A signal detection and interval identification method based on adaptive threshold

Abstract

The application discloses a signal detection and interval identification method based on an adaptive threshold, and belongs to the field of communication signal processing. The application extracts a target number of data points from original data by performing value extraction processing on original frequency and power spectrum density data, keeps the characteristics and trends of the data, and combines mean value smoothing operation to suppress high-frequency noise, so that the power spectrum is more smooth. An adaptive threshold is calculated according to the local characteristics of the signal, and the threshold is flexibly adjusted according to the signal characteristics of different regions. The local characteristics of the power spectrum density and the dynamically set adaptive threshold are used. The minimum signal-to-noise ratio threshold is set according to channel analysis, and the ratio of the maximum power of the signal to the minimum power of the noise is calculated, false alarm judgment is performed, and the real signal interval is screened out. The power characteristics, signal-to-noise ratio and interval distance of adjacent signal intervals are analyzed, the signal intervals meeting the conditions are merged, the integrity of the signal segment is ensured, and the accuracy of signal detection is improved.

Description

Technical Field

[0001] This invention relates to a signal detection and interval identification method based on adaptive threshold, belonging to the field of communication signal processing, and is applicable to signal detection and processing applications such as wireless communication, radar detection, and satellite communication. Technical Background

[0002] With the rapid development of technologies such as wireless communication, radar detection, and satellite communication, signal detection plays an increasingly important role in various fields. The task of signal detection is typically to extract effective signals from complex signal environments, remove noise and false alarms, and accurately identify the start and end points of signals. Against this backdrop, signal detection technology is widely used in communication systems, electronic warfare, wireless sensor networks, radar monitoring, medical imaging, satellite positioning, and many other fields.

[0003] In wireless communication systems, signal detection technology is used to improve communication quality and ensure accurate information transmission. In radar detection systems, signal detection is fundamental for target identification and tracking. Especially in fields such as satellite communication and long-range radar detection, the signal detection capability directly determines the reliability and sensitivity of the system. Therefore, how to effectively perform signal detection has become a key issue in research and application across various technical fields.

[0004] Existing signal detection methods primarily rely on fixed threshold decisions, i.e., setting a fixed threshold; when the signal strength exceeds this threshold, it is considered a valid signal; otherwise, it is regarded as noise or a false alarm. However, this method has significant limitations in many complex environments, especially when multiple types of signals coexist or in noisy environments, easily leading to false alarms (false signals) and missed alarms (failure to detect valid signals). Therefore, improving the accuracy and robustness of signal detection has become an urgent problem to be solved in the field of signal detection.

[0005] Currently, although many signal detection methods have been proposed and widely applied in various fields, most methods still suffer from the limitation of fixed thresholds, making it difficult to adapt to dynamically changing signal environments. Fixed threshold methods may lead to missed detections of certain signals or misidentification of noise as signals, thus affecting detection accuracy. Furthermore, in high-noise environments or situations with multiple signal interference, existing methods struggle to balance false alarms and missed alarms, especially in systems sensitive to false alarms, where false alarms result in wasted system resources and reduced processing capacity. Simultaneously, traditional methods exhibit poor adaptability to both broadband and narrowband signals, failing to effectively handle situations with significant differences in signal type, leading to inaccurate detection of target signals. Summary of the Invention

[0006] To address the shortcomings of existing signal detection methods, such as fixed threshold limitations, the challenge of balancing false alarms and missed alarms, and insufficient detection accuracy in complex noise environments, this invention aims to provide a signal detection and interval identification method based on adaptive thresholds. By dynamically adjusting the detection threshold and combining it with the power spectrum characteristics of the signal, adaptive detection of different types of signals can be achieved. By utilizing false alarm and missed alarm suppression and signal segment merging strategies, the detection results are optimized, ensuring signal integrity and improving detection accuracy and reliability. This method is particularly suitable for fine signal identification in multi-signal environments and complex noise backgrounds.

[0007] The objective of this invention is achieved through the following technical solution.

[0008] This invention discloses a signal detection and interval identification method based on adaptive thresholding. It calculates the power spectral density of the signal from collected signal data to obtain the power distribution of the signal in the frequency domain. By sampling the original frequency and power spectral density data, a target number of data points are extracted from the original data while preserving the data's characteristics and trends. Mean smoothing is then applied to suppress high-frequency noise, resulting in a smoother power spectrum and improving the accuracy of subsequent adaptive threshold calculation and the reliability of signal detection. Based on this, an adaptive threshold is calculated according to the signal's local characteristics (such as the maximum and minimum signal values ​​and the signal-to-noise ratio). The threshold is flexibly adjusted according to the signal characteristics of different regions, achieving accurate differentiation between broadband signals, narrowband signals, and noisy regions. The local characteristics of the power spectral density and the dynamically set adaptive threshold further optimize the differentiation between signal and noise, improving the accuracy of signal detection. A minimum signal-to-noise ratio threshold is set based on channel analysis, and the ratio of the maximum signal power to the minimum noise power is calculated for false alarm judgment, thereby filtering out the true signal intervals. To address the potential issue of signal missegmentation, this invention analyzes the power characteristics, signal-to-noise ratio, and interval spacing of adjacent signal intervals. It then merges signal intervals that meet the specified conditions, ensuring signal segment integrity and improving signal detection accuracy. After false alarm removal and signal merging, an optimized signal interval is obtained, which is the merged effective signal interval (the lowest and highest frequencies of the signal). This enhances the accuracy and reliability of signal detection, effectively reducing false positives and false negatives.

[0009] This invention discloses a signal detection and interval identification method based on adaptive threshold, comprising the following steps:

[0010] Step 1: Using the acquired signal data, calculate the power spectral density (PSD) of the signal using the Fast Fourier Transform (FFT) method to obtain the power distribution of the signal in the frequency domain.

[0011]

[0012] P(f)=|X(f)|2 (2)

[0013] Where P(f) is the power spectral density, and X(f) is the frequency domain representation of the signal, representing the energy distribution of the signal in the frequency domain. The frequency characteristics of the signal are obtained according to equations (1) and (2), providing basic data for subsequent threshold setting and signal interval detection.

[0014] Step 2: Extract the target number of data points from the original data by sampling the original frequency and power spectral density data, while preserving the characteristics and trends of the data; suppress high-frequency noise through mean smoothing operation to make the power spectrum smoother, thereby improving the accuracy of subsequent adaptive threshold calculation and the reliability of signal detection.

[0015] The specific steps for implementing mean smoothing are as follows:

[0016] Step 2.1: Calculate the position of each target data point by using the ratio between the number of target points and the number of original points.

[0017]

[0018] Step 2.2: Then, calculate the index position of the original data points using the ratio:

[0019] originalIndices=ratio×(numPoints-1) (4)

[0020] Calculate the index position of the original data point using a ratio. Since the target data point may not fall exactly at the position of the original data point, it is necessary to determine its two nearest index points in the original data:

[0021]

[0022] in, Indicates rounding down. This indicates rounding up. `indexLow` and `indexHigh` represent the two nearest indices of the target data point in the original data.

[0023] Step 2.3: For each target data point, extract the frequency and power spectral density values ​​from the original data based on the calculated index. Specifically, the frequency and power spectral density of the target data point are obtained by linearly weighted averaging two adjacent original data points.

[0024]

[0025] Among them, f interpolated (j) and P interpolated (j) represent the frequency and power spectral density after decomposition, respectively, ω jThe interpolation weights represent the position of the target data point within the original data points, and are calculated using the following formula:

[0026] ω j =originalIndices(j)-floor(originalIndices(j)) (7)

[0027] Step 2.4: Handle boundary issues to avoid out-of-bounds errors. When the target data point is located at the boundary of the original data (e.g., the minimum index is less than 0 or the maximum index exceeds the data length), boundary handling is required:

[0028] If indexLow < 0, then set indexLow = 0; if indexHigh > N-1, then set indexHigh = N-1; when indexLow = indexHigh, directly take the corresponding frequency and power spectral density.

[0029] f interpolated (j)=f indexLow ,P interpolated (j)=P indexLow (8)

[0030] By handling boundary issues, we ensure that the indices of all target data points are within a valid range, thus avoiding array out-of-bounds errors.

[0031] Step 2.5: To further reduce noise and enhance signal smoothness, the power spectrum after sampling is smoothed by mean averaging. The smoothing operation reduces the impact of high-frequency noise by averaging N_smooth adjacent data points, resulting in a smoother power spectrum and thus improving the stability of signal detection.

[0032]

[0033] According to formulas (8) and (9), the frequency array f after de-smoothing is obtained. interpolated and power spectral density array P interpolated The interpolation results will be used in subsequent adaptive threshold setting and signal detection processes.

[0034] Step 3: Determine the adaptive threshold using local signal characteristics to flexibly address the different needs of broadband signals, narrowband signals, and noisy regions. Classify the region based on the signal and noise characteristics within each window, and then calculate different adaptive thresholds for different types of signal regions. The local characteristics include the maximum and minimum signal values ​​and the signal-to-noise ratio.

[0035] The specific implementation steps for step 3 are as follows:

[0036] Step 3.1: For each frequency point, by calculating the maximum and minimum values ​​of the signal data within its window, and combining this with the number of points within the window that are greater than the global mean, determine whether the window belongs to a broadband signal and noise region or a mixed signal and noise region.

[0037] If the difference between the maximum and minimum values ​​within the window is less than the set signal-to-noise ratio threshold SET snr If the number of points greater than the mean is close to the window size, then the window is determined to belong to the wideband signal area; if the signal difference within the window is less than the set signal-to-noise ratio threshold SET... snr However, if the number of points greater than the mean is small, the window is determined to be a noise area; while the rest is a region where signal and noise are mixed.

[0038] Step 3.2: Different threshold setting methods are used for different types of signal regions. For wideband signal regions, the threshold is taken as the global mean; for noisy regions, the threshold is increased by increasing the local maximum value; and for regions where signal and noise are mixed, the threshold is flexibly adjusted by calculating the local increment and the signal ratio.

[0039] The complete formula for calculating the adaptive threshold is as follows:

[0040]

[0041] Among them, T adaptive (i) represents the adaptive threshold for the i-th data point, E{data} is the global mean, max{window_data} is the maximum signal power within the window, and δ i The difference between the local maximum and the global mean is given by percent(i), where percent(i) is the proportion of points in the window that are greater than the mean.

[0042] The signal-to-noise ratio threshold SET is set externally to adapt to the environment and scenario requirements. snr Compared with the adaptive threshold adjustment method described above, this step can flexibly cope with different types of signals and noise backgrounds, significantly improving the accuracy and robustness of signal detection.

[0043] Step 4: By utilizing the local features of the power spectral density and a dynamically set adaptive threshold, signals and noise are accurately distinguished, improving the accuracy of signal detection. Formula (11) is used to detect the start and end positions of the signal:

[0044]

[0045] Among them, P i Let start be the power value at the i-th frequency point in the power spectral density. idx and end idx These represent the start and end indices of the signal interval, respectively.

[0046] Step 5: Based on channel analysis, pre-set the minimum signal-to-noise ratio (MIN) that meets the scenario and requirements. snr The ratio of the maximum power of the signal to the minimum power of the noise is calculated to determine false alarms. The formula for calculating the false alarm decision is:

[0047]

[0048] Among them, P max (i) represents the maximum power of the i-th signal segment. and These are the minimum noise values ​​on both sides of the signal segment.

[0049] If the false alarm judgment value V i Less than the set minimum signal-to-noise ratio, i.e., V i <MIN snr If the signal segment is false, it is determined to be a false alarm and the signal segment needs to be removed.

[0050] Step 6: By analyzing the characteristics of adjacent signal intervals and merging intervals that meet the conditions, the signal detection results are optimized, ensuring the integrity of the signal segment. This mechanism dynamically determines whether two signal intervals need to be merged based on the power characteristics, signal-to-noise ratio, and interval spacing of adjacent signal intervals. The signal interval is determined based on the dynamic judgment result, avoiding incorrect signal segmentation and improving the accuracy of signal detection.

[0051] The criteria for signal segment merging are based on two main conditions: the noise intensity ratio between signal segments and the result of the false alarm decision. Specifically, a merging decision value C is set. i This value depends on the relationship between the minimum noise and the average signal value of adjacent signal segments. The decision condition for merging is:

[0052]

[0053] Among them, C i Let i be the merging decision value of the i-th signal segment. P represents the minimum noise power to the right of the i-th signal segment. sigmean (i) and P sigmean (i+1) represent the average signal power of the i-th and i+1-th signal segments, respectively.

[0054] If the merged decision value C i Greater than the set threshold 1 / MIN snr And the false alarm judgment value V i Less than the set minimum signal-to-noise ratio MIN snrIf the two adjacent signal intervals are determined to be merging, a merged signal interval is obtained. In multi-signal environments and complex noise backgrounds, the merging step helps improve the accuracy of signal detection, avoids signal missegmentation or missed detection, and improves the precision and reliability of detection.

[0055] Step 7: The signal range obtained in Step 6 is the effective signal range after merging (the lowest and highest frequencies of the signal). The signal range after false alarm removal and signal merging improves the accuracy and reliability of signal detection and reduces false alarms and false negatives.

[0056] Beneficial effects:

[0057] 1. This invention discloses a signal detection and interval identification method based on adaptive thresholding. By dynamically adjusting the detection threshold, it adapts to changes in the signal's spectral characteristics and noise level, achieving accurate power spectrum detection. Compared to traditional fixed threshold methods, this method can automatically adapt to different signal-to-noise ratio environments, accurately distinguish between broadband signals, narrowband signals, and composite signals, effectively reduce false alarm rates and missed alarm rates, improve the reliability and effectiveness of the signal detection system, ensure stable signal detection even under complex background noise conditions, and enhance the accuracy and robustness of signal detection.

[0058] 2. This invention discloses a signal detection and interval identification method based on adaptive thresholds. By introducing signal interval detection and false alarm suppression mechanisms, it filters out the true signal frequency bands and removes noise interference intervals. This method fully considers the power characteristics and signal-to-noise ratio distribution of signal intervals during the detection process, resulting in more accurate signal segmentation, reduced false alarms caused by misjudgment, and improved detection accuracy. It is particularly suitable for fine signal identification tasks in complex spectrum environments.

[0059] 3. This invention discloses a signal detection and interval recognition method based on adaptive thresholds. Through a signal segment merging algorithm, it performs feature analysis and dynamically merges adjacent signal intervals, improving signal integrity and avoiding multiple detection errors caused by signal missegmentation. This method is particularly suitable for multi-signal environments, effectively optimizing detection results, making signal intervals more continuous, reducing missed detections caused by signal segment breaks, and improving the overall accuracy and robustness of signal detection.

[0060] 4. This invention discloses a signal detection and interval identification method based on adaptive thresholds. By adjusting the signal-to-noise ratio (SNR) threshold, adaptively adjusting the threshold calculation rules, and optimizing the signal segment merging strategy, this method can flexibly adapt to different signal environments. It possesses strong scalability and can be applied to multiple signal processing fields such as wireless communication, radar systems, and spectrum monitoring. It can still ensure high-precision signal detection in low SNR environments, thereby improving the overall performance of the signal analysis system. Attached Figure Description

[0061] Figure 1 This is the raw power spectrum of the satellite's actual data.

[0062] Figure 2 The power spectrum after sampling and smoothing in step 2;

[0063] Figure 3 A schematic diagram of the data type segmentation results and adaptive decision threshold in step 3;

[0064] Figure 4 The results of the preliminary signal interval detection in step 4;

[0065] Figure 5 The final detection and interval recognition results of the complete algorithm are shown in the image.

[0066] Figure 6 This is a flowchart of a signal detection and interval recognition method based on adaptive threshold according to the present invention;

[0067] Figure 7 The diagram shows the effect and results of the algorithm of this invention in detecting and identifying broadband signals. Specific implementation methods

[0068] To enable those skilled in the art to more deeply understand the implementation ideas of the present invention, the technical solutions in the embodiments of the present invention will be described in detail and clearly below with reference to the accompanying drawings. The embodiments use satellite signal data collected in the field to demonstrate the effect of each processing step.

[0069] Example 1

[0070] This embodiment is applied to the field of satellite communication. Utilizing satellite signal data collected in the field, the signal detection method of this invention is used for signal analysis and detection, ultimately accurately distinguishing satellite signals from noise. The specific experimental steps, parameters used, and results are described in detail below.

[0071] This experiment uses satellite signal data collected in the field, sourced from the signal receiving section of the satellite communication system. Signals were acquired in different frequency bands and time windows, and the data files are in .dat format, containing the signal frequency and corresponding power spectral density data. Due to the complexity of the satellite communication environment, the data contains various signal types (such as broadband and narrowband signals) and strong background noise.

[0072] In this experiment, a satellite signal dataset containing seven narrowband signals and strong noise was selected. This dataset includes multiple signals in the frequency range of 4.274 GHz to 4.276 GHz, each with a different power spectral density value, and is also accompanied by strong background noise.

[0073] like Figure 6 As shown in the figure, this embodiment discloses a signal detection and interval recognition method based on adaptive threshold, and the specific implementation steps are as follows:

[0074] Step 1: Read the satellite signal data collected in the field and extract the frequency data f and power spectral density data pxx. This process allows us to understand the spectral distribution of the original signal and provides a basis for subsequent processing.

[0075] like Figure 1 The image shows the raw power spectrum of the satellite signal, illustrating the power distribution within the frequency range of 4.274 GHz to 4.276 GHz. Due to strong noise and blurred boundaries between signal and noise, further processing is required.

[0076] Step 2: To reduce the amount of data and improve processing efficiency, we perform a sampling and smoothing operation on the original signal power spectrum data imported in Step 1. The goal of sampling is to reduce the number of data points from 10001 in the original data to 2000, while ensuring that the main characteristics of the signal are not affected. Smoothing helps to reduce the impact of peak noise to some extent.

[0077] During the sampling process, we first calculate the position of the target data point, originalIndices(j), as the index of the corresponding frequency point in the original data. Then, we extract the corresponding frequency and power spectral density values ​​from the original data using a linear interpolation method, preserving the main characteristics of the signal.

[0078] The location of the target data point, originalIndices(j), is calculated according to the following formula:

[0079]

[0080] The interpolation calculation formula is:

[0081]

[0082] Where, ω j f represents the interpolation weights, indicating the position of the target data point in the original data. indexLow and P indexLow For the frequency and power spectral density of the low index point, f indexHigh and P indexHigh For the high index point, the frequency and power spectral density.

[0083] Even after desampling, high-power noise may still exist in the power spectrum, which can affect subsequent signal classification and detection. To further suppress noise, a mean smoothing operation is used to smooth the power spectrum data. Specifically, a smoothing window of size 7 is used to calculate the average value within the neighborhood of each data point.

[0084]

[0085] Figure 2 The power spectrum after sampling and smoothing is shown. Noise in the signal is effectively suppressed, and the signal curve becomes smoother and clearer, which facilitates subsequent signal detection and threshold calculation.

[0086] Step 3, threshold calculation, is a crucial step in signal detection. Based on the smoothed power spectrum data, the adaptive threshold calculation method proposed in this invention dynamically adjusts the threshold at each frequency point according to the local characteristics of the signal. This method determines the data type (wideband signal, noise, or signal-to-noise mixture) at the center frequency of the window by analyzing the difference between the maximum and minimum values ​​of the signal within each window, as well as the ratio of signal to noise, and sets the threshold value according to different rules.

[0087] First, set the window length to 40 (i.e., 2% of the input data length) and the signal-to-noise ratio threshold to SET. snr The value is 6dB. If the difference between the maximum and minimum values ​​within the window is less than the set signal-to-noise ratio threshold SET... snr If the number of points greater than the mean is close to the size of the window, then the window is considered to belong to the wideband signal area; if the difference between the maximum and minimum values ​​within the window is less than SET... snr If the number of points greater than the mean is small, then the window is considered to be in a noise zone; if both signal and noise exist in the window, then the window is considered to be a mixed signal and noise zone.

[0088] The formula for calculating the adaptive threshold is as follows:

[0089]

[0090] Among them, T adaptive (i) represents the adaptive threshold for the i-th data point, E{data} is the global mean, max{window_data} is the maximum signal power within the window, and δ i The value is the difference between the local maximum and the global mean, and percent(i) is the proportion of points in the window that are greater than the mean.

[0091] Figure 3The dashed line represents the result calculated using an adaptive threshold, indicating a threshold line dynamically adjusted based on the local characteristics of the signal and the background noise. This threshold line adjusts with changes in signal strength and noise, effectively distinguishing between signal and noise regions and ensuring accurate signal detection. The adaptive threshold line dynamically adjusts its position according to changes in signal and noise, successfully improving the distinction between signal and noise and ensuring accurate signal detection.

[0092] Step 4, signal interval detection, is one of the core steps in signal detection. In this process, an adaptive threshold is used to analyze the power spectrum, distinguishing signal intervals from noise intervals based on a set threshold. First, all points exceeding the threshold are identified and marked as signal intervals. The start and end points of each signal interval are calculated to determine the boundary of each valid signal interval. For overlapping signal intervals, a merging mechanism helps to merge them, avoiding misjudgments.

[0093]

[0094] Among them, P i Let start be the power value at the i-th frequency point in the power spectral density. idx and end idx These represent the start and end indices of the signal interval, respectively.

[0095] Figure 4 The results of signal interval detection and the signal-to-noise ratio differentiation are demonstrated. The final signal interval map shows the precise location of each signal, and the use of adaptive thresholding significantly improves the accuracy of signal detection.

[0096] Step 5: After initially confirming the signal range, further screening is required, proceeding to the false alarm and missed alarm suppression stage. This is achieved by analyzing the signal-to-noise ratio (SNR) V of the detected signal range. i And thresholds, further filtering out false alarms and missed alarms.

[0097] The specific method involves setting a minimum signal-to-noise ratio (MIN). snr This determines which signal ranges are valid. In this embodiment, a minimum signal-to-noise ratio (MIN) is set. snr =4dB. The formula for determining false alarm and missed alarm suppression is:

[0098]

[0099] Among them, P max (i) represents the maximum power of the i-th signal segment. and These are the minimum noise values ​​on both sides of the signal segment.

[0100] If the false alarm judgment value V iGreater than the set minimum signal-to-noise ratio (V) i >MIN snr If the signal segment is valid, it is considered valid; otherwise, it needs to be removed.

[0101] Step 6: After reducing false alarms through the above steps, it is also necessary to avoid signal missegmentation. After determination, adjacent signal segments are merged. This is done by calculating the merging decision value C. i And false alarm judgment value V i This is used to determine whether to merge adjacent signal intervals. Specifically, when the merge decision value C... i The false alarm value V is greater than the set threshold. i Less than the set minimum signal-to-noise ratio MIN snr At this time, the signal segment merging operation will be triggered.

[0102] In this embodiment, the threshold is set to 1 / MIN. snr .

[0103] Combined judgment value C i Calculated using the following formula:

[0104]

[0105] When C i The false alarm value V is greater than the set threshold. i Less than MIN snr When this happens, the signal segment merging operation will be triggered, connecting the two signal segments.

[0106] Figure 5 The upper half shows the correspondence between the signal interval and the power spectrum obtained after the complete algorithm analysis and processing, while the lower half shows the final output signal detection and interval identification results. Compared with the signal intervals obtained from the initial screening in step 4, after false alarm and missed alarm suppression and signal segment merging, the signal intervals are more accurate, noise and invalid signals are effectively eliminated, and the boundaries of the signal intervals are clearer, avoiding missegmentation and missed detection.

[0107] At this point, the signal detection and interval identification using actual satellite communication signals in Example 1 above has been completely completed. The flowchart is as follows: Figure 6 As shown in this embodiment, a signal detection and interval recognition method based on adaptive threshold is disclosed. After value sampling and smoothing, the algorithm complexity is reduced while retaining signal characteristics. Through adaptive threshold calculation, the detection threshold is dynamically adjusted to flexibly cope with different signal and noise backgrounds, thereby improving the distinguishability between signals and noise. The signal interval detection accurately identifies the effective signal interval, and the suppression of false alarms and missed alarms effectively filters false alarm signals. Finally, the signal detection results are further optimized by signal segment merging, ensuring the accuracy and reliability of the detection.

[0108] Through observation Figure 5 It can be observed that the present invention exhibits excellent detection and interval identification performance for narrowband signals under low signal-to-noise ratio conditions. For actual sampled satellite broadband signals (such as OFDM signals), using the same algorithm parameters as in Example 1, analysis was performed, and the resulting detection and identification results are as follows: Figure 7 As shown.

[0109] Through observation Figure 7 It can be observed that for broadband signals with low signal-to-noise ratios, adaptive threshold calculation can effectively detect the frequency range of the signal. False alarm and missed alarm suppression operations are indispensable steps in the entire signal detection process, further filtering out false signals that do not meet the criteria and ensuring that only valid signals are retained. Finally, the signal segment merging operation merges adjacent signal intervals, effectively avoiding mis-segmentation of the signal, thereby further improving the accuracy of the detection results and ultimately ensuring high-precision signal detection and frequency range identification.

[0110] This invention provides a signal detection and interval identification method based on adaptive thresholds. By flexibly setting the threshold, it achieves efficient detection of both broadband and narrowband signals. This method can dynamically adjust the detection threshold, reduce false alarms, merge adjacent signals, and improve detection accuracy, making it suitable for complex signal environments. Through power spectrum analysis, this invention can minimize false detections and missed detections while ensuring high efficiency, demonstrating strong application prospects.

[0111] The above detailed description further illustrates the purpose and technical solution of the invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A signal detection and interval identification method based on adaptive threshold, characterized in that: Includes the following steps, Step 1: Using the acquired signal data, calculate the power spectral density (PSD) of the signal using the Fast Fourier Transform (FFT) method to obtain the power distribution of the signal in the frequency domain. Step 2: Extract the target number of data points from the original data by sampling the original frequency and power spectral density data, while preserving the characteristics and trends of the data; suppress high-frequency noise through mean smoothing operation to make the power spectrum smoother; Step 3: Based on the maximum and minimum values ​​of the signal and the signal-to-noise ratio within each window, classify broadband signals, mixed signal-to-noise signals, and noisy regions, and calculate different adaptive thresholds for different types of signal regions; Step 4: By utilizing the local features of the power spectral density and dynamically set adaptive thresholds, signals and noise can be accurately distinguished, thereby improving the accuracy of signal detection. The complete formula for calculating the adaptive threshold is as follows. in, For the adaptive threshold of the i-th data point, The global mean. The maximum signal power within the window. The difference between the local maximum and the global mean. This represents the proportion of points within the window that are greater than the mean. Step 5: Based on channel analysis, pre-set the minimum signal-to-noise ratio that meets the scenario and requirements. Calculate the ratio of the maximum power of the signal to the minimum power of the noise to determine false alarms; Step 6: Based on the power characteristics, signal-to-noise ratio, and interval spacing of adjacent signal intervals, merge the intervals that meet the conditions to optimize the signal detection results and ensure the integrity of the signal segment; Step 7: The signal range obtained in step 6 is the effective signal range after merging.

2. The signal detection and interval identification method based on adaptive threshold as described in claim 1, characterized in that: In step 1, Step 1: Using the acquired signal data, calculate the power spectral density (PSD) of the signal using the Fast Fourier Transform (FFT) method to obtain the power distribution of the signal in the frequency domain. in, For power spectral density, The frequency domain representation of the signal represents the energy distribution of the signal in the frequency domain; the frequency characteristics of the signal are obtained according to equations (1) and (2), providing basic data for subsequent threshold setting and signal interval detection.

3. The signal detection and interval identification method based on adaptive threshold as described in claim 1, characterized in that: In step 2, The specific steps for implementing mean smoothing are as follows: Step 2.1: Calculate the position of each target data point based on the ratio between the target number of points and the original number of points; Step 2.2: Then, calculate the index position of the original data points using the ratio: Calculate the index position of the original data point using a ratio; since the target data point may not fall exactly at the position of the original data point, it is necessary to determine its two nearest index points in the original data: in, This indicates rounding down. This indicates rounding up. and These represent the two indices in the original data where the target data point is closest to it. Step 2.3: For each target data point, extract the frequency and power spectral density values ​​from the original data according to the calculated index; the frequency and power spectral density of the target data point are obtained by linearly weighted averaging two adjacent original data points; in, and These are the frequency and power spectral density after decomposition, respectively. The interpolation weights represent the position of the target data point within the original data points, and are calculated using the following formula: Step 2.4: Handle boundary issues to avoid exceeding limits; when the target data point is located on the boundary of the original data, boundary processing is required: like Then set ;like Then set ;when At that time, directly take the corresponding frequency and power spectral density: By handling boundary issues, we ensure that the indices of all target data points are within the valid range, thus avoiding array out-of-bounds errors. Step 2.5: To further reduce noise and enhance signal smoothness, the power spectrum after sampling is subjected to mean smoothing; the smoothing operation is performed by... The average value of adjacent data points is taken to reduce the influence of high-frequency noise, making the power spectrum smoother and thus improving the stability of signal detection. According to formulas (8) and (9), the frequency array after sampling smoothing is obtained. and power spectral density array .

4. The signal detection and interval identification method based on adaptive threshold as described in claim 3, characterized in that: The specific implementation steps for step 3 are as follows: Step 3.1: For each frequency point, by calculating the maximum and minimum values ​​of the signal data within its window, and combining the number of points within the window that are greater than the global mean, determine whether the window belongs to a broadband signal and noise region or a mixed signal and noise region. Step 3.2: Different threshold setting methods are used for different types of signal areas; for wideband signal areas, the threshold is taken as the global mean; for noisy areas, the threshold is increased by increasing the local maximum value; and for areas where signals and noise are mixed, the threshold is flexibly adjusted by calculating the local increment and the signal ratio.

5. The signal detection and interval identification method based on adaptive threshold as described in claim 1, characterized in that: In step 4, Use formula (11) to detect the start and end positions of the signal: in, Let be the power value at the i-th frequency point in the power spectral density. and These represent the start and end indices of the signal interval, respectively.

6. The signal detection and interval identification method based on adaptive threshold as described in claim 1, characterized in that: In step 5, The formula for calculating false alarm decisions is: in, The maximum power of the m-th signal segment. and These are the minimum noise values ​​on both sides of the signal segment; If the false alarm judgment value Less than the set minimum signal-to-noise ratio, i.e. If the signal segment is false, it is determined to be a false alarm and the signal segment needs to be removed.

7. The signal detection and interval identification method based on adaptive threshold as described in claim 1, characterized in that: In step 6, The criteria for signal segment merging are based on two main conditions: the noise intensity ratio between signal segments and the result of the false alarm decision; a merging decision value is set. This value depends on the relationship between the minimum noise and the average signal of adjacent signal segments; the decision condition for merging is: in, This is the merging decision value for the m-th signal segment. Let be the minimum noise power to the right of the m-th signal segment. and These are the average signal power of the m-th and (m+1)-th signal segments, respectively; If the merged judgment value Greater than the set threshold And false alarm judgment value Less than the set minimum signal-to-noise ratio If the two adjacent signal intervals can be merged, then the merged signal interval can be obtained.

Images