Adaptive signal waveform recognition method, medium and device

Abstract

The present invention provides an adaptive signal waveform recognition method, medium and device, which is divided into two parts: signal waveform positioning and signal waveform recognition. Signal waveform positioning uses a center diffusion algorithm with rules to more accurately screen out all waveform segments to be identified, and then adaptively adjusts the prior waveform template according to the size and shape of the screened waveform to be identified, and then interpolates and resamples the waveform to be identified and the adaptively adjusted waveform template, and then performs peak center calibration, positive value and rebalancing on the resampled waveform to eliminate the systematic error caused by the waveform position; waveform recognition regards waveform data as probability and transforms it to meet the three axioms required for probability, and uses KL divergence to measure the distribution difference between the probabilistic waveform data and template data, so as to identify the waveform model of the waveform to be measured. The present invention can directly identify the waveform data collected by the sensor and output the relevant parameters of the waveform.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to an adaptive signal waveform recognition method, medium and device. Background Art

[0002] Traditional recognition algorithms typically use the L2 norm as the error function and assume a Gaussian parameter distribution. However, any signal deformation, whether proportional or degradation, can significantly affect the final recognition results. Furthermore, while recent machine learning, especially deep learning, has achieved tremendous success in many fields, an effective deep network relies on a large amount of high-quality labeled data to ensure convergence, which is often costly. When data volumes are limited and lack annotations, supervised learning is no longer the optimal solution. In this case, extracting relevant features from the signal itself and distinguishing between similar data types is crucial for solving these problems. Furthermore, in addition to the impact of noise, communication signal recognition also presents the challenge of distinguishing between similar and heterogeneous signals. This distinction sometimes requires practical experience and reliance on prior knowledge. Therefore, an effective recognition method based on unlabeled, small-sample data must be able to capture the characteristics of each waveform and possess the ability to perform rechecks based on a certain amount of prior knowledge. Summary of the Invention

[0003] The present invention aims to provide an adaptive signal waveform recognition method, medium and device. The method is based on sub-algorithms such as KL divergence definition similarity, regular central diffusion, and template adaptive scaling. It only needs to select a suitable waveform template for different hardware devices without modifying other parts of the algorithm. The waveform data collected by the sensor can be directly recognized and the relevant parameters of the waveform can be output.

[0004] The present invention provides an adaptive signal waveform recognition method, comprising the following steps:

[0005] Step 1: Select a representative waveform template based on the acquired data;

[0006] Step 2: obtaining the original data containing the waveform to be identified, and preprocessing the original data to obtain useful data;

[0007] Step 3: Perform peak detection on useful data;

[0008] Step 4: Starting from each detected peak, perform regular center diffusion to capture the waveform to be measured as completely as possible;

[0009] Step 5: Adaptively adjust the waveform template according to the intercepted waveform to be measured so that the waveform template and the waveform to be measured are in the same proportion;

[0010] Step 6: interpolate the waveform to be measured and the waveform template after adaptive adjustment to obtain an interpolation function of the waveform to be measured and an interpolation function of the waveform template after adaptive adjustment respectively;

[0011] Step 7: resample the waveform to be measured and the waveform template after adaptive adjustment according to the interpolation function;

[0012] Step 8: Process the resampled waveform to be measured and the adaptively adjusted waveform template;

[0013] Step 9: Perform probability transformation on the processed waveform to be measured and the waveform template;

[0014] Step 10: Calculate the KL divergence of the waveform to be measured and the waveform template after the probability transformation, and compare it with a preset threshold. For the waveform templates that meet the preset threshold, the waveform template with the smallest KL divergence value is selected as the preliminary recognition result of the waveform to be measured;

[0015] Step 11: Recheck the preliminary recognition results to obtain the final recognition results.

[0016] Furthermore, in step 2, the method of preprocessing the original data includes:

[0017] Select valuable data slices, remove noise and perform smoothing to obtain useful data.

[0018] Furthermore, in step 3, the method of performing peak detection on useful data includes:

[0019] Perform first-order difference on the useful data and detect all possible peaks based on monotonicity.

[0020] Furthermore, in step 4, the regular central diffusion refers to:

[0021] First diffuse the center, then intercept the waveform to be measured, remove the tail, and then equalize to compensate.

[0022] Furthermore, in step 5, the waveform template is adaptively adjusted according to the width and height data of the intercepted waveform to be measured.

[0023] Furthermore, in step 8, the method for processing the resampled waveform to be measured and the adaptively adjusted waveform template includes:

[0024] Peak center calibration;

[0025] Positive value;

[0026] and balance processing.

[0027] The present invention also provides a computer terminal storage medium storing computer terminal executable instructions, wherein the computer terminal executable instructions are used to execute the above-mentioned adaptive signal waveform recognition method.

[0028] The present invention further provides a computing device, comprising:

[0029] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the adaptive signal waveform recognition method as described above.

[0030] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0031] 1. The present invention does not rely on a large number of labeled samples, is low-cost and easy to deploy.

[0032] 2. The present invention has strong interpretability: the recognition principles are strictly mathematically guaranteed, rather than a black box system.

[0033] 3. The present invention has strong maintainability: for waveforms that are very similar or even difficult for the human eye to distinguish the difference, new priori rules can be added at any time based on practical experience to recheck in order to reduce the false detection rate.

[0034] 4. The present invention has strong versatility: it can be used for identification based on any type of hardware detection equipment. For different hardware devices, only the waveform template data needs to be updated without making any changes to other parts of the algorithm.

[0035] 5. The present invention has strong generalization capability: it can well identify both low-power and high-power waveforms, and has strong adaptability to certain degrees of deformation and degradation of the waveform to be measured.

[0036] 6. The present invention has strong scalability: for the recognition requirements of new waveforms, corresponding waveform templates can be added at any time without changing other parts of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings in the embodiments will be briefly introduced below. It should be understood that the following drawings only illustrate certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without paying any creative work.

[0038] Figure 1 Flowchart of the adaptive signal waveform recognition method in an embodiment of the present invention.

[0039] Figure 2Flowchart of a center diffusion algorithm with rules in an embodiment of the present invention.

[0040] Figure 3 Flowchart of adaptive scaling of a waveform to be measured and a waveform template in an embodiment of the present invention.

[0041] Figure 4 3 is a comparison diagram of the recognition process and the final recognition result in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0043] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0044] Example

[0045] like Figure 1 、 Figure 2 、 Figure 3 As shown, this embodiment proposes an adaptive signal waveform recognition method, including the following steps:

[0046] Step 1: Select a representative waveform template based on the acquired data;

[0047] Step 2: Obtain the original data containing the waveform to be identified, pre-process the original data, select valuable data slices, remove noise and perform smoothing to obtain useful data;

[0048] Step 3: Perform peak detection on the useful data. First-order difference can be performed on the useful data to detect all possible peaks based on monotonicity.

[0049] Step 4: Starting from each detected peak, perform regular center diffusion to capture the waveform to be tested as completely as possible:

[0050] The core idea of ​​the rule-based center diffusion algorithm is to start from the approximate peak (which may not be the peak in the strict sense, as it has a certain robustness effect on the degradation and deformation of the wave) and diffuse to both sides. Ideally, the algorithm stops when it reaches the two ends of the wave. However, due to the different background noise, noise intensity, and signal power of different signal data, there are usually jitters of varying strengths at both ends of the wave, and the recognition effect of the traditional center diffusion algorithm is unsatisfactory. Therefore, by adding rules to it, first diffuse the center, then intercept the waveform to be tested, remove the tail, and then equalize to compensate, so that the waveform to be tested can be accurately intercepted. In detail, if Figure 2 As shown, the center diffusion algorithm with rules is as follows:

[0051] Starting from the approximate peak obtained by determining the monotonicity based on the first-order difference in step 3, the waveform data to be measured is diffused to both sides. The diffused waveform data needs to meet the requirements of minimum window length and left-right balance. At the same time, in order to avoid the jitter caused by a small amount of noise and the interference of background noise, the inverse monotonicity tolerance on the left and right sides and the rules for removing the tail and compensating by monotonicity in the tail neighborhood are set to ensure the quality of the intercepted waveform. The final effective output of the rule-based central diffusion algorithm is the accurately intercepted waveform to be measured. This step is the basis for subsequent waveform template adaptation and differentiation measurement;

[0052] Step 5: Adaptively adjust the waveform template according to the width and height data of the intercepted waveform to be measured so that the waveform template and the waveform to be measured are in the same proportion;

[0053] Step 6: interpolate the waveform to be measured and the waveform template after adaptive adjustment to obtain an interpolation function of the waveform to be measured and an interpolation function of the waveform template after adaptive adjustment respectively;

[0054] Step 7: Resample the waveform to be measured and the waveform template after adaptive adjustment according to the interpolation function:

[0055] like Figure 3 As shown, the intercepted waveform to be measured is obtained by the rule-based central diffusion algorithm in step 4, and the longitudinal scaling ratio is adaptively determined based on the ratio of the original number of sampling points of the intercepted waveform to be measured and the original number of sampling points of the waveform template (acquired by the same hardware model using the same parameters). At the same time, in order to reduce or eliminate factors that may affect the subsequent distribution, such as the phase difference between the original sampling points of the same model waveform at different times, it is considered to interpolate the two separately to obtain an interpolation function, and sample the same number of points at equal intervals. After the above resampling, the waveform template and the waveform to be identified are at the same ratio, and the subsequent recognition algorithm can be performed.

[0056] Step 8: Processing the resampled waveform to be measured and the adaptively adjusted waveform template, including:

[0057] Peak center calibration to eliminate position effects;

[0058] Positive value to ensure the rationality of subsequent probability transformation operations;

[0059] and balancing processing to ensure the symmetry of the bottom end of the waveform and eliminate the skew effect on the probability transformation;

[0060] Step 9: Perform probability transformation on the processed waveform to be measured and the waveform template, so that the waveform data after probability transformation satisfies the three axioms of probability, thereby capturing the distribution characteristics of the waveform.

[0061] Step 10: Calculate the KL divergence of the waveform to be measured and the waveform template after the probability transformation, and compare it with a preset threshold. For the waveform templates that meet the preset threshold, the waveform template with the smallest KL divergence value is selected as the preliminary recognition result of the waveform to be measured;

[0062] The core idea of ​​calculating the similarity between the waveform to be tested and the waveform template based on KL divergence is to use KL divergence to measure the distribution difference between the waveform to be tested and the waveform template. Kullback-Leibler (KL) divergence:

[0063]

[0064] is an effective measure to quantify the difference between two distributions because:

[0065]

[0066] Therefore, the KL divergence is always non-negative, and the KL divergence is 0 if and only if the two distributions are completely identical. Based on this characteristic, the smaller the KL divergence between the waveform to be tested and the waveform template after probability transformation, the smaller the difference between the two in terms of distribution, that is, the more similar the waveforms are. To ensure the credibility of the recognition results, the present invention requires peak center calibration, positive value correction, and balancing of the waveform to be identified and the waveform template before calculating the KL divergence.

[0067] Step 11: Recheck the preliminary recognition results to obtain the final recognition results. For example, for some waveforms, if there is prior knowledge gained from practice, after obtaining the preliminary recognition results, relevant rules can be written based on the prior knowledge to recheck and obtain the final recognition results based on the recheck results.

[0068] The final recognition result is Figure 4 As shown. Figure 4It can be seen that when the optical power is low, degradation is more likely to occur (the 13th waveform from left to right). By rechecking the results of the KL divergence calculation with appropriate prior rules, the false detection rate can be reduced (in terms of distribution, this waveform belongs to the same type as the 11th waveform, and the rule recheck has assigned it to the correct category).

[0069] In addition, in some embodiments, a computer terminal storage medium is provided, storing computer terminal executable instructions, wherein the computer terminal executable instructions are used to execute the adaptive signal waveform recognition method described in the above embodiments. Examples of computer storage media include magnetic storage media (e.g., floppy disks, hard disks, etc.), optical recording media (e.g., CD-ROMs, DVDs, etc.), or memories such as memory cards, ROMs, or RAMs. Computer storage media can also be distributed on network-connected computer systems, such as in an application store.

[0070] In some embodiments, a computing device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the adaptive signal waveform recognition method described in the above embodiments. Examples of computing devices include a PC, a tablet computer, a smartphone, or a PDA.

[0071] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. An adaptive signal waveform recognition method, characterized in that: The steps include: Step 1: Select a representative waveform template based on the acquired data; Step 2: obtaining the original data containing the waveform to be identified, and preprocessing the original data to obtain useful data; Step 3: Perform peak detection on useful data; Step 4: Starting from each detected peak, perform regular center diffusion to capture the waveform to be measured as completely as possible; Step 5: Adaptively adjust the waveform template according to the intercepted waveform to be measured so that the waveform template and the waveform to be measured are in the same proportion; Step 6: interpolate the waveform to be measured and the waveform template after adaptive adjustment to obtain an interpolation function of the waveform to be measured and an interpolation function of the waveform template after adaptive adjustment respectively; Step 7: resample the waveform to be measured and the waveform template after adaptive adjustment according to the interpolation function; Step 8: Process the resampled waveform to be measured and the adaptively adjusted waveform template; Step 9: Perform probability transformation on the processed waveform to be measured and the waveform template; Step 10: Calculate the KL divergence of the waveform to be measured and the waveform template after the probability transformation, and compare it with a preset threshold. For the waveform templates that meet the preset threshold, the waveform template with the smallest KL divergence value is selected as the preliminary recognition result of the waveform to be measured; Step 11: Recheck the preliminary recognition results to obtain the final recognition results.

2. The adaptive signal waveform recognition method according to claim 1, characterized in that: In step 2, the method for preprocessing the original data includes: Select valuable data slices, remove noise and perform smoothing to obtain useful data.

3. The adaptive signal waveform recognition method according to claim 1, wherein: In step 3, the method for performing peak detection on useful data includes: Perform first-order difference on the useful data and detect all possible peaks based on monotonicity.

4. The adaptive signal waveform recognition method according to claim 1, wherein: In step 4, the regular central diffusion refers to: First diffuse the center, then intercept the waveform to be measured, remove the tail, and then equalize to compensate.

5. The adaptive signal waveform recognition method according to claim 1, wherein: In step 5, the waveform template is adaptively adjusted according to the width and height data of the intercepted waveform to be measured.

6. The adaptive signal waveform recognition method according to claim 1, characterized in that: In step 8, the method for processing the resampled waveform to be measured and the adaptively adjusted waveform template includes: Peak center calibration; Positive value; and balance processing.

7. A computer terminal storage medium storing computer terminal executable instructions, characterized in that: The computer terminal executable instructions are used to execute the adaptive signal waveform recognition method according to any one of claims 1 to 6.

8. A computing device, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the adaptive signal waveform recognition method according to any one of claims 1 to 6.

Images