Airplane radio frequency cable in-situ detection system and method
By combining multi-parameter fusion analysis and intelligent diagnostic models with components such as display control terminals and signal generation units, rapid and accurate detection of radio frequency cables is achieved, solving the problems of low detection efficiency and insufficient accuracy in existing technologies, and improving the flexibility and fault location accuracy of the detection system.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-04-07
AI Technical Summary
In existing technologies, radio frequency cable testing is inefficient, environmentally limited, and lacks accuracy, making it difficult to achieve rapid and accurate fault assessment and diagnosis.
By employing a multi-parameter fusion analysis and intelligent diagnostic model, and combining a display control terminal, a signal generation unit, a directional coupler, an adaptive filtering unit, a multi-parameter analysis unit, and an intelligent diagnostic unit, in-situ detection of radio frequency cables is achieved, and convolutional neural networks are used for fault classification and location.
It enables rapid and accurate testing of radio frequency cables, improves the flexibility and accuracy of testing, reduces the impact of electromagnetic interference, and enhances the practicality of the testing system and the accuracy of fault location.
Smart Images
Figure CN121805894A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of radio frequency cable testing technology for avionics systems, and specifically to an in-situ testing system and method for aircraft radio frequency cables. Background Technology
[0002] Radio frequency (RF) cables, as an important medium for RF signal transmission, are widely used in equipment such as aircraft, ships, and missiles. Their transmission performance is greatly affected by factors such as laying path, bending radius, shielding layer damage, extrusion deformation, and joint oxidation. Moreover, with the increase of usage time, performance degradation will occur, directly affecting signal attenuation, transmission, and even the realization and operation of equipment functions.
[0003] To avoid the above problems, a rapid, accurate, and reliable testing method is needed, capable of accurate and intelligent assessment and diagnosis of RF cable faults. Existing testing methods have the following shortcomings: 1) Low detection efficiency: Traditional detection methods require disassembling the radio frequency cable for visual inspection and laboratory testing, which is inefficient and prone to causing secondary damage.
[0004] 2) Limited testing environment: For radio frequency cables that have already been laid, the difficulty of disassembling the cables is due to factors such as the laying path and narrow channels, making it impossible to conduct comprehensive testing and evaluation of the radio frequency cables.
[0005] 3) Low detection accuracy: Existing in-situ detection technologies (such as time domain reflectometry) are sensitive to terminal load, limited by the multi-node topology of the aircraft and complex electromagnetic interference, and rely on the personal experience of the testing personnel for assessment, making it difficult to accurately locate faults.
[0006] Therefore, there is an urgent need for an in-situ detection method that requires no disassembly, has good environmental adaptability, and high detection accuracy. Summary of the Invention
[0007] This invention provides an in-situ testing system and method for aircraft radio frequency cables. Through multi-parameter fusion analysis and intelligent diagnostic models, it can effectively solve the problems of low testing efficiency, limited testing environment, and insufficient accuracy of traditional testing methods, and achieve rapid and accurate testing of the transmission performance of aircraft radio frequency cables, significantly improving the reliability and ease of maintenance of aircraft radio frequency cables.
[0008] Technical solution: In a first aspect, this application provides an in-situ inspection system for aircraft radio frequency cables, comprising a display and control terminal, a signal generation unit, a directional coupler, an adaptive filtering unit, a multi-parameter analysis unit, and an intelligent diagnostic unit, wherein: The display control terminal is connected to the signal generation unit and the intelligent diagnostic unit via network cables to ensure low latency in signal transmission between modules; the signal generation unit, directional coupler, adaptive filtering unit, and multi-parameter analysis unit are connected via low-loss RF cables to ensure high fidelity in RF signal transmission; and the multi-parameter analysis unit and the intelligent diagnostic unit are connected via network to ensure real-time and stable data transmission.
[0009] Specifically, the display control terminal serves as the human-machine interface, receiving and transmitting control commands and displaying detection and diagnostic results; the signal transmission unit generates the sweep frequency signal and pulse signal required for RF cable detection; the directional coupler separates the forward signal from the reflected signal; the adaptive filtering unit suppresses environmental electromagnetic noise signals and dynamically adjusts the filtering parameters; the multi-parameter analysis unit synchronously acquires detection signals and calculates time-domain parameters, frequency-domain parameters, and phase changes; and the intelligent diagnostic unit uses a convolutional neural network to classify and locate faults.
[0010] Specifically, the multi-parameter analysis unit includes a signal acquisition and reception module, a signal frequency conversion module, an analog-to-digital conversion module, and a digital processing module; The signal acquisition and receiving module is connected to the frequency conversion module via a coaxial cable to ensure stable signal transmission; the signal frequency conversion module is connected to the analog-to-digital conversion module via a microstrip line to ensure complete and correct signal transmission; and the analog-to-digital conversion module is connected to the digital processing module via a data bus to ensure high-speed processing of digital signals.
[0011] Specifically, the signal acquisition and receiving module is equipped with an RF port, which is responsible for acquiring and receiving detection signals; the signal frequency conversion module down-converts the detection signals to generate fixed, low-frequency intermediate frequency analog signals; the analog-to-digital conversion module converts the intermediate frequency analog signals into digital signals; and the digital signal processing module performs mathematical calculations on the digital signals to obtain time-domain parameters, frequency-domain parameters, and phase parameters.
[0012] Specifically, the intelligent diagnostic unit includes a data acquisition module, a preprocessing module, a fusion processing module, a model training module, and a fault diagnosis module.
[0013] Specifically, the data acquisition module collects time-domain waveform parameters, frequency-domain parameters, and phase parameters; the preprocessing module formats and standardizes the acquired parameters to facilitate subsequent data fusion and computational analysis; the fusion processing module includes data fusion and feature fusion to increase the reliability of model diagnostic results; the model training module includes large-scale general data pre-training, domain-adaptive fine-tuning, task-specific joint fine-tuning, and model validation to improve the robustness of the final diagnostic performance; and the fault diagnosis module generates a three-dimensional fault topology map through the trained model and labels the fault type, location, and confidence level.
[0014] Specifically, the fusion processing module uses wavelet transform threshold denoising to remove random noise from the parameter signal while retaining key features. At the same time, it normalizes data of different dimensions to eliminate the influence of dimensions and facilitate model convergence. Feature fusion extracts key features that can characterize the state of the RF cable from the preprocessed data to form a high-dimensional joint feature vector. Principal component analysis is used to reduce the dimensionality of the high-dimensional joint features, remove redundant information, and retain the core features.
[0015] Secondly, this application provides a method for in-situ testing of aircraft radio frequency cables, the method comprising: Step S01: Reliably connect the cable under test to the RF interface of the testing system; Step S02: Send control commands through the display control terminal to cause the signal generation unit to generate a standard detection signal, which may include a linear sweep frequency signal and a pulse signal; Step S03: The detection signal is injected into the cable under test through a directional coupler; Step S04: The reflected signal returned by the RF cable under test is transmitted again through the directional coupler to the adaptive filtering unit for filtering processing, and then output to the multi-parameter analysis unit; Step S05: The multi-parameter analysis module acquires the signal in real time, performs down-conversion and analog-to-digital conversion on the reflected signal, and calculates the time-domain parameters, frequency-domain parameters, and phase parameters; Step S06: Input the time domain parameters, frequency domain parameters and phase parameters into the intelligent diagnostic unit for preprocessing and data fusion. Through the trained CNN-LSTM model, output a three-dimensional fault topology map and fault type, location and confidence level.
[0016] Specifically, step S06 includes: Step S061: Preprocess the time-domain parameters, frequency-domain parameters, and phase parameters; Step S062: Perform channel fusion and feature fusion on the time domain and frequency domain parameters, while ensuring that the time domain and frequency domain parameters are aligned in time or frequency; Step S063: Design the CNN-LSTM model; Step S064: Train the CNN-LSTM model; Step S065: Evaluate the trained CNN-LSTM model.
[0017] In summary, this invention provides an in-situ detection system and method for aircraft radio frequency cables, which has the following advantages compared to existing technologies: By designing the detection system into independently detachable functional modules, it is more flexible and convenient in actual use, while facilitating subsequent functional expansion and daily maintenance, thus improving practicality; the detection system adopts a double-layer metal shielding structure to reduce the impact of electromagnetic interference inside the aircraft on the detection signal, and uses an adaptive filtering algorithm to automatically adjust the bandpass filter parameters to achieve noise reduction processing of the original signal, ensuring signal separation and effective detection; by using a convolutional neural network algorithm to fuse the analysis results of multi-dimensional parameters such as time-domain parameters and frequency-domain parameters, a fault feature map of the radio frequency cable is constructed, enabling accurate identification of typical faults such as short circuits, open circuits, and impedance mismatches, effectively improving the accuracy of radio frequency cable detection and fault location. Attached Figure Description
[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0019] Figure 1 A schematic diagram of the structure of an in-situ detection system for aircraft radio frequency cables according to a preferred embodiment of the present invention; Figure 2 A flowchart of a preferred embodiment of the radio frequency cable detection method of the present invention; Figure 3 A flowchart of the intelligent diagnostic model in a preferred embodiment of the present invention. Detailed Implementation
[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0021] Example 1 like Figure 1 As shown, this invention provides an in-situ inspection system for aircraft radio frequency cables, including a display and control terminal, a signal generation unit, a directional coupler, an adaptive filtering unit, a multi-parameter analysis unit, and an intelligent diagnostic unit, wherein: The display control terminal is connected to the signal generation unit and the intelligent diagnostic unit via network cables to ensure low latency in signal transmission between modules; the signal generation unit, directional coupler, adaptive filtering unit, and multi-parameter analysis unit are connected via low-loss RF cables to ensure high fidelity in RF signal transmission; and the multi-parameter analysis unit and the intelligent diagnostic unit are connected via network to ensure real-time and stable data transmission.
[0022] Specifically, the display control terminal serves as a human-machine interface, capable of receiving and transmitting control commands and displaying detection and diagnostic results; the signal transmission unit generates the sweep frequency signal and pulse signal required for RF cable detection; the directional coupler separates the forward signal from the reflected signal; the adaptive filtering unit suppresses environmental electromagnetic noise signals and dynamically adjusts filtering parameters; the multi-parameter analysis unit synchronously acquires detection signals and calculates time-domain parameters, frequency-domain parameters, and phase changes; and the intelligent diagnostic unit classifies and locates faults based on convolutional neural networks.
[0023] The multi-parameter analysis unit mainly includes a signal acquisition and receiving module, a signal frequency conversion module, an analog-to-digital conversion module, and a digital processing module. The signal acquisition and receiving module is connected to the frequency conversion module via a coaxial cable to ensure stable signal transmission; the signal frequency conversion module is connected to the analog-to-digital conversion module via a microstrip line to ensure complete and accurate signal transmission; and the analog-to-digital conversion module is connected to the digital processing module via a data bus to ensure high-speed digital signal processing.
[0024] Specifically, the signal acquisition and receiving module is equipped with an RF port, which is responsible for acquiring and receiving detection signals; the signal frequency conversion module can down-convert the detection signal to generate a fixed, low-frequency intermediate frequency analog signal; the analog-to-digital conversion module can convert the intermediate frequency analog signal into a digital signal; and the digital signal processing module can perform mathematical calculations on the digital signal to obtain time-domain parameters, frequency-domain parameters, and phase parameters, etc.
[0025] The intelligent diagnostic unit includes a data acquisition module, a preprocessing module, a fusion processing module, a model training module, and a fault diagnosis module.
[0026] Specifically, the data acquisition module can collect time-domain waveform parameters, frequency-domain parameters, and phase parameters; the preprocessing module can format and standardize the acquired parameters to facilitate subsequent data fusion and computational analysis; the fusion processing module includes data fusion and feature fusion to increase the reliability of model diagnostic results; the model training module includes large-scale general data pre-training, domain-adaptive fine-tuning, task-specific joint fine-tuning, and model validation to improve the robustness of the final diagnostic performance; and the fault diagnosis module can generate a three-dimensional fault topology map through the trained model and label the fault type, location, and confidence level.
[0027] The data fusion method employs wavelet transform threshold denoising, which removes random noise from parameter signals while retaining key features (such as fault reflections). It also normalizes data of different dimensions, eliminating dimensional influences and facilitating model convergence. Feature fusion extracts key features characterizing the RF cable's state from the preprocessed data (time-domain parameters: reflection peak amplitude, location, width, rise time, waveform entropy, etc.; frequency-domain parameters: return loss S11 amplitude and phase, insertion loss S21 amplitude and phase, bandwidth, resonant frequency, quality factor, group delay variation, etc.), forming a high-dimensional joint feature vector. Principal component analysis is then used to reduce the dimensionality of this joint feature vector, eliminating redundant information and retaining core features. This reduces the burden on subsequent models and improves model generalization ability.
[0028] The model is based on a CNN-LSTM hybrid architecture. In the pre-training phase, a large dataset of historical, simulation, and analog data is used for pre-training, facilitating the CNN part to learn and extract general time-domain / frequency-domain features, and the LSTM part to learn to understand the general laws of phase changes, thus initializing feature extraction. In the domain-adaptive fine-tuning phase, a dataset of laboratory test data for aircraft RF cables is used for training and fine-tuning, enabling the model to adapt to the unique signal characteristics of aircraft RF cables. In the task-specific joint fine-tuning phase, small-scale, high-quality measured and maintenance verification data are used to optimize the model. A total loss function is established using weighted cross-entropy loss and smoothing loss functions: Total Loss = λ1 * Classification Loss + λ2 * Localization Loss, where λ1 and λ2 are hyperparameters determined by Bayesian optimization to balance the importance of fault classification and localization tasks. Through joint training, the model can accurately match the fault modes and localization accuracy requirements in the actual aircraft RF cable detection process. In the model validation phase, hierarchical cross-validation and multi-model ensemble validation methods are used to ensure the representativeness and accuracy of the model.
[0029] Specifically, the working principle of the in-situ detection system for aircraft radio frequency cables provided by this invention is as follows: The cable under test is connected to the detection system. The display control terminal sends control commands through the network cable to cause the signal generation unit to generate a standard detection signal. The standard detection signal is transmitted to the directional coupler through the coaxial cable and then divided into a forward signal and a reverse signal. The forward signal enters the cable under test through the RF port, while the reverse signal serves as a reference signal and passes through the adaptive filtering unit to the multi-parameter analysis unit for subsequent analysis and calculation. The forward signal returns after passing through the cable under test, forming a reflected signal, which is transmitted to the adaptive filtering unit through the directional coupler for noise reduction. The processed reflected signal enters the multi-parameter analysis unit for down-conversion, analog-to-digital conversion, etc., and calculates time-domain parameters, frequency-domain parameters, and phase parameters in combination with the reference signal. The intelligent diagnostic unit performs data processing and fusion analysis on the various parameters generated by the multi-parameter analysis unit, and generates a three-dimensional fault topology map based on the constructed model, and marks the fault type, location, and confidence level.
[0030] Example 2 like Figure 2 As shown, this invention provides an in-situ inspection method for aircraft radio frequency cables, applied to the in-situ inspection system for aircraft radio frequency cables provided in the above embodiments. The method includes: Step S01: Reliably connect the cable under test to the RF interface of the testing system; Step S02: Send control commands through the display control terminal to cause the signal generation unit to generate a standard detection signal, which may include a linear sweep frequency signal and a pulse signal; The frequency range of the linear sweep signal is set to 2MHz to 6GHz, the number of scan points is set to 1600, and the output power is 0dBm. The pulse signal has a pulse width of 5–10 ns, a rise time of 100 ps, a pulse amplitude of ±7V, a pulse repetition frequency of 100 kHz, and the parameters of the sweep frequency signal and the pulse signal can be dynamically adjusted and adapted.
[0031] Step S03: The detection signal is injected into the cable under test through a directional coupler; Step S04: The reflected signal returned by the RF cable under test is transmitted again through the directional coupler to the adaptive filtering unit for filtering processing, and then output to the multi-parameter analysis unit; Step S05: The multi-parameter analysis module acquires signals in real time, performs down-conversion and analog-to-digital conversion on the reflected signals, and calculates time-domain parameters, frequency-domain parameters, and phase parameters.
[0032] Step S06: Input time domain parameters, frequency domain parameters and phase parameters into the intelligent diagnostic unit for preprocessing and data fusion. Through the trained CNN-LSTM model, output a three-dimensional fault topology map and fault type, location and confidence level.
[0033] Specifically, such as Figure 3 As shown, step S06 includes: Step S061: Preprocess the time domain parameters, frequency domain parameters, and phase parameters, including data cleaning, format conversion, and standardization, to ensure that the data format is suitable for input into the CNN-LSTM model and that features from different data sources are comparable.
[0034] Preferably, wavelet transform is used to denoise the time-domain waveform, with the wavelet basis function being db4, the decomposition level being 3, the thresholding method being a general threshold, and the threshold selection rule being a soft threshold.
[0035] Preferably, a Savitzky-Golay filter is used for frequency domain smoothing, with a window length of 11 and a polynomial order of 3.
[0036] Step S062: Perform channel fusion and feature fusion on the time domain and frequency domain parameters, while ensuring that the time domain and frequency domain parameters are aligned in time or frequency.
[0037] Preferably, the time-domain parameters include reflection peak amplitude, position, width, rise time, waveform entropy, etc., and the frequency-domain parameters include return loss S11 amplitude and phase, insertion loss S21 amplitude and phase, bandwidth, resonant frequency, quality factor, group delay variation, etc.
[0038] Preferably, feature fusion is performed using principal component analysis, retaining principal components with a variance contribution rate of over 85%.
[0039] Step S063: Design the CNN-LSTM model.
[0040] Preferably, the input layer is used to receive time-domain and frequency-domain parameters, the convolutional layer is used to extract local features of the data, the pooling layer is used to reduce computational complexity and extract more abstract features, the fusion layer combines time-domain and frequency-domain parameters, the fully connected layer is used for classification or regression tasks, and the output layer is used for classification output of fault detection.
[0041] Preferably, three one-dimensional convolutional layers are set, with the number of filters being 64, 128, and 256, and the convolutional kernel sizes being 7, 5, and 3, respectively. Each convolutional layer is followed by a max pooling layer with a pooling window size of 2 and a stride of 2.
[0042] Preferably, two LSTM layers are set up with 128 and 64 hidden units respectively. The first layer is configured to return the complete sequence, and the second layer is configured to return only the final state. Each layer includes a dropout mechanism with a dropout rate of 0.2.
[0043] Preferably, the output results of the output layer are classified as: normal, open circuit, short circuit, impedance mismatch, poor contact, insulation aging, shielding layer damage, and partial discharge.
[0044] Step S064: Train the CNN-LSTM model.
[0045] Preferably, the data is divided into training set, validation set and test set, an appropriate loss function is selected for task classification, the Adam optimizer is used to optimize the CNN-LSTM model, the training set is used to train the model, and the performance is monitored and the model structure is adjusted during the training process.
[0046] Step S065: Evaluate the trained CNN-LSTM model.
Claims
1. An in-situ inspection system for aircraft radio frequency cables, characterized in that, It includes a display control terminal, a signal generation unit, a directional coupler, an adaptive filtering unit, a multi-parameter analysis unit, and an intelligent diagnostic unit, wherein: The display control terminal is connected to the signal generation unit and the intelligent diagnostic unit via network cables to ensure low latency in signal transmission between modules; the signal generation unit, directional coupler, adaptive filtering unit, and multi-parameter analysis unit are connected via low-loss RF cables to ensure high fidelity in RF signal transmission; and the multi-parameter analysis unit and the intelligent diagnostic unit are connected via network to ensure real-time and stable data transmission.
2. The system according to claim 1, characterized in that, The display control terminal serves as the human-machine interface, receiving and transmitting control commands and displaying detection and diagnostic results; the signal transmission unit generates the sweep frequency signal and pulse signal required for radio frequency cable detection. The directional coupler separates the forward and reflected signals; the adaptive filtering unit suppresses environmental electromagnetic noise and dynamically adjusts the filtering parameters; the multi-parameter analysis unit synchronously acquires the detection signal and calculates time-domain parameters, frequency-domain parameters, and phase changes; and the intelligent diagnostic unit classifies and locates faults based on a convolutional neural network.
3. The system according to claim 1, characterized in that, The multi-parameter analysis unit includes a signal acquisition and reception module, a signal frequency conversion module, an analog-to-digital conversion module, and a digital processing module; The signal acquisition and receiving module is connected to the frequency conversion module via a coaxial cable to ensure stable signal transmission; the signal frequency conversion module is connected to the analog-to-digital conversion module via a microstrip line to ensure complete and correct signal transmission; and the analog-to-digital conversion module is connected to the digital processing module via a data bus to ensure high-speed processing of digital signals.
4. The system according to claim 3, characterized in that, The signal acquisition and receiving module is equipped with an RF port, which is responsible for acquiring and receiving detection signals; the signal frequency conversion module down-converts the detection signals to generate fixed, low-frequency intermediate frequency analog signals; the analog-to-digital conversion module converts the intermediate frequency analog signals into digital signals; and the digital signal processing module performs mathematical calculations on the digital signals to obtain time-domain parameters, frequency-domain parameters, and phase parameters.
5. The system according to claim 1, characterized in that, The intelligent diagnostic unit includes a data acquisition module, a preprocessing module, a fusion processing module, a model training module, and a fault diagnosis module.
6. The system according to claim 5, characterized in that, The data acquisition module collects time-domain waveform parameters, frequency-domain parameters, and phase parameters; the preprocessing module formats and standardizes the acquired parameters to facilitate subsequent data fusion and computational analysis; the fusion processing module includes data fusion and feature fusion to increase the reliability of model diagnostic results; the model training module includes large-scale general data pre-training, domain-adaptive fine-tuning, task-specific joint fine-tuning, and model validation to improve the robustness of the final diagnostic performance; the fault diagnosis module generates a three-dimensional fault topology map through the trained model and labels the fault type, location, and confidence level.
7. The system according to claim 5, characterized in that, The fusion processing module uses wavelet transform threshold denoising to remove random noise from the parameter signal while retaining key features. At the same time, it normalizes data of different dimensions to eliminate the influence of dimensions and facilitate model convergence. Feature fusion extracts key features that characterize the state of the RF cable from the preprocessed data, forming a high-dimensional joint feature vector. Principal component analysis is then used to reduce the dimensionality of the high-dimensional joint features, eliminating redundant information and retaining the core features.
8. A method for in-situ testing of aircraft radio frequency cables, characterized in that, The method is applied to the aircraft radio frequency cable in-situ inspection system according to any one of claims 1 to 7, and the method includes: Step S01: Reliably connect the cable under test to the RF interface of the testing system; Step S02: Send control commands through the display control terminal to cause the signal generation unit to generate a standard detection signal, which may include a linear sweep frequency signal and a pulse signal; Step S03: The detection signal is injected into the cable under test through a directional coupler; Step S04: The reflected signal returned by the RF cable under test is transmitted again through the directional coupler to the adaptive filtering unit for filtering processing, and then output to the multi-parameter analysis unit; Step S05: The multi-parameter analysis module acquires the signal in real time, performs down-conversion and analog-to-digital conversion on the reflected signal, and calculates the time-domain parameters, frequency-domain parameters, and phase parameters; Step S06: Input the time domain parameters, frequency domain parameters and phase parameters into the intelligent diagnostic unit for preprocessing and data fusion. Through the trained CNN-LSTM model, output a three-dimensional fault topology map and fault type, location and confidence level.
9. The method according to claim 8, characterized in that, Step S06 includes: Step S061: Preprocess the time-domain parameters, frequency-domain parameters, and phase parameters; Step S062: Perform channel fusion and feature fusion on the time domain and frequency domain parameters, while ensuring that the time domain and frequency domain parameters are aligned in time or frequency; Step S063: Design the CNN-LSTM model; Step S064: Train the CNN-LSTM model; Step S065: Evaluate the trained CNN-LSTM model.