Signal fault identification method and system based on multi-sensor space-time alignment

Abstract

The invention relates to the field of data processing, in particular to a signal fault identification method and system based on multi-sensor space-time alignment, and the method comprises the steps: collecting the real-time signal intensity of an unmanned plane and the distance data from a signal source; calculating a primary conventional change degree of the signal intensity and a distance adjustment factor, and normalizing the product of the two to obtain a conventional change degree; clustering the signal intensity data by using a hierarchical clustering algorithm; and screening suspected abnormal clusters from the clustering result according to a preset first cutting rule, calculating the average conventional change degree of the signal intensity in each suspected abnormal cluster, and marking the cluster with the average conventional change degree smaller than or equal to a conventional threshold value as an abnormal cluster so as to identify abnormal signal data. According to the method, whether the suspected abnormal cluster is the abnormal cluster can be accurately judged, misjudgment and missed judgment are avoided, and the accuracy of abnormal signal identification is improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing. More specifically, the present invention relates to a signal fault identification method and system based on multi-sensor spatio-temporal alignment. Background Art

[0002] With the continuous progress of unmanned aerial vehicles (UAVs), they are widely used in many fields such as agricultural monitoring (e.g., monitoring the growth status of crops), environmental monitoring (such as monitoring air quality, water quality, etc.), and geographical surveying (obtaining topographic and geomorphic information, etc.). Among the performance indicators of UAVs, signal stability is particularly crucial. Signal faults, especially abnormally low signal intensity, can easily lead to the loss of contact of UAVs, mission failures, or safety accidents. Therefore, it is very important to quickly identify and effectively locate these signal anomalies, which can significantly reduce the probability of accidents and ensure the smooth completion of UAV missions.

[0003] The prior art, such as the patent application document with the publication number CN116010841A, discloses a signal modulation automatic recognition method based on deep learning. The signal modulation automatic recognition method includes: extracting signal features using wavelet transform; clustering the signal feature data using the agglomerative hierarchical clustering algorithm to classify the signal feature data; setting the loss function of the generative adversarial network using the identity loss and the least squares loss; and training the generative adversarial network using the signal features extracted by wavelet transform and the class labels obtained by the agglomerative hierarchical clustering algorithm to obtain the modulation mode.

[0004] In the above signal modulation automatic recognition method, the signal data is clustered using the agglomerative hierarchical clustering algorithm to divide the signal into different categories. However, due to the influence of external interference and other factors, misclassification may occur when clustering the signal data, resulting in a certain degree of interference and deviation in the clustering results, and causing errors in the judgment of the signal state. Summary of the Invention

[0005] To solve the above technical problem that misclassification may occur when clustering signal data, resulting in errors in the judgment of the signal state, the present invention provides solutions in the following aspects.

[0006] In the first aspect, a signal fault identification method based on multi-sensor spatio-temporal alignment includes: Collecting the real-time signal intensity data of the UAV during operation and the distance data between the UAV and the signal source, and converting the collected data into digital signals; Calculating the primary conventional change degree of the signal intensity data and the distance adjustment factor, and normalizing the product of the primary conventional change degree and the distance adjustment factor to obtain the conventional change degree of the signal intensity data; Using the hierarchical clustering algorithm to cluster the signal intensity data to obtain the clustering result; After obtaining the clustering result, all suspected abnormal clusters are screened from the clustering result according to a preset first pruning rule. For each suspected abnormal cluster, the average regular change degree of all signal strength data within the cluster is calculated, and the suspected abnormal cluster with an average regular change degree less than or equal to the regular threshold is marked as an abnormal cluster, thereby identifying all abnormal signal data.

[0007] The present invention first analyzes the collected data, calculates the primary regular change degree and the distance adjustment factor of the signal strength data, and normalizes the product of the two to obtain the regular change degree, comprehensively considering the self-change law of the signal strength and the influence of the distance from the signal source on the signal strength, which can more comprehensively and accurately evaluate the change of the signal strength and provide a more reliable basis for the subsequent identification of abnormal signals; then uses the hierarchical clustering algorithm to cluster the signal strength data, which can automatically group the data according to the similarity of the signal strength data; finally, first screen the suspected abnormal clusters through preset rules (such as cluster size, signal range), and then further verify based on the average regular change degree, which can accurately judge whether the suspected abnormal cluster is really an abnormal cluster, avoiding misjudgment and missed judgment, and improving the accuracy of abnormal signal identification.

[0008] Preferably, the process of obtaining the primary regular change degree includes: Select any signal strength data as the target signal. Taking the target signal as the center, select a set number of data to construct the target sequence of the target signal, calculate the mean value of the absolute value of the difference between the target signal and its adjacent signal strength data, and use the mean value of the absolute value of the difference as the local difference of the target signal; calculate the difference between the maximum value and the minimum value of the signal strength data in the target sequence of the target signal, and use the difference between the maximum value and the minimum value as the overall difference of the target signal. Normalize the product of the local difference and the overall difference to obtain the primary regular change degree of the target signal.

[0009] By analyzing the local difference and the overall difference of the target signal, not only can the short-term change trend of the signal be understood, but also the long-term trend or extreme value difference of the signal can be understood. Multiplying and normalizing the local and global differences avoids the limitations of a single index and more comprehensively describes the dynamic change degree of the signal.

[0010] Preferably, the process of obtaining the distance adjustment factor includes: Obtain the numerical mean value of all signal strength data with the same distance from the signal source of the drone at the sampling moment corresponding to the target signal in the historical signal strength data. Take the absolute value of the difference between the target signal and the numerical mean value as the first parameter, take the distance from the drone to the signal source at the sampling moment corresponding to the target signal as the second parameter, and take the ratio of the second parameter to the first parameter as the distance adjustment factor of the target signal.

[0011] By calculating the distance adjustment factor, the target signal is further corrected to more accurately reflect the relationship between the signal strength and the distance between the drone and the signal source.

[0012] Preferably, the hierarchical clustering is the agglomerative hierarchical clustering algorithm.

[0013] Preferably, after obtaining the abnormal clusters, it further includes: For each signal strength data marked as an abnormal cluster, according to the distance data between the drone and the signal source, find the signal source position information where abnormal signal strength appears within a preset distance range, and further formulate a maintenance decision.

[0014] Preferably, the first pruning rule includes: presetting a merging distance threshold, and retaining the clustering clusters with a merging distance less than the preset merging distance threshold; Presetting a threshold for the number of data points within a cluster, and pruning the clustering clusters with the number of data points within the cluster greater than the threshold for the number of data points within the cluster.

[0015] Preferably, the first pruning rule further includes: Presetting a threshold for the mean value after normalization of the data within a cluster, and pruning the clustering clusters with the mean value after normalization of the data within the cluster greater than the threshold for the mean value after normalization of the data within the cluster.

[0016] By removing abnormal, few-data-point, and abnormally-averaged clustering clusters, the main and representative clustering clusters can be made more obvious.

[0017] Preferably, the process of obtaining the primary conventional change degree includes: Select any one signal strength data as the target signal. Taking the target signal as the center, select a set number of data to construct the target sequence of the target signal. Calculate the mean of the absolute values of the differences between the target signal and its adjacent signal strength data, and take this mean of the absolute values of the differences as the local difference of the target signal; calculate the difference between the maximum and minimum values of the signal strength data in the target sequence of the target signal, and take this difference between the maximum and minimum values as the overall difference of the target signal; Perform standardization processing on the local difference and the overall difference, and then calculate the Euclidean distance between the processed local difference and the overall difference, and take the Euclidean distance as the primary conventional change degree of the target signal.

[0018] By calculating the Euclidean distance between the processed local difference and the overall difference and taking this distance as the primary conventional change degree, it can more comprehensively reflect the difference degree between the two vectors (i.e., the standardized local difference and the overall difference), considering the comprehensive influence of the differences in each dimension.

[0019] Preferably, the process of obtaining the distance adjustment factor includes: Obtain the numerical mean of the signal strength data in the historical signal strength data that are all at the same distance from the signal source as the UAV at the sampling moment corresponding to the target signal. Take the absolute value of the difference between the target signal and the numerical mean as the first parameter, and take the distance between the UAV and the signal source at the sampling moment corresponding to the target signal as the second parameter; Take the product of the first parameter and the attenuation coefficient as the negative exponent of the exponential function, perform operations on the exponential function, and multiply the obtained result by the second parameter to obtain the distance adjustment factor of the target signal.

[0020] By considering the non-linear characteristic of signal attenuation through the exponential function, it can more accurately reflect the real situation of the signal strength changing with distance, thereby more precisely correcting the target signal to make it closer to the actual signal propagation.

[0021] In a second aspect, a signal fault identification system based on multi-sensor spatio-temporal alignment includes: a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, any one of the signal fault identification methods based on multi-sensor spatio-temporal alignment is implemented.

[0022] The beneficial effects of the present invention are: By fusing signal strength data and distance data, combining dynamic quantization evaluation (conventional change degree) and distance adjustment factor, the evaluation of the signal change degree is made more accurate and comprehensive; Then, the hierarchical clustering algorithm is used to cluster the signal strength data, and further through multi-step screening and judgment, the false alarm rate can be effectively reduced and the accuracy of signal fault identification can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flowchart of the method from step S1 to step S4 in a signal fault identification method based on multi-sensor spatio-temporal alignment according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention.

[0025] Refer to Figure 1 , a signal fault identification method based on multi-sensor spatio-temporal alignment includes steps S1 - S4, specifically as follows: S1: Collect the real-time signal strength data of the UAV during operation and the distance data between the UAV and the signal source, and convert the collected data into digital signals.

[0026] In one embodiment, a wireless communication module is used to collect signal strength data emitted by a drone device during operation. The wireless communication module can capture the strength of the signal when the drone interacts with the surrounding communication environment, and this data reflects information such as the communication quality of the drone.

[0027] With the help of a GPS module, positioning data of the drone during operation is obtained. The GPS module accurately determines the position of the drone in the geographical space by receiving satellite signals, usually represented in the form of coordinates such as longitude and latitude.

[0028] The acquisition of the above two types of data needs to be carried out within the same time period to ensure that the signal strength data and the positioning data correspond to each other in the time dimension and can accurately reflect the state of the drone at a specific moment. The acquisition duration is set to one hour, and the acquisition frequency is five times per second, that is, five data are acquired per second.

[0029] An analog-to-digital conversion device is used to digitally convert the collected signal strength data and positioning data. The role of the analog-to-digital conversion device is to convert analog signals (such as the original signals output by the wireless communication module and the GPS module) into digital signals that can be processed by a computer, obtaining digital representations of these two types of data, which is convenient for subsequent storage, processing, and analysis in a computer system.

[0030] Based on the positioning data of the drone during operation, the distance data between the drone and the signal source is calculated in real time (such as the formula for calculating the distance between two points based on longitude and latitude).

[0031] S2: Calculate the primary conventional change degree of the signal strength data and the distance adjustment factor, and normalize the product of the primary conventional change degree and the distance adjustment factor to obtain the conventional change degree of the signal strength data.

[0032] When the drone is operating, the signal strength data may be low due to a long flight distance. This low signal strength caused by normal flight is close to the values of some abnormally low data and is difficult to directly distinguish. To distinguish these two types of data, first calculate the primary conventional change degree of each signal strength data, and then combine the distance data from the signal source corresponding to the signal strength data and historical data to optimize the primary conventional change degree to obtain the conventional change degree, which is used as the basis for distinguishing between normal low and abnormally low data.

[0033] In one embodiment, any one signal strength data is selected as the target signal. Centered on the target signal, at least 5 data are selected on each of its two sides as references, for a total of 10 surrounding reference points, and then the target sequence of the target signal is constructed. It should be noted that if the number of data on one side is less than 5 (for example, the data is at the start or end of the data segment), it is supplemented from the other side to ensure that there are a total of 10 surrounding reference points.

[0034] Further, calculate the mean of the absolute values of the differences between the target signal and its immediately adjacent signal strength data, and use the mean of the absolute values of the differences as the local difference of the target signal. Then calculate the difference between the maximum and minimum values of the signal strength data in the target sequence of the target signal, and use it as the overall difference of the target signal.

[0035] Further, normalize the product of the calculated local difference and overall difference to obtain the primary conventional change degree of the target signal. The calculated primary conventional change degree satisfies the relational expression:

[0036] In the formula, is the primary conventional change degree of the target signal, is the local difference of the target signal, is the overall difference of the target signal; where, is the first hyperparameter, and its value is 0.001, which is used to prevent or from taking a value of 0; represents the exponential function with the natural number e as the base.

[0037] It should be noted that if the target signal is at the end or start of the collected data, it may have only one adjacent data point, then is the absolute value of the numerical difference between the target signal and this one adjacent point.

[0038] Among them, if the difference between the target signal and the adjacent data points on both sides is small, then intuitively, this numerical change is more in line with the general and conventional change pattern. Therefore, the credibility that the numerical change of this target signal belongs to the conventional change is greater; if the overall difference of the target signal is smaller, it indicates that the numerical differences of the data around the target signal are smaller, indicating that the changes of these data are relatively stable and more likely to belong to the conventional fluctuations.

[0039] In summary, when and are smaller, the primary conventional change degree of the target signal is greater.

[0040] In another embodiment, after calculating the local difference and the overall difference of the target signal, the two can also be standardized first (such as Z-score standardization), and then the Euclidean distance between the two is calculated as the primary conventional change degree of the target signal.

[0041] The above-mentioned distance metric more intuitively reflects the and relative difference between. The greater the distance, the greater the difference between and is, and the higher the primary conventional change degree.

[0042] Furthermore, according to the above calculation method of the primary conventional change degree of the target signal, the primary conventional change degrees of all the collected signal strength data are calculated in the same way.

[0043] Quantifying the possibility that the signal strength data belongs to the conventional change by calculating the primary conventional change degree. However, the primary conventional change degree of the signal strength data with a relatively low level may also be low. It is not possible to well distinguish the signal strength data with a relatively low level and the signal strength data with an abnormally low level only based on the primary conventional change degree.

[0044] There is usually a certain relationship between the signal strength data when the drone device is working and the distance data of the drone from the signal source. That is, the signal strength data will attenuate as the distance between the drone and the signal source increases.

[0045] In order to more accurately evaluate the conventional change degree of the signal strength data, it is necessary to analyze the numerical characteristics of each signal strength data and the corresponding distance data from the signal source. Optimize the primary conventional change degree of each signal strength data to obtain the conventional change degree of each signal strength data.

[0046] In one embodiment, still taking the above-mentioned target signal as an example, obtain the numerical mean of all the signal strength data in the historical signal strength data that have the same distance from the signal source as the drone corresponding to the target signal at the sampling moment. Take the absolute value of the difference between the target signal and the calculated numerical mean as the first parameter, take the distance of the drone from the signal source at the sampling moment corresponding to the target signal as the second parameter, and take the ratio of the second parameter to the first parameter as the distance adjustment factor of the target signal.

[0047] The distance adjustment factor of the above-mentioned target signal satisfies the relationship:

[0048] In the formula, is the distance adjustment factor of the target signal, is the second parameter (the distance of the drone from the signal source at the sampling moment corresponding to the target signal), is the first parameter ( is the value of the target signal, is the average value of the signal intensity data in the historical signal intensity data that are consistent with the distance of the UAV from the signal source at the sampling moment corresponding to the target signal); where is the second hyperparameter, with a value of 0.01, to prevent the denominator from being zero.

[0049] Among them, the larger it is, if the value of the target signal is low, that is is large, then the distance adjustment factor decreases.

[0050] In another embodiment, considering that when the signal difference is small, the change of the adjustment factor is not obvious enough and may not accurately reflect the subtle change of the signal intensity.

[0051] Therefore, the distance adjustment factor of the target signal is calculated by exponential decay, that is, the relational expression is satisfied as:

[0052] In the formula, is the distance adjustment factor of the target signal, is the second parameter (the distance of the UAV from the signal source at the sampling moment corresponding to the target signal), is the first parameter ( is the value of the target signal, is the average value of the signal intensity data in the historical signal intensity data that are consistent with the distance of the UAV from the signal source at the sampling moment corresponding to the target signal); is the attenuation coefficient, which controls the speed at which the adjustment factor decreases with the signal difference.

[0053] Finally, the product of the primary conventional change degree and the distance adjustment factor is normalized to obtain the conventional change degree of the target signal, that is, the relational expression is satisfied as:

[0054] In the formula, is the conventional change degree of the target signal, is the primary conventional change degree of the target signal, is the distance adjustment factor of the target signal, represents the normalization process.

[0055] The above reflects the basic change characteristics of the signal itself, while considers the influence of the distance factor on the signal. Multiplying the two can comprehensively consider the internal change of the signal and the external distance factor, so as to more comprehensively describe the change degree of the signal.

[0056] Furthermore, according to the above calculation method of the conventional change degree of the target signal, the conventional change degrees of all signal strength data can be obtained in the same way.

[0057] S3: Use the hierarchical clustering algorithm to cluster the signal strength data to obtain the clustering result.

[0058] In one embodiment, the signal strength data corresponding to every 5 minutes on the time axis is selected and clustered according to the prior art of the agglomerative hierarchical clustering algorithm to obtain a preliminary clustering result. The agglomerative hierarchical clustering algorithm is a commonly used clustering method that gradually merges similar data points to form clusters.

[0059] S4: After obtaining the clustering result, according to the preset first pruning rule, all suspected abnormal clusters are screened out from the clustering result. For each suspected abnormal cluster, calculate the average conventional change degree of all signal strength data within the cluster, and mark the suspected abnormal clusters with an average conventional change degree less than or equal to the conventional threshold as abnormal clusters, thereby identifying all abnormal signal data.

[0060] In the clustering of signal strength data, abnormal clusters usually contain a relatively small number of data points. This is because abnormal data is often sporadic and isolated, unlike normal data which has a tendency to cluster. For example, when a drone is flying normally, the signal strength remains relatively stable, and the data points will concentrate within a certain range to form a normal cluster; while when there is interference or equipment failure, a small number of abnormally low signal strength data points may be generated, forming an abnormal cluster.

[0061] In one embodiment, the preliminary clustering result is pruned according to three preset thresholds. The specific pruning process is as follows: Set the merging distance threshold during the clustering process (set to 2 in the embodiments of the present invention). If the merging distance between two clusters is greater than 2.0, it is considered that they should not be merged.

[0062] In addition, since the number of data points in abnormal clusters is usually small, a threshold for the number of data points within a cluster (set to 5 in the embodiments of the present invention) can also be set to prune data clusters with a relatively large number of data points.

[0063] It should be noted that since abnormally low signal strength values need to be detected, a threshold for the mean value of the normalized data within a cluster (set to 0.5 in the embodiments of the present invention) is set to distinguish between data clusters with relatively large and small signal strength values, and data clusters with relatively large values are pruned.

[0064] After pruning the preliminary clustering result with the three thresholds set above, the resulting clusters are suspected abnormal clusters.

[0065] After the first cut described above, although some suspected abnormal clusters are obtained, some interference items may be included. In actual signal strength data, even if the number of data points is small and the normalized mean is low, it does not necessarily mean that the data is abnormal. In some cases, the change in signal strength may be caused by normal environmental factors or small fluctuations in the device operating state. For example, during the flight of a drone, due to factors such as minor adjustments in flight attitude and slight interference from the surrounding environment, the signal strength may show a brief and regular low situation.

[0066] Through secondary cutting, we can further screen out those clusters with a relatively small average conventional change degree. These clusters are more likely to be caused by abnormal reasons such as device failures and severe interference, resulting in low signal strength.

[0067] In one embodiment, calculate the average conventional change degree of all signal strength data within all suspected abnormal clusters, and set a conventional threshold (set to 0.35 in the embodiment of the present invention). If the average conventional change degree within a suspected abnormal cluster is greater than the conventional threshold, it means that the data within this cluster belongs to the small signal strength data caused by conventional changes with a greater credibility. Then, such clusters need to be cut off. After this cut, the final abnormal clusters are obtained.

[0068] After the above two cuts, the finally obtained clusters are marked as abnormal clusters.

[0069] After identifying all the abnormal clusters, set a preset distance range according to the communication range of the drone, the effective coverage range of the signal source, and practical experience. For example, if the effective communication radius of the drone is 500 meters, then the preset distance range can be set to 500 meters.

[0070] Traverse the signal strength data of each abnormal cluster, and check the distance between the corresponding drone and each signal source. If the distance between a certain signal source and the drone is within the preset distance range, record the location information of the signal source. The location information can include the longitude and latitude coordinates of the signal source, the specific address, etc.

[0071] According to the location and distribution of the abnormal signal sources, formulate a reasonable maintenance plan. For example, if the abnormal signal sources are relatively concentrated, a centralized maintenance operation can be arranged to improve the maintenance efficiency.

[0072] The system includes a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the signal fault identification method based on multi-sensor spatio-temporal alignment according to the first aspect of the present invention is implemented.

[0073] The system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface, whose settings and functions are known in the art and thus will not be elaborated herein.

[0074] It should be noted that, for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent shall be subject to the appended claims.

Claims

1. A signal fault identification method based on multi-sensor spatiotemporal alignment, characterized in that: include: Collect the real-time signal strength data of the drone during operation and the distance data between the drone and the signal source, and convert the collected data into digital signals; Calculating the primary regular variation degree and the distance adjustment factor of the signal strength data, and normalizing the product of the primary regular variation degree and the distance adjustment factor to obtain the regular variation degree of the signal strength data; The signal strength data is clustered using a hierarchical clustering algorithm to obtain clustering results; After obtaining the clustering results, all suspected abnormal clusters are screened out from the clustering results according to the preset first clipping rule. For each suspected abnormal cluster, the average regular variation degree of all signal strength data in the cluster is calculated, and the suspected abnormal clusters whose average regular variation degree is less than or equal to the regular threshold are marked as abnormal clusters, thereby identifying all abnormal signal data.

2. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 1 is characterized in that: The process of obtaining the primary routine variation degree includes: Select any signal strength data as the target signal, take the target signal as the center, select a set number of data to construct a target sequence of the target signal, calculate the average of the absolute values ​​of the differences between the target signal and its adjacent signal strength data, and use the average of the absolute values ​​of the differences as the local difference of the target signal; calculate the difference between the maximum and minimum values ​​of the signal strength data in the target sequence of the target signal, and use the difference between the maximum and minimum values ​​as the overall difference of the target signal; The product of the local difference and the overall difference is normalized to obtain the primary conventional variation degree of the target signal.

3. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 2 is characterized in that: The process of obtaining the distance adjustment factor includes: The numerical mean of all signal strength data that are consistent with the distance between the drone and the signal source at the sampling moment corresponding to the target signal in the historical signal strength data is obtained, the absolute value of the difference between the target signal and the numerical mean is used as the first parameter, the distance between the drone and the signal source at the sampling moment corresponding to the target signal is used as the second parameter, and the ratio of the second parameter to the first parameter is used as the distance adjustment factor of the target signal.

4. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 3 is characterized in that: The hierarchical clustering is an agglomerative hierarchical clustering algorithm.

5. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 4 is characterized in that: After getting the abnormal cluster, it also includes: For each signal strength data marked as an abnormal cluster, the location information of the signal source with abnormal signal strength within the preset distance range is found according to the distance data between the drone and the signal source, and further maintenance decisions are made.

6. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 5 is characterized in that: The first clipping rule includes: presetting a merge distance threshold, and retaining clusters whose merge distance is less than the preset merge distance threshold; A threshold for the number of data points within a cluster is preset, and clusters whose number of data points within the cluster is greater than the threshold are pruned.

7. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 6 is characterized in that: The first clipping rule also includes: A threshold of the normalized mean value of the intra-cluster data is preset, and clusters whose normalized mean value of the intra-cluster data is greater than the threshold of the normalized mean value of the intra-cluster data are pruned.

8. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 1 is characterized in that: The process of obtaining the primary routine variation degree includes: Select any signal strength data as the target signal, take the target signal as the center, select a set number of data to construct a target sequence of the target signal, calculate the average of the absolute values ​​of the differences between the target signal and its adjacent signal strength data, and use the average of the absolute values ​​of the differences as the local difference of the target signal; calculate the difference between the maximum and minimum values ​​of the signal strength data in the target sequence of the target signal, and use the difference between the maximum and minimum values ​​as the overall difference of the target signal; The local difference and the overall difference are standardized, and then the Euclidean distance between the processed local difference and the overall difference is calculated, and the Euclidean distance is used as the primary conventional variation degree of the target signal.

9. The signal fault identification method based on multi-sensor spatiotemporal alignment according to claim 2 is characterized in that: The process of obtaining the distance adjustment factor includes: Obtain the numerical mean of all signal strength data that are consistent with the distance between the drone and the signal source at the sampling time corresponding to the target signal in the historical signal strength data, use the absolute value of the difference between the target signal and the numerical mean as the first parameter, and use the distance between the drone and the signal source at the sampling time corresponding to the target signal as the second parameter; The product of the first parameter and the attenuation coefficient is used as the negative exponent of the exponential function, the exponential function is operated, and the obtained result is multiplied by the second parameter to obtain the distance adjustment factor of the target signal.

10. A signal fault identification system based on multi-sensor spatiotemporal alignment, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the signal fault identification method based on multi-sensor spatiotemporal alignment according to any one of claims 1 to 9 is implemented.

Images