Regression test simulation waveform classification method based on Mama neural network

Abstract

The invention discloses a regression test error simulation waveform automatic classification method based on a Mama neural network architecture, and belongs to the field of integrated circuits, and the method comprises the steps: analyzing a simulation waveform file obtained through a regression test, and converting a key signal waveform into an integer waveform discrete sequence; normalizing the numerical range of the waveform sequence to a 0-1 interval, intercepting a fixed length, inputting the fixed length into a Mama neural network model, and outputting two parameters, namely the size of a block and the initial position of the block, by the network; dividing waveforms by using two characteristic parameters output by the network, calculating the thickness, fluctuation degree and mean value characteristics of each waveform, and calculating overall characteristics; using an integer multiple of the size of a waveform block as a window, performing fast Fourier transform on the waveform, and calculating frequency domain features; and merging the extracted features into a multi-dimensional feature vector, inputting the multi-dimensional feature vector into a support vector machine, and outputting to obtain a final classification result. The regression test failure reason diagnosis method can effectively improve the accuracy of regression test failure reason diagnosis, replace a manual triage process, and improve the verification efficiency.

Description

Technical Field

[0001] The invention belongs to the field of integrated circuits, and in particular relates to a regression test simulation waveform classification method based on a Mamba neural network. Background Art

[0002] Regression testing is one of the key steps in integrated circuit design and verification. Its purpose is to verify through simulation whether the circuit still maintains its original function after modification. As the scale of integrated circuit design increases, the amount of data generated by regression testing is also growing rapidly, especially in the analysis and classification of simulation waveforms, which brings great challenges to engineers. Traditional regression test error analysis methods rely on manual inspection and experience judgment, which is time-consuming and error-prone, and it is difficult to cope with the massive amount of waveform data in complex circuits.

[0003] Since digital circuit design often introduces structures such as integrators, accumulators or feedback loops, their simulation waveforms often contain many potential periodic features or block features, which are of great significance for fault diagnosis. However, existing classification methods are usually unable to effectively extract the potential features of the waveform, resulting in an inability to accurately distinguish the error type. Therefore, how to efficiently and accurately classify regression test error waveforms through automation technology is an urgent problem to be solved to improve the efficiency and accuracy of regression test analysis.

[0004] As an efficient learning model, neural networks have been widely used in many fields such as image processing and speech recognition. In integrated circuit testing and verification, the application of neural networks can effectively improve the speed and accuracy of data processing. In recent years, the Mamba neural network architecture has gradually become a popular choice in waveform classification problems due to its advantages in processing complex waveform data. Traditional neural network architectures often fail to fully consider the time characteristics and frequency domain characteristics of waveforms, resulting in unsatisfactory classification effects and processing speeds when faced with large amounts of test data.

[0005] In addition, most existing methods rely on a single classifier to classify waveforms, lacking a way to fuse multi-dimensional features, and are unable to cope with the processing requirements of highly complex waveforms. Especially in the case of small sample data sets, the classification performance may be severely limited and cannot be used as a general solution. Therefore, how to combine the time domain block features and frequency domain features of the waveform and use an efficient parallel network structure to accelerate the classification process has become an important research direction in the current regression test error simulation waveform classification.

[0006] In order to solve the above problems, the present invention proposes a regression test simulation waveform classification method based on Mamba neural network. By processing the waveform data in blocks, extracting the time domain block features and frequency domain features, and combining the support vector machine (SVM) for classification, the method can achieve efficient and accurate classification results on a small sample data set, thereby effectively improving the accuracy of regression test error analysis, replacing the cumbersome manual triage process, and improving verification efficiency. Summary of the invention

[0007] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a regression test simulation waveform classification method based on Mamba neural network. The method takes into account the block characteristics of the simulation waveform in the regression test, and first parses the key signal waveform in the regression test into a discrete sequence. Then, the Mamba neural network architecture is adopted to divide the waveform into blocks by taking advantage of the global receptive field and dynamic weighting. The time domain features are calculated using the block results, and the waveforms are classified using a support vector machine in combination with the frequency domain features. In order to improve the classification efficiency, a parallel network structure design is adopted to accelerate the classification process, and efficient classification of large-scale simulation waveforms can be achieved under a small sample data set, thereby improving the accuracy of diagnosis of the cause of regression test failure. The method proposed in the present invention can be used as a universal solution to replace the traditional manual triage process and effectively improve the verification efficiency.

[0008] The present invention adopts the following technical solutions to achieve the above-mentioned invention object: The regression test simulation waveform classification method based on Mamba neural network includes the following steps: Step 1: parse the simulation waveform file obtained by the regression test to obtain an integer discrete sequence of key signal waveforms for subsequent processing and classification of the waveforms.

[0009] Step 2: Normalize the numerical range of the waveform sequence to the interval of 0-1, cut off the fixed length of 2048 and input it into the Mamba neural network model. The network inference outputs two parameters: block size and block starting position.

[0010] Step three, use the two characteristic parameters output by the network to divide the waveform into blocks, calculate the thickness, fluctuation degree, mean and other characteristics of each waveform, and count the overall characteristics.

[0011] Step 4: Use integer multiples of the waveform block size as a window to perform fast Fourier transform on the waveform and calculate frequency domain features such as the ratio of high-frequency and low-frequency energy and the concentration of spectrum peak energy.

[0012] Step 5: Combine the features extracted from the above process to obtain a multi-dimensional feature vector, input it into the support vector machine, and output the final classification result.

[0013] Preferably, the method for processing the key signal waveform in step 1 is: At the rising edge of the base frequency clock, read the data bit value of the key signal, and convert the binary signal value at the same time into a decimal integer for storage. Repeat the above steps to obtain a discrete sequence of a key waveform signal.

[0014] Preferably, the method of normalizing the waveform sequence in step 2 and inputting it into the Mamba neural network model is: Divide each value in the waveform sequence by the maximum value in the sequence to normalize each value in the sequence to the range of 0-1. Truncate the waveform sequence to a length of 2048, and fill the end of the sequence with a value of 0 if the length is insufficient. Input the processed waveform sequence of length 2048 into the Mamba neural network model, and the model outputs two parameters: block size L and block starting position K.

[0015] Preferably, the method of waveform segmentation and feature extraction in step 3 is: According to the output results of the model, the waveform is divided into waveform blocks of the same size starting from the block starting position K according to the block size L, and the waveform before the block starting position and the waveform whose final length is less than L are discarded. The thickness, fluctuation degree, mean and other characteristics of each waveform block are calculated, and based on the calculation results, the overall block thickness mean, block thickness fluctuation standard deviation, block fluctuation degree mean, and block mean standard deviation are statistically calculated.

[0016] Preferably, the frequency domain feature calculation method in step 4 is: Using N times the block size L (N is as large as possible and L * N does not exceed the total length of the waveform) output by the model as a window, perform fast Fourier transform on the waveform sequence in the window to obtain an amplitude sequence of length L * N. Calculate the ratio of high-frequency and low-frequency energy of the amplitude sequence and the energy concentration at the spectral peak position of the integer multiple of the amplitude sequence N.

[0017] Preferably, the method for the support vector machine in step 5 to process features and obtain classification results is: The seven eigenvalues ​​of block size L, block thickness mean, block thickness fluctuation standard deviation, block fluctuation degree mean, block mean standard deviation, high-low frequency energy ratio, and spectrum peak energy concentration are combined into a 7-dimensional feature vector and input into the support vector machine to finally obtain the output key signal waveform sequence classification result.

[0018] The present invention adopts the above technical solution and has the following beneficial effects: (1) The present invention proposes an automatic classification method for regression test error simulation waveforms based on the Mamba neural network architecture. By dividing the simulation waveforms in the regression test into blocks, extracting the time domain block features and frequency domain features of the waveforms, and combining them with support vector machines for classification, efficient and accurate regression test waveform error classification is achieved. Compared with traditional manual triage methods, the present invention can greatly improve classification efficiency, reduce manual intervention, and improve the accuracy of diagnosis of regression test error causes.

[0019] (2) The present invention adopts the Mamba neural network architecture, which can effectively capture long-distance dependencies and complex timing patterns in waveforms. It takes advantage of the global receptive field and dynamic weighting, and provides better generalization ability, classification accuracy and computational efficiency than traditional convolutional neural networks, transformer networks, etc. The Mamba network can train an efficient classification model on a small sample data set, and is particularly suitable for the automated classification of regression test waveforms.

[0020] (3) The present invention has strong versatility and can be used as a solution to replace manual triage. By using waveform training sets with different block sizes to train the Mamba block model, when faced with new waveforms, only the support vector machine (SVM) needs to be retrained to complete the classification. Since the input feature dimension of SVM is low and the training process is fast, this method can be quickly deployed and efficiently applied in practical applications, and is suitable for different types of regression test waveform classification tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 1 is a flow chart of an automatic classification method for regression test error simulation waveforms based on the Mamba neural network architecture of the present invention; Figure 2 It is a structural diagram of the Mamba neural network architecture of the present invention; Figure 3 It is a structural diagram of a single core layer in the Mamba neural network architecture of the present invention; Figure 4 It is a schematic diagram of the block waveform thickness, fluctuation degree and mean value characteristics of the present invention; Figure 5 It is a schematic diagram of obtaining the final classification result by inputting the feature vector of the present invention into the support vector machine. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should belong to the protection scope of the present invention.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Embodiment: The present invention is further described below in conjunction with the accompanying drawings. Figure 1 As shown, it includes 5 steps.

[0025] 1. Read the data bit values ​​of the key signal at the rising edge of the base frequency clock, and convert the binary signal value at the same time into a decimal integer for storage. Repeat the above steps to obtain a discrete sequence of a key waveform signal.

[0026] The simulation waveform is usually a complex data stream composed of multiple signals, which change periodically and synchronously with the baseband clock. The reason for the failure of the regression test is often reflected in certain key signals. The present invention synchronously reads the values ​​of all data bits of the key signal at the rising edge of each baseband clock.

[0027] After sampling at each rising edge of the clock, the acquired signal data bits are binary (i.e., each bit has only two possible states: "0" or "1"). In order to facilitate subsequent processing and analysis, these binary numbers are converted into decimal integers for storage. By looping the above steps, a discrete sequence of a key waveform signal is obtained.

[0028] 2. Divide each value in the waveform sequence by the maximum value in the sequence to normalize each value in the sequence to the range of 0-1. Truncate the sequence that is too long to 2048, and add 0 to the end of the sequence if the length is insufficient, to ensure that the length of the waveform is 2048 before entering the network.

[0029] The overall structure of the Mamba waveform block network is as follows Figure 2 As shown in the figure, the waveform is reversed and then input into the first Mamba core layer from two directions. The output of the core layer is activated using the ReLU activation function and then input into the next core layer. After being passed through multiple Mamba core layers, it is merged into one Mamba core layer. The final output of the core layer is passed through two linear layers to obtain two characteristic parameters: block size and block start position.

[0030] The structure of a single Mamba core layer is as follows Figure 3 As shown in the figure, the input sequence is first divided into multiple local windows, so that SSM is performed in different directions while retaining the global SSM operation. The core layer implements spatial and channel attention modules before patch merging to enhance the integration of directional features and reduce redundancy. In addition, a strategy is adopted to select the most effective scanning direction for each layer, thereby optimizing computational efficiency.

[0031] Combining the above technical methods, the network architecture maximizes the global perception ability of the model and simulates the process of artificial waveform segmentation. In addition, through such a setting, parallel computing can be achieved within a single Mamba core layer and between core layers, greatly improving the reasoning efficiency of the model.

[0032] 3. According to the output of the Mamba block model in step 2, the waveform is divided into waveform blocks of the same length L in sequence starting from the block starting position K, with the block size L as the step length, and the waveform before the block starting position K and the last waveform with a length less than L are discarded.

[0033] Calculate the thickness, fluctuation degree and mean characteristics of each waveform. The specific meanings of the above characteristics are as follows Figure 4 As shown in the figure, the thickness T of the block is the difference between the second largest value and the second smallest value in the waveform block, which represents the thickness of the waveform block; the mean E of the block is the mean of the waveform block after removing the maximum and minimum values, which represents the average height of the waveform block; the fluctuation degree D of the block is the mean of the absolute value of the gradient of the waveform block sequence value, which represents the fluctuation degree of the waveform block. Figure 4 The difference between waveform blocks with larger and smaller D values ​​is shown in the figure. It can be seen that the waveform blocks with smaller D values ​​appear sparser.

[0034] Assuming that the entire waveform can be divided into m blocks, each block size is L, the following will count the block thickness mean of all waveform blocks , Block thickness fluctuation standard deviation , block fluctuation mean , block mean standard deviation .

[0035] Their specific calculation methods are as follows: Block thickness mean : , block thickness fluctuation standard deviation : , block volatility mean : , block mean standard deviation : , so far all the time domain block features have been calculated.

[0036] 4. Use N times the block size L output by the Mamba block model in step 2 as a window (N is an integer, N is as large as possible and L * N does not exceed the total length of the waveform), perform fast Fourier transform on the waveform sequence in the window, and obtain an amplitude sequence of length L * N Since the amplitude sequence is centrally symmetric, the first L * N / 2 values ​​are retained.

[0037] Calculate the ratio of high and low frequency energy in the amplitude sequence , the energy concentration at the peak position of the integer multiple of the amplitude sequence N As frequency domain features. Their specific calculation methods are as follows: Ratio of high to low frequency energy : , , , spectral peak energy concentration : , , so far all the frequency domain features have been calculated.

[0038] 5. In the present invention, in order to achieve efficient classification of regression test error simulation waveforms, multiple eigenvalues ​​are combined into a eigenvector after feature extraction and input into a support vector machine (SVM) model for classification. Figure 5 As shown, the block size L and the block thickness mean , Block thickness fluctuation standard deviation , block fluctuation mean , block mean standard deviation , the ratio of high and low frequency energy , spectral peak energy concentration These seven eigenvalues ​​are combined into a 7-dimensional feature vector and input into the support vector machine. This 7-dimensional feature vector contains key information from the time domain to the frequency domain and can fully describe all aspects of the waveform. In this way, the complex information of the waveform is compressed and provided to the subsequent classification algorithm in the form of a numerical vector.

[0039] The present invention adopts RBF kernel (radial basis function kernel) as the kernel function of support vector machine. RBF kernel is a commonly used kernel function, which can effectively handle nonlinear classification problems. It improves the performance of the classifier by mapping the input data to a high-dimensional space, in which the nonlinear relationship between data points can be linearly segmented. In addition, the present invention adopts GridSearchCV combined with cross-validation to automatically search for the optimal hyperparameter combination to obtain the optimal SVM model. The final model obtains the classification result of the simulation waveform of the key signal of the regression test according to the input feature vector.

[0040] It should be noted that the above content only illustrates the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications all fall within the protection scope of the claims of the present invention.

Claims

1. A regression test simulation waveform classification method based on Mamba neural network, characterized in that: The following steps are involved: Step 1: parse the simulation waveform file obtained by regression testing and convert the key signal waveform into a discrete sequence of integer waveforms for subsequent processing and classification of the waveforms; Step 2: Normalize the numerical range of the waveform sequence to the interval of 0-1, intercept the fixed length and input it into the Mamba neural network model, and the network outputs two parameters: block size and block starting position; Step 3: Use the two characteristic parameters output by the network to divide the waveform, calculate the thickness, fluctuation degree and mean characteristics of each waveform, and calculate the overall characteristics; Step 4, using integer multiples of the waveform block size as a window, performing fast Fourier transform on the waveform and calculating the frequency domain features; Step 5: The features extracted from the above process are combined into a multi-dimensional feature vector and input into the support vector machine to output the final classification result.

2. The regression test simulation waveform classification method based on Mamba neural network according to claim 1, characterized in that: The method for processing the key signal waveform in step 1 is: read each data bit value of the key signal at the rising edge of the baseband clock, convert it into a decimal integer for storage, and repeat the above steps to obtain a discrete sequence of a key waveform signal.

3. The regression test simulation waveform classification method based on Mamba neural network according to claim 1 is characterized in that: The method for normalizing the waveform sequence in step 2 and inputting it into the Mamba neural network model is as follows: divide each value in the waveform sequence by the maximum value in the sequence to normalize each value in the sequence to the range of 0-1, truncate the waveform sequence to a length of 2048, and fill in 0 values ​​at the end of the sequence if the length is insufficient. Input the processed waveform sequence into the Mamba neural network model, and the model outputs two parameters: block size L and block starting position K.

4. The regression test simulation waveform classification method based on Mamba neural network according to claim 1 is characterized in that: The method of waveform segmentation and feature extraction in step three is: according to the output results of the model, start the waveform from the block starting position K, and divide it into waveform blocks of the same size according to the block size L, discard the waveform before the block starting position and the last waveform with a length less than L, calculate the thickness, fluctuation degree and mean characteristics of each waveform block, and based on the calculation results, statistically calculate the overall block thickness mean, block thickness fluctuation standard deviation, block fluctuation degree mean and block mean standard deviation.

5. The regression test simulation waveform classification method based on Mamba neural network according to claim 1 is characterized in that: The frequency domain feature calculation method in step 4 is: use N times the block size L of the model output as a window, perform fast Fourier transform on the waveform sequence in the window, obtain an amplitude sequence of length L*N, calculate the ratio of high and low frequency energy of the amplitude sequence, and the energy concentration at the spectral peak position of the integer multiple of the amplitude sequence N.

6. The regression test simulation waveform classification method based on Mamba neural network according to claim 1 is characterized in that: The method for the support vector machine in step five to process features and obtain classification results is as follows: the seven eigenvalues ​​of block size L, block thickness mean, block thickness fluctuation standard deviation, block fluctuation degree mean, block mean standard deviation, high-low frequency energy ratio, and spectral peak energy concentration are combined into a 7-dimensional feature vector, which is input into the support vector machine to finally obtain the output key signal waveform sequence classification result.

Images