Intelligent Setting Method and System for Resolution Bandwidth of Spectrum Testing of Aerospace Products

Abstract

A method and system for intelligently setting the resolution bandwidth for spectrum testing of aerospace products includes: determining the main factors for automatic configuration of spectrum items based on commonly used parameters that need to be configured for testing of product spectrum test items; extracting factors influencing resolution bandwidth setting and constructing feature vectors based on the characteristics of historical test data of product spectrum test items; establishing a neural network model for resolution bandwidth setting, training the neural network model using a training sample set to obtain a resolution bandwidth prediction model; and using the resolution bandwidth prediction model to set the resolution bandwidth of the spectrum test items. The present invention extracts factors influencing resolution bandwidth setting for spectrum test items based on the characteristics of historical test data of product spectrum items, leverages the advantages of machine learning methods in parameter optimization, and uses neural network technology to fit a nonlinear model of resolution bandwidth and its influencing factors, thereby achieving intelligent setting of resolution bandwidth for spectrum test items.

Description

Technical Field

[0001] The invention relates to an intelligent setting method for resolution bandwidth of spectrum test of aerospace products, belonging to the technical field of automatic testing of electrical properties. Background Art

[0002] With the increasing maturity of digital and intelligent technologies, aerospace product testing systems have been upgraded. During the ground testing and verification stage of aerospace products, most spectrum testing items can be automated, and testing efficiency has been greatly improved.

[0003] When testing spectrum items in existing test systems, test engineers must first manually test the individual units and select appropriate parameters based on their actual performance and instrument performance. These parameters must then be entered into the test system. However, aerospace products are numerous and limited in number, requiring manual setting of spectrum item test parameters for each product, hindering further improvement in testing efficiency. Furthermore, aerospace product testing involves numerous test stages and projects, with limited instrumentation resources. Reusing parameters after replacing instruments and equipment in different test stages can lead to parameter mismatches, resulting in inaccurate test data that fails to truly reflect the performance of individual units. This can lead to secondary testing, a waste of time and resources, and hinders schedule and cost control. Summary of the Invention

[0004] The present invention addresses the following technical issues: It overcomes the shortcomings of existing technologies and provides an intelligent method for setting resolution bandwidth for spectrum testing of aerospace products. This method utilizes machine learning techniques to establish a neural network model for setting resolution bandwidth for spectrum testing of aerospace products. The model is trained based on accumulated test data, fitting the nonlinear relationship between resolution bandwidth, measured signal characteristics, and other instrument parameters to find the optimal resolution bandwidth setting. This method enables fully automated testing of spectrum items and further improves product testing efficiency.

[0005] The technical solution of the present invention is:

[0006] A method for intelligently setting resolution bandwidth for spectrum testing of aerospace products, comprising:

[0007] Determine the main factors for automatic configuration of spectrum items based on the common parameters that need to be configured for product spectrum test items;

[0008] Based on the characteristics of historical test data of product spectrum test items, the factors affecting the resolution bandwidth setting are extracted and the feature vector is constructed;

[0009] Establishing a neural network model for resolution bandwidth setting, training the neural network model using a training sample set, and obtaining a resolution bandwidth prediction model;

[0010] The resolution bandwidth prediction model is used to set the resolution bandwidth of the spectrum test item.

[0011] Furthermore, the common parameters that need to be configured for the spectrum test item of the product include: carrier frequency, bandwidth, built-in attenuation, reference level, amplitude resolution, resolution bandwidth, and video bandwidth;

[0012] Carrier frequency and bandwidth are the input conditions of the test;

[0013] The settings of built-in attenuation, reference level, and amplitude resolution are directly affected by the output power of the measured signal. The parameter values are linear functions of the output power of the measured signal.

[0014] The built-in attenuation is used to protect the mixer inside the spectrum analyzer. The minimum value of the built-in attenuation is equal to the maximum output power of the measured signal minus the maximum input power of the mixer.

[0015] The reference level is equal to the maximum output power of the measured signal plus b, where b>0;

[0016] The amplitude resolution is equal to the maximum output power of the measured signal minus the minimum output power of the measured signal, divided by 10;

[0017] The video bandwidth and resolution bandwidth are set to a fixed multiple relationship.

[0018] Furthermore, the main factor for automatic configuration of spectrum test items is the resolution bandwidth.

[0019] Furthermore, the factors affecting the extracted resolution bandwidth setting include: the center frequency f of the measured signal c , bandwidth s, reference level l, attenuation a, sweep time t and index requirements.

[0020] Furthermore, the factors affecting the resolution bandwidth are used as the input vector factors for setting the resolution bandwidth neural network model, and the input feature vector is expressed as:

[0021] X=[f c slati]

[0022] Normalize the feature vector to obtain the normalized feature vector; collect historical test data and formulate the data feature vector X k , k=1,2,…,m, where m is the maximum number of samples; and mark the samples and set the label value, then the vector X k and their corresponding labels as the sample set of the neural network model, 70% of the sample set is extracted as the training sample set, and the remaining 30% is used as the test sample set.

[0023] Furthermore, the normalized eigenvector is calculated as follows:

[0024]

[0025] Among them, f min and f max is the spectrum analyzer test frequency, s min and s max is the spectrum analyzer test bandwidth, l min and l max is the spectrum analyzer reference level, a min and a max is the internal attenuation of the spectrum analyzer, t min and t max is the spectrum analyzer scanning time, i min and i max These are the common index requirements for traveling wave tube amplifiers.

[0026] Furthermore, a neural network model for resolution bandwidth setting is established, and the neural network model is trained using the training sample set to obtain a resolution bandwidth prediction model, specifically:

[0027] A fully connected neural network, i.e., a neural network model for resolution bandwidth setting, is constructed. The network model structure includes an input layer, three hidden layers, and an output layer. Based on the constructed feature vector, the input layer vector dimension is 6, and the output layer uses softmax classification, outputting the setting probabilities of 10 types of resolution bandwidth. The training sample set is input into the neural network model, and the number of training iterations, the number of batch samples, and the learning rate are set. The network model is continuously trained and optimized to obtain a resolution bandwidth prediction model.

[0028] Furthermore, the resolution bandwidth prediction model is used to set the resolution bandwidth of the spectrum test item, specifically: the obtained resolution bandwidth prediction model is embedded in the product test system, and the center frequency f of the measured signal is read. c , bandwidth s, automatically identify the parameters reference level l, attenuation a, sweep time t and index requirement i, generate real-time feature vectors, obtain the setting probability of the preset 10 types of resolution bandwidths, and take the maximum probability.

[0029] Furthermore, the present invention also provides an intelligent resolution bandwidth setting system for spectrum testing of aerospace products, comprising:

[0030] Main factor determination module: determines the main factors for automatic configuration of spectrum items based on the common parameters that need to be configured for the product spectrum test items; the common parameters that need to be configured for the product spectrum test items include: carrier frequency, bandwidth, built-in attenuation, reference level, amplitude resolution, resolution bandwidth, and video bandwidth; carrier frequency and bandwidth are the input conditions for the test; the settings of built-in attenuation, reference level, and amplitude resolution are directly affected by the output power of the measured signal, and their parameter values are a linear function of the output power of the measured signal; the built-in attenuation is used to protect the mixer inside the spectrum analyzer, and the minimum value of the built-in attenuation is equal to the maximum output power of the measured signal minus the maximum input power of the mixer; the reference level is equal to the maximum output power of the measured signal plus b, b>0; the amplitude resolution is equal to the maximum output power of the measured signal minus the minimum output power of the measured signal, and then divided by 10; the video bandwidth and resolution bandwidth are set in a fixed multiple relationship; the main factor for automatic configuration of spectrum test items is the resolution bandwidth;

[0031] Feature vector construction module: According to the characteristics of the historical test data of the product spectrum test project, the factors affecting the resolution bandwidth setting are extracted and the feature vector is constructed; the factors affecting the resolution bandwidth setting extracted include: the center frequency f of the measured signal c , bandwidth s, reference level l, attenuation a, sweep time t and index requirements;

[0032] The factors affecting the resolution bandwidth are used as the input vector factors for setting the resolution bandwidth neural network model, and the input feature vector is expressed as:

[0033] X=[f c slati]

[0034] Neural network model establishment and training module: establishes a neural network model for resolution bandwidth setting, trains the neural network model using a training sample set, and obtains a resolution bandwidth prediction model;

[0035] Resolution bandwidth setting module: uses the resolution bandwidth prediction model to set the resolution bandwidth of the spectrum test item.

[0036] The beneficial effects of the present invention compared with the prior art are:

[0037] (1) This invention uses machine learning technology to establish a neural network model for setting resolution bandwidth for spectrum testing of aerospace products. The model is trained based on accumulated test data, fitting the nonlinear relationship between resolution bandwidth, measured signal characteristics, and other instrument parameters to find the optimal resolution bandwidth setting, thereby achieving fully automated testing of spectrum items and further improving product testing efficiency.

[0038] (2) Based on the characteristics of historical test data of aerospace product spectrum projects and engineering practices, the present invention innovatively utilizes the advantages of machine learning methods in parameter optimization, uses neural network technology to fit the nonlinear model of resolution bandwidth and its influencing factors, and realizes the intelligent setting of resolution bandwidth for spectrum project testing.

[0039] (3) The present invention embeds the trained neural network model into the aerospace product test system, and intelligently sets the resolution bandwidth of the spectrum items of aerospace products, eliminating the tedious manual setting of spectrum test parameters, avoiding test errors introduced by parameter reuse when the spectrum analyzer is not in use, ensuring the consistency of the test baseline, and further improving the test efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Flow chart of the method of the present invention;

[0041] Figure 2 This is the structure diagram of the neural network model. DETAILED DESCRIPTION

[0042] The specific embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0043] The present invention conducts research on parameter settings for spectrum test items. Through a series of experiments, the common parameters that need to be configured for spectrum testing, including built-in attenuation, reference level and amplitude resolution, resolution bandwidth, video bandwidth, etc., are analyzed, and the resolution bandwidth, the main factor for automatic configuration of spectrum items, is found.

[0044] This invention will take the aerospace product traveling wave tube amplifier test as an example, combined with Figure 1 The specific process of the present invention is described.

[0045] Step 1: Identify the key factors for spectrum project automation

[0046] The common parameters that need to be configured for the wave tube amplifier spectrum project test are shown in Table 1:

[0047] Table 1 Common configuration parameters for spectrum project testing

[0048]

[0049] These include carrier frequency, bandwidth, built-in attenuation, reference level and amplitude resolution, resolution bandwidth, and video bandwidth. Carrier frequency and bandwidth are test input conditions. Parameters like built-in attenuation, reference level, and amplitude resolution are directly affected by the output power of the measured signal. They have a linear functional relationship and can be automatically identified, calculated, and set by the test system. Video bandwidth can be set to a fixed multiple of resolution bandwidth. Resolution bandwidth, on the other hand, requires comprehensive consideration of the measured signal's characteristics, test parameters, instrument noise floor, sweep time, and instrument performance. It has a nonlinear functional relationship with these factors and is the primary factor in automatic spectrum test configuration.

[0050] Step 2: Extract the factors affecting the resolution bandwidth setting and construct the feature vector

[0051] Based on the characteristics of the historical test data of the traveling wave tube amplifier spectrum project, Keysight's 9030B-50G was selected as the test instrument. The factors affecting the resolution bandwidth setting include: the center frequency of the measured signal f c , bandwidth s, reference level l, attenuation a, sweep time t, and index requirement i. Table 2 below is an example of statistical data on influencing factors.

[0052] Table 2: Example of statistical data on factors affecting resolution bandwidth setting

[0053]

[0054] The factors affecting the resolution bandwidth are used as input vector factors for setting the resolution bandwidth neural network model, and the input feature vector is expressed as:

[0055] X=[f c slati] (1)

[0056] In order to weaken the influence of different feature units and scales, the features are normalized to obtain the normalized feature vector. The calculation formula is:

[0057]

[0058] Among them, f min and f max is the spectrum analyzer test frequency, with a minimum value of 3Hz and a maximum value of 50GHz, s min and s max The test bandwidth of the spectrum analyzer is set to 1 Hz for the minimum and 25 GHz for the maximum. min and l max is the reference level of the spectrum analyzer, with a minimum value of -25dBm and a maximum value of 25dBm. min and a maxis the internal attenuation of the spectrum analyzer, with a minimum value of -80dBm and a maximum value of 0dBm, respectively. min and t max is the spectrum analyzer scanning time, with a minimum value of 0.1ms and a maximum value of 60s, i min and i max These are the common index requirements for traveling wave tube amplifiers, with a minimum value of -80dBc and a maximum value of -35dBc.

[0059] Collect historical test data of traveling wave tube amplifiers and formulate data feature vector X k (k=1,2,…,m), where m is the maximum number of samples. And mark the samples and set the label value, then the vector X k and their corresponding labels as the sample set of the neural network model, 70% of the sample set is extracted as the training sample set, and the remaining 30% is used as the test sample set.

[0060] Step 3: Use the training sample set to train the neural network model to obtain the resolution bandwidth prediction model

[0061] Construct a fully connected neural network, the network model structure is as follows Figure 2 As shown in Figure 3, it includes one input layer, three hidden layers, and one output layer. The network training parameter settings are shown in Table 3.

[0062] According to the feature vector constructed in step 2, the input layer vector dimension is 6. The output layer uses softmax classification and outputs the setting probability of 10 types of resolution bandwidth (3Hz, 10Hz, 30Hz, 100Hz, 300Hz, 1kHz, 3kHz, 10kHz, 30kHz, 100kHz). The calculation formula is:

[0063]

[0064] Where X is the input vector, W 10×6 is the weight, b is the bias term, is the probability of softmax output.

[0065] Then the probability of each type of resolution bandwidth is:

[0066]

[0067] Output of the neural network There will definitely be differences from the real data Y, so the key to establishing a neural network model for resolution bandwidth setting is to find a suitable weight value W so that the model can accurately reflect the nonlinear relationship between the resolution bandwidth setting value and the factors affecting the resolution bandwidth. Input the information of the influencing factors into the neural network model, set the number of training iterations, the number of batch samples and the learning rate, and obtain the output of the neural network. Take the cross entropy loss to evaluate the function Y and The difference in distribution between them is continuously trained and optimized, and the weight value W is adjusted so that Infinitely approach Y. When the error between them reaches 10 -3 Stop training when .

[0068] Table 3 Network training parameter settings

[0069]

[0070] Step 4: Use the prediction model to set the resolution bandwidth of the spectrum test item

[0071] The prediction model obtained in step 3 is embedded in the traveling wave tube amplifier test system. The parameter table of the tested product is read to obtain the center frequency f and bandwidth s. The parameter reference level l, attenuation a, scanning time t and index requirement i are automatically identified. The real-time feature vector x is generated to obtain the setting probability y = softmax (W T x+b), and take the resolution bandwidth with the largest probability as the input of the test system.

[0072] This embodiment runs the test system embedded with the model to measure more than 100 traveling wave tube amplifiers covering the L, S, X, and Ka bands. The measurement is compared with the manually preset spectrum parameter method, and 10 different spectrum graphs are compared for each single machine.

[0073] Table 4 Comparison of different spectra

[0074]

[0075] As can be seen from Table 4, the test system embedded with the neural network model saves about 90% of the setup time, avoids test errors introduced by parameter reuse when not using a spectrum analyzer, achieves the optimization goal of the method, and realizes intelligent testing of spectrum projects.

[0076] The present invention uses machine learning to establish a suitable resolution bandwidth neural network model prediction model to quickly and accurately estimate the resolution bandwidth of aerospace product spectrum testing, solving the problem of manual setting of resolution bandwidth and realizing fully automatic testing of spectrum items. It is a technological advancement in the intelligentization of test systems.

[0077] The above is one of the best embodiments of the present invention, and other embodiments are possible. The scope of protection of the present invention is not limited to this. Any person familiar with the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, and they should fall within the scope of protection of the claims attached to the present invention.

[0078] Parts of the present invention that are not described in detail belong to common knowledge among those skilled in the art.

Claims

1. A method for intelligently setting the resolution bandwidth for spectrum testing of aerospace products, characterized in that include: Determine the main factors for automatic configuration of spectrum items based on the common parameters that need to be configured for product spectrum test items; Based on the characteristics of historical test data of product spectrum test items, the factors affecting the resolution bandwidth setting are extracted and the feature vector is constructed; Establishing a neural network model for resolution bandwidth setting, training the neural network model using a training sample set, and obtaining a resolution bandwidth prediction model; Setting the resolution bandwidth of a spectrum test item using the resolution bandwidth prediction model; Common parameters that need to be configured for the spectrum test items of the product include: carrier frequency, bandwidth, built-in attenuation, reference level, amplitude resolution, resolution bandwidth, and video bandwidth; Carrier frequency and bandwidth are the input conditions of the test; The settings of built-in attenuation, reference level, and amplitude resolution are directly affected by the output power of the measured signal. The parameter values are linear functions of the output power of the measured signal. The built-in attenuation is used to protect the mixer inside the spectrum analyzer. The minimum value of the built-in attenuation is equal to the maximum output power of the measured signal minus the maximum input power of the mixer. The reference level is equal to the maximum output power of the measured signal plus b, where b>0; The amplitude resolution is equal to the maximum output power of the measured signal minus the minimum output power of the measured signal, divided by 10; The video bandwidth and resolution bandwidth are set to a fixed multiple relationship; The main factor for automatic configuration of spectrum test items is resolution bandwidth; The factors affecting the extracted resolution bandwidth setting include: the center frequency f of the measured signal c , bandwidth s, reference level l, attenuation a, sweep time t and index requirements; The factors affecting the resolution bandwidth are used as the input vector factors for setting the resolution bandwidth neural network model, and the input feature vector is expressed as: X=[f c s l a t i] Normalize the feature vector to obtain the normalized feature vector; collect historical test data and formulate the data feature vector X k , k=1,2,…,m, where m is the maximum number of samples; and mark the samples and set the label value, then the vector X k and its corresponding labels as the sample set of the neural network model; Normalized eigenvector, calculated as: Among them, f min and f max is the spectrum analyzer test frequency, s min and s max is the spectrum analyzer test bandwidth, l min and l max is the spectrum analyzer reference level, a min and a max is the internal attenuation of the spectrum analyzer, t min and t max is the spectrum analyzer scanning time, i min and i max These are the common index requirements for traveling wave tube amplifiers; A neural network model for resolution bandwidth setting is established, and the neural network model is trained using the training sample set to obtain a resolution bandwidth prediction model. Specifically: A fully connected neural network, i.e., a neural network model for resolution bandwidth setting, is constructed, wherein the network model structure includes an input layer, three hidden layers, and an output layer; based on the constructed feature vector, the input layer vector dimension is 6, and the output layer uses softmax classification, outputting the setting probabilities of 10 types of resolution bandwidth; a training sample set is input into the neural network model, and the number of training iterations, the number of batch samples, and the learning rate are set. The network model is continuously trained and optimized to obtain a resolution bandwidth prediction model; The resolution bandwidth prediction model is used to set the resolution bandwidth of the spectrum test item, specifically: the obtained resolution bandwidth prediction model is embedded in the product test system, and the center frequency f of the measured signal is read. c , bandwidth s, automatically identify the parameters reference level l, attenuation a, sweep time t and index requirement i, generate real-time feature vectors, obtain the setting probability of the preset 10 types of resolution bandwidths, and take the maximum probability.

2. The method for intelligently setting resolution bandwidth for spectrum testing of aerospace products according to claim 1, characterized in that: 70% of the sample set is extracted as the training sample set, and the remaining 30% is used as the testing sample set.

3. An intelligent setting system for resolution bandwidth of spectrum test of aerospace products, characterized by include: Main factor determination module: determines the main factors for automatic configuration of spectrum items based on the common parameters that need to be configured for the product spectrum test items; The common parameters that need to be configured for the spectrum test items of the product include: carrier frequency, bandwidth, built-in attenuation, reference level, amplitude resolution, resolution bandwidth, and video bandwidth; carrier frequency and bandwidth are the input conditions of the test; the settings of built-in attenuation, reference level, and amplitude resolution are directly affected by the output power of the measured signal, and their parameter values are a linear function of the output power of the measured signal; the built-in attenuation is used to protect the mixer inside the spectrum analyzer, and the minimum built-in attenuation value is equal to the maximum output power of the measured signal minus the maximum input power of the mixer; the reference level is equal to the maximum output power of the measured signal plus b, where b>0; the amplitude resolution is equal to the maximum output power of the measured signal minus the minimum output power of the measured signal divided by 10; the video bandwidth and resolution bandwidth are set to a fixed multiple relationship; the main factor for automatic configuration of spectrum test items is the resolution bandwidth; Feature vector construction module: According to the characteristics of the historical test data of the product spectrum test project, the factors affecting the resolution bandwidth setting are extracted and the feature vector is constructed; the factors affecting the resolution bandwidth setting extracted include: the center frequency f of the measured signal c , bandwidth s, reference level l, attenuation a, sweep time t and index requirements; The factors affecting the resolution bandwidth are used as the input vector factors for setting the resolution bandwidth neural network model, and the input feature vector is expressed as: X=[f c s l a t i] Neural network model establishment and training module: establishes a neural network model for resolution bandwidth setting, trains the neural network model using a training sample set, and obtains a resolution bandwidth prediction model; A resolution bandwidth setting module: uses the resolution bandwidth prediction model to set the resolution bandwidth of the spectrum test item; Normalize the feature vector to obtain the normalized feature vector; collect historical test data and formulate the data feature vector X k , k=1,2,…,m, where m is the maximum number of samples; and mark the samples and set the label value, then the vector X k and its corresponding labels as the sample set of the neural network model, 70% of the sample set is extracted as the training sample set, and the remaining 30% is used as the test sample set; Normalized eigenvector, calculated as: Among them, f min and f max is the spectrum analyzer test frequency, s min and s max is the spectrum analyzer test bandwidth, l min and l max is the spectrum analyzer reference level, a min and a max is the internal attenuation of the spectrum analyzer, t min and t max is the spectrum analyzer scanning time, i min and i max These are the common index requirements for traveling wave tube amplifiers; A neural network model for resolution bandwidth setting is established, and the neural network model is trained using the training sample set to obtain a resolution bandwidth prediction model. Specifically: A fully connected neural network, i.e., a neural network model for resolution bandwidth setting, is constructed, wherein the network model structure includes an input layer, three hidden layers, and an output layer; based on the constructed feature vector, the input layer vector dimension is 6, and the output layer uses softmax classification, outputting the setting probabilities of 10 types of resolution bandwidth; a training sample set is input into the neural network model, and the number of training iterations, the number of batch samples, and the learning rate are set. The network model is continuously trained and optimized to obtain a resolution bandwidth prediction model; The resolution bandwidth prediction model is used to set the resolution bandwidth of the spectrum test item, specifically: the obtained resolution bandwidth prediction model is embedded in the product test system, and the center frequency f of the measured signal is read. c , bandwidth s, automatically identify the parameters reference level l, attenuation a, sweep time t and index requirement i, generate real-time feature vectors, obtain the setting probability of the preset 10 types of resolution bandwidths, and take the maximum probability.

Images