A method and system for evaluating the accuracy of chip quality analysis models

Abstract

This invention relates to a method and system for evaluating the accuracy of a chip quality analysis model, belonging to the field of chip quality assessment technology, and addresses the problem of the lack of existing methods for quantitatively evaluating the accuracy of quality analysis models. The method includes the following steps: performing thermal simulation on the circuit modules corresponding to the chip performance parameters; sorting the performance parameters according to the thermal simulation results to obtain a standard parameter sequence; generating a model calculation sequence based on the performance parameter scoring results obtained from the chip quality analysis model; calculating the similarity between the standard parameter sequence and the model calculation sequence; and evaluating the accuracy of the chip quality analysis model based on the similarity. This achieves a quantitative, accurate, and objective evaluation of the accuracy of the chip quality analysis model.

Description

Technical Field

[0001] This invention relates to the field of chip quality assessment technology, and in particular to a method and system for assessing the accuracy of chip quality analysis models. Background Technology

[0002] To ensure chip quality, chips undergo multiple rounds of testing during production. Electrical performance, overall performance, and reliability test data are obtained at each stage. This process generates a large volume of multi-dimensional test data. To fully extract the quality information hidden within this high-dimensional chip test data, researchers need to analyze the data based on its characteristics and test categories, establishing multiple chip quality analysis models for defect identification and quality grading. The quality analysis models differ for different types of chips, making the quantitative evaluation of their accuracy crucial.

[0003] Existing chip quality analysis models yield inconsistent results and vary in accuracy. Current technology lacks a quantitative evaluation method to assess the accuracy of these quality analysis models. Summary of the Invention

[0004] Based on the above analysis, the embodiments of the present invention aim to provide a method and system for evaluating the accuracy of chip quality analysis models, in order to solve the problem of the lack of existing methods for quantitatively evaluating the accuracy of quality analysis models.

[0005] On one hand, embodiments of the present invention provide a method for evaluating the accuracy of a chip quality analysis model, comprising the following steps:

[0006] Thermal simulation is performed on the circuit modules corresponding to the chip performance parameters, and the performance parameters are sorted according to the thermal simulation results to obtain a standard parameter sequence.

[0007] Generate a model calculation sequence based on the performance parameter scoring results obtained from the chip quality analysis model;

[0008] Calculate the similarity between the standard parameter sequence and the model calculation sequence, and evaluate the accuracy of the chip quality analysis model based on the similarity.

[0009] Furthermore, the step of sorting the performance parameters according to the thermal simulation results to obtain a standard parameter sequence includes:

[0010] The temperature range is determined based on the highest thermal simulation temperature of all circuit modules, and the temperature range is divided into multiple temperature intervals arranged in order of temperature value.

[0011] Calculate the thermal simulation area of ​​each circuit module based on the temperature range to which the highest thermal simulation temperature of each circuit module belongs;

[0012] The standard parameter sequence is obtained by sorting the performance parameters of each circuit module according to the highest thermal simulation temperature and thermal simulation area.

[0013] Furthermore, the circuit area within the temperature range to which the highest temperature belongs is calculated as the thermal simulation area of ​​the circuit module.

[0014] Furthermore, based on the highest thermal simulation temperature and thermal simulation area of ​​each circuit module, the performance parameters corresponding to the circuit modules are sorted to obtain a standard parameter sequence, including:

[0015] The performance parameters of each circuit module are grouped according to the temperature range to which the highest thermal simulation temperature of each circuit module belongs.

[0016] Based on the thermal simulation area of ​​the circuit module corresponding to the performance parameters, the performance parameters within the same group are sorted to obtain a standard parameter sequence.

[0017] Furthermore, the similarity between the standard parameter sequence and the model-calculated sequence is calculated using the following method:

[0018] If the length of the standard parameter sequence is less than or equal to the first threshold, then according to the formula Calculate the similarity sim(S,P) between the standard parameter sequence and the model-calculated sequence, where n represents the sequence length and S... 1:i ∩P 1:i This indicates the number of identical elements between the standard parameter sequence and the model calculation sequence;

[0019] If the length of the standard parameter sequence is greater than the first threshold, then the first d elements of the standard parameter sequence and the model calculation sequence are taken based on the sequence weights, according to the formula. Calculate the similarity between the standard parameter sequence and the model-calculated sequence; where S represents the standard parameter sequence, P represents the model-calculated sequence, d represents the sequence depth, p represents the weight parameter, and A i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

[0020] Furthermore, the sequence depth d and weight parameter p are determined as follows:

[0021] Calculation based on formula Calculate the sequence weight W at depth d. SIM (1:d);

[0022] The weight parameter p is selected as the weight parameter and sequence depth when the sequence weight is greater than the second threshold and the sequence weight is greater than the third threshold. The similarity between the standard parameter sequence and the model-calculated sequence is then calculated.

[0023] Furthermore, based on the performance parameter scoring results obtained from the chip quality analysis model, a model calculation sequence is generated, including:

[0024] Obtain the performance parameter scoring results from the chip quality analysis model, sort the performance parameters in descending order of fault scores, and obtain the model calculation sequence.

[0025] On the other hand, embodiments of the present invention provide a system for evaluating the accuracy of a chip quality analysis model, comprising the following modules:

[0026] A standard parameter sequence generation module is used to perform thermal simulation on the circuit modules corresponding to the chip performance parameters, and sort the performance parameters according to the thermal simulation results to obtain a standard parameter sequence.

[0027] The model calculation sequence generation module is used to generate a model calculation sequence based on the performance parameter scoring results obtained from the chip quality analysis model.

[0028] The accuracy evaluation module is used to calculate the similarity between the standard parameter sequence and the model calculation sequence, and to evaluate the accuracy of the chip quality analysis model based on the similarity.

[0029] Furthermore, the standard parameter sequence generation module includes:

[0030] The temperature range division module is used to determine the temperature range based on the highest thermal simulation temperature of all circuit modules, and divide the temperature range into multiple temperature ranges arranged in order of temperature value.

[0031] The thermal simulation area calculation module is used to calculate the thermal simulation area of ​​each circuit module based on the temperature range to which the highest thermal simulation temperature of each circuit module belongs.

[0032] The parameter sorting module is used to sort the performance parameters of each circuit module according to the highest thermal simulation temperature and thermal simulation area to obtain a standard parameter sequence.

[0033] Furthermore, the accuracy evaluation module calculates the similarity between the standard parameter sequence and the model-calculated sequence in the following manner:

[0034] If the length of the standard parameter sequence is less than or equal to the first threshold, then according to the formula Calculate the similarity sim(S,P) between the standard parameter sequence and the model-calculated sequence, where n represents the sequence length and S... 1:i ∩P 1:i This indicates the number of identical elements between the standard parameter sequence and the model calculation sequence;

[0035] If the length of the standard parameter sequence is greater than the first threshold, then the first d elements of the standard parameter sequence and the model calculation sequence are taken based on the sequence weights, according to the formula. Calculate the similarity between the standard parameter sequence and the model-calculated sequence; where S represents the standard parameter sequence, P represents the model-calculated sequence, d represents the sequence depth, p represents the weight parameter, and A i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

[0036] Compared with existing technologies, this invention constructs a standard parameter sequence for a chip model using thermal simulation, and builds a model calculation sequence based on the analysis results of the chip quality analysis model. By calculating the similarity between the standard parameter sequence and the model calculation sequence, the accuracy of the chip quality analysis model can be quantitatively, accurately, and objectively evaluated. A sequence overlap comparison method is used to calculate the similarity between the standard parameter sequence and the actual model calculation sequence. Evaluation depth and weight assignment are introduced during the comparison process. In the weight assignment process, the length of the top parameter can be selected and assigned a larger weight. This sequence overlap comparison method can select the sequence length and sequence weight based on the actual sequence content, and then calculate the similarity between the standard parameter sequence and the model calculation sequence, thereby achieving a rapid and reasonable evaluation of the accuracy of the chip quality analysis model.

[0037] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description

[0038] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0039] Figure 1 This is a flowchart illustrating a method for evaluating the accuracy of a chip quality analysis model according to an embodiment of the present invention.

[0040] Figure 2 This is a block diagram of a system for evaluating the accuracy of a chip quality analysis model according to an embodiment of the present invention;

[0041] Figure 3 This is a mathematical relationship diagram between parameters in an embodiment of the present invention. Detailed Implementation

[0042] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0043] To ensure chip quality, chips undergo multiple rounds of testing during production. Electrical performance, overall performance, and reliability test data are obtained at each stage. This process generates a large volume of multi-dimensional test data. To fully extract the quality information hidden within this high-dimensional chip test data, researchers need to analyze the data based on its characteristics and test categories, establishing multiple chip quality analysis models for defect identification and quality grading. The quality analysis models differ for different types of chips, making the quantitative evaluation of their accuracy crucial.

[0044] Existing chip quality analysis models yield inconsistent results and vary in accuracy. Current technology lacks a quantitative evaluation method to assess the accuracy of these quality analysis models. To fill this technological gap, one specific embodiment of the present invention discloses a method for evaluating the accuracy of chip quality analysis models, such as... Figure 1 As shown, it includes the following steps:

[0045] S1. Perform thermal simulation on the circuit modules corresponding to the chip performance parameters, and sort the performance parameters according to the thermal simulation results to obtain a standard parameter sequence;

[0046] S2. Generate a model calculation sequence based on the performance parameter scoring results obtained from the chip quality analysis model;

[0047] S3. Calculate the similarity between the standard parameter sequence and the model calculation sequence, and evaluate the accuracy of the chip quality model based on the similarity.

[0048] By using thermal simulation to construct a standard parameter sequence for the chip model, and constructing a model calculation sequence based on the analysis results of the chip quality analysis model, the accuracy of the chip quality analysis model can be quantitatively, accurately, and objectively evaluated by calculating the similarity between the standard parameter sequence and the model calculation sequence.

[0049] The impact of temperature increases on chip performance is significant. Elevated temperatures lead to decreased current and increased circuit delay, affecting circuit performance, and also reduce circuit reliability. In short, thermal issues have received increasing attention. Studies show that 55% of device failures are caused by temperatures exceeding specified values. Therefore, a standard parameter sequence is constructed based on the thermal simulation structure of the chip circuit.

[0050] During implementation, the approach begins with the chip's circuitry and layout, marking the performance parameters corresponding to the internal circuitry. For example, for a typical ADC / DAC chip, the performance parameters tested mainly include: resolution, full-scale range, error offset, least significant bit, differential nonlinearity error, integral nonlinearity error, signal-to-noise ratio, and total harmonic distortion (THD). Another example is a low-noise amplifier module in an RF circuit, which requires testing performance parameters such as gain, gain flatness, noise figure, noise figure flatness, and power consumption at a specific frequency band. The circuit modules corresponding to these performance parameters are extracted, and thermal simulation is performed on each module to obtain the simulation results.

[0051] During implementation, thermal simulation analysis of the circuit can be performed using ANSYS software. To obtain the temperature distribution of the circuit layout, the corresponding circuit layout must first be exported from the Cadence platform to GDSII format. Then, the GDSII file is converted to an ANSYS-compatible file (ANF format) using LinkCAD software, and thermal simulation is performed within ANSYS software. The specific thermal simulation model construction and simulation process are existing technologies and will not be elaborated upon here.

[0052] Specifically, step S1, which involves sorting the performance parameters according to the thermal simulation results to obtain a standard parameter sequence, includes:

[0053] S11. Determine the temperature range based on the highest thermal simulation temperature of all circuit modules, and divide the temperature range into multiple temperature intervals arranged in order of temperature value.

[0054] After obtaining the thermal simulation results for the circuit module corresponding to each performance parameter, first, obtain the highest temperature from the thermal simulation results, labeled as T. Divide the temperature range of 0 to T into x temperature intervals, arranging each interval from highest to lowest temperature. During implementation, the value of x can be set according to different chip types or calculation accuracy requirements. For example, if the highest temperature is 50℃, divide 0 to 50℃ into 10 intervals: the first interval is 50-45℃, the second interval is 45-40℃, and so on.

[0055] S12. Calculate the thermal simulation area of ​​each circuit module based on the temperature range to which the highest thermal simulation temperature of each circuit module belongs.

[0056] Specifically, the circuit area within the temperature range to which the highest temperature belongs is calculated as the thermal simulation area of ​​the circuit module.

[0057] For example, the highest temperature of a certain circuit module in thermal simulation is 47℃, which falls within the first temperature range. The temperature range of the first range is 50-45℃. Therefore, the circuit area within the first temperature range in the thermal simulation results of this circuit module is calculated as the corresponding thermal simulation area of ​​the circuit module.

[0058] Understandably, performing thermal simulation on a circuit module can yield its thermal distribution map. The circuit area within the first temperature range of the thermal distribution map is taken as the corresponding thermal simulation area of ​​that circuit module.

[0059] S13. Based on the highest thermal simulation temperature and thermal simulation area of ​​each circuit module, sort the performance parameters corresponding to the circuit modules to obtain a standard parameter sequence.

[0060] Specifically, based on the highest thermal simulation temperature and thermal simulation area of ​​each circuit module, the performance parameters corresponding to the circuit modules are sorted to obtain a standard parameter sequence, including:

[0061] S131. Group the performance parameters of each circuit module according to the temperature range to which the highest thermal simulation temperature of each circuit module belongs.

[0062] The higher the temperature and the larger the heat distribution area, the more critical the parameters become to chip quality. Therefore, the performance parameters of each circuit module are first grouped according to the temperature range to which the highest temperature in the thermal simulation belongs. If the highest temperature in the thermal simulation belongs to the first temperature range, the performance parameters of that circuit module are assigned to the first group; if the highest temperature in the thermal simulation belongs to the second temperature range, the performance parameters of that circuit module are assigned to the second group, and so on.

[0063] S132. Sort the performance parameters in the same group according to the thermal simulation area of ​​the circuit module corresponding to the performance parameters to obtain a standard parameter sequence.

[0064] After grouping the performance parameters, within each group, they are sorted based on the thermal simulation area of ​​the circuit modules corresponding to the performance parameters. Performance parameters corresponding to circuit modules with larger thermal simulation areas are placed before those corresponding to circuit modules with smaller thermal simulation areas. For example, performance parameters a and b belong to the same group. Since the thermal simulation area of ​​the circuit module corresponding to performance parameter a is larger than that of performance parameter b, performance parameter a is placed before performance parameter b. After sorting the performance parameters within the same group according to their thermal simulation areas, a standard parameter sequence is obtained, as shown in Table 1. The standard parameter sequence is denoted as S, where S = {S1, S2, ... S...} n} 1×n Where n represents the sequence length, S i The smaller the value of parameter i, the better S i The more important the impact on chip quality.

[0065] Table 1. Thermal distribution of each module in the chip thermal analysis model.

[0066]

[0067] After obtaining the standard parameter sequence, the analysis structure of the chip quality analysis model is acquired, and the model calculation sequence is constructed. Specifically, the model calculation sequence is generated based on the performance parameter scoring results obtained from the chip quality analysis model, including:

[0068] Obtain the performance parameter scoring results from the chip quality analysis model, sort the performance parameters in descending order of fault scores, and obtain the model calculation sequence.

[0069] During implementation, the data analysis results of the chip test data analysis model are first used to extract the performance parameter scoring results from the analysis model and save them in a .csv format that is easy to statistically analyze. The chip performance parameter scoring matrix is ​​defined as X.

[0070]

[0071] X is an m×k dimensional matrix, where k represents the total number of chips and m represents the total number of chip performance parameters.

[0072] Matrix element x ij This represents the score result of the i-th parameter of the j-th chip. Different fault types are represented by different integers, with 0 to N representing chip quality from zero faults to complete faults.

[0073] The fault score results of the performance parameters of k chips are statistically analyzed, and a fault score matrix Y is established, where y i This represents the sum or mean of the elements in the i-th row of matrix X.

[0074]

[0075] The performance parameters corresponding to each element in matrix Y are sorted in descending order of element value to generate the model calculation sequence P.

[0076] To accurately calculate the similarity between the two sequences, step S3 uses the following method to calculate the similarity between the standard parameter sequence and the model-calculated sequence:

[0077] If the length of the standard parameter sequence is less than or equal to the first threshold, then according to the formula Calculate the similarity sim(S,P) between the standard parameter sequence and the model-calculated sequence, where n represents the sequence length and S... 1:i ∩P 1:i A represents the number of identical elements between the standard parameter sequence and the model computation sequence. i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

[0078] In implementation, if the length of the standard parameter sequence is less than the first threshold (e.g., the first threshold is set to 100), the amount of data to be calculated is relatively small. In this case, the entire standard parameter sequence can be extracted and its similarity calculated with the model's calculated sequence. The specific calculation formula is as follows: That is, the similarity between two sequences is measured by accumulating the overlap between the two sequences at different depths.

[0079] If the length of the standard parameter sequence is greater than the first threshold, then the first d elements of the standard parameter sequence and the model calculation sequence are taken based on the sequence weights, according to the formula. Calculate the similarity between the standard parameter sequence and the model-calculated sequence; where S represents the standard parameter sequence, P represents the model-calculated sequence, d represents the sequence depth, p represents the weight parameter, and A i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

[0080] If the standard parameter sequence is long, in order to further improve computational efficiency, a corresponding weight is assigned to each layer depth, and similarity calculation is performed on a portion of the sequence based on the weight, thereby reducing the amount of computation and improving the evaluation efficiency.

[0081] First, define the weight of the i-th element as w. i ,

[0082] w i =(1-p)p i-1 ,0<p<1(1)

[0083] Where p represents the weight parameter, which indicates the magnitude of the decrease in weight between adjacent elements.

[0084] w i It follows a geometric series distribution, as n→∞ Therefore, weight

[0085] Since the matching of two elements in the P and S sequences at depth d does not affect the consistency result before d, but does affect the consistency of the sequences after d, the matching of two elements (P...) at depth d... d ,S d The sequence d (p) affects every sequence after d+1. When the sequence of the first d elements is selected as the evaluation of the overall sequence similarity, d affects all sequences after d+1, and similarly, the sequence of the first d elements affects the sequences after d+1. Therefore, at depth d, the element (P) d ,S d The weight of the impact on the overall sequence is:

[0086]

[0087] The sum of the weights of sequences from depth 1 to d is:

[0088]

[0089] Based on formulas (1) and (3), the weight w of the sequence of the first d elements in the whole sequence can be obtained. SIM (1:d),

[0090]

[0091] The specific derivation process is as follows:

[0092] first,

[0093] Where 1≤j≤d, j+1≤i≤d.

[0094]

[0095] therefore,

[0096] Specifically, for accurate calculation, the weight w SIM (1:d) should exceed the third threshold. For example, the third threshold could be 0.85.

[0097] During implementation, by selecting different p values, w can be plotted. SIM The relationship between (1:d) and d, and the mathematical relationship among the three are as follows: Figure 3 As shown.

[0098] Depend on Figure 3 It can be seen that when w is determined SIM When the ratio is (1:d), a larger p parameter requires a larger evaluation depth d to achieve the desired global weight percentage for the prefix sequence. The p parameter determines the magnitude of the weight decrease; a smaller p parameter results in a larger weight decrease and greater weight differences between adjacent elements. When p = 0, only the highest-ranking element is considered; when p approaches 1, the weight decrease becomes flat, and the evaluation depth becomes arbitrarily large. Therefore, the selection of the p parameter depends on the length of the actual sequence. A p value that is too small will distort the sequence similarity calculation results. Table 2 lists w for different p and d values. SIM (1:d) Reference results.

[0099] Table 2w SIM (1:d), reference values ​​for p and d

[0100]

[0101] In practice, to ensure the accuracy of the calculation results, the weight parameter p should be greater than the second threshold, for example, the second threshold could be 0.95. That is, the value of d when the weight parameter p is greater than the second threshold and the sequence weight is greater than the third threshold is selected as the weight parameter and sequence depth, and the similarity between the standard parameter sequence and the model-calculated sequence is calculated.

[0102] Once the parameters p and d are determined, according to the formula... The similarity between the standard parameter sequence and the model-calculated sequence is calculated; therefore, a subset of sequences is selected for similarity calculation to improve the efficiency of similarity calculation. Here, S represents the standard parameter sequence, P represents the model-calculated sequence, d represents the sequence depth, p represents the weight parameter, and A... i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

[0103] During implementation, the consistency of the standard sequence S and the model-calculated sequence P at different depths d is calculated by comparing the element overlap. The comparison process is shown in Table 3.

[0104] Table 3. Calculation of Sequence Overlap and Consistency

[0105]

[0106] For different chip quality analysis models, the higher the similarity between the model's calculation sequence and the standard parameter sequence, the higher the accuracy of its analysis results, thus quantitatively and objectively evaluating the accuracy of different quality analysis models.

[0107] One specific embodiment of the present invention discloses a system for evaluating the accuracy of a chip quality analysis model, such as... Figure 2 As shown, the system includes the following modules:

[0108] A standard parameter sequence generation module is used to perform thermal simulation on the circuit modules corresponding to the chip performance parameters, and sort the performance parameters according to the thermal simulation results to obtain a standard parameter sequence.

[0109] The model calculation sequence generation module is used to generate a model calculation sequence based on the performance parameter scoring results obtained from the chip quality analysis model.

[0110] The accuracy evaluation module is used to calculate the similarity between the standard parameter sequence and the model calculation sequence, and to evaluate the accuracy of the chip quality analysis model based on the similarity.

[0111] Preferably, the standard parameter sequence generation module includes:

[0112] The temperature range division module is used to determine the temperature range based on the highest thermal simulation temperature of all circuit modules, and divide the temperature range into multiple temperature ranges arranged in order of temperature value.

[0113] The thermal simulation area calculation module is used to calculate the thermal simulation area of ​​each circuit module based on the temperature range to which the highest thermal simulation temperature of each circuit module belongs.

[0114] The parameter sorting module is used to sort the performance parameters of each circuit module according to the highest thermal simulation temperature and thermal simulation area to obtain a standard parameter sequence.

[0115] Preferably, the accuracy evaluation module calculates the similarity between the standard parameter sequence and the model-calculated sequence in the following manner:

[0116] If the length of the standard parameter sequence is less than or equal to the first threshold, then according to the formula Calculate the similarity sim(S,P) between the standard parameter sequence and the model-calculated sequence, where n represents the sequence length and S... 1:i ∩P 1:i This indicates the number of identical elements between the standard parameter sequence and the model calculation sequence;

[0117] If the length of the standard parameter sequence is greater than the first threshold, then the first d elements of the standard parameter sequence and the model calculation sequence are taken based on the sequence weights, according to the formula. Calculate the similarity between the standard parameter sequence and the model-calculated sequence; where S represents the standard parameter sequence, P represents the model-calculated sequence, d represents the sequence depth, p represents the weight parameter, and A i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

[0118] The above-described method and system embodiments are based on the same principles, and their related aspects can be referenced from each other to achieve the same technical effects. For specific implementation processes, please refer to the foregoing embodiments, which will not be repeated here.

[0119] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0120] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for evaluating the accuracy of a chip quality analysis model, characterized in that, Includes the following steps: Thermal simulation is performed on the circuit modules corresponding to the chip performance parameters, and the performance parameters are sorted according to the thermal simulation results to obtain a standard parameter sequence. Generate a model calculation sequence based on the performance parameter scoring results obtained from the chip quality analysis model; Calculate the similarity between the standard parameter sequence and the model calculation sequence, and evaluate the accuracy of the chip quality analysis model based on the similarity. The step of sorting the performance parameters according to the thermal simulation results to obtain a standard parameter sequence includes: The temperature range is determined based on the highest thermal simulation temperature of all circuit modules, and the temperature range is divided into multiple temperature intervals arranged in order of temperature value. Calculate the thermal simulation area of ​​each circuit module based on the temperature range to which the highest thermal simulation temperature of each circuit module belongs; The standard parameter sequence is obtained by sorting the performance parameters of each circuit module according to the highest thermal simulation temperature and thermal simulation area.

2. The method for evaluating the accuracy of a chip quality analysis model according to claim 1, characterized in that, The circuit area within the temperature range to which the highest temperature belongs is calculated as the thermal simulation area of ​​the circuit module.

3. The method for evaluating the accuracy of a chip quality analysis model according to claim 1, characterized in that, Based on the highest thermal simulation temperature and thermal simulation area of ​​each circuit module, the performance parameters corresponding to the circuit modules are sorted to obtain a standard parameter sequence, including: The performance parameters of each circuit module are grouped according to the temperature range to which the highest thermal simulation temperature of each circuit module belongs. Based on the thermal simulation area of ​​the circuit module corresponding to the performance parameters, the performance parameters within the same group are sorted to obtain a standard parameter sequence.

4. The method for evaluating the accuracy of a chip quality analysis model according to claim 1, characterized in that, The similarity between the standard parameter sequence and the model-calculated sequence is calculated using the following method: If the length of the standard parameter sequence is less than or equal to the first threshold, then according to the formula Calculate the similarity sim(S,P) between the standard parameter sequence and the model-calculated sequence, where n represents the sequence length and S... 1:i ∩P 1:i This indicates the number of identical elements between the standard parameter sequence and the model calculation sequence; If the length of the standard parameter sequence is greater than the first threshold, then the first d elements of the standard parameter sequence and the model calculation sequence are taken based on the sequence weights, according to the formula. Calculate the similarity between the standard parameter sequence and the model-calculated sequence; where S represents the standard parameter sequence, P represents the model-calculated sequence, d represents the sequence depth, p represents the weight parameter, and A i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

5. The method for evaluating the accuracy of a chip quality analysis model according to claim 4, characterized in that, The sequence depth d and weight parameter p are determined as follows: Calculation based on formula Calculate the sequence weight W at depth d. SIM (1:d); The weight parameter p is selected as the weight parameter and sequence depth when the sequence weight is greater than the second threshold and the sequence weight is greater than the third threshold. The similarity between the standard parameter sequence and the model-calculated sequence is then calculated.

6. The method for evaluating the accuracy of a chip quality analysis model according to claim 1, characterized in that, The model calculation sequence is generated based on the performance parameter scoring results obtained from the chip quality analysis model, including: Obtain the performance parameter scoring results from the chip quality analysis model, sort the performance parameters in descending order of fault scores, and obtain the model calculation sequence.

7. A system for evaluating the accuracy of a chip quality analysis model, characterized in that, Includes the following modules: A standard parameter sequence generation module is used to perform thermal simulation on the circuit modules corresponding to the chip performance parameters, and sort the performance parameters according to the thermal simulation results to obtain a standard parameter sequence. The model calculation sequence generation module is used to generate a model calculation sequence based on the performance parameter scoring results obtained from the chip quality analysis model. An accuracy evaluation module is used to calculate the similarity between the standard parameter sequence and the model calculation sequence, and to evaluate the accuracy of the chip quality analysis model based on the similarity. The standard parameter sequence generation module includes: The temperature range division module is used to determine the temperature range based on the highest thermal simulation temperature of all circuit modules, and divide the temperature range into multiple temperature ranges arranged in order of temperature value. The thermal simulation area calculation module is used to calculate the thermal simulation area of ​​each circuit module based on the temperature range to which the highest thermal simulation temperature of each circuit module belongs. The parameter sorting module is used to sort the performance parameters of each circuit module according to the highest thermal simulation temperature and thermal simulation area to obtain a standard parameter sequence.

8. The system for evaluating the accuracy of a chip quality analysis model according to claim 7, characterized in that, The accuracy evaluation module calculates the similarity between the standard parameter sequence and the model-calculated sequence in the following manner: If the length of the standard parameter sequence is less than or equal to the first threshold, then according to the formula Calculate the similarity sim(S,P) between the standard parameter sequence and the model-calculated sequence, where n represents the sequence length and S... 1:i ∩P 1:i This indicates the number of identical elements between the standard parameter sequence and the model calculation sequence; If the length of the standard parameter sequence is greater than the first threshold, then the first d elements of the standard parameter sequence and the model calculation sequence are taken based on the sequence weights, according to the formula. Calculate the similarity between the standard parameter sequence and the model-calculated sequence; where S represents the standard parameter sequence, P represents the model-calculated sequence, d represents the sequence depth, p represents the weight parameter, and A i This represents the consistency between the standard parameter sequence and the model calculation sequence at depth i.

Images