate of the to-be-tested product; according to the unqualified product data, generating a defective product environment association trend; according to the predicted product qualified rate and the qualified rate of the to-be-detected product, determining predicted deviation production equipment; generating an equipment difference index in combination with the real-time operation data of the prediction deviation production equipment and the performance data of the to-be-tested product; the defective product environment correlation trend, the equipment difference index and the equipment parameter real-time standard data are combined, and multiple monitoring standards are established.

Description

Technical Field

[0001] The present invention relates to the field of production monitoring, and in particular to a method and system for intelligently monitoring diode production quality based on multi-source data analysis. Background Art

[0002] A diode is an electronic component made of semiconductor material with unidirectional conductivity. It is an indispensable basic component of modern electronic circuits. In the circuit, the rectifier diode can convert alternating current into direct current, providing a stable DC power supply for electronic equipment. The voltage regulator diode can use the reverse breakdown characteristic to stabilize the circuit voltage, ensuring that the electronic components work in a stable voltage environment. The light-emitting diode emits light when it is forward-conducted, and is widely used in indicator lights, display screens and lighting fields.

[0003] The production quality monitoring of diodes is usually carried out after the product is produced. It is not handled until unqualified products are found. This monitoring method is difficult to predict the appearance of unqualified products in advance, does not fully consider the individual differences between the same type of equipment, ignores the connection between the defect data of unqualified products and production, and is difficult to trace the root cause of the unqualified products. As a result, the ability to prevent similar problems in subsequent production is weak. Summary of the invention

[0004] In order to solve the above technical problems, a method for intelligent monitoring of diode production quality based on multi-source data analysis is provided. This technical solution solves the problem raised in the above background technology that it is difficult to predict the occurrence of unqualified products in advance, does not fully consider the individual differences between equipment of the same type, ignores the connection between unqualified product defect data and production, and is difficult to trace the root cause of the unqualified products, which in turn leads to a weak ability to prevent similar problems in subsequent production.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0006] A diode production quality intelligent monitoring method based on multi-source data analysis, comprising:

[0007] Obtain and analyze diode production plans to generate standard performance thresholds and standard shape data;

[0008] Collect historical data of production equipment, obtain real-time standard data of equipment parameters, and establish a prediction model for the product qualification rate of the equipment;

[0009] Obtain the real-time environmental data and process parameter data of each production equipment, combine it with the product qualification rate prediction model of the equipment, and generate the predicted product qualification rate of each production equipment;

[0010] Collect the performance data and appearance data of the product to be tested, and compare them with the standard performance threshold and standard shape data respectively, to generate the data of unqualified products and the qualified rate of the products to be tested;

[0011] Generate defective product environmental correlation trends based on real-time environmental data of production equipment and unqualified product data;

[0012] Determine the production equipment for forecast deviation based on the forecast product qualification rate and the qualified rate of the product to be tested;

[0013] Obtain the real-time operation data of the predicted deviation production equipment, combine the performance data and appearance data of the product to be tested, and generate the equipment difference index;

[0014] Combined with the real-time standard data of equipment parameters, the environmental correlation trend of defective products and the equipment difference index, multiple monitoring standards are established to monitor the real-time data of the equipment.

[0015] Preferably, the establishment of a product qualification rate prediction model for the equipment specifically includes:

[0016] Obtaining the production process of the diode and the production operation manual of each production equipment, extracting the characteristic steps of each production process and the reference parameters of each production equipment, wherein the reference parameters of the production equipment include equipment parameters and environmental parameters;

[0017] Classify the historical data of each production equipment according to process parameters and equipment environment data, and generate process parameter classification data and equipment environment data classification data;

[0018] Based on time, the process parameter classification data and equipment environment data classification data are sorted out, and the corresponding relationship between different process parameters and product qualification rate, as well as the corresponding relationship between different equipment environment data and product qualification rate are listed;

[0019] According to the corresponding relationship between different process parameters and product qualification rate, and the corresponding relationship between different equipment environment data and product qualification rate, obtain the product qualification rate change values ​​corresponding to the process parameter difference and the equipment environment data difference respectively and calculate the average value, obtain the change degree of the product qualification rate corresponding to the change of process parameters, and the change degree of the product qualification rate corresponding to the change of equipment environment data, and record them as the sensitivity of product qualification rate to process parameters and the sensitivity of product qualification rate to production equipment environment respectively;

[0020] Record the equipment corresponding to the characteristic step of the production process as the characteristic equipment, classify the historical data of the characteristic equipment by week as the time interval, and generate time period historical data, wherein the time period historical data includes the time series data of the process parameters of the characteristic equipment and the product quality data;

[0021] Analyze the historical data of the time period, take one hour as an analysis node, correspond the product quality data according to the analysis node, draw the product quality change curve with the analysis node as the horizontal axis and the product quality data as the vertical axis, and generate the time fluctuation trend of the characteristic equipment;

[0022] Generate real-time standard data of equipment parameters according to the time fluctuation trend of the characteristic equipment and the reference parameters of the production equipment, wherein the real-time standard data of equipment parameters includes real-time standard data of equipment parameters and real-time standard data of equipment environment;

[0023] Taking the real-time standard data of equipment parameters as the benchmark, respectively compare the real-time environmental data and process parameter data of each production equipment with the real-time standard data of equipment parameters, obtain the proportion of production equipment whose real-time environmental data and process parameter data are both greater than the real-time standard data of equipment parameters in the total number of production equipment, and generate the first product prediction qualification rate of the production equipment;

[0024] Based on the product qualification rate of production equipment, analyze the historical production data of production equipment to obtain the process interaction index and process environment impact index;

[0025] The preliminary product qualification rate relationship of the equipment is generated by comprehensively considering the sensitivity of the product qualification rate to the process parameters, the sensitivity of the product qualification rate to the production equipment environment, the process mutual influence index and the process environment influence index;

[0026] The electrostatic field strength and magnetic induction strength of the magnetic field in which each production equipment is located are used as the production environment in which the production equipment is located, and the real-time alignment accuracy, processing time and packaging pressure of each production equipment are used as process parameter values, which are substituted into the product qualification rate relationship of the production equipment to obtain the predicted qualification rate of the second product;

[0027] Taking the predicted qualified rate of the first product and the predicted qualified rate of the second product as input variables, a product qualified rate prediction model of the equipment is established;

[0028] Among them, the relationship between the qualified rate of production equipment products is as follows:

[0029]

[0030] In the formula, y represents the predicted qualified rate of the second product, b i represents the sensitivity of the product qualification rate of the i-th process to the process parameters, x i represents the value of the i-th process parameter, a ij represents the process interaction index between the i-th process and the j-th process, x j represents the jth process parameter value, b j represents the sensitivity of the product qualification rate of the jth process to the process parameters, b ilrepresents the process environment impact index between the ith process and the production environment l, x l It represents the specific value of the production environment l in which the production equipment is located, n represents the total number of processes required for the production equipment to generate products, and m represents the total number of environments that affect the production of products by the production equipment.

[0031] Preferably, generating defective product environment association trends specifically includes:

[0032] According to the diode production goals, clarify the use scenarios of the diodes and obtain the test environment thresholds of the diodes;

[0033] Sorting out the data of unqualified products to obtain production data and defect data of unqualified products, wherein the defect data includes appearance characteristic values, forward voltage values, reverse current values, and breakdown voltage values;

[0034] Classify the unqualified products according to the defect data into single-defect unqualified products and multiple-defect unqualified products;

[0035] According to the production data of the unqualified products, the defect data of the single-defect unqualified products are analyzed to obtain the relevant production data corresponding to the defect data;

[0036] Combine the relevant production data corresponding to the defect data and the data of multi-defective unqualified products to determine the cause of the multi-defective unqualified products;

[0037] According to the test environment threshold of the diode, the real-time test environment is adjusted to obtain the specific values ​​of the unqualified products under different test data, and the specific value changes of the unqualified products are compared with the change range of the test data to obtain the degree of change of the defect data of the unqualified products under different test environment data;

[0038] Combined with the relevant production data corresponding to the defect data and the degree of change in the defect data of unqualified products under different test environment data, the data relationship between the product defect type and the environment is determined, and the defect product environment association trend is generated.

[0039] Preferably, the determining of the causes of the occurrence of multi-defective unqualified products specifically includes:

[0040] According to the data of unqualified products, the data of unqualified products with multiple defects are classified to generate defect type combination data, defect severity data and defect frequency data respectively;

[0041] Using defect type combination data as the main judgment criterion, determine the problematic process corresponding to the multi-defect unqualified products;

[0042] Based on the defect severity data, determine whether the defects of the multi-defect unqualified products are serious. If so, conduct comprehensive quality traceability and process improvement for the multi-defect unqualified products. If not, adjust the production process parameters of the problem process corresponding to the multi-defect unqualified products.

[0043] Based on the defect frequency data, determine whether the defects of multi-defect unqualified products show periodic changes. If so, it means that the operating procedures of the production line operators are not standardized or there are quality problems with the production raw materials. It is necessary to further analyze the problem in combination with the periodicity of the defects. If not, it means that the defect is a normal loss in the production process.

[0044] Preferably, the generating device difference index specifically includes:

[0045] Obtain the product qualification rate of the forecast deviation production equipment;

[0046] According to the working time and type of the equipment, the historical data of the production equipment is classified to generate a group of equipment with the same working condition, wherein the equipment type and working time in the same group are the same;

[0047] Obtain and compare the standard deviation of product qualification rates of equipment groups with the same working conditions. If the standard deviation is greater than 0.02, it is determined that there is a difference between the equipment and the rest of the equipment in the same group. The equipment with the difference is determined, and the corresponding operating data and product data of the equipment with the difference are obtained.

[0048] According to the generated product data, the performance characteristic data and appearance characteristic data of the products generated by the difference equipment are extracted, and compared with the standard performance threshold and the standard shape data respectively to obtain the product performance difference and the product shape data difference;

[0049] Obtain the equipment standard process parameters, combine them with the operation data corresponding to the difference equipment, subtract the operation data corresponding to the difference equipment from the equipment operation data in the equipment standard process parameters, and obtain the difference of the equipment operation process parameters;

[0050] The ratios of product performance difference and equipment operation process parameter difference, as well as product shape data difference and equipment operation process parameter difference are calculated respectively to generate the equipment difference index.

[0051] Preferably, the establishment of multiple monitoring standards to monitor the real-time data of production equipment specifically includes:

[0052] The real-time standard data of equipment parameters, including the real-time standard data of equipment parameters and the real-time standard data of equipment environment, are used as the first standard data, and compared with the real-time environment data and process parameter data of the production equipment to determine whether they are lower than the first standard data. If so, a warning is issued in advance to notify the production line operator to adjust the equipment process parameters. If not, the product qualification rate of the production equipment is continuously predicted;

[0053] Acquire real-time test environment information through sensors, combine with defective product environment correlation trends corresponding to the current test environment, generate performance data corresponding to the test environment data as second standard data, obtain performance data of the product to be tested in the test environment data through measuring instruments, compare with the second standard data, and determine whether it maintains a consistent change trend with the second standard data. If so, it means that the product is a substandard product and a warning is issued in advance. If not, continue to test the remaining products to be tested;

[0054] Multiply the equipment difference index and the equipment standard process parameters to generate equipment process adaptation data as the third standard data, and compare it with the real-time process parameters of the production equipment to determine whether there are differences with the third standard data. If so, it means that the products produced by the equipment are at risk of being unqualified, and a warning is issued in advance. If not, continue to monitor the real-time process parameters of the equipment.

[0055] Furthermore, a diode production quality intelligent monitoring system based on multi-source data analysis is proposed, including:

[0056] A data acquisition module, which is used to collect real-time environmental data and process parameters of production equipment, historical data of each production equipment, performance data and appearance data of the product to be tested, and transmit the data to the feature extraction module, the data analysis module and the model generation module;

[0057] A feature extraction module, which is used to extract key features from the received data, extract characteristic steps of the production process, extract defect data of single-defect unqualified products, and transmit the data to the data analysis module;

[0058] A data analysis module, which is used to organize the received data, generate a preliminary product qualification rate relationship of the equipment, analyze the relationship between the equipment operation data, product data and environmental data, and transmit the data to the model generation module and the quality monitoring module;

[0059] A model generation module, wherein the model generation module is used to establish a prediction model according to the received data, generate prediction data through the prediction model and real-time data, and transmit the prediction data to the quality monitoring module;

[0060] A quality monitoring module, which is used to generate monitoring standard data based on the received data, establish multiple monitoring standards, and transmit the data to the abnormal warning module;

[0061] An abnormal warning module is used to compare the received data, determine whether there is an abnormal situation and whether a warning needs to be issued based on the comparison result, and transmit the data to the intelligent decision-making module;

[0062] The intelligent decision-making module is used to analyze the received data, provide suggestions for the production process and formulate a maintenance plan for the production process equipment based on the data analysis results and the data transmitted by the abnormal warning module.

[0063] Preferably, the feature extraction module includes:

[0064] A first feature extraction unit, the first feature extraction unit is used to compare the process production flow of each production equipment and extract the feature steps in the production flow;

[0065] A second feature extraction unit, the second feature extraction unit is used to extract features according to the use environment of the diode and obtain a test environment threshold of the diode;

[0066] The third feature extraction unit is used to analyze the defect data of the unqualified products, extract the defect features of the unqualified products, and classify the unqualified products.

[0067] Preferably, the data analysis module includes:

[0068] A first analysis unit, the first analysis unit is used to analyze the product qualification rate of the production equipment based on the process parameters corresponding to the characteristic steps and the environmental data of the production equipment, and obtain the change relationship index between each process parameter and the product qualification rate, and the change relationship index between each production environment data of the production equipment and the product qualification rate;

[0069] A second analysis unit, the second analysis unit is used to adjust the real-time test environment according to the test environment threshold of the diode, obtain the degree of change of defect data of unqualified products under different test environment data, and generate a defect product environment correlation trend in combination with relevant production data corresponding to the defect data;

[0070] The third analysis unit is used to generate an equipment difference index according to product performance differences, product shape data differences and equipment operation process parameter differences.

[0071] Preferably, the quality monitoring module includes:

[0072] A first quality monitoring module, which is used to predict the product qualification rate of each production equipment according to the product qualification rate prediction model of the equipment, and compare it with the actual product qualification rate of the production equipment to determine whether the quality of the product is abnormal;

[0073] A second quality monitoring module, which is used to generate performance data corresponding to the real-time test environment data according to the real-time test environment information and the defective product environment correlation trend, and compare it with the performance data of the product to be tested in the real-time test environment data to determine whether the product quality is abnormal;

[0074] The third quality monitoring module is used to generate equipment process adaptation data according to equipment standard process parameters and equipment difference index, and compare it with the real-time process parameters of the production equipment to determine whether there is any abnormality in the quality of the product.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] The present invention proposes an intelligent monitoring method for diode production quality based on multi-source data analysis. According to the historical data of each production equipment, a product qualification rate prediction model of the equipment is established. In combination with the real-time environmental data and process parameters of each production equipment, the predicted product qualification rate of each production equipment is generated. The performance data and appearance data of the product to be tested are obtained, and compared with the standard performance threshold and the standard shape data respectively, to generate unqualified product data and the qualification rate of the product to be tested. According to the real-time environmental data of the production equipment and the unqualified product data, a defective product environment correlation trend is generated. According to the predicted product qualification rate and the qualification rate of the product to be tested, the predicted deviation production equipment is determined. In combination with the real-time operation data of the predicted deviation production equipment, the performance data and appearance data of the product to be tested, an equipment difference index is generated. The defective product environment correlation trend, the equipment difference index and the product qualification rate prediction model of the equipment are combined to establish multiple monitoring standards. In this way, the appearance of unqualified products can be effectively predicted, the individual differences between the same type of equipment are fully considered, the problem links in the production are accurately located, the root cause of the unqualified products is traced back, and the prevention capability of similar problems in subsequent production is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 This is a flow chart of the diode production quality intelligent monitoring method based on multi-source data analysis proposed by the present invention;

[0078] Figure 2 A flowchart of the steps of establishing a product qualification rate prediction model for equipment in the present invention;

[0079] Figure 3A flowchart of the steps for generating defective product environment correlation trends in the present invention;

[0080] Figure 4 A flowchart of the steps for determining the cause of the occurrence of multi-defective unqualified products in the present invention;

[0081] Figure 5 A flow chart of the steps of generating a device difference index in the present invention;

[0082] Figure 6 A flowchart of the steps for establishing multiple monitoring standards in the present invention;

[0083] Figure 7 This is a structural diagram of the diode production quality intelligent monitoring system based on multi-source data analysis proposed by the present invention. DETAILED DESCRIPTION

[0084] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art may think of other obvious variations.

[0085] Reference Figure 1 As shown, a diode production quality intelligent monitoring method based on multi-source data analysis includes:

[0086] Acquire and analyze diode production targets to generate standard performance thresholds and standard shape data;

[0087] Collect historical data of production equipment, obtain real-time standard data of equipment parameters, and establish a prediction model for the product qualification rate of the equipment;

[0088] Obtain the real-time environmental data and process parameters of each production equipment, and generate the predicted product qualification rate of each production equipment by combining the product qualification rate prediction model of the equipment;

[0089] Collect the performance data and appearance data of the product to be tested, and compare them with the standard performance threshold and standard shape data respectively, to generate the data of unqualified products and the qualified rate of the products to be tested;

[0090] Generate defective product environmental correlation trends based on real-time environmental data of production equipment and unqualified product data;

[0091] Determine the production equipment for forecast deviation based on the forecast product qualification rate and the qualified rate of the product to be tested;

[0092] Obtain the real-time operation data of the predicted deviation production equipment, combine the performance data and appearance data of the product to be tested, and generate the equipment difference index;

[0093] Combined with the real-time standard data of equipment parameters, the environmental correlation trend of defective products and the equipment difference index, multiple monitoring standards are established to monitor the real-time data of the equipment.

[0094] This solution establishes a product qualification rate prediction model for each production equipment based on the historical data of each production equipment, generates a predicted product qualification rate for each production equipment in combination with the real-time environmental data and process parameters of each production equipment, obtains the performance data and appearance data of the product to be tested, and compares them with the standard performance threshold and standard shape data respectively, generates unqualified product data and the qualification rate of the product to be tested, generates defective product environment association trends based on the real-time environmental data of the production equipment and the unqualified product data, determines the predicted deviation production equipment based on the predicted product qualification rate and the qualification rate of the product to be tested, generates the equipment difference index in combination with the real-time operation data of the predicted deviation production equipment and the performance data and appearance data of the product to be tested, combines the defective product environment association trends, the equipment difference index and the real-time standard data of equipment parameters, and establishes multiple monitoring standards.

[0095] Reference Figure 2 As shown, a prediction model for the product qualification rate of the equipment is established, which specifically includes:

[0096] Obtaining the production process of the diode and the production operation manual of each production equipment, extracting the characteristic steps of each production process and the reference parameters of each production equipment, wherein the reference parameters of the production equipment include equipment parameters and environmental parameters;

[0097] Classify the historical data of each production equipment according to process parameters and equipment environment data, and generate process parameter classification data and equipment environment data classification data;

[0098] Based on time, the process parameter classification data and equipment environment data classification data are sorted out, and the corresponding relationship between different process parameters and product qualification rate, as well as the corresponding relationship between different equipment environment data and product qualification rate are listed;

[0099] According to the corresponding relationship between different process parameters and product qualification rate, and the corresponding relationship between different equipment environment data and product qualification rate, obtain the product qualification rate change values ​​corresponding to the process parameter difference and the equipment environment data difference respectively and calculate the average value, obtain the change degree of the product qualification rate corresponding to the change of process parameters, and the change degree of the product qualification rate corresponding to the change of equipment environment data, and record them as the sensitivity of product qualification rate to process parameters and the sensitivity of product qualification rate to production equipment environment respectively;

[0100] Record the equipment corresponding to the characteristic step of the production process as the characteristic equipment, classify the historical data of the characteristic equipment by week as the time interval, and generate time period historical data, wherein the time period historical data includes the time series data of the process parameters of the characteristic equipment and the product quality data;

[0101] Analyze the historical data of the time period, take one hour as an analysis node, correspond the product quality data according to the analysis node, draw the product quality change curve with the analysis node as the horizontal axis and the product quality data as the vertical axis, and generate the time fluctuation trend of the characteristic equipment;

[0102] According to the time fluctuation trend of the characteristic equipment and the reference parameters of the production equipment, the real-time standard data of the equipment parameters are generated, and the real-time standard data of the equipment parameters include the real-time standard data of the equipment parameters and the real-time standard data of the equipment environment. The process mutual influence index represents the relationship between the mutual influence between the processes and the product qualification rate, and the process environment influence index represents the relationship between the mutual influence between the process and the environment and the product qualification rate. Through the historical production data, the specific value of the change in the product qualification rate of other equipment caused by the change in the process parameters of the same equipment under the same other conditions is obtained, and the specific value of the change in the product qualification rate of other equipment is divided by the change value of the process parameter to generate the process mutual influence index. Under the same other conditions, the specific value of the change in the environmental data of the equipment is obtained, which causes the specific value of the change in the product qualification rate of the equipment. The specific value of the change in the product qualification rate of the equipment is divided by the data change value of the environment where the equipment is located to generate the process environment influence index;

[0103] Taking the real-time standard data of equipment parameters as the benchmark, respectively compare the real-time environmental data and process parameter data of each production equipment with the real-time standard data of equipment parameters, obtain the proportion of production equipment whose real-time environmental data and process parameter data are both greater than the real-time standard data of equipment parameters in the total number of production equipment, and generate the first product prediction qualification rate of the production equipment;

[0104] Based on the product qualification rate of production equipment, analyze the historical production data of production equipment to obtain the process interaction index and process environment impact index;

[0105] The preliminary product qualification rate relationship of the equipment is generated by comprehensively considering the sensitivity of the product qualification rate to the process parameters, the sensitivity of the product qualification rate to the production equipment environment, the process mutual influence index and the process environment influence index;

[0106] The electrostatic field strength and magnetic induction strength of the magnetic field in which each production equipment is located are used as the production environment in which the production equipment is located, and the real-time alignment accuracy, processing time and packaging pressure of each production equipment are used as process parameter values, which are substituted into the product qualification rate relationship of the production equipment to obtain the predicted qualification rate of the second product;

[0107] Taking the predicted qualified rate of the first product and the predicted qualified rate of the second product as input variables, a product qualified rate prediction model of the equipment is established;

[0108] Among them, the relationship between the qualified rate of production equipment products is as follows:

[0109]

[0110] In the formula, y represents the predicted qualified rate of the second product, b i represents the sensitivity of the product qualification rate of the i-th process to the process parameters, x i represents the value of the i-th process parameter, a ij represents the process interaction index between the i-th process and the j-th process, x j represents the jth process parameter value, b j represents the sensitivity of the product qualification rate of the jth process to the process parameters, b il represents the process environment impact index between the ith process and the production environment l, x l It represents the specific value of the production environment l in which the production equipment is located, n represents the total number of processes required for the production equipment to generate products, and m represents the total number of environments that affect the production of products by the production equipment.

[0111] It is understandable that the production process of diodes includes cutting process, oxidation process, photolithography process, diffusion process, electrode preparation and packaging welding. In the various process production processes of diodes, although the production equipment maintains a constant temperature and humidity to reduce the impact of the environment on the production process, some environmental impacts on production are inevitable, such as static electricity and electromagnetic interference. The friction of raw materials and the movement of people will generate static electricity, causing the semiconductor structure inside the diode chip to be broken down. The operation of large equipment will generate electromagnetic radiation, affecting the precision equipment required for diode production. The transient pulse signal generated by external electromagnetic interference will interfere with the normal operation of the timing chip, causing the exposure time to be longer or shorter than the set value. If the exposure time is too long If the photoresist is overexposed, the pattern lines will become thicker and the resolution will be reduced. Conversely, if the exposure time is insufficient, the pattern cannot be fully formed, and the lines will be missing or blurred, which will affect the three-dimensional structural accuracy of the diode chip. Because the production of diodes is a delicate process, there is a mutual influence between processes. Taking the diffusion process and the photolithography process as an example, the PN junction depth and impurity distribution formed by the diffusion process directly affect the accuracy requirements of the photolithography process. In diode manufacturing, if the PN junction depth after diffusion does not meet the design standards, the alignment accuracy of the photolithography pattern will require higher requirements. The photolithography pattern needs to fall precisely on the area where the PN junction is located, otherwise it will cause problems such as short circuits or open circuits in the circuit.

[0112] Reference Figure 3 As shown, the defective product environment correlation trend is generated, including:

[0113] According to the diode production goals, clarify the use scenarios of the diodes and obtain the test environment thresholds of the diodes;

[0114] Sorting out the data of unqualified products to obtain production data and defect data of unqualified products, wherein the defect data includes appearance characteristic values, forward voltage values, reverse current values, and breakdown voltage values;

[0115] Classify the unqualified products according to the defect data into single-defect unqualified products and multiple-defect unqualified products;

[0116] According to the production data of the unqualified products, the defect data of the single-defect unqualified products are analyzed to obtain the relevant production data corresponding to the defect data;

[0117] Combine the relevant production data corresponding to the defect data and the data of multi-defective unqualified products to determine the cause of the multi-defective unqualified products;

[0118] According to the test environment threshold of the diode, the real-time test environment is adjusted to obtain the specific values ​​of the unqualified products under different test data, and the specific value changes of the unqualified products are compared with the change range of the test data to obtain the degree of change of the defect data of the unqualified products under different test environment data;

[0119] Combined with the relevant production data corresponding to the defect data and the degree of change in the defect data of unqualified products under different test environment data, the data relationship between the product defect type and the environment is determined, and the defect product environment association trend is generated.

[0120] It is understandable that for a diode, a qualified product needs to ensure that all indicators pass the test indicators perfectly. Therefore, there will be products that are eliminated due to single indicator failure and products that are eliminated due to multiple indicators failure. Unqualified products with defects in both forward voltage and reverse current values ​​are due to problems in the manufacturing process of the PN junction, such as uneven doping or abnormal diffusion process, which results in a higher voltage being required to conduct when forward biased, and cannot effectively prevent current from passing when reverse biased. For unqualified products whose forward voltage and breakdown voltage values ​​do not meet the standards at the same time, there are problems with the material or structure of the diode, for example, the quality of the semiconductor material is poor or the depth of the PN junction does not meet the requirements. When the forward voltage is abnormal, it means that the diffusion process of the majority carriers is affected, and the abnormal breakdown voltage is because the PN junction cannot withstand the normal reverse electric field strength and is easily broken down. For products with defects in both reverse current and breakdown voltage values, it is due to impurities inside the diode. The content is too high or there are defects in the packaging process. Excessive impurities will form a leakage channel when reverse biased, increase the reverse current, and more easily cause breakdown when the reverse voltage increases. If the forward voltage, reverse current and breakdown voltage of a defective product do not meet the standards, this usually means that the product has systemic problems in the entire electrical performance, which is caused by errors in multiple links in the production process, such as poor quality of raw materials, serious deviation of process parameters from standards or equipment failure. The appearance characteristic value defects include package cracks, pin bending or oxidation. Package cracks will cause the chip to be affected by external factors such as moisture and impurities, thereby changing its electrical properties and changing the forward voltage. Pin oxidation will increase contact resistance. When reverse biased, this additional resistance will cause deviations in the measured value of the reverse current. Poor quality of packaging materials or problems in the packaging process such as bubbles and stratification will affect the electric field distribution of the diode, thereby reducing the breakdown voltage.

[0121] Reference Figure 4 As shown in the figure, the reasons for the occurrence of multi-defective unqualified products are as follows:

[0122] According to the data of unqualified products, the data of unqualified products with multiple defects are classified to generate defect type combination data, defect severity data and defect frequency data respectively;

[0123] Using defect type combination data as the main judgment criterion, determine the problematic process corresponding to the multi-defect unqualified products;

[0124] Based on the defect severity data, determine whether the defects of the multi-defect unqualified products are serious. If so, conduct comprehensive quality traceability and process improvement for the multi-defect unqualified products. If not, adjust the production process parameters of the problem process corresponding to the multi-defect unqualified products.

[0125] Based on the defect frequency data, determine whether the defects of multi-defect unqualified products show periodic changes. If so, it means that the operating procedures of the production line operators are not standardized or there are quality problems with the production raw materials. It is necessary to further analyze the problem in combination with the periodicity of the defects. If not, it means that the defect is a normal loss in the production process.

[0126] It is understandable that the defect type combination data counts the situations where different defect types appear on the same diode at the same time. For example, some diodes are found to have both abnormal forward voltage values ​​and package cracks. This indicates that there is some correlation between the two defects. Perhaps the heat generated by the abnormal forward voltage value causes excessive thermal stress in the package material and cracks, or the package cracks cause the internal structure of the diode to be affected by external factors, resulting in abnormal forward voltage values. Defect severity data is usually classified according to the degree of impact of the defect on the function, performance and reliability of the diode. For example, minor defects include tiny scratches on the appearance, unclear markings, etc., which have a great impact on the electrical performance and basic usage of the diode. There is no impact at all. General defects such as slight oxidation of the pins will affect the welding quality and the stability of the electrical connection to a certain extent, but it can still be used normally through simple treatment. Serious defects such as internal short circuit or open circuit of the chip will directly cause the diode to not work properly and affect the performance of the entire circuit. The defect frequency data can intuitively show which defects are the most common and which are relatively rare. For example, if the defect frequency of excessive reverse leakage current is found to be high, it means that there are factors affecting the reverse withstand voltage performance of the diode in the production process, such as oxide layer quality problems, improper chip surface treatment, etc. It is necessary to improve the monitoring of these factors to improve the quality stability of the generated products.

[0127] Reference Figure 5 As shown, the device difference index is generated, including:

[0128] Obtain the product qualification rate of the forecast deviation production equipment;

[0129] According to the working time and type of the equipment, the historical data of the production equipment is classified to generate a group of equipment with the same working condition, wherein the equipment type and working time in the same group are the same;

[0130] Obtain and compare the standard deviation of product qualification rates of equipment groups with the same working conditions. If the standard deviation is greater than 0.02, it is determined that there is a difference between the equipment and the rest of the equipment in the same group. The equipment with the difference is determined, and the corresponding operating data and product data of the equipment with the difference are obtained.

[0131] According to the generated product data, the performance characteristic data and appearance characteristic data of the products generated by the difference equipment are extracted, and compared with the standard performance threshold and the standard shape data respectively to obtain the product performance difference and the product shape data difference;

[0132] Obtain the equipment standard process parameters, combine them with the operation data corresponding to the difference equipment, subtract the operation data corresponding to the difference equipment from the equipment operation data in the equipment standard process parameters, and obtain the difference of the equipment operation process parameters;

[0133] The ratios of product performance difference and equipment operation process parameter difference, as well as product shape data difference and equipment operation process parameter difference are calculated respectively to generate the equipment difference index.

[0134] It is understandable that the same model of equipment may have differences in setting equipment parameters due to different operators. Even if the operation is carried out according to the same process standards, the actual operating performance of the equipment may be inconsistent with the standard parameters due to personal operating habits, deviations in parameter understanding, etc. For example, in the chip diffusion process, a slight deviation in the setting of diffusion temperature and time will cause changes in the impurity concentration and distribution inside the chip, thereby affecting the electrical properties of the diode such as the forward voltage drop and reverse breakdown voltage. This process will be further aggravated by differences in equipment maintenance and parts replacement, resulting in performance differences between equipment of the same working time and the same equipment type.

[0135] Reference Figure 6 As shown, multiple monitoring standards are established to monitor the real-time data of production equipment, including:

[0136] The real-time standard data of equipment parameters, including the real-time standard data of equipment parameters and the real-time standard data of equipment environment, are used as the first standard data, and compared with the real-time environment data and process parameter data of the production equipment to determine whether they are lower than the first standard data. If so, a warning is issued in advance to notify the production line operator to adjust the equipment process parameters. If not, the product qualification rate of the production equipment is continuously predicted;

[0137] Acquire real-time test environment information through sensors, combine with defective product environment correlation trends corresponding to the current test environment, generate performance data corresponding to the test environment data as second standard data, obtain performance data of the product to be tested in the test environment data through measuring instruments, compare with the second standard data, and determine whether it maintains a consistent change trend with the second standard data. If so, it means that the product is a substandard product and a warning is issued in advance. If not, continue to test the remaining products to be tested;

[0138] Multiply the equipment difference index and the equipment standard process parameters to generate equipment process adaptation data as the third standard data, and compare it with the real-time process parameters of the production equipment to determine whether there are differences with the third standard data. If so, it means that the products produced by the equipment are at risk of being unqualified, and a warning is issued in advance. If not, continue to monitor the real-time process parameters of the equipment.

[0139] It is understandable that the production quality monitoring of diodes involves multiple aspects, from processing technology to testing and inspection. By taking the real-time standard data of equipment parameters as the first monitoring standard, the production quality of diodes can be monitored from the process production aspect. By taking the performance data corresponding to the real-time test environment data as the second standard data, the production quality of diodes can be monitored from the testing and inspection link. By taking the equipment process adaptation data as the third standard data, the production quality of diodes can be further monitored in terms of process production, thereby reducing the impact of differences between equipment on the quality monitoring of diode process production.

[0140] Further, see Figure 7 As shown, according to the same inventive concept as the diode production quality intelligent monitoring method based on multi-source data analysis, this solution proposes a diode production quality intelligent monitoring system based on multi-source data analysis, including:

[0141] A data acquisition module, which is used to collect real-time environmental data and process parameters of production equipment, historical data of each production equipment, performance data and appearance data of the product to be tested, and transmit the data to the feature extraction module, the data analysis module and the model generation module;

[0142] A feature extraction module, which is used to extract key features from the received data, extract characteristic steps of the production process, extract defect data of single-defect unqualified products, and transmit the data to the data analysis module;

[0143] A data analysis module, which is used to organize the received data, generate a preliminary product qualification rate relationship of the equipment, analyze the relationship between the equipment operation data, product data and environmental data, and transmit the data to the model generation module and the quality monitoring module;

[0144] A model generation module, wherein the model generation module is used to establish a prediction model according to the received data, generate prediction data through the prediction model and real-time data, and transmit the prediction data to the quality monitoring module;

[0145] A quality monitoring module, which is used to generate monitoring standard data based on the received data, establish multiple monitoring standards, and transmit the data to the abnormal warning module;

[0146] An abnormal warning module is used to compare the received data, determine whether there is an abnormal situation and whether a warning needs to be issued based on the comparison result, and transmit the data to the intelligent decision-making module;

[0147] The intelligent decision-making module is used to analyze the received data, provide suggestions for the production process and formulate a maintenance plan for the production process equipment based on the data analysis results and the data transmitted by the abnormal warning module.

[0148] Furthermore, the feature extraction module includes:

[0149] A first feature extraction unit, the first feature extraction unit is used to compare the process production flow of each production equipment and extract the feature steps in the production flow;

[0150] A second feature extraction unit, the second feature extraction unit is used to extract features according to the use environment of the diode and obtain a test environment threshold of the diode;

[0151] The third feature extraction unit is used to analyze the defect data of the unqualified products, extract the defect features of the unqualified products, and classify the unqualified products.

[0152] Furthermore, the data analysis module includes:

[0153] A first analysis unit, the first analysis unit is used to analyze the product qualification rate of the production equipment based on the process parameters corresponding to the characteristic steps and the environmental data of the production equipment, and obtain the change relationship index between each process parameter and the product qualification rate, and the change relationship index between each production environment data of the production equipment and the product qualification rate;

[0154] A second analysis unit, the second analysis unit is used to adjust the real-time test environment according to the test environment threshold of the diode, obtain the degree of change of defect data of unqualified products under different test environment data, and generate a defect product environment correlation trend in combination with relevant production data corresponding to the defect data;

[0155] The third analysis unit is used to generate an equipment difference index according to product performance differences, product shape data differences and equipment operation process parameter differences.

[0156] Furthermore, the quality monitoring module includes:

[0157] A first quality monitoring module, which is used to predict the product qualification rate of each production equipment according to the product qualification rate prediction model of the equipment, and compare it with the actual product qualification rate of the production equipment to determine whether the quality of the product is abnormal;

[0158] A second quality monitoring module, which is used to generate performance data corresponding to the real-time test environment data according to the real-time test environment information and the defective product environment correlation trend, and compare it with the performance data of the product to be tested in the real-time test environment data to determine whether the product quality is abnormal;

[0159] The third quality monitoring module is used to generate equipment process adaptation data according to equipment standard process parameters and equipment difference index, and compare it with the real-time process parameters of the production equipment to determine whether there is any abnormality in the quality of the product.

[0160] In summary, the advantages of the present invention are: it can effectively predict the occurrence of unqualified products, fully consider the individual differences between equipment of the same type, can accurately locate the problem links in production, facilitate tracing the root cause of the unqualified products, and improve the ability to prevent similar problems in subsequent production.

[0161] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions only describe the principles of the present invention. The present invention may be subject to various changes and improvements without departing from the spirit and scope of the present invention. These changes and improvements fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for intelligent monitoring of diode production quality based on multi-source data analysis, characterized in that: include: Obtain and analyze diode production plans to generate standard performance thresholds and standard shape data; Collect historical data of production equipment, obtain real-time standard data of equipment parameters, and establish a prediction model for the product qualification rate of the equipment; Obtain the real-time environmental data and process parameter data of each production equipment, combine it with the product qualification rate prediction model of the equipment, and generate the predicted product qualification rate of each production equipment; Collect the performance data and appearance data of the product to be tested, and compare them with the standard performance threshold and standard shape data respectively, to generate the data of unqualified products and the qualified rate of the products to be tested; Generate defective product environmental correlation trends based on real-time environmental data of production equipment and unqualified product data; Determine the production equipment for forecast deviation based on the forecast product qualification rate and the qualified rate of the product to be tested; Obtain the real-time operation data of the predicted deviation production equipment, combine the performance data and appearance data of the product to be tested, and generate the equipment difference index; Combined with the real-time standard data of equipment parameters, the environmental correlation trend of defective products and the equipment difference index, multiple monitoring standards are established to monitor the real-time data of the equipment.

2. The method for intelligently monitoring diode production quality based on multi-source data analysis according to claim 1, characterized in that: The establishment of a product qualification rate prediction model for the equipment specifically includes: Obtaining the production process of the diode and the production operation manual of each production equipment, extracting the characteristic steps of each production process and the reference parameters of each production equipment, wherein the reference parameters of the production equipment include equipment parameters and environmental parameters; Classify the historical data of each production equipment according to process parameters and equipment environment data, and generate process parameter classification data and equipment environment data classification data; Based on time, the process parameter classification data and equipment environment data classification data are sorted out, and the corresponding relationship between different process parameters and product qualification rate, as well as the corresponding relationship between different equipment environment data and product qualification rate are listed; According to the corresponding relationship between different process parameters and product qualification rate, and the corresponding relationship between different equipment environment data and product qualification rate, obtain the product qualification rate change values ​​corresponding to the process parameter difference and the equipment environment data difference respectively and calculate the average value, obtain the change degree of the product qualification rate corresponding to the change of process parameters, and the change degree of the product qualification rate corresponding to the change of equipment environment data, and record them as the sensitivity of product qualification rate to process parameters and the sensitivity of product qualification rate to production equipment environment respectively; Record the equipment corresponding to the characteristic step of the production process as the characteristic equipment, classify the historical data of the characteristic equipment by week as the time interval, and generate time period historical data, wherein the time period historical data includes the time series data of the process parameters of the characteristic equipment and the product quality data; Analyze the historical data of the time period, take one hour as an analysis node, correspond the product quality data according to the analysis node, draw the product quality change curve with the analysis node as the horizontal axis and the product quality data as the vertical axis, and generate the time fluctuation trend of the characteristic equipment; Generate device parameter real-time standard data according to the time fluctuation trend of the characteristic device and the reference parameters of the production device, wherein the device parameter real-time standard data includes device parameter real-time standard data and device environment real-time standard data; Taking the real-time standard data of equipment parameters as the benchmark, respectively compare the real-time environmental data and process parameter data of each production equipment with the real-time standard data of equipment parameters, obtain the proportion of production equipment whose real-time environmental data and process parameter data are both greater than the real-time standard data of equipment parameters in the total number of production equipment, and generate the first product prediction qualification rate of the production equipment; Based on the product qualification rate of production equipment, analyze the historical production data of production equipment to obtain the process interaction index and process environment impact index; The preliminary product qualification rate relationship of the equipment is generated by comprehensively considering the sensitivity of the product qualification rate to the process parameters, the sensitivity of the product qualification rate to the production equipment environment, the process mutual influence index and the process environment influence index; The electrostatic field strength and magnetic induction strength of the magnetic field in which each production equipment is located are used as the production environment in which the production equipment is located, and the real-time alignment accuracy, processing time and packaging pressure of each production equipment are used as process parameter values, which are substituted into the product qualification rate relationship of the production equipment to obtain the predicted qualification rate of the second product; Taking the predicted qualified rate of the first product and the predicted qualified rate of the second product as input variables, a product qualified rate prediction model of the equipment is established; Among them, the relationship between the qualified rate of production equipment products is as follows: In the formula, y represents the predicted qualified rate of the second product, b i represents the sensitivity of the product qualification rate of the i-th process to the process parameters, x i represents the value of the i-th process parameter, a ij represents the process interaction index between the i-th process and the j-th process, x j represents the jth process parameter value, b j represents the sensitivity of the product qualification rate of the jth process to the process parameters, b il represents the process environment impact index between the ith process and the production environment l, x l It represents the specific value of the production environment l in which the production equipment is located, n represents the total number of processes required for the production equipment to generate products, and m represents the total number of environments that affect the production of products by the production equipment.

3. The method for intelligently monitoring diode production quality based on multi-source data analysis according to claim 2 is characterized in that: The generating of defective product environment correlation trend specifically includes: According to the diode production goals, clarify the use scenarios of the diodes and obtain the test environment thresholds of the diodes; Sorting out the data of unqualified products to obtain production data and defect data of unqualified products, wherein the defect data includes appearance characteristic values, forward voltage values, reverse current values, and breakdown voltage values; Classify the unqualified products according to the defect data into single-defect unqualified products and multiple-defect unqualified products; According to the production data of the unqualified products, the defect data of the single-defect unqualified products are analyzed to obtain the relevant production data corresponding to the defect data; Combine the relevant production data corresponding to the defect data and the data of multi-defective unqualified products to determine the cause of the multi-defective unqualified products; According to the test environment threshold of the diode, the real-time test environment is adjusted to obtain the specific values ​​of the unqualified products under different test data, and the specific value changes of the unqualified products are compared with the change range of the test data to obtain the degree of change of the defect data of the unqualified products under different test environment data; Combined with the relevant production data corresponding to the defect data and the degree of change in the defect data of unqualified products under different test environment data, the data relationship between the product defect type and the environment is determined, and the defect product environment association trend is generated.

4. The method for intelligently monitoring diode production quality based on multi-source data analysis according to claim 3 is characterized in that: The reasons for determining the occurrence of multi-defective unqualified products specifically include: According to the data of unqualified products, the data of unqualified products with multiple defects are classified to generate defect type combination data, defect severity data and defect frequency data respectively; Using defect type combination data as the main judgment criterion, determine the problematic process corresponding to the multi-defect unqualified products; Based on the defect severity data, determine whether the defects of the multi-defect unqualified products are serious. If so, conduct comprehensive quality traceability and process improvement for the multi-defect unqualified products. If not, adjust the production process parameters of the problem process corresponding to the multi-defect unqualified products. Based on the defect frequency data, determine whether the defects of multi-defect unqualified products show periodic changes. If so, it means that the operating procedures of the production line operators are not standardized or there are quality problems with the production raw materials. It is necessary to further analyze the problem in combination with the periodicity of the defects. If not, it means that the defect is a normal loss in the production process.

5. The method for intelligently monitoring diode production quality based on multi-source data analysis according to claim 4 is characterized in that: The generating device difference index specifically includes: Obtain the product qualification rate of the forecast deviation production equipment; According to the working time and type of the equipment, the historical data of the production equipment is classified to generate a group of equipment with the same working condition, wherein the equipment type and working time in the same group are the same; Obtain and compare the standard deviation of product qualification rates of equipment groups with the same working conditions. If the standard deviation is greater than 0.02, it is determined that there is a difference between the equipment and the rest of the equipment in the same group. The equipment with the difference is determined, and the corresponding operating data and product data of the equipment with the difference are obtained. According to the generated product data, the performance characteristic data and appearance characteristic data of the products generated by the difference equipment are extracted, and compared with the standard performance threshold and the standard shape data respectively to obtain the product performance difference and the product shape data difference; Obtain the equipment standard process parameters, combine them with the operation data corresponding to the difference equipment, subtract the operation data corresponding to the difference equipment from the equipment operation data in the equipment standard process parameters, and obtain the difference of the equipment operation process parameters; The ratios of product performance difference and equipment operation process parameter difference, as well as product shape data difference and equipment operation process parameter difference are calculated respectively to generate the equipment difference index.

6. The method for intelligently monitoring diode production quality based on multi-source data analysis according to claim 5, characterized in that: The establishment of multiple monitoring standards to monitor the real-time data of production equipment specifically includes: The real-time standard data of equipment parameters, including the real-time standard data of equipment parameters and the real-time standard data of equipment environment, are used as the first standard data, and compared with the real-time environment data and process parameter data of the production equipment to determine whether they are lower than the first standard data. If so, a warning is issued in advance to notify the production line operator to adjust the equipment process parameters. If not, the product qualification rate of the production equipment is continuously predicted; Acquire real-time test environment information through sensors, combine with defective product environment correlation trends corresponding to the current test environment, generate performance data corresponding to the test environment data as second standard data, obtain performance data of the product to be tested in the test environment data through measuring instruments, compare with the second standard data, and determine whether it maintains a consistent change trend with the second standard data. If so, it means that the product is a substandard product and a warning is issued in advance. If not, continue to test the remaining products to be tested; Multiply the equipment difference index and the equipment standard process parameters to generate equipment process adaptation data as the third standard data, and compare it with the real-time process parameters of the production equipment to determine whether there are differences with the third standard data. If so, it means that the products produced by the equipment are at risk of being unqualified, and a warning is issued in advance. If not, continue to monitor the real-time process parameters of the equipment.

7. An intelligent monitoring system for diode production quality based on multi-source data analysis, characterized in that: The method for intelligently monitoring diode production quality based on multi-source data analysis according to any one of claims 1 to 6 comprises: A data acquisition module, which is used to collect real-time environmental data and process parameters of production equipment, historical data of each production equipment, performance data and appearance data of the product to be tested, and transmit the data to the feature extraction module, the data analysis module and the model generation module; A feature extraction module, which is used to extract key features from the received data, extract characteristic steps of the production process, extract defect data of single-defect unqualified products, and transmit the data to the data analysis module; A data analysis module, which is used to organize the received data, generate a preliminary product qualification rate relationship of the equipment, analyze the relationship between the equipment operation data, product data and environmental data, and transmit the data to the model generation module and the quality monitoring module; A model generation module, wherein the model generation module is used to establish a prediction model according to the received data, generate prediction data through the prediction model and real-time data, and transmit the prediction data to the quality monitoring module; A quality monitoring module, which is used to generate monitoring standard data based on the received data, establish multiple monitoring standards, and transmit the data to the abnormal warning module; An abnormal warning module is used to compare the received data, determine whether there is an abnormal situation and whether a warning needs to be issued based on the comparison result, and transmit the data to the intelligent decision-making module; The intelligent decision-making module is used to analyze the received data, provide suggestions for the production process and formulate a maintenance plan for the production process equipment based on the data analysis results and the data transmitted by the abnormal warning module.

8. The diode production quality intelligent monitoring system based on multi-source data analysis according to claim 7 is characterized in that: The feature extraction module comprises: A first feature extraction unit, the first feature extraction unit is used to compare the process production flow of each production equipment and extract the feature steps in the production flow; A second feature extraction unit, the second feature extraction unit is used to extract features according to the use environment of the diode and obtain a test environment threshold of the diode; The third feature extraction unit is used to analyze the defect data of the unqualified products, extract the defect features of the unqualified products, and classify the unqualified products.

9. The diode production quality intelligent monitoring system based on multi-source data analysis according to claim 8 is characterized in that: The data analysis module includes: A first analysis unit, the first analysis unit is used to analyze the product qualification rate of the production equipment based on the process parameters corresponding to the characteristic steps and the environmental data of the production equipment, and obtain the change relationship index between each process parameter and the product qualification rate, and the change relationship index between each production environment data of the production equipment and the product qualification rate; A second analysis unit, the second analysis unit is used to adjust the real-time test environment according to the test environment threshold of the diode, obtain the degree of change of defect data of unqualified products under different test environment data, and generate a defect product environment correlation trend in combination with relevant production data corresponding to the defect data; The third analysis unit is used to generate an equipment difference index according to product performance differences, product shape data differences and equipment operation process parameter differences.

10. The diode production quality intelligent monitoring system based on multi-source data analysis according to claim 9, characterized in that: The quality monitoring module includes: A first quality monitoring module, which is used to predict the product qualification rate of each production equipment according to the product qualification rate prediction model of the equipment, and compare it with the actual product qualification rate of the production equipment to determine whether the quality of the product is abnormal; A second quality monitoring module, which is used to generate performance data corresponding to the real-time test environment data according to the real-time test environment information and the defective product environment correlation trend, and compare it with the performance data of the product to be tested in the real-time test environment data to determine whether the product quality is abnormal; The third quality monitoring module is used to generate equipment process adaptation data according to equipment standard process parameters and equipment difference index, and compare it with the real-time process parameters of the production equipment to determine whether there is any abnormality in the quality of the product.

Images