Multi-information-domain intelligent testing method of intelligent oscilloscope

Abstract

The invention discloses a multi-information-domain intelligent testing method for an intelligent oscilloscope, which comprises the following steps: firstly, collecting a signal to be tested in a real-time file, storing the signal into a DDR (Double Data Rate), then carrying out time-frequency analysis on collected data in the DDR, and identifying a time-frequency graph based on a trained YOLO neural network to obtain a center frequency and a Span parameter of a time-frequency event; the mixing frequency and the extraction multiplying power of the down converter are adjusted in a self-adaptive mode according to the parameters; frequency spectrums of collected signals are shifted to a base band through down-conversion, and meanwhile the sampling rate of the frequency spectrums is reduced so that higher frequency resolution can be achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of digital oscilloscopes, and more specifically, relates to a multi-information domain intelligent testing method for an intelligent oscilloscope. Background Art

[0002] With the rapid development of modern electronic information technology, the measurement signals of digital oscilloscopes present complex and diverse characteristics. The signal bandwidth is getting wider and wider, the transmission rate is getting higher and higher, and the non-stationary characteristics are also growing rapidly. Simply analyzing the signal from the perspective of time domain can no longer meet the needs, and the analysis capabilities of frequency domain and time-frequency domain are becoming more and more important. People hope that the measuring instrument can not only analyze the time domain characteristics of the signal, but also analyze the characteristics of other domains such as frequency domain, time-frequency domain, amplitude domain, etc.

[0003] However, in traditional electronic measurement platforms, the acquisition of high-speed signals often faces hardware performance limitations, especially when recording for a long time, the storage and processing speed becomes a bottleneck. At the same time, with the sharp increase in the amount of test signal data, how to effectively store and process these large-capacity data becomes a challenge. In addition, the complexity and diversity of measurement signals are becoming increasingly prominent. In order to improve the efficiency and accuracy of data analysis, the introduction of more intelligent automation technology is also urgent. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a multi-information domain intelligent testing method for an intelligent oscilloscope, which can realize efficient time-frequency analysis of large amounts of data under a delay of seconds, and focus on and personalize the analysis of information-intensive areas containing time-frequency events, thereby improving the accuracy and efficiency of data analysis.

[0005] To achieve the above-mentioned object of the invention, the present invention provides a multi-information domain intelligent testing method for an intelligent oscilloscope, characterized in that it comprises the following steps:

[0006] (1) Set the digital oscilloscope parameters;

[0007] Input the signal to be tested into the digital oscilloscope, and then adjust the time base and amplitude of the digital oscilloscope, adjust the time base to the real-time position, and adjust the amplitude of the digital oscilloscope so that the amplitude of the signal to be tested is greater than 6 grids and does not exceed the full screen;

[0008] (2) Collect the signal to be tested;

[0009] The signal to be tested is collected in real-time mode for a collection time of T, and then the collected data is stored in DDR;

[0010] T=D / F s

[0011] Among them, F sis the sampling rate of the digital oscilloscope, and D is the storage depth of DDR;

[0012] (3) Read the collected data from DDR to generate a two-dimensional time-frequency graph;

[0013] (4) Train the YOLO network model;

[0014] Each two-dimensional time-frequency graph is used as the input of the YOLO network model, and the output is the coordinates (x, y), area (w, h) and signal category of the time-frequency event in the two-dimensional time-frequency graph. The trained YOLO network model is obtained through repeated iterative training.

[0015] (5) Intelligent recognition of time-frequency graphs;

[0016] The industrial computer reads a set of collected data of length L from the DDR to generate a two-dimensional time-frequency graph;

[0017] Input the two-dimensional time-frequency graph into the trained YOLO network model to obtain the coordinates (x, y), area (w, h) and signal category of the time-frequency event;

[0018] (6) Calculate the central frequency f of the time-frequency event c and frequency width Span;

[0019] (7) Update the parameters of the digital oscilloscope;

[0020] The center frequency f c The frequency width Span is sent to the down converter to adjust the mixing frequency of the down converter to f c , the extraction ratio is

[0021] (8) The data collected in step (2) is down-converted by the down-converter, the spectrum is moved to the baseband, and the sampling rate becomes F' s =Span; then perform a 1k-point FFT transform on the down-converted signal to obtain the signal's spectrum. Finally, upload the spectrum to the display and automatically display the analysis view belonging to the signal type. The analysis view includes frequency domain analysis, modulation domain analysis and protocol analysis.

[0022] The object of the invention of the present invention is achieved in this way:

[0023] The invention discloses a multi-information domain intelligent testing method for an intelligent oscilloscope. The method comprises the following steps: firstly, a signal to be tested is collected in real time and stored in a DDR, then a time-frequency analysis is performed on the collected data in the DDR, and a time-frequency diagram is identified based on a trained YOLO neural network to obtain a central frequency and a Span parameter of a time-frequency event, and a mixing frequency and a decimation ratio of a down-converter are adaptively adjusted according to the parameters; the spectrum of the collected signal is moved to a baseband through down-conversion, and the sampling rate is reduced to achieve a higher frequency resolution.

[0024] At the same time, the multi-information domain intelligent testing method of an intelligent oscilloscope of the present invention also has the following beneficial effects:

[0025] (1) Automatic detection of time-frequency events is achieved by identifying the time-frequency distribution of the data collected in DDR;

[0026] (2) By automatically adjusting the mixing frequency and decimation ratio of the down-converter, the spectrum of the acquired signal is moved to the baseband, achieving automatic spectrum tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is the multi-information domain intelligent test principle diagram of the intelligent oscilloscope;

[0028] Figure 2 It is a schematic diagram of a two-dimensional time-frequency diagram;

[0029] Figure 3 It is a time-frequency event identification diagram in a two-dimensional time-frequency diagram;

[0030] Figure 4 It is a flow chart of the down converter entering the processing of the collected data;

[0031] Figure 5 is an analysis view of a linear frequency modulation signal. DETAILED DESCRIPTION

[0032] The specific implementation of the present invention is described below in conjunction with the accompanying drawings so that those skilled in the art can better understand the present invention. It should be noted that in the following description, when the detailed description of known functions and designs may dilute the main content of the present invention, these descriptions will be omitted here.

[0033] Example

[0034] Figure 1 It is the multi-information domain intelligent test principle diagram of the intelligent oscilloscope.

[0035] In this embodiment, if Figure 1 As shown, the present invention provides a multi-information domain intelligent testing method for an intelligent oscilloscope, comprising the following steps:

[0036] (1) Set the digital oscilloscope parameters;

[0037] Input the signal to be tested into the digital oscilloscope, and then adjust the time base and amplitude of the digital oscilloscope, adjust the time base to the real-time position, and adjust the amplitude of the digital oscilloscope so that the amplitude of the signal to be tested is greater than 6 grids and does not exceed the full screen;

[0038] (2) Collect the signal to be tested;

[0039] The signal to be tested is collected in real-time mode for a collection time of T, and then the collected data is stored in DDR;

[0040] T=D / F s

[0041] Among them, F s is the sampling rate of the digital oscilloscope, and D is the storage depth of DDR;

[0042] (3) Generate time-frequency graph;

[0043] (3.1) Set the parameters of time-frequency analysis, including the width W and height H of the time-frequency graph;

[0044] (3.2) Calculate the data length L and address step S of the collected data read by the industrial computer from the DDR;

[0045]

[0046] (3.3) The industrial computer reads the collected data of length L from the DDR with address step S, and reads out n groups of collected data in total;

[0047] (3.4) Perform H-point FFT transformation on each set of collected data, a total of W times, to obtain a spectrum matrix of size W*H;

[0048] (3.5), the industrial computer maps each set of spectrum matrices into a two-dimensional time-frequency graph, and obtains a total of n two-dimensional time-frequency graphs;

[0049] (4) Train the YOLO network model;

[0050] Each two-dimensional time-frequency graph is used as the input of the YOLO network model, and the output is the coordinates (x, y), area (w, h) and signal category of the time-frequency event in the two-dimensional time-frequency graph. The trained YOLO network model is obtained through repeated iterative training.

[0051] (5) Intelligent recognition of time-frequency graphs;

[0052] The industrial computer reads a set of collected data of length L from the DDR, and then generates a two-dimensional time-frequency graph according to steps (3.4) and (3.5);

[0053] In this embodiment, the two-dimensional time-frequency diagram is as follows: Figure 2 As shown, the horizontal axis of the time-frequency diagram is time, the vertical axis is frequency, and the color and brightness represent energy intensity.

[0054] Input the two-dimensional time-frequency graph into the trained YOLO network model to obtain the coordinates (x, y), area (w, h) and signal category of the time-frequency event;

[0055] In this embodiment, if Figure 3 As shown in the figure, the box in the figure is the identified time-frequency event, the coordinates of the lower left corner of the box are (x, y), the width of the box is w, and the height is h.

[0056] (6) Calculate the central frequency f of the time-frequency event c and frequency width Span;

[0057]

[0058] (7) Update the parameters of the digital oscilloscope;

[0059] The center frequency f c The frequency width Span is sent to the down converter to adjust the mixing frequency of the down converter to f c , the extraction ratio is

[0060] (8) The data collected in step (2) is down-converted by the down-converter, the spectrum is moved to the baseband, and the sampling rate becomes F' s = Span; In this embodiment, the down converter processes the collected data into the following process: Figure 4 As shown, a baseband signal with a reduced sampling rate is obtained through mixing and decimation, and finally a 1k-point FFT transform is performed on the down-converted baseband signal to obtain the signal spectrum. Finally, the spectrum is uploaded to the display and the analysis view belonging to the signal type is automatically displayed. The analysis view includes frequency domain analysis, modulation domain analysis and protocol analysis.

[0061] In this embodiment, taking the linear frequency modulation signal as an example, the center frequency of the time-frequency event is identified to be 100 MHz, and the frequency bandwidth range is 100 MHz. Then, the mixing frequency and the decimation ratio of the down converter are updated. The linear frequency modulation signal is processed by step (8) to obtain the spectrum of the signal. Then, the spectrum is uploaded to the display for display, and the analysis view of the linear frequency modulation signal is displayed at the same time, such as Figure 5 As shown, the upper figure is the time-frequency diagram of the linear frequency modulation signal, the middle figure is the spectrum diagram of the linear frequency modulation signal, and the lower figure is the time domain waveform of the linear frequency modulation signal.

[0062] Although the above describes the illustrative specific embodiments of the present invention to facilitate those skilled in the art to understand the present invention, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations using the concept of the present invention are protected.

Claims

1. A multi-information domain intelligent testing method for an intelligent oscilloscope, characterized in that: The following steps are involved: (1) Set the digital oscilloscope parameters; Input the signal to be tested into the digital oscilloscope, and then adjust the time base and amplitude of the digital oscilloscope, adjust the time base to the real-time position, and adjust the amplitude of the digital oscilloscope so that the amplitude of the signal to be tested is greater than 6 grids and does not exceed the full screen; (2) Collect the signal to be tested; The signal to be tested is collected in real-time mode for a collection time of T, and then the collected data is stored in DDR; T=D / F s Among them, F s is the sampling rate of the digital oscilloscope, and D is the storage depth of DDR; (3) Read the collected data from DDR to generate a two-dimensional time-frequency graph; (4) Train the YOLO network model; Each two-dimensional time-frequency graph is used as the input of the YOLO network model, and the output is the coordinates (x, y), area (w, h) and signal category of the time-frequency event in the two-dimensional time-frequency graph. The trained YOLO network model is obtained through repeated iterative training. (5) Intelligent recognition of time-frequency graphs; The industrial computer reads a set of collected data of length L from the DDR to generate a two-dimensional time-frequency graph; Input the two-dimensional time-frequency graph into the trained YOLO network model to obtain the coordinates (x, y), area (w, h) and signal category of the time-frequency event; (6) Calculate the central frequency f of the time-frequency event c and frequency width Span; (7) Update the parameters of the digital oscilloscope; The center frequency f c The frequency width Span is sent to the down converter to adjust the mixing frequency of the down converter to f c , the extraction ratio is (8) The data collected in step (2) is down-converted by the down-converter, the spectrum is moved to the baseband, and the sampling rate becomes F' s =Span; then perform a 1k-point FFT transform on the down-converted signal to obtain the signal's spectrum. Finally, upload the spectrum to the display and automatically display the analysis view belonging to the signal type. The analysis view includes frequency domain analysis, modulation domain analysis and protocol analysis.

2. The multi-information domain intelligent testing method of an intelligent oscilloscope according to claim 1, characterized in that: The method for reading the collected data and generating a two-dimensional time-frequency graph in step (3) is: (2.1) Set the parameters of time-frequency analysis, including the width W and height H of the time-frequency graph; (2.2) Calculate the data length L and address step S of the collected data read by the industrial computer from the DDR; (2.3) The industrial computer reads the collected data of length L from the DDR with address step S, and reads out n groups of collected data in total; (2.4) Perform H-point FFT transformation on each set of collected data, a total of W times, to obtain a spectrum matrix of size W*H; (2.5), the industrial computer maps each set of spectrum matrices into a two-dimensional time-frequency graph, and obtains a total of n two-dimensional time-frequency graphs.

3. The multi-information domain intelligent testing method of an intelligent oscilloscope according to claim 1, characterized in that: The central frequency f of the time-frequency event c The calculation formula for the sum frequency width Span is:

Images