Power amplifier operating characteristic simulation method and device, computer equipment and medium

Abstract

The invention relates to the technical field of power amplifier characteristic simulation, and discloses a power amplifier working characteristic simulation method and device, computer equipment and a medium, and the method comprises the steps: obtaining a plurality of different input data of a power amplifier, intercepting time sequence data of continuous time periods with different lengths, dividing the time sequence data into training data and test data; training a neural network based on an attention mechanism based on the training data to obtain a model with nonlinear characteristics of the power amplifier; testing the power amplifier nonlinear characteristic model based on the test data to obtain a power amplifier working characteristic simulation model; the real-time input data of the power amplifier are obtained, the working characteristic simulation output result of the power amplifier is obtained based on the real-time input data and the working characteristic simulation model of the power amplifier, the analysis capability of the complex relation between the input signal and the output signal is remarkably improved, and the simulation precision and efficiency of the power amplifier are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power amplifier characteristic simulation, and in particular to a method, device, computer equipment and medium for simulating the working characteristics of a power amplifier. Background Art

[0002] Power amplifiers are widely used in fields such as communications, audio, and signal processing, and their performance directly affects the overall efficiency and signal quality of the system. In practical applications, power amplifiers often operate in nonlinear regions, where their output characteristics are complex and difficult to accurately simulate using traditional linear models. Traditional simulation methods mainly rely on physical models and empirical formulas, which face many challenges when dealing with nonlinear characteristics and can easily lead to inaccurate simulation results, thus affecting system performance. As the performance requirements for power amplifiers continue to increase, traditional physical models and empirical formulas have become difficult to meet the simulation requirements for their complex characteristics.

[0003] With the rapid development of computer technology, especially the rise of artificial intelligence and deep learning technology, researchers have begun to explore the use of these emerging technologies to solve the shortcomings of traditional simulation methods. However, existing deep learning methods still have certain limitations in practical applications, especially in processing variable-length input signals and real-time reasoning. There are problems such as insufficient accuracy and low efficiency in the nonlinear region of the power amplifier due to the single input signal. Summary of the invention

[0004] In view of this, the present invention provides a method, apparatus, computer device and medium for simulating the operating characteristics of a power amplifier to solve the problems of insufficient simulation accuracy and low efficiency in the nonlinear region of the power amplifier in the prior art.

[0005] In a first aspect, the present invention provides a method for simulating operating characteristics of a power amplifier, the method comprising:

[0006] Acquire multiple different input data of the power amplifier;

[0007] Extract time series data of continuous time periods of different lengths from multiple different input data, and divide the time series data into training data and test data;

[0008] A neural network based on the attention mechanism is trained based on the training data to obtain a model with nonlinear characteristics of the power amplifier;

[0009] Testing a model with nonlinear characteristics of a power amplifier based on the test data to obtain a simulated model of working characteristics of the power amplifier after the test;

[0010] Real-time input data of the power amplifier is acquired, and a power amplifier operating characteristic simulation output result is obtained based on the real-time input data and a power amplifier operating characteristic simulation model.

[0011] The power amplifier operating characteristic simulation method provided by the present invention can accurately capture the complex dynamic characteristics of the power amplifier in the nonlinear region by extracting and analyzing the time series data characteristics of the input signal in the continuous time period, and significantly improve the analytical ability of the complex relationship between the input signal and the output signal. Automated feature extraction and modeling are achieved through the deep learning algorithm. The extracted features better reflect the nonlinear characteristics of the power amplifier, greatly improve the accuracy and efficiency of the power amplifier simulation, and solve the problem of insufficient simulation accuracy and low efficiency in the nonlinear region of the power amplifier in the prior art.

[0012] In an optional implementation, extracting time series data of continuous time periods of different lengths from a plurality of different input data includes:

[0013] Select sliding windows of different lengths and use them to extract time series data of continuous time periods of different lengths from multiple input data.

[0014] The power amplifier operating characteristic simulation method provided by the present invention selects sliding windows of different lengths, and uses sliding windows of different lengths to intercept time series data of continuous time periods of different lengths from multiple input data, thereby achieving the purpose of intercepting time series data of adaptive lengths, and providing conditions for subsequent use of time series data to train the model. The time series data of adaptive length is suitable for modeling the characteristics of different types of power amplifiers, has low dependence on data, and can be flexibly adjusted according to the operating characteristics of different amplifiers.

[0015] In an optional implementation, the attention mechanism-based neural network includes a Masked CNN module, a BiLSTM module, and a Length-Aware Attention module;

[0016] Based on the training data, the neural network based on the attention mechanism is trained to obtain a model with nonlinear characteristics of the power amplifier including:

[0017] Input the training data into the Masked CNN module for feature extraction and local feature fusion to obtain the fused features;

[0018] The BiLSTM module is used to perform time series feature analysis on the fused features to obtain the nonlinear characteristics of the power amplifier;

[0019] The Length-Aware Attention module is used to focus the nonlinear characteristics of the power amplifier and output the training results of the nonlinear characteristics of the power amplifier. The neural network that outputs the training results of the nonlinear characteristics of the power amplifier is used as a model with the nonlinear characteristics of the power amplifier.

[0020] In an optional implementation, the formula for inputting the training data into the Masked CNN module for feature extraction and local feature fusion is as follows:

[0021] H cnn =CNN(x)⊙mask;

[0022] Among them, CNN(x) is the output of the pre-signal through the CNN network, mask is the mask operation, which is used to filter the input features and is related to the effective output length. cnn It is the final output of the Masked CNN module;

[0023] The formula for using the BiLSTM module to analyze the time series features of the fused features is as follows:

[0024]

[0025] Among them, α i is the attention coefficient, exp(a i ) is for a i An exponential function is applied to convert the score values ​​into positive numbers and enhance the larger scores, mask i is the mask function, T is the total length of the sequence or the number of time steps, h i is the fused feature, a i is the eigenvalue, i is the sequence position currently being calculated, and j is all the positions in the sequence traversed during the summation process of the formula.

[0026] The method for simulating the working characteristics of a power amplifier provided by the present invention comprises the following steps: inputting training data into a Masked CNN module for feature extraction and local feature fusion to obtain fused features; adopting a BiLSTM module to perform time series feature analysis on the fused features to obtain nonlinear features of the power amplifier; adopting a Length-Aware Attention module to perform feature focusing on the nonlinear features of the power amplifier and outputting a training result of the nonlinear characteristics of the power amplifier; using a neural network that outputs the training result of the nonlinear characteristics of the power amplifier as a model with nonlinear characteristics of the power amplifier; automatically processing massive data through a deep learning algorithm, reducing manual intervention and the use of empirical formulas, and improving the efficiency of model training and reasoning; processing features through the Masked CNN module, the BiLSTM module and the Length-Aware Attention module significantly improves the simulation accuracy of the nonlinear regional characteristics of the power amplifier, can better adapt to complex signal environments, and reduce the errors of traditional models under extreme working conditions.

[0027] In an optional implementation, training a neural network based on an attention mechanism based on training data further includes:

[0028] The back propagation algorithm and Adam optimizer are used to optimize parameters and minimize the loss function of the model with nonlinear characteristics of the power amplifier.

[0029] In an optional implementation, using a back propagation algorithm and an Adam optimizer to optimize parameters and minimize a loss function for a model having nonlinear characteristics of a power amplifier includes:

[0030] The parameter optimization formula of the Masked CNN module in the back propagation algorithm is as follows:

[0031]

[0032] The parameter optimization formula of the Length-Aware Attention module is as follows:

[0033]

[0034] in, represents the calculated loss, represents partial derivative operation;

[0035] The formula of the loss function is as follows:

[0036]

[0037] Among them, smoothL1(x) represents the loss function, x=|ytrue-ypred|, which represents the absolute value of the difference between the predicted value ypred and the true value ytrue, and otherwise represents other cases except |x|<1.

[0038] The power amplifier operating characteristic simulation method provided by the present invention calculates the gradient of the loss function with respect to the weights of each layer in the neural network through back propagation, thereby guiding the updating of the weights, enabling the neural network to learn the mapping relationship between input data and output data, and realizing the parameter optimization and model adjustment of the model with nonlinear characteristics of the power amplifier.

[0039] In an optional implementation, the power amplifier operating characteristic simulation method further includes: obtaining an actual output result corresponding to the real-time input data, and comparing the actual output result with the simulated output result to verify the simulation accuracy of the power amplifier operating characteristic simulation model.

[0040] The power amplifier operating characteristic simulation method provided by the present invention obtains the actual output result corresponding to the real-time input data, and compares the actual output result with the simulated output result, thereby achieving the purpose of verifying the simulation accuracy of the power amplifier operating characteristic simulation model and providing conditions for improving the model simulation accuracy.

[0041] In a second aspect, the present invention provides a power amplifier operating characteristic simulation device, the device comprising:

[0042] An input data acquisition module, used to acquire a plurality of different input data of the power amplifier;

[0043] A training data interception module is used to intercept time series data of continuous time periods of different lengths from multiple different input data, and divide the time series data into training data and test data;

[0044] A training module, used for training a neural network based on an attention mechanism based on training data to obtain a model with nonlinear characteristics of a power amplifier;

[0045] A testing module, used to test a model having nonlinear characteristics of a power amplifier based on the test data to obtain a simulation model of working characteristics of the power amplifier after the test;

[0046] The simulation module is used to obtain real-time input data of the power amplifier, and obtain a simulation output result of the nonlinear working characteristic of the power amplifier based on the real-time input data and the power amplifier working characteristic simulation model.

[0047] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, computer instructions are stored in the memory, and the processor executes the power amplifier operating characteristic simulation method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0048] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the power amplifier operating characteristic simulation method of the first aspect or any corresponding embodiment thereof.

[0049] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the method for simulating the operating characteristics of a power amplifier according to the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 is a schematic flow chart of a method for simulating operating characteristics of a power amplifier according to an embodiment of the present invention;

[0052] Figure 2 is a flow chart of another method for simulating operating characteristics of a power amplifier according to an embodiment of the present invention;

[0053] Figure 3 is a schematic flow chart of another method for simulating operating characteristics of a power amplifier according to an embodiment of the present invention;

[0054] Figure 4 is a schematic diagram of the structure of another neural network based on the attention mechanism according to an embodiment of the present invention;

[0055] Figure 5 is a schematic structural diagram of a power amplifier according to an embodiment of the present invention;

[0056] Figure 6 is an effect diagram on an online test set using a power amplifier operating characteristic simulation method according to an embodiment of the present invention;

[0057] Figure 7 is a structural block diagram of a device for simulating operating characteristics of a power amplifier according to an embodiment of the present invention;

[0058] Figure 8 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0060] Deep learning algorithms can automatically learn complex features from data without manually defining rules, so they excel in capturing nonlinear characteristics. By building a model based on deep learning, the complex relationship between input signals and output signals can be effectively extracted, thereby achieving accurate simulation of the operating characteristics of the amplifier. An embodiment of the present invention provides a neural network based on an attention mechanism, specifically a Masked CNN+BiLSTM+Length-Aware Attention neural network, which implements automated feature extraction and modeling through the Masked CNN+BiLSTM+Length-Aware Attention neural network, thereby greatly improving the accuracy and efficiency of power amplifier simulation.

[0061] According to an embodiment of the present invention, an embodiment of a method for simulating the operating characteristics of a power amplifier is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0062] In this embodiment, a method for simulating the operating characteristics of a power amplifier is provided, which can be used in a computer terminal, such as a central processing unit, a server, etc. Figure 1 is a flow chart of a method for simulating operating characteristics of a power amplifier according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0063] Step S101, obtaining a plurality of different input data of a power amplifier.

[0064] Specifically, the structural diagram of the power amplifier is as follows: Figure 5As shown, the power amplifier of this embodiment is selected as a radio frequency (RF) amplifier, including capacitors C1, C2 and C6, a comparator U1, an inductor L2 and a resistor B1, wherein one end of the capacitor C2 is connected to a data input terminal J1 (i.e., RFInput), the other end of the capacitor C2 is connected to the input terminal of the comparator U1, the output terminal of the comparator U1 is connected to one end of the capacitor C6 and one end of the inductor L2, the other end of the inductor L2 is connected to one end of the resistor B1 and one end of the capacitor C1, the other end of the capacitor C1 is grounded, and the other end of the resistor B1 is connected to an external voltage source J3 (i.e., V DD ), the other end of the capacitor C6 is connected to the data output terminal J2 (ie, RFOutput), and the backside paddle refers to a structure used for heat dissipation in an integrated circuit (IC) package. It is a metal area located at the bottom of the package for enhancing thermal conductivity.

[0065] The input data and output data of the power amplifier are collected through multiple experiments (the input data and output data of the power amplifier collected in this embodiment are about 80,000), and the data of the power amplifier working in the linear region are removed by data cleaning to ensure the quality of subsequent training data and test data. After data cleaning, scatter plot analysis, statistical characteristics analysis, numerical correlation analysis, etc. are performed to further organize the data and provide a basis for the design of the neural network model.

[0066] Step S102, extracting time series data of continuous time periods of different lengths from a plurality of different input data, and dividing the time series data into training data and test data.

[0067] Specifically, in the cleaned data file, combined with the data analysis situation, the continuous time series input of the random window is flexibly selected as the training data and test data for the input of the neural network model, and the mask setting in the model layer is performed according to the size of the window to avoid the influence of the fill value on the model effect. For example, a sliding window with a length of 90 to 120 is selected to intercept the amplitude signal of continuous time of different lengths as the time series data, and the time series data is divided into training data and test data according to the ratio. In this embodiment, the training data and the test data are divided according to the ratio of 7:3.

[0068] Step S103: training a neural network based on an attention mechanism based on training data to obtain a model having nonlinear characteristics of a power amplifier.

[0069] For example, Figure 4As shown, this embodiment uses a network structure of Masked CNN+BiLSTM+Length-AwareAttention to deeply explore the timing features, and uses deep learning technology to automatically extract high-level features from the preprocessed data. The high-level features include the relationship between the input and output of the power amplifier and the timing correlation of the input signal. Figure 4 In the figure, data1, data2, data3…data100 are all training data, Masked CNN represents a framework based on Faster-RCNN, in which a fully connected segmentation network is added after the basic feature network, 1DPOOL represents a feature pool for storing features, multiple LSTMs (Long Short-Term Memory) are connected in series to form a unidirectional network, and two unidirectional networks in opposite directions form a BiLSTM, and Length-Aware Attention is an attention mechanism designed to adapt to encoding based on expected length to generate high-quality summaries with desired length.

[0070] Step S104: testing the power amplifier nonlinear characteristic model based on the test data to obtain a tested power amplifier operating characteristic simulation model.

[0071] Specifically, the test data is used to test the trained model with nonlinear characteristics of the power amplifier. The model extracts features and outputs results for time series data of different window scales, thereby simulating the signal output of the amplifier. By adopting the Masked CNN module and the Length-Aware Attention module, the test results are compatible with signal inputs of time periods of different lengths within a certain range, and the error rate can reach less than 0.02. The use of simple CNN and attention cannot achieve compatibility with the adaptation length. The model that has completed the test is used as a simulation model of the working characteristics of the power amplifier.

[0072] Step S105 , acquiring real-time input data of the power amplifier, and obtaining a power amplifier operating characteristic simulation output result based on the real-time input data and the power amplifier operating characteristic simulation model.

[0073] Specifically, real-time input data of different data acquisition groups of the power amplifier are obtained, and the input signals of continuous time periods of different lengths are intercepted by a sliding window for simulation. The output results of the power amplifier working characteristics simulation model are compared with the actual output results of the power amplifier to verify the accuracy and effect of the power amplifier working characteristics simulation model.

[0074] The power amplifier operating characteristic simulation method provided in this embodiment can accurately capture the complex dynamic characteristics of the power amplifier in the nonlinear region by extracting and analyzing the time series data features of the input signal in the continuous time period, and significantly improve the ability to analyze the complex relationship between the input signal and the output signal. Automated feature extraction and modeling are achieved through deep learning algorithms. The extracted features better reflect the nonlinear characteristics of the power amplifier, greatly improve the accuracy and efficiency of the power amplifier simulation, and solve the problem of insufficient simulation accuracy and low efficiency in the nonlinear region of the power amplifier in the prior art.

[0075] In this embodiment, a method for simulating the operating characteristics of a power amplifier is provided, which can be used in a computer terminal, such as a central processing unit, a server, etc. Figure 2 is a flow chart of a method for simulating operating characteristics of a power amplifier according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0076] Step S201, obtaining a plurality of different input data of the power amplifier. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0077] Step S202, extracting time series data of continuous time periods of different lengths from a plurality of different input data, and dividing the time series data into training data and test data.

[0078] Specifically, the above step S202 includes:

[0079] Step a. Select sliding windows of different lengths and use sliding windows of different lengths to intercept time series data of continuous time periods of different lengths from multiple input data.

[0080] Exemplarily, a sliding window with a length of 90 to 120 is selected, and the sliding window with a length of 90 to 120 is used to intercept the amplitude signal of continuous time as time series data, that is, the model predicts the final output amplitude signal of the time period.

[0081] Step S203: training a neural network based on an attention mechanism based on training data to obtain a model having nonlinear characteristics of a power amplifier.

[0082] Specifically, Figure 4 As shown, the neural network based on the attention mechanism is a neural network structure of Masked CNN+BiLSTM+Length-Aware Attention, including a Masked CNN module, a BiLSTM module and a Length-Aware Attention module. The above step S203 includes:

[0083] Step S2031, input the training data into the Masked CNN module for feature extraction and local feature fusion to obtain fused features.

[0084] Specifically, we first use Masked CNN (based on the Faster-RCNN framework, with a fully connected segmentation network added after the basic feature network) to achieve local feature fusion. The training data is input into the Masked CNN module for feature extraction and local feature fusion. The formula is as follows:

[0085] H cnn =CNN(x)⊙mask(1);

[0086] Among them, CNN(x) is the output of the pre-signal (pre-signal refers to the original data or features input into the convolutional neural network (CNN)) through the CNN network, mask is the mask operation, which filters the input features and is related to the effective output length, H cnn It is the final output of the Masked CNN module;

[0087] Step S2032: Use the BiLSTM module to perform time series feature analysis on the fused features to obtain nonlinear features of the power amplifier.

[0088] Specifically, the formula for using the BiLSTM module (BiLSTM is a bidirectional long short-term memory network, which is an improved recurrent neural network) to perform time series feature analysis on the fused features is as follows:

[0089]

[0090] Among them, α i is the attention coefficient, exp(a i ) is for a i An exponential function is applied to convert the score values ​​into positive numbers and enhance the larger scores, mask i is the mask function, T is the total length of the sequence or the number of time steps, h i is the fused feature, a i is the eigenvalue, i is the sequence position currently being calculated, and j is all the positions in the sequence traversed during the summation process of the formula.

[0091] Step S2033, using the Length-Aware Attention module to perform feature focusing on the nonlinear characteristics of the power amplifier and output the nonlinear characteristic training result of the power amplifier, and using the neural network that outputs the nonlinear characteristic training result of the power amplifier as a model with nonlinear characteristics of the power amplifier.

[0092] Specifically, the Length-Aware Attention module is an attention mechanism designed to adapt to the encoding based on the expected length to generate a high-quality summary with the required length, that is, to generate the focused nonlinear characteristics of the power amplifier. After the training result of the nonlinear characteristics of the power amplifier is output, the neural network training of Masked CNN+BiLSTM+Length-Aware Attention becomes a model with the nonlinear characteristics of the power amplifier.

[0093] Step S2034, using the back propagation algorithm and Adam optimizer to perform parameter optimization and loss function minimization on the model with nonlinear characteristics of the power amplifier.

[0094] Adam optimizer (Adaptive Moment Estimation) is an optimization algorithm for deep learning, which aims to adaptively adjust the learning rate of each parameter to improve the convergence speed and generalization ability of the model. For example, in model training, Adam optimizer is used for supervised parameter optimization, and the initial learning rate is 1e -4 The learning rate is dynamically adjusted through the cosine annealing mechanism. Ten rounds of training were conducted based on about 80,000 training data. After the training, the trained model with the nonlinear characteristics of the power amplifier was saved locally.

[0095] In some optional implementations, the above step S2034 includes:

[0096] The parameter optimization formula of the Masked CNN module in the back propagation algorithm is as follows:

[0097]

[0098] The parameter optimization formula of the Length-Aware Attention module is as follows:

[0099]

[0100] in, represents the calculated loss, represents partial derivative operation;

[0101] In order to improve the stability of convergence, this embodiment uses smooth L1 as the loss function. The formula of the loss function is as follows:

[0102]

[0103] Among them, smoothL1(x) represents the loss function, x=|ytrue-ypred|, which represents the absolute value of the difference between the predicted value ypred and the true value ytrue, and otherwise represents other cases except |x|<1.

[0104] By using the smooth L1 loss function, the influence of a small number of outliers can be avoided and the final convergence effect can be improved.

[0105] The power amplifier operating characteristics simulation method provided in this embodiment calculates the gradient of the loss function with respect to the weights of each layer in the neural network through back propagation, thereby guiding the updating of the weights, enabling the neural network to learn the mapping relationship between input data and output data, and realizing the optimization of model parameters and model adjustment with nonlinear characteristics of the power amplifier.

[0106] Step S204: Test the power amplifier nonlinear characteristic model based on the test data to obtain a power amplifier operating characteristic simulation model that has been tested. Figure 1 Step S104 of the illustrated embodiment will not be described in detail here.

[0107] Step S205, obtaining the real-time input data of the power amplifier, and obtaining the power amplifier operating characteristic simulation output result based on the real-time input data and the power amplifier operating characteristic simulation model. Figure 1 Step S105 of the illustrated embodiment will not be described in detail here.

[0108] The power amplifier operating characteristics simulation method provided in this embodiment selects sliding windows of different lengths, and uses sliding windows of different lengths to intercept time series data of continuous time periods of different lengths from multiple input data, thereby achieving the purpose of intercepting time series data of adaptive lengths, and providing conditions for subsequent use of time series data to train the model. The time series data of adaptive length is suitable for modeling the characteristics of different types of power amplifiers, and has low dependence on data, and can be flexibly adjusted according to the operating characteristics of different amplifiers. The training data is input into the Masked CNN module for feature extraction and local feature fusion to obtain the fused features; the BiLSTM module is used to perform time series feature analysis on the fused features to obtain the nonlinear features of the power amplifier; the Length-Aware Attention module is used to focus the nonlinear features of the power amplifier and output the training results of the nonlinear characteristics of the power amplifier. The neural network that outputs the training results of the nonlinear characteristics of the power amplifier is used as a model with the nonlinear characteristics of the power amplifier. The deep learning algorithm is used to automatically process massive data, reduce manual intervention and the use of empirical formulas, and improve the efficiency of model training and reasoning. The processing of features by the Masked CNN module, the BiLSTM module and the Length-Aware Attention module significantly improves the simulation accuracy of the nonlinear regional characteristics of the power amplifier, can better adapt to complex signal environments, and reduce the errors of traditional models under extreme working conditions.

[0109] In this embodiment, a method for simulating the operating characteristics of a power amplifier is provided, which can be used in the above-mentioned mobile terminals, such as mobile phones, tablet computers, etc. Figure 3 is a flow chart of a method for simulating operating characteristics of a power amplifier according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:

[0110] Step S301, obtaining a plurality of different input data of a power amplifier.

[0111] Step S302: extract time series data of different lengths of continuous time periods from multiple different input data, and divide the time series data into training data and test data. Figure 2 Step S202 of the illustrated embodiment will not be described in detail here.

[0112] Step S303: Train the neural network based on the attention mechanism based on the training data to obtain a model with nonlinear characteristics of the power amplifier. Figure 2 Step S203 of the illustrated embodiment will not be described in detail here.

[0113] Step S304: Test the power amplifier nonlinear characteristic model based on the test data to obtain a power amplifier operating characteristic simulation model that has been tested. Figure 2 Step S204 of the illustrated embodiment will not be described in detail here.

[0114] Step S305, obtaining the real-time input data of the power amplifier, and obtaining the power amplifier operating characteristic simulation output result based on the real-time input data and the power amplifier operating characteristic simulation model. Figure 2 Step S205 of the illustrated embodiment is not described in detail here. Step S306: obtaining the actual output result corresponding to the real-time input data, and comparing the actual output result with the simulated output result to verify the simulation accuracy of the power amplifier operating characteristic simulation model.

[0115] Specifically, the real-time input data of different data acquisition groups of the power amplifier are obtained, and the input signals of continuous time periods of different lengths are intercepted by a sliding window for simulation. The output results of the power amplifier working characteristic simulation model are compared with the actual output results of the power amplifier to verify the accuracy and effect of the power amplifier working characteristic simulation model. Figure 6 As shown in the figure, the horizontal axis Samples is the number of training data samples, and the vertical axis is the actual output result TrueAmplitude and the simulated output result PredictedAmplitude. Figure 6 It can be seen that the simulated output result obtained by using the power amplifier operating characteristic simulation method provided in this embodiment is substantially consistent with the actual output result.

[0116] The power amplifier operating characteristics simulation method provided in this embodiment obtains the actual output results corresponding to the real-time input data, and compares the actual output results with the simulated output results, thereby achieving the purpose of verifying the simulation accuracy of the power amplifier operating characteristics simulation model and providing conditions for improving the model simulation accuracy.

[0117] This embodiment realizes automatic feature extraction and modeling through deep learning algorithms, greatly improving the accuracy and efficiency of power amplifier simulation. By extracting and analyzing the timing characteristics of input signals in continuous time periods, the present invention can accurately capture the complex dynamic characteristics of power amplifiers in nonlinear regions and significantly improve the ability to analyze the complex relationship between input and output signals.

[0118] In this embodiment, a power amplifier operating characteristic simulation device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made are not repeated here. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0119] This embodiment provides a power amplifier operating characteristic simulation device, such as Figure 7 As shown, including:

[0120] The input data acquisition module 701 is used to acquire a plurality of different input data of the power amplifier.

[0121] The training data interception module 702 is used to intercept time series data of continuous time periods of different lengths from a plurality of different input data, and divide the time series data into training data and test data.

[0122] The training module 703 is used to train the neural network based on the attention mechanism based on the training data to obtain a model with nonlinear characteristics of the power amplifier.

[0123] The testing module 704 is used to test the power amplifier nonlinear characteristic model based on the test data to obtain a power amplifier operating characteristic simulation model after the test.

[0124] The simulation module 705 is used to obtain real-time input data of the power amplifier, and obtain a simulation output result of the nonlinear working characteristic of the power amplifier based on the real-time input data and the power amplifier working characteristic simulation model.

[0125] In some optional implementations, the training data interception module 702 includes:

[0126] The sliding window capture data unit is used to select sliding windows of different lengths and use sliding windows of different lengths to capture time series data of continuous time periods of different lengths from multiple input data*.

[0127] In some optional implementations, the neural network based on the attention mechanism includes a Masked CNN module, a BiLSTM module, and a Length-Aware Attention module; the training module 703 includes:

[0128] The feature extraction and fusion unit is used to input the training data into the Masked CNN module for feature extraction and local feature fusion to obtain the fused features.

[0129] The feature analysis unit is used to perform time series feature analysis on the fused features using the BiLSTM module to obtain the nonlinear features of the power amplifier.

[0130] The feature focusing unit is used to perform feature focusing on the nonlinear features of the power amplifier by using the Length-Aware Attention module and output the training results of the nonlinear characteristics of the power amplifier, and use the neural network that outputs the training results of the nonlinear characteristics of the power amplifier as a model with the nonlinear characteristics of the power amplifier.

[0131] Among them, the formula for inputting training data into the Masked CNN module for feature extraction and local feature fusion is as follows:

[0132] H cnn =CNN(x)⊙mask(1);

[0133] Among them, CNN(x) is the output of the pre-signal through the CNN network, mask is the mask operation, which is used to filter the input features and is related to the effective output length. cnn It is the final output of the Masked CNN module;

[0134] The formula for using the BiLSTM module to analyze the time series features of the fused features is as follows:

[0135]

[0136] Among them, α i is the attention coefficient, exp(a i ) is for a i An exponential function is applied to convert the score values ​​into positive numbers and enhance the larger scores, mask i is the mask function, T is the total length of the sequence or the number of time steps, h i is the fused feature, a i is the eigenvalue, i is the sequence position currently being calculated, and j is all the positions in the sequence traversed during the summation process of the formula.

[0137] The optimization unit is used to optimize parameters and minimize the loss function of the model with nonlinear characteristics of the power amplifier by using the back propagation algorithm and the Adam optimizer.

[0138] Among them, the parameter optimization formula of the Masked CNN module in the back propagation algorithm is as follows:

[0139]

[0140] The parameter optimization formula of the Length-Aware Attention module is as follows:

[0141]

[0142] in, represents the calculated loss, represents partial derivative operation;

[0143] The formula of the loss function is as follows:

[0144]

[0145] Among them, smoothL1(x) represents the loss function, x=|ytrue-ypred|, which represents the absolute value of the difference between the predicted value ypred and the true value ytrue, and otherwise represents other cases except |x|<1.

[0146] In some optional implementations, the power amplifier operating characteristic simulation device further includes:

[0147] The simulation accuracy verification module is used to obtain the actual output result corresponding to the real-time input data, and compare the actual output result with the simulated output result to verify the simulation accuracy of the power amplifier working characteristic simulation model.

[0148] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0149] The power amplifier operating characteristic simulation device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0150] The embodiment of the present invention also provides a computer device having the above Figure 7 The power amplifier operating characteristics simulation device is shown.

[0151] See also Figure 8 , Figure 8 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 8As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 8 A processor 10 is taken as an example.

[0152] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0153] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0154] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0155] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0156] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 8 The example of connecting through bus is taken in the following.

[0157] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0158] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0159] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0160] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for simulating the operating characteristics of a power amplifier, characterized in that: The method comprises: Acquire multiple different input data of the power amplifier; Extracting time series data of continuous time periods of different lengths from the plurality of different input data, and dividing the time series data into training data and test data; Training a neural network based on an attention mechanism based on the training data to obtain a model having nonlinear characteristics of a power amplifier; Testing a model having nonlinear characteristics of a power amplifier based on the test data to obtain a simulated model of working characteristics of the power amplifier after the test; Real-time input data of the power amplifier is acquired, and a power amplifier operating characteristic simulation output result is obtained based on the real-time input data and the power amplifier operating characteristic simulation model.

2. The method according to claim 1, characterized in that Extracting time series data of continuous time periods of different lengths from the plurality of different input data includes: Sliding windows of different lengths are selected, and the sliding windows of different lengths are used to intercept time series data of continuous time periods of different lengths from multiple input data.

3. The method according to claim 1, characterized in that The neural network based on the attention mechanism includes a Masked CNN module, a BiLSTM module and a Length-Aware Attention module; Based on the training data, a neural network based on the attention mechanism is trained to obtain a power amplifier nonlinear characteristic model including: Inputting the training data into the Masked CNN module for feature extraction and local feature fusion to obtain fused features; The BiLSTM module is used to perform time series feature analysis on the fused features to obtain nonlinear features of the power amplifier; The Length-Aware Attention module is used to perform feature focusing on the nonlinear characteristics of the power amplifier and output a training result of the nonlinear characteristics of the power amplifier, and a neural network that outputs the training result of the nonlinear characteristics of the power amplifier is used as a model with nonlinear characteristics of the power amplifier.

4. The method according to claim 3, characterized in that The formula for inputting the training data into the Masked CNN module for feature extraction and local feature fusion is as follows: H cnn =CNN(x)⊙mask; Among them, CNN(x) is the output of the pre-signal through the CNN network, mask is the mask operation, which is used to filter the input features and is related to the effective output length. cnn It is the final output of the Masked CNN module; The formula for performing time series feature analysis on the fused features using the BiLSTM module is as follows: Among them, α i is the attention coefficient, exp(a i ) is for a i An exponential function is applied to convert the score values ​​into positive numbers and enhance the larger scores, mask i is the mask function, T is the total length of the sequence or the number of time steps, h i is the fused feature, a i is the eigenvalue, i is the sequence position currently being calculated, and j is all the positions in the sequence traversed during the summation process of the formula.

5. The method according to claim 3, characterized in that: Training a neural network based on an attention mechanism based on the training data further includes: The back propagation algorithm and Adam optimizer are used to perform parameter optimization and loss function minimization on the model with nonlinear characteristics of the power amplifier.

6. The method according to claim 5, characterized in that The method of using the back propagation algorithm and the Adam optimizer to optimize the parameters and minimize the loss function of the model with nonlinear characteristics of the power amplifier includes: The parameter optimization formula of the Masked CNN module in the back propagation algorithm is as follows: The parameter optimization formula of the Length-Aware Attention module is as follows: in, represents the calculated loss, represents partial derivative operation; The formula of the loss function is as follows: Among them, smoothL1(x) represents the loss function, x=|ytrue-ypred|, which represents the absolute value of the difference between the predicted value ypred and the true value ytrue, and otherwise represents other cases except |x|<1.

7. The method according to claim 1, characterized in that The method further includes: obtaining an actual output result corresponding to the real-time input data, and comparing the actual output result with the simulated output result to verify the simulation accuracy of the power amplifier operating characteristic simulation model.

8. A power amplifier operating characteristic simulation device, characterized in that: The device comprises: An input data acquisition module, used to acquire a plurality of different input data of the power amplifier; A training data interception module, used to intercept time series data of continuous time periods of different lengths from the multiple different input data, and divide the time series data into training data and test data; A training module, used for training a neural network based on an attention mechanism based on the training data to obtain a model having nonlinear characteristics of a power amplifier; A testing module, used to test a model having nonlinear characteristics of a power amplifier based on the test data to obtain a simulation model of working characteristics of the power amplifier after the test; The simulation module is used to obtain real-time input data of the power amplifier, and obtain a power amplifier operating characteristic simulation output result based on the real-time input data and the power amplifier operating characteristic simulation model.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the power amplifier operating characteristic simulation method according to any one of claims 1 to 7 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the power amplifier operating characteristic simulation method according to any one of claims 1 to 7.

Images