Waveform verification method and device, model training method and device, equipment, and medium

Abstract

The present disclosure provides a waveform verification method and a training method of a model thereof, a device, equipment and a medium. The waveform verification method comprises: obtaining a time information sequence and an amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence and the amplitude information in the amplitude information sequence correspond one by one; taking the time information sequence as input data and the amplitude information sequence as expected output data, generating a training sample containing the input data and the expected output data; and training a recurrent neural network model to be trained based on the training sample by using a machine learning algorithm to obtain a waveform verification model. Thus, the accuracy of waveform verification can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of signal processing, and more specifically, to a waveform verification and its model training method, apparatus, device, and medium. Background Technology

[0002] Periodic signals are signals that repeat their previous trend of change at regular intervals over time. These signals have wide applications in control, audio, and communication. Common periodic signals include sine waves, square waves, sawtooth waves, triangle waves, and composite periodic signals formed by superimposing one or more of these.

[0003] Traditional methods for verifying digital periodic waveform signals in related technologies mainly focus on both the time and frequency domains, combining signal processing and mathematical analysis to extract and classify waveform features. Time-domain waveform verification methods include the RMS (root mean square) method and threshold comparison methods. The RMS method classifies and verifies waveforms by quantifying their energy characteristics and comparing the relationship between the root mean square and peak values. The threshold comparison method classifies and verifies waveforms based on the waveform amplitude distribution characteristics by statistically analyzing the proportion of sampling points within each threshold interval. Frequency-domain waveform verification primarily uses the FFT (Fast Fourier Transform) method, which converts the time-domain signal into a frequency-domain signal and then classifies and verifies the waveform by analyzing the spectral distribution and harmonic components. However, the accuracy of these digital periodic waveform signal verification methods is relatively low. Summary of the Invention

[0004] The main purpose of this disclosure is to provide methods, apparatus, equipment, and media for waveform verification and model training, so as to solve the technical problem of low accuracy of existing digital periodic waveform signal verification methods, and to achieve the technical effect of improving the accuracy of waveform verification.

[0005] To achieve the above objectives, a first aspect of this disclosure proposes a training method for a waveform verification model, comprising: Obtain the time information sequence and amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence; Using time information sequences as input data and amplitude information sequences as expected output data, training samples containing input data and expected output data are generated. A machine learning algorithm is used to train the recurrent neural network model to be trained based on training samples, resulting in a waveform verification model.

[0006] In some possible implementations, acquiring the time information sequence and amplitude information sequence of a digital periodic waveform signal includes: For each waveform type in a preset set of waveform types, obtain the time information sequence and amplitude information sequence of the digital periodic waveform signal of that waveform type to obtain the corresponding time information sequence and amplitude information sequence; and Using machine learning algorithms, a recurrent neural network model to be trained is trained based on training samples to obtain a waveform verification model, including: A machine learning algorithm is used to train a recurrent neural network model based on training samples determined by the time information sequence and amplitude information sequence corresponding to each waveform type in the waveform type set, thereby obtaining a waveform verification model.

[0007] In some possible implementations, acquiring the time information sequence and amplitude information sequence of a digital periodic waveform signal includes: For each waveform type in a preset set of waveform types, obtain the time information sequence and amplitude information sequence of the digital periodic waveform signal of that waveform type to obtain the corresponding time information sequence and amplitude information sequence; and Using machine learning algorithms, a recurrent neural network model to be trained is trained based on training samples to obtain a waveform verification model, including: A machine learning algorithm is used to train a recurrent neural network model based on training samples determined by the time information sequence and amplitude information sequence corresponding to each waveform type in the waveform type set, thereby obtaining a waveform verification model.

[0008] In some possible implementations, acquiring the time information sequence and amplitude information sequence of a digital periodic waveform signal includes: Determine whether the digital periodic waveform signal meets the preset generation conditions; In response to the digital periodic waveform signal meeting the preset generation conditions, the time information sequence and amplitude information sequence of the digital periodic waveform signal are generated based on the signal characteristics of the digital periodic waveform signal. In response to the digital periodic waveform signal not meeting the preset generation conditions, the digital periodic waveform signal output by the digital chip is acquired, and the time information sequence and amplitude information sequence of the digital periodic waveform signal are determined based on the digital periodic waveform signal output by the digital chip.

[0009] In some possible implementations, the preset generation conditions include at least one of the following: The signal noise of the digital periodic waveform signal is less than or equal to a preset noise threshold; The similarity between the signal features of a digital periodic waveform signal and the corresponding reference features of the digital periodic waveform signal is greater than or equal to a preset similarity threshold, wherein the reference features represent the waveform features of the waveform type to which the digital periodic waveform signal belongs.

[0010] In some possible implementations, a machine learning algorithm is used to train the recurrent neural network model to be trained based on training samples to obtain a waveform verification model, including: Machine learning algorithms are used to adjust the model parameters of the recurrent neural network model to be trained based on training samples; Determine whether the accuracy of the recurrent neural network model after parameter adjustment is greater than or equal to the preset accuracy threshold; In response to a precision greater than or equal to a precision threshold, the recurrent neural network model with adjusted model parameters is determined as the waveform verification model.

[0011] In some possible implementations, the recurrent neural network model includes a long short-term memory network.

[0012] Secondly, embodiments of this disclosure provide a waveform verification method, including: Acquire the signal to be verified output by the digital chip; The time information sequence of the signal to be verified is input into the pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model. The waveform verification model is trained using the method described in the first aspect above. Based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model, it is determined whether the signal to be verified passes the waveform verification.

[0013] In some possible implementations, determining whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model includes: Determine the error information between the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model; Based on the relationship between the error represented by the error information and the preset error threshold, it is determined whether the digital periodic waveform signal passes waveform verification.

[0014] Thirdly, embodiments of this disclosure provide a training apparatus for a waveform verification model, comprising: The first acquisition unit is configured to acquire the time information sequence and amplitude information sequence of the digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence; The generation unit is configured to take the time information sequence as input data and the amplitude information sequence as the expected output data, and generate training samples containing the input data and the expected output data. The training unit is configured to use a machine learning algorithm to train the recurrent neural network model to be trained based on the training samples, and obtain a waveform verification model.

[0015] In some possible implementations, acquiring the time information sequence and amplitude information sequence of a digital periodic waveform signal includes: For each waveform type in a preset set of waveform types, obtain the time information sequence and amplitude information sequence of the digital periodic waveform signal of that waveform type to obtain the corresponding time information sequence and amplitude information sequence; and Using machine learning algorithms, a recurrent neural network model to be trained is trained based on training samples to obtain a waveform verification model, including: A machine learning algorithm is used to train a recurrent neural network model based on training samples determined by the time information sequence and amplitude information sequence corresponding to each waveform type in the waveform type set, thereby obtaining a waveform verification model.

[0016] In some possible implementations, acquiring the time information sequence and amplitude information sequence of a digital periodic waveform signal includes: Determine whether the digital periodic waveform signal meets the preset generation conditions; In response to the digital periodic waveform signal meeting the preset generation conditions, the time information sequence and amplitude information sequence of the digital periodic waveform signal are generated based on the signal characteristics of the digital periodic waveform signal. In response to the digital periodic waveform signal not meeting the preset generation conditions, the digital periodic waveform signal output by the digital chip is acquired, and the time information sequence and amplitude information sequence of the digital periodic waveform signal are determined based on the digital periodic waveform signal output by the digital chip.

[0017] In some possible implementations, the preset generation conditions include at least one of the following: The signal noise of the digital periodic waveform signal is less than or equal to a preset noise threshold; The similarity between the signal features of a digital periodic waveform signal and the corresponding reference features of the digital periodic waveform signal is greater than or equal to a preset similarity threshold, wherein the reference features represent the waveform features of the waveform type to which the digital periodic waveform signal belongs.

[0018] In some possible implementations, a machine learning algorithm is used to train the recurrent neural network model to be trained based on training samples to obtain a waveform verification model, including: Machine learning algorithms are used to adjust the model parameters of the recurrent neural network model to be trained based on training samples; Determine whether the accuracy of the recurrent neural network model after parameter adjustment is greater than or equal to the preset accuracy threshold; In response to a precision greater than or equal to a precision threshold, the recurrent neural network model with adjusted model parameters is determined as the waveform verification model.

[0019] In some possible implementations, the recurrent neural network model includes a long short-term memory network.

[0020] Fourthly, embodiments of this disclosure provide a waveform verification apparatus, comprising: The second acquisition unit is configured to acquire the signal to be verified output by the digital chip; The input unit is configured to: input the time information sequence of the signal to be verified into a pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model, wherein the waveform verification model is trained using any of the methods described in the first aspect above; The verification unit is configured to determine whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model.

[0021] In some possible implementations, determining whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model includes: Determine the error information between the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model; Based on the relationship between the error represented by the error information and the preset error threshold, it is determined whether the digital periodic waveform signal passes waveform verification.

[0022] Fifthly, embodiments of this disclosure provide an electronic device, including: Memory, used to store computer programs; A processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, it implements the training method for waveform verification of the first aspect of the present disclosure or the method for waveform verification of the second aspect of the present disclosure, in any one of the embodiments.

[0023] In a sixth aspect, embodiments of this disclosure provide a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, it implements the training method for waveform verification of the first aspect of this disclosure or the method of waveform verification of the second aspect of this disclosure in any one of the embodiments.

[0024] In a seventh aspect, embodiments of this disclosure provide a computer program including computer-readable code, wherein when the computer program instructions are executed by a processor, they implement the training method for waveform verification of the first aspect of this disclosure or the method for waveform verification of the second aspect of this disclosure in any one of the embodiments.

[0025] The technical solutions provided by the embodiments of this disclosure may include the following beneficial effects: In this disclosure, training samples are generated based on the time information sequence and amplitude information sequence of a digital periodic waveform signal. Then, a machine learning algorithm is used to train a recurrent neural network model to be trained based on the training samples, thereby obtaining a waveform verification model and achieving the technical effect of improving the accuracy of waveform verification. Attached Figure Description

[0026] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of the disclosure and to make other features, objects, and advantages of the disclosure more apparent. The illustrative embodiments of the disclosure, along with their descriptions, are used to explain the disclosure and do not constitute an undue limitation thereof. In the drawings: Figure 1 A flowchart illustrating a training method for a waveform verification model provided in this embodiment of the disclosure; Figure 2 A flowchart of another training method for a waveform verification model provided in an embodiment of this disclosure; Figure 3 A flowchart of a waveform verification method provided in this embodiment of the disclosure; Figure 4 is a comparison diagram of the amplitude information sequences predicted before and after training of a waveform verification model provided in an embodiment of this disclosure; Figure 5 A schematic diagram of the structure of a training device for a waveform verification model provided in an embodiment of this disclosure; Figure 6 This is a schematic diagram of the structure of a waveform verification device provided in an embodiment of the present disclosure; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present disclosure, the technical solutions of the present disclosure will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present disclosure, and not all embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present disclosure.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this disclosure described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] In this disclosure, the terms "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings. These terms are primarily for the purpose of better describing this disclosure and its embodiments, and are not intended to limit the indicated devices, elements, or components to having a specific orientation, or to be constructed and operated in a specific orientation.

[0030] Furthermore, in addition to indicating location or positional relationship, some of the aforementioned terms may also have other meanings. For example, the term "above" may also be used in certain circumstances to indicate a dependency or connection. Those skilled in the art can understand the specific meaning of these terms in this disclosure according to the specific circumstances.

[0031] Furthermore, the terms "installation," "setup," "equipped with," "connection," "linked," and "socketing" should be interpreted broadly. For example, "connection" can be a fixed connection, a detachable connection, or an integral structure; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of the above terms in this disclosure according to the specific circumstances.

[0032] Figure 1This is a flowchart illustrating a training method for a waveform verification model provided in an embodiment of this disclosure. This method can be applied to one or more electronic devices such as smartphones, laptops, desktop computers, portable computers, and servers. Furthermore, the execution entity of this method can be hardware or software. When the execution entity is hardware, it can be one or more of the aforementioned electronic devices. For example, a single electronic device can execute this method, or multiple electronic devices can cooperate with each other to execute this method. When the execution entity is software, this method can be implemented as multiple software programs or software modules, or as a single software program or software module. No specific limitations are made here.

[0033] like Figure 1 As shown, the method specifically includes: Step 101: Obtain the time information sequence and amplitude information sequence of the digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence.

[0034] In this embodiment, the digital periodic waveform signal can be a digital signal that repeats the same waveform at fixed intervals. Furthermore, the digital periodic waveform signal can represent the correspondence between multiple time information and multiple amplitude information. As an example, the digital periodic waveform signal can include at least one of the following types of waveform signals: sine wave, square wave, sawtooth wave, triangle wave, and composite periodic waveform (e.g., a waveform signal formed by superimposing at least one of the following waveform signals: sine wave, square wave, sawtooth wave, and triangle wave).

[0035] A time information sequence can be multiple time information items arranged in chronological order. The time information in a time information sequence can represent the sampling moments of a digital periodic waveform signal. Each time information item in a time information sequence typically represents a different sampling moment. As an example, a time information sequence can be a sequence consisting of "100ms, 200ms, 300ms...". The time information in the above time information sequence can be one of the following: 100ms, 200ms, or 300ms.

[0036] An amplitude information sequence can be multiple amplitude values ​​arranged in chronological order. The amplitude values ​​in the sequence can represent any of the following: voltage, current, level, temperature, etc. The amplitude represented by each value in the sequence typically (but not necessarily) changes according to a fixed period. For example, an amplitude information sequence can be a sequence of "0.5V, 0.8V, 0.5V..." where the amplitude values ​​can be one of the following: 0.5V, 0.8V, 0.5V.

[0037] Here, the time information in the time information sequence corresponds to the amplitude information at the corresponding position in the amplitude information sequence. For example, for a digital periodic waveform signal with a waveform type of sine wave, <10ns, 5.9V>, <20ns, 9.5V>, <30ns, 9.5V>, <40ns, 5.9V>, <50ns, 0V>, the time information sequence can be represented as "10ns, 20ns, 30ns, 40ns, 50ns", and the amplitude information sequence as "5.9V, 9.5V, 9.5V, 5.9V, 0V". The two correspond one-to-one according to their positions; that is, the time information "10ns" corresponds to the amplitude information "5.9V", indicating that the amplitude at 10nm is 5.9V; the time information "20ns" corresponds to the amplitude information "9.5V", indicating that the amplitude at 20ns is 9.5V, and so on.

[0038] In some alternative implementations, scripting tools such as Python and Perl, used to generate digital periodic waveform signals, can be employed to generate sequences containing <time information and amplitude information>, thereby obtaining the time information sequence and amplitude information sequence of the digital periodic waveform signal. Alternatively, the digital periodic waveform signal output from an actual digital system (such as a digital chip) can be acquired to obtain the time information sequence and amplitude information sequence of the digital periodic waveform signal.

[0039] Step 102: Using the time information sequence as input data and the amplitude information sequence as expected output data, generate training samples containing the input data and expected output data.

[0040] In this embodiment, training samples are used to associate input data and expected output data. During the training process of the waveform verification model, the model can be obtained through training with a large number of training samples. As an example, one of the training samples can be "10ns, 20ns, 30ns, 40ns, 50ns; 5.9V, 9.5V, 9.5V, 5.9V, 0V". The input data can be "10ns, 20ns, 30ns, 40ns, 50ns", and the expected output data can be "5.9V, 9.5V, 9.5V, 5.9V, 0V".

[0041] Step 103: Using a machine learning algorithm, the recurrent neural network model to be trained is trained based on the training samples to obtain the waveform verification model.

[0042] In this embodiment, the machine learning algorithm can be used to adjust the model parameters of the recurrent neural network model to be trained. As an example, the machine learning algorithm may include, but is not limited to, at least one of the following: gradient descent, stochastic gradient descent, and Adam optimization algorithm.

[0043] The recurrent neural network (RNN) model to be trained can be an untrained RNN or a trained RNN model that has not yet met the training termination condition. As an example, the recurrent neural network model can include, but is not limited to, at least one of the following: Simple RNN, Gated Recurrent Unit (GRU).

[0044] In some optional implementations of this embodiment, the recurrent neural network model includes a Long Short-Term Memory (LSTM) network.

[0045] Here, Long Short-Term Memory (LSTM) networks can control the flow of information through gating mechanisms such as forget gates, input gates, and output gates. They can capture dependencies in long sequences of data, solving the gradient explosion or vanishing gradient problems of traditional RNNs. As an example, an LSM network can include memory units, hidden state units, and forget gate units.

[0046] It is understandable that among the above-mentioned optional implementation methods, LSTM can solve the gradient explosion or gradient vanishing problems in long sequence data processing compared to traditional RNNs. It can more accurately capture the long-distance dependencies of the time information sequence of digital periodic waveform signals (such as the amplitude correlation at different time points within the waveform period). For example, when processing long-period composite waveforms, traditional RNNs may fail to learn the amplitude correlation between early and later time points due to gradient problems, while LSTM can retain key historical information through gating mechanisms, enabling the Long Short-Term Memory network to learn the "time information-amplitude information" pattern more accurately. Therefore, waveform verification models trained based on LSTM have higher accuracy in predicting amplitude information sequences when processing long-time-series digital periodic waveform signals, and their performance is superior to traditional RNN models, which can further improve the reliability of waveform verification.

[0047] The waveform verification model can be a recurrent neural network model that meets the training termination condition after training.

[0048] Training termination conditions can be used to determine whether to end the training process. As an example, training termination conditions may include, but are not limited to, at least one of the following: the number of training epochs has reached a preset number of epochs, the loss function value has converged to a threshold range, and the error (e.g., relative error, absolute error) between the predicted time information sequence and the actual time information sequence is lower than a preset error.

[0049] In some optional implementations, the input data and expected output data from the training samples can be input into the recurrent neural network model to be trained. The actual output data (i.e., the predicted amplitude information sequence) of the recurrent neural network model to be trained can be obtained through forward propagation. The loss function value between the actual output data and the expected output data can be calculated. The model parameters of the recurrent neural network model to be trained can be updated based on the loss function value using the backpropagation algorithm. Subsequently, the forward and backpropagation processes are repeated based on the updated model parameters until the above training termination condition is met, and a waveform verification model is obtained.

[0050] In some optional implementations of this embodiment, the time information sequence and amplitude information sequence of the digital periodic waveform signal can be obtained in the following manner: For each waveform type in the preset waveform type set, the time information sequence and amplitude information sequence of the digital periodic waveform signal of that waveform type are obtained, thus obtaining the time information sequence and amplitude information sequence corresponding to that waveform type.

[0051] As an example, the preset waveform type set can be a collection of pre-defined waveform types that include various digital periodic waveform signals. A waveform type can be a category of digital periodic waveform signals. For example, the preset waveform type set may include sine waves, square waves, sawtooth waves, triangle waves, trapezoidal waves, and composite periodic waveforms. Typically, for digital periodic waveform signals of different waveform types, the amplitude information sequences corresponding to the same time information sequence will differ.

[0052] The digital periodic waveform signal of this waveform type is a digital periodic waveform signal belonging to a certain waveform type in the preset waveform type set.

[0053] Based on this, the waveform validation model can be obtained by using machine learning algorithms and training samples to train the recurrent neural network model to be trained: A machine learning algorithm is used to train a recurrent neural network model based on training samples determined by the time information sequence and amplitude information sequence corresponding to each waveform type in the waveform type set, thereby obtaining a waveform verification model.

[0054] In some optional implementations, each waveform type in the waveform type set can correspond to at least one or more digital periodic waveform signals. When each waveform type corresponds to multiple digital periodic waveform signals, these signals can be the same or different. This allows for the acquisition of multiple training samples containing time information sequences and amplitude information sequences corresponding to each waveform type in the waveform type set. Furthermore, a machine learning algorithm can be used to train a recurrent neural network model based on these training samples to obtain a waveform validation model.

[0055] It is understandable that in the above optional implementation methods, by generating corresponding training samples for each waveform type in the preset waveform type set, the training samples cover the "time information-amplitude information" patterns of multiple different waveform types. Based on the training samples corresponding to these multiple different waveform types, the recurrent neural network model can be trained to learn the characteristics of digital periodic waveform signals of multiple different waveform types. Compared with training with only a single waveform type, this can improve the adaptability of the waveform verification model to the signals to be verified of different waveform types, and the number of verifiable waveform types is greater, the versatility is higher, and the limitations of waveform verification caused by a single waveform type are reduced.

[0056] In some optional implementations of this embodiment, the waveform verification model can be obtained by using a machine learning algorithm to train the recurrent neural network model to be trained based on training samples, in the following manner: The first step is to use machine learning algorithms to adjust the model parameters of the recurrent neural network model to be trained based on the training samples.

[0057] The model parameters may include, but are not limited to, at least one of the following: weight parameters and bias parameters. As an example, in the case of an LSTM recurrent neural network model, the model parameters may include: the weight matrix and bias vector of the forget gate, input gate, and output gate.

[0058] In some alternative implementations, machine learning algorithms can be used to calculate the error (such as mean square error) between the actual output data and the expected output data of the recurrent neural network model to be trained, and the model parameters of the recurrent neural network model to be trained can be modified according to the direction and magnitude of the deviation to reduce the error.

[0059] The second step is to determine whether the accuracy of the recurrent neural network model after the model parameters have been adjusted is greater than or equal to the preset accuracy threshold.

[0060] The aforementioned accuracy can be the degree of matching between the predicted amplitude information sequence output by the recurrent neural network model after adjusting the model parameters and the expected output data (i.e., the actual amplitude information sequence). As an example, model accuracy can be determined based on at least one of the following: error rate (e.g., absolute error, relative error) or accuracy.

[0061] The preset accuracy threshold can be used to determine whether the accuracy of the recurrent neural network model meets the standard.

[0062] The third step is to determine the recurrent neural network model with adjusted model parameters as the waveform verification model if the accuracy is greater than or equal to the accuracy threshold.

[0063] Here, when the accuracy is greater than or equal to the accuracy threshold, the recurrent neural network model with adjusted model parameters (i.e., the recurrent neural network model with accuracy greater than or equal to the accuracy threshold) can be determined as the waveform verification model.

[0064] It is understandable that, among the above optional implementation methods, determining the recurrent neural network model with the accuracy target (i.e., the accuracy is greater than or equal to the preset accuracy threshold) as the waveform verification model can ensure that the final waveform verification model has a more stable and reliable prediction ability. Therefore, when the waveform verification model is used for waveform verification, the accuracy of waveform verification can be improved.

[0065] It should be noted that, where there is no conflict, the technical features described in different alternative implementations can be included in the same embodiment. For the sake of brevity, they will not be elaborated here.

[0066] Based on the embodiments of this disclosure, training samples are generated by the time information sequence and amplitude information sequence of digital periodic waveform signals. Then, a machine learning algorithm is used to train the recurrent neural network model to be trained based on the training samples, thereby obtaining a waveform verification model. Thus, when the waveform verification model is used for waveform verification, the accuracy of waveform verification can be improved.

[0067] Figure 2 This is a flowchart illustrating another training method for a waveform verification model provided in an embodiment of this disclosure. Figure 2 As shown, the method specifically includes: Step 201: Determine whether the digital periodic waveform signal meets the preset generation conditions. If yes, proceed to step 202; otherwise, proceed to step 203.

[0068] In this embodiment, the preset generation conditions can be used to determine whether a digital periodic waveform signal can be directly generated by a script tool (specifically, by generating a time information sequence and an amplitude information sequence to generate a digital periodic waveform signal). As an example, the preset generation conditions may include at least one of the following: no noise or low noise, and the periodic characteristics can be described by a mathematical expression.

[0069] In some optional implementations of this embodiment, the preset generation conditions may include at least one of the following: The first requirement is that the signal noise of the digital periodic waveform signal is less than or equal to a preset noise threshold.

[0070] Among them, the signal noise of a digital periodic waveform signal can be the intensity of the interference signal that affects the purity of the digital periodic waveform signal.

[0071] The preset noise threshold can be used to determine whether a digital periodic waveform signal can be directly generated by a script tool (specifically, by generating a time information sequence and an amplitude information sequence to generate a digital periodic waveform signal).

[0072] The second item is that the similarity between the signal features of the digital periodic waveform signal and the corresponding reference features of the digital periodic waveform signal is greater than or equal to a preset similarity threshold, wherein the reference features represent the waveform features of the waveform type to which the digital periodic waveform signal belongs.

[0073] The signal characteristics may include, but are not limited to, at least one of the following: peak value, valley value, period duration, etc.

[0074] The reference characteristics for a digital periodic waveform signal can be the standard characteristic parameters of the waveform type to which the digital periodic waveform signal belongs (e.g., a sine wave). As an example, the reference characteristics for a sine wave may include: peak value 1V, frequency 50Hz, and phase 0rad; as another example, the reference characteristics for a digital square wave may include: peak value 0.8V, frequency 100Hz, and duty cycle 50%.

[0075] Similarity can be defined as the degree of similarity between the signal characteristics of a digital periodic waveform signal and the characteristics of a reference signal. As an example, similarity can be expressed as a percentage or a numerical value in the range of 0-1.

[0076] A preset similarity threshold can be used to determine whether a digital periodic waveform signal can be directly generated by a script tool (specifically, by generating a time information sequence and an amplitude information sequence to generate the digital periodic waveform signal). As an example, the preset similarity threshold can be 95%, 90%, etc.

[0077] It is understandable that, among the above-mentioned optional implementation methods, the training sample generation strategy is determined by at least one of the following: signal noise, signal features, and similarity to benchmark features. For example, if the signal noise is less than or equal to a preset threshold, it indicates that the signal purity is high, and the training samples generated based on the directly generated sequences (time information sequences and amplitude information sequences) are relatively accurate; if the feature similarity meets the standard, it indicates that the signal is close to the standard waveform, and the generated sequences (time information sequences and amplitude information sequences) have little deviation from the actual waveform. Therefore, it is determined that using either "generating sequences" or "acquiring signals" to generate training samples can, to some extent, balance the efficiency and accuracy of training sample generation.

[0078] Step 202: Based on the signal characteristics of the digital periodic waveform signal, generate the time information sequence and amplitude information sequence of the digital periodic waveform signal. Then, continue to execute step 204.

[0079] In this embodiment, when the digital periodic waveform signal meets the preset generation conditions, the time information sequence and amplitude information sequence of the digital periodic waveform signal can be generated based on the signal characteristics of the digital periodic waveform signal.

[0080] Signal characteristics may include, but are not limited to, at least one of the following: peak value, valley value, period duration, etc.

[0081] The time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence.

[0082] In some alternative implementations, scripting tools such as Python and Perl can be used to generate digital periodic waveform signals. Based on the signal characteristics of the digital periodic waveform signal, a sequence containing <time information and amplitude information> can be generated, thereby obtaining the time information sequence and amplitude information sequence of the digital periodic waveform signal.

[0083] Step 203: Acquire the digital periodic waveform signal output by the digital chip. Based on the digital periodic waveform signal output by the digital chip, determine the time information sequence and amplitude information sequence of the digital periodic waveform signal. Then, continue to execute step 204.

[0084] In this embodiment, when acquiring the digital periodic waveform signal output by the digital chip, the time information sequence and amplitude information sequence of the digital periodic waveform signal can be determined based on the actual digital periodic waveform signal output by the digital chip.

[0085] Step 204: Using the time information sequence as input data and the amplitude information sequence as the expected output data, generate training samples containing both input data and expected output data. Then, proceed to step 205.

[0086] In this embodiment, step 204 and Figure 1 Step 102 in the corresponding embodiment is basically the same, and will not be repeated here.

[0087] Step 205: Using a machine learning algorithm, the recurrent neural network model to be trained is trained based on the training samples to obtain the waveform verification model.

[0088] In this embodiment, step 205 and Figure 1 Step 103 in the corresponding embodiment is basically the same, and will not be repeated here.

[0089] It should be noted that, in addition to the contents described above, this embodiment may also include... Figure 1 The corresponding technical features described in the corresponding embodiments, thereby achieving Figure 1 The waveform shown verifies the technical effectiveness of the training method for the model. Please refer to [link / reference needed]. Figure 1 The relevant descriptions are presented concisely and will not be elaborated upon here.

[0090] Based on the embodiments of this disclosure, when the preset generation conditions are met, the time information sequence and amplitude information sequence of the digital periodic waveform signal are generated based on the signal characteristics of the digital periodic waveform signal. This eliminates the need for signal acquisition, saving acquisition time and cost, and the generated training samples have high accuracy. When the preset generation conditions are not met, the actual output signal of the digital chip is acquired and the training samples are determined, ensuring that training samples can still be obtained even in scenarios with high signal noise and unclear features. This ensures that the waveform verification model adapts to different scenario requirements, improving the model's generalization ability and verification accuracy.

[0091] Optionally, the ease of generating a digital periodic waveform signal can be determined based on the noise signal of the digital periodic waveform signal, a preset noise threshold, and the similarity between the signal characteristics of the digital periodic waveform signal and the reference characteristics corresponding to the digital periodic waveform signal.

[0092] As an example, the difference between a preset noise threshold and the noise level represented by the noise signal can be calculated. Then, based on this difference and the aforementioned similarity, the difficulty of generating a digital periodic waveform signal can be determined. Different methods can be used to obtain training samples corresponding to the digital periodic waveform signal for different levels of difficulty. For example, for digital periodic waveform signals that are difficult to generate, the corresponding training samples can be obtained using the method described in step 203 above; for digital periodic waveform signals that are easy to generate, the corresponding training samples can be obtained using the method described in step 202 above. Furthermore, the aforementioned difficulty level can be represented numerically. For example, a larger value indicates that it is more difficult to generate the digital periodic waveform signal. Therefore, based on the numerical value representing the difficulty level, the proportion of training samples obtained using the methods described in steps 202 and 203 above can be determined. For example, if the difficulty level after normalization is "1", indicating that the digital periodic waveform signal is difficult to generate, then step 203 is used to obtain all training samples; if the difficulty level after normalization is "0", indicating that the digital periodic waveform signal is easy to generate, then step 202 is used to obtain all training samples; if the difficulty level after normalization is "0.5", indicating that the difficulty of generating the digital periodic waveform signal is moderate, then step 203 is used to generate 50% of the training samples, and step 202 is used to generate the other 50% of the training samples. Then, the recurrent neural network to be trained is trained based on the generated training samples to obtain the waveform validation model.

[0093] Figure 3 This is a flowchart illustrating a waveform verification method provided in an embodiment of this disclosure. This method can be applied to one or more electronic devices such as smartphones, laptops, desktop computers, portable computers, and servers. Furthermore, the execution entity of this method can be hardware or software. When the execution entity is hardware, it can be one or more of the aforementioned electronic devices. For example, a single electronic device can execute this method, or multiple electronic devices can cooperate with each other to execute this method. When the execution entity is software, this method can be implemented as multiple software programs or software modules, or as a single software program or software module. No specific limitations are made here.

[0094] like Figure 3 As shown, the method specifically includes: Step 301: Obtain the signal to be verified output by the digital chip.

[0095] In this embodiment, the signal to be verified can be a signal actually output by a digital chip. Through subsequent waveform verification, it can be determined whether it belongs to a waveform type in a preset waveform type set.

[0096] Step 302: Input the time information sequence of the signal to be verified into the pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model.

[0097] In this embodiment, the waveform verification model is trained using any of the above-mentioned waveform verification model training methods.

[0098] Step 303: Based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model, determine whether the signal to be verified passes the waveform verification.

[0099] In this embodiment, waveform verification can be used to determine whether the signal actually output by the digital chip belongs to a preset waveform type set.

[0100] In some optional implementations of this embodiment, the following method can be used to determine whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model: The first step is to determine the error information between the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model.

[0101] The aforementioned error information can represent the error between the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model. As an example, the aforementioned error information may include, but is not limited to, at least one of the following: absolute error, mean square error, average absolute error, and maximum deviation. As yet another example, the aforementioned error information can be calculated based on at least one of the above.

[0102] The second step is to determine whether the digital periodic waveform signal passes waveform verification based on the relationship between the error represented by the error information and the preset error threshold.

[0103] The error threshold represents the critical value at which a digital periodic waveform signal passes waveform verification. If the error indicated by the error information is greater than the preset error threshold, the digital periodic waveform signal is determined to have failed waveform verification; if the error indicated by the error information is less than or equal to the preset error threshold, the digital periodic waveform signal is determined to have passed waveform verification.

[0104] It is understandable that determining whether a digital periodic waveform signal passes waveform verification by using the magnitude relationship between the error represented by the error information and the preset error threshold can improve the efficiency and accuracy of waveform verification.

[0105] Based on the embodiments of this disclosure, training samples are generated based on the time information sequence and amplitude information sequence of a digital periodic waveform signal. Then, a machine learning algorithm is used to train the recurrent neural network model to be trained based on the training samples, thereby obtaining a waveform verification model. This waveform verification model is then used for waveform verification, thereby improving the accuracy of waveform verification.

[0106] An exemplary embodiment of this disclosure will be described below. However, it should be noted that the following content is only used to understand the technical solutions of the embodiments of this disclosure and does not constitute a limitation on the protection scope of the embodiments of this disclosure.

[0107] In realizing this disclosure, the inventors discovered that RNNs, due to their recurrent network structure's ability to store, remember, and utilize previous information, possess unique advantages in processing sequential data and are widely used in tasks such as speech recognition and natural language processing. However, as the length of sequential data continues to increase, RNNs frequently encounter gradient explosion or vanishing gradient problems during processing, making it difficult for the network structure to learn long-distance dependencies.

[0108] Furthermore, the inventors discovered that LSTM, by introducing a gating mechanism, can control the flow of information, allowing the network structure to selectively remember or forget information. This enables it to specifically and effectively capture long-distance dependencies, solving the gradient explosion or vanishing problem encountered by RNNs when processing long sequence data. A standard LSTM typically includes several key components: a memory unit, a hidden state unit, a forget gate unit, an input gate unit, a candidate memory unit, and an output gate unit. The memory unit stores long-term memory, the hidden state unit transmits short-term memory, the forget gate unit determines how much past information is forgotten, the input gate unit determines how much new information is received, the candidate memory unit updates the memory unit based on new information, and the output gate unit determines how much information is output from the memory unit.

[0109] In view of this, the embodiments of this disclosure acquire digital periodic waveform signals of commonly used waveform types such as sine waves, square waves, sawtooth waves, triangular waves, and composite periodic waveforms, then train them in an LSTM model to generate a corresponding neural network model, and finally call it in the verification project of the digital chip to compare the amplitude information sequence of the actually acquired signal with the amplitude information sequence predicted by the LSTM neural network model, thereby realizing the verification of the digital periodic waveform signal.

[0110] The waveform types of digital periodic waveform signals in digital chips include sine waves, square waves, sawtooth waves, triangular waves, and composite periodic waveforms formed by superimposing one or more of these. When verifying these waveforms, the following two cases can be divided according to the strength of the noise or the ease of generating the periodic waveform sequence: In scenario one, when there is no noise, very little noise, or periodic waveform sequences are easily generated, the verification of digital periodic waveform signals can proceed as follows: First, obtain an ideal digital periodic waveform signal. This can be achieved by automatically generating a sequence of <time information, amplitude information> (i.e., the aforementioned time information sequence and amplitude information sequence) based on signal characteristics such as peak value, frequency, and phase, using a script tool. Then, divide these sequence sets proportionally into a training set (including training samples) and a test set (including test samples). Next, build an LSTM neural network model (i.e., the recurrent neural network model to be trained). Train the LSTM neural network model using the training set and test the trained LSTM neural network model using the test set. Once the testing accuracy is met, the signal can be called upon in the verification project. Compare the actual amplitude information sequence of the collected data (i.e., the signal to be verified output by the digital chip) with the data predicted by the LSTM neural network model (i.e., the amplitude information sequence predicted by the waveform verification model), thereby completing the verification of the digital periodic waveform signal.

[0111] Here, taking a periodic ideal sine wave as an example, a set of <time information, amplitude information> sequences can be generated using script tools based on the actual peak value, frequency, and phase of the sine wave in the project. The waveforms of the LSTM neural network model at the beginning of training and after training meets the test accuracy are as follows: Figure 4A and Figure 4B As shown.

[0112] In scenario two, when there is significant noise or the periodic waveform sequence is difficult to generate, the verification of the digital periodic waveform signal can proceed as follows: First, build an LSTM neural network model. Then, continuously input the actual acquired data (i.e., the digital periodic waveform signal output by the digital chip) into the neural network model (i.e., the recurrent neural network model to be trained) for training. When the matching degree (e.g., accuracy) between the actual acquired data and the data predicted by the LSTM neural network model reaches a threshold, the training is considered complete. Subsequently, the amplitude information sequence of the actual acquired data (i.e., the signal to be verified) is compared with the data predicted by the LSTM neural network model (i.e., the amplitude information sequence predicted by the waveform verification model), thereby completing the waveform verification of the digital periodic waveform signal.

[0113] Unlike the process described in Scenario 1 above, this process directly uses the digital periodic waveform signals collected from the project to be verified to train the LSTM neural network model. Training is considered complete only when the matching degree between the amplitude information sequence of the actually collected data (i.e., the signal to be verified) and the data predicted by the LSTM neural network model (i.e., the amplitude information sequence predicted by the waveform verification model) reaches a threshold (i.e., the error is lower than a preset error). Subsequent comparison and verification can proceed in the same manner as described above. Here, if the data matching degree consistently fails to reach the threshold, it indicates a problem with the digital periodic waveform signals in the project to be verified, resulting in the LSTM neural network model failing to extract periodic features, and thus the verification can be directly deemed unsuccessful.

[0114] It should be noted that, in addition to the contents described above, this embodiment may also include the technical features described in the above embodiments, thereby achieving the technical effects of the waveform verification model training method or waveform verification method shown above. Please refer to the above description for details. For the sake of brevity, it will not be elaborated here.

[0115] This disclosure proposes a verification method for digital periodic waveform signals based on LSTM neural networks. It can automatically extract waveform features using machine learning, making the identification of different digital periodic waveform signals more accurate. In particular, it is more accurate and flexible than traditional time-domain and frequency-domain verification methods for verifying periodic waveforms with noise interference, waveform distortion, or complex superposition.

[0116] Figure 5 This is a schematic diagram of the structure of a training device for a waveform verification model provided in an embodiment of this disclosure. Specifically, it includes: The first acquisition unit 501 is configured to acquire the time information sequence and amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence. The generation unit 502 is configured to: take the time information sequence as input data and the amplitude information sequence as expected output data, and generate training samples containing the input data and the expected output data. Training unit 503 is configured to: use machine learning algorithms to train the recurrent neural network model to be trained based on training samples to obtain a waveform verification model.

[0117] In some possible implementations, acquiring the time information sequence and amplitude information sequence of a digital periodic waveform signal includes: For each waveform type in a preset set of waveform types, obtain the time information sequence and amplitude information sequence of the digital periodic waveform signal of that waveform type to obtain the corresponding time information sequence and amplitude information sequence; and Using machine learning algorithms, a recurrent neural network model to be trained is trained based on training samples to obtain a waveform verification model, including: A machine learning algorithm is used to train a recurrent neural network model based on training samples determined by the time information sequence and amplitude information sequence corresponding to each waveform type in the waveform type set, thereby obtaining a waveform verification model.

[0118] In some possible implementations, acquiring the time information sequence and amplitude information sequence of a digital periodic waveform signal includes: Determine whether the digital periodic waveform signal meets the preset generation conditions; In response to the digital periodic waveform signal meeting the preset generation conditions, the time information sequence and amplitude information sequence of the digital periodic waveform signal are generated based on the signal characteristics of the digital periodic waveform signal. In response to the digital periodic waveform signal not meeting the preset generation conditions, the digital periodic waveform signal output by the digital chip is acquired, and the time information sequence and amplitude information sequence of the digital periodic waveform signal are determined based on the digital periodic waveform signal output by the digital chip.

[0119] In some possible implementations, the preset generation conditions include at least one of the following: The signal noise of the digital periodic waveform signal is less than or equal to a preset noise threshold; The similarity between the signal features of a digital periodic waveform signal and the corresponding reference features of the digital periodic waveform signal is greater than or equal to a preset similarity threshold, wherein the reference features represent the waveform features of the waveform type to which the digital periodic waveform signal belongs.

[0120] In some possible implementations, a machine learning algorithm is used to train the recurrent neural network model to be trained based on training samples to obtain a waveform verification model, including: Machine learning algorithms are used to adjust the model parameters of the recurrent neural network model to be trained based on training samples; Determine whether the accuracy of the recurrent neural network model after parameter adjustment is greater than or equal to the preset accuracy threshold; In response to a precision greater than or equal to a precision threshold, the recurrent neural network model with adjusted model parameters is determined as the waveform verification model.

[0121] In some possible implementations, the recurrent neural network model includes a long short-term memory network.

[0122] The waveform verification model training device provided in this embodiment can execute the corresponding steps of the waveform verification model training methods described above, thereby achieving the technical effects of the waveform verification model training methods described above. The waveform verification model training device and the waveform verification model training methods can refer to and cite each other in terms of specific implementation and technical effects. For the sake of brevity, they will not be elaborated here.

[0123] Figure 6 This is a schematic diagram of the structure of a waveform verification device provided in an embodiment of this disclosure. Specifically, it includes: The second acquisition unit 601 is configured to acquire the verification signal output by the digital chip; The input unit 602 is configured to: input the time information sequence of the signal to be verified into a pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model, wherein the waveform verification model is trained using any of the above-mentioned waveform verification model training methods; The verification unit 603 is configured to determine whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model.

[0124] In some possible implementations, determining whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model includes: Determine the error information between the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model; Based on the relationship between the error represented by the error information and the preset error threshold, it is determined whether the digital periodic waveform signal passes waveform verification.

[0125] The waveform verification device provided in this embodiment can execute the corresponding steps of the waveform verification methods described above, thereby achieving the technical effects of the waveform verification methods described above. The waveform verification device and the waveform verification methods can refer to and cite each other in terms of specific implementation and technical effects. For the sake of brevity, they will not be elaborated here.

[0126] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. Figure 7 The illustrated electronic device 700 includes at least one processor 701, a memory 702, at least one network interface 704, and other user interfaces 703. The various components in the electronic device 700 are coupled together via a bus system 705. It is understood that the bus system 705 is used to implement communication between these components. In addition to a data bus, the bus system 705 also includes a power bus, a control bus, and a status signal bus. However, for clarity, ... Figure 7The general labeled all buses as Bus System 705.

[0127] The user interface 703 may include a display, keyboard, or clicking device (e.g., mouse, trackball, touchpad, or touchscreen).

[0128] It is understood that the memory 702 in this embodiment can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 702 described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0129] In some implementations, memory 702 stores elements, executable units or data structures, or subsets thereof, or extended sets thereof: operating system 7021 and application program 7022.

[0130] The operating system 7021 includes various system programs, such as a framework layer, a core library layer, and a driver layer, used to implement various basic business functions and handle hardware-based tasks. The application program 7022 includes various applications, such as a media player and a browser, used to implement various application functions. The program implementing the method of this disclosure embodiment can be included in the application program 7022.

[0131] In this embodiment, by calling the program or instructions stored in memory 702, specifically the program or instructions stored in application program 7022, processor 701 executes the method steps provided in each method embodiment, including, for example: Obtain the time information sequence and amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence; Using time information sequences as input data and amplitude information sequences as expected output data, training samples containing input data and expected output data are generated. A machine learning algorithm is used to train the recurrent neural network model to be trained based on training samples, resulting in a waveform verification model.

[0132] or, Acquire the signal to be verified output by the digital chip; The time information sequence of the signal to be verified is input into the pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model. The waveform verification model is trained using any waveform verification model training method. Based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model, it is determined whether the signal to be verified passes the waveform verification.

[0133] The methods disclosed in the above embodiments of this disclosure can be applied to or implemented by processor 701. Processor 701 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in processor 701 or by instructions in the form of software. The processor 701 may be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software units in the decoding processor. The software units may be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory 702. Processor 701 reads the information in memory 702 and, in conjunction with its hardware, completes the steps of the above method.

[0134] It is understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described above, or combinations thereof.

[0135] For software implementation, the techniques described herein can be implemented by units that perform the functions described above. The software code can be stored in memory and executed by a processor. The memory can be implemented within the processor or external to the processor.

[0136] The electronic device provided in this embodiment may be as follows: Figure 7 The electronic device shown can execute all the steps of the training method or waveform verification method of each waveform verification model described above, thereby achieving the technical effects of the training method or waveform verification method of each waveform verification model described above. For details, please refer to the relevant descriptions above. For the sake of brevity, it will not be elaborated here.

[0137] This disclosure also provides a storage medium (computer-readable storage medium). This storage medium stores one or more programs. The storage medium may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive; the memory may also include combinations of the above types of memory.

[0138] When one or more programs in the storage medium can be executed by one or more processors to implement the above-mentioned waveform verification model training method or waveform verification method executed on the electronic device side.

[0139] The processor described above is used to execute a training program or waveform verification program for a waveform verification model stored in memory, to implement the following steps of a waveform verification model training method or waveform verification method executed on the electronic device side: Obtain the time information sequence and amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence; Using time information sequences as input data and amplitude information sequences as expected output data, training samples containing input data and expected output data are generated. A machine learning algorithm is used to train the recurrent neural network model to be trained based on training samples, resulting in a waveform verification model.

[0140] or, Acquire the signal to be verified output by the digital chip; The time information sequence of the signal to be verified is input into the pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model. The waveform verification model is trained using any waveform verification model training method. Based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model, it is determined whether the signal to be verified passes the waveform verification.

[0141] Furthermore, the computer program product provided in this disclosure embodiment may include computer-readable code that, when executed on a device, causes a processor in the device to implement the steps of a waveform verification model training method or a waveform verification method executed on the electronic device side: Obtain the time information sequence and amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence; Using time information sequences as input data and amplitude information sequences as expected output data, training samples containing input data and expected output data are generated. A machine learning algorithm is used to train the recurrent neural network model to be trained based on training samples, resulting in a waveform verification model.

[0142] or, Acquire the signal to be verified output by the digital chip; The time information sequence of the signal to be verified is input into the pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model. The waveform verification model is trained using any waveform verification model training method. Based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model, it is determined whether the signal to be verified passes the waveform verification.

[0143] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0144] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0145] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0146] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A training method for a waveform verification model, characterized in that, The method includes: Acquire the time information sequence and amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence; Using the time information sequence as input data and the amplitude information sequence as expected output data, training samples containing the input data and the expected output data are generated. Using machine learning algorithms, a recurrent neural network model to be trained is trained based on the training samples to obtain a waveform verification model.

2. The method according to claim 1, characterized in that, The acquisition of the time information sequence and amplitude information sequence of the digital periodic waveform signal includes: For each waveform type in a preset set of waveform types, obtain the time information sequence and amplitude information sequence of the digital periodic waveform signal of that waveform type to obtain the corresponding time information sequence and amplitude information sequence; and The process employs a machine learning algorithm to train a recurrent neural network model based on the training samples, resulting in a waveform verification model, including: Using machine learning algorithms, a recurrent neural network model is trained based on training samples determined by the time information sequence and amplitude information sequence corresponding to each waveform type in the waveform type set, to obtain a waveform verification model.

3. The method according to claim 1, characterized in that, The acquisition of the time information sequence and amplitude information sequence of the digital periodic waveform signal includes: Determine whether the digital periodic waveform signal meets the preset generation conditions; In response to the digital periodic waveform signal meeting the preset generation conditions, a time information sequence and an amplitude information sequence of the digital periodic waveform signal are generated based on the signal characteristics of the digital periodic waveform signal. In response to the digital periodic waveform signal not meeting the preset generation conditions, the digital periodic waveform signal output by the digital chip is acquired, and the time information sequence and amplitude information sequence of the digital periodic waveform signal are determined based on the digital periodic waveform signal output by the digital chip.

4. The method according to claim 3, characterized in that, The preset generation conditions include at least one of the following: The signal noise of the digital periodic waveform signal is less than or equal to a preset noise threshold; The similarity between the signal features of the digital periodic waveform signal and the reference features corresponding to the digital periodic waveform signal is greater than or equal to a preset similarity threshold, wherein the reference features represent the waveform features of the waveform type to which the digital periodic waveform signal belongs.

5. The method according to claim 1, characterized in that, The process employs a machine learning algorithm to train a recurrent neural network model based on the training samples, resulting in a waveform verification model, including: Using machine learning algorithms, the model parameters of the recurrent neural network model to be trained are adjusted based on the training samples; Determine whether the accuracy of the recurrent neural network model after the model parameters are adjusted is greater than or equal to a preset accuracy threshold; In response to the accuracy being greater than or equal to the accuracy threshold, the recurrent neural network model with the adjusted model parameters is determined as the waveform verification model.

6. The method according to any one of claims 1-5, characterized in that, The recurrent neural network model includes a long short-term memory network.

7. A waveform verification method, characterized in that, The method includes: Acquire the signal to be verified output by the digital chip; The time information sequence of the signal to be verified is input into a pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model, wherein the waveform verification model is trained using the method described in any one of claims 1-6 above. Based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model, it is determined whether the signal to be verified passes the waveform verification.

8. The method according to claim 7, characterized in that, The step of determining whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model includes: Determine the error information between the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model; Based on the magnitude relationship between the error represented by the error information and the preset error threshold, it is determined whether the digital periodic waveform signal passes waveform verification.

9. A training device for a waveform verification model, characterized in that, The device includes: The first acquisition unit is configured to acquire a time information sequence and an amplitude information sequence of a digital periodic waveform signal, wherein the time information in the time information sequence corresponds one-to-one with the amplitude information in the amplitude information sequence; The generation unit is configured to: take the time information sequence as input data and the amplitude information sequence as expected output data, and generate training samples containing the input data and the expected output data; The training unit is configured to: use a machine learning algorithm to train the recurrent neural network model to be trained based on the training samples, and obtain a waveform verification model.

10. A waveform verification device, characterized in that, The device includes: The second acquisition unit is configured to acquire the signal to be verified output by the digital chip; The input unit is configured to: input the time information sequence of the signal to be verified into a pre-trained waveform verification model to obtain the amplitude information sequence predicted by the waveform verification model, wherein the waveform verification model is trained using the method described in any one of claims 1-6 above; The verification unit is configured to determine whether the signal to be verified passes waveform verification based on the actual amplitude information sequence of the signal to be verified and the amplitude information sequence predicted by the waveform verification model.

11. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing a computer program stored in the memory, wherein when the computer program is executed, it implements the method described in any one of claims 1-8.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1-8.

Images