Method and device for detecting anomaly of soil moisture monitoring data based on time sequence

Abstract

The invention provides an anomaly detection method and device for soil moisture monitoring data based on a time series, and the method comprises the steps: obtaining soil moisture values of a plurality of time steps collected by a soil moisture sensor; performing moisture change detection based on the soil moisture values of the plurality of time steps, and determining a soil moisture change result of a target time step; determining the response minimum rainfall of the soil moisture sensor based on the measurement depth and the measurement precision of the soil moisture sensor; based on the soil moisture change result and the response minimum rainfall, a data detection result of a first soil moisture value of the target time step is determined, and the data detection result is used for representing whether the soil moisture change result of the target time step is caused by rainfall or not; according to the invention, the reliability of soil moisture monitoring data can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of soil moisture monitoring, and particularly to an abnormal detection method and device for soil moisture monitoring data based on time series. Background Art

[0002] Soil moisture, as an important part of water resources, is one of the most important factors in terrestrial ecosystems. Therefore, it is crucial to monitor the soil moisture status.

[0003] All along, there have been many methods for monitoring soil moisture, such as the most commonly used oven-drying method, resistance method, and gamma-ray method, etc. These methods can accurately measure the soil moisture value at a single point, but the measurement accuracy is easily affected by human activities, weather conditions, and the monitoring range. The single-point measurement results cannot meet the application requirements.

[0004] It can be seen that the soil moisture monitoring methods in the related art have the technical problem of low data reliability easily caused by external factor interference. Summary of the Invention

[0005] The present invention provides an abnormal detection method and device for soil moisture monitoring data based on time series, which are used to solve the defect of low data reliability of the soil moisture monitoring method in the prior art easily caused by external factor interference, and to improve the reliability of soil moisture monitoring data.

[0006] The present invention provides an abnormal detection method for soil moisture monitoring data based on time series, including the following steps.

[0007] Obtain the soil moisture values at multiple time steps collected by a soil moisture sensor; perform moisture change detection based on the soil moisture values at the multiple time steps to determine the soil moisture change result at a target time step; determine the minimum rainfall response of the soil moisture sensor based on the measurement depth and measurement accuracy of the soil moisture sensor; determine the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

[0008] An anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, which performs moisture change detection based on the soil moisture values at multiple time steps to determine the soil moisture change result at the target time step, includes: determining a first soil moisture value at the target time step, a second soil moisture value at the previous time step of the target time step, and a third soil moisture value at the time step 24 hours before the target time step; when the first soil moisture value is greater than the second soil moisture value, determining that the soil moisture change result at the target time step meets the first precipitation condition, where the first precipitation condition is used to indicate that there is an increase in the target soil moisture value; determining the difference between the first soil moisture value and the third soil moisture value; when the difference is greater than the moisture deviation threshold, determining that the soil moisture change result at the target time step meets the second precipitation condition, where the second precipitation condition is used to indicate that the target soil moisture value exceeds the daily value.

[0009] An anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, which determines the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the response minimum rainfall, includes: obtaining the actual rainfall at the target time step; when the actual rainfall is less than the response minimum rainfall and the soil moisture change result at the target time step meets the first precipitation condition and the second precipitation condition, determining that the data detection result of the first soil moisture value at the target time step is data anomaly.

[0010] An anomaly detection method for soil moisture monitoring data based on time series provided by the present invention. After determining the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the response minimum rainfall, the method further includes: when the data detection result is data anomaly, performing time series frequency spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step; the performing time series frequency spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step includes: determining a first ratio between the first soil moisture value at the target time step and the second soil moisture value at the previous time step of the target time step; when the first ratio is greater than the rising threshold or less than the falling threshold, determining that the first soil moisture value meets the first peak condition; determining a second ratio between the second soil moisture value and the fourth soil moisture value at the next time step of the target time step; when the absolute value of the second derivative of the second ratio is within a preset numerical range, determining that the first soil moisture value meets the second peak condition; determining the variance and average value of the soil moisture values between the time step 12 hours before the target time step and the time step 12 hours after the target time step; when the absolute value of the third ratio of the variance to the average value is less than 1, determining that the first soil moisture value meets the third peak condition; when the first soil moisture value meets the first peak condition, the second peak condition, and the third peak condition, determining that the first soil moisture value at the target time step is an abnormal peak.

[0011] An anomaly detection method for soil moisture monitoring data based on time series provided by the present invention. The performing time series frequency spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step includes: determining the relative change and absolute change between the first soil moisture value at the target time step and the second soil moisture value at the previous time step of the target time step; when the relative change is greater than the first change threshold and the absolute change is greater than the second change threshold, determining that the first soil moisture value meets the first interruption condition; determining the average value of the first derivatives of the soil moisture values of all time steps within 24 hours centered on the target time step; determining the product of the average value and the preset interruption weight; when the first derivative of the first soil moisture value is greater than the product, determining that the first soil moisture value meets the second interruption condition; based on the second derivative of the first soil moisture value, determining whether the first soil moisture value meets the third interruption condition; when the first soil moisture value meets the first interruption condition, the second interruption condition, and the third interruption condition, determining that the first soil moisture value at the target time step is an abnormal interruption.

[0012] An anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, the third interruption condition is represented by the following formula: and Wherein, represents the second derivative of the first soil moisture value, represents the fourth soil moisture value at the next time step of the target time step, represents the fifth soil moisture value at the second next time step of the target time step.

[0013] The present invention also provides an anomaly detection device for soil moisture monitoring data based on time series, including the following modules: An acquisition module for acquiring soil moisture values at multiple time steps collected by a soil moisture sensor; a moisture detection module for performing moisture change detection based on the soil moisture values at the multiple time steps to determine the soil moisture change result at the target time step; a determination module for determining the minimum rainfall response of the soil moisture sensor based on the measurement depth and measurement accuracy of the soil moisture sensor; a precipitation detection module for determining the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, wherein the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, it implements the anomaly detection method for soil moisture monitoring data based on time series as described in any one of the above.

[0015] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the anomaly detection method for soil moisture monitoring data based on time series as described in any one of the above.

[0016] The present invention also provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the anomaly detection method for soil moisture monitoring data based on time series as described in any one of the above.

[0017] The anomaly detection method and device for soil moisture monitoring data based on time series provided by the present invention can monitor the soil moisture status in real time by obtaining the soil moisture values collected by the soil moisture sensor within multiple time steps; determine its minimum rainfall response based on the measurement depth and measurement accuracy of the soil moisture sensor, which helps to determine the performance limitations and applicable ranges of the soil moisture sensor, thereby improving the accuracy of data interpretation; determine the data detection result by combining the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall, which can further verify the rationality of the data and reduce misjudgments caused by sensor errors or environmental factors; and improve the accuracy and reliability of the soil moisture monitoring data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] To more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art one by one. Obviously, the following drawings are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 It is a schematic flowchart of the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention.

[0020] Figure 2 It is a schematic diagram of the verification of the threshold dynamic range of the target site provided by the present invention.

[0021] Figure 3 It is a schematic diagram of the indirect verification of the auxiliary data provided by the present invention.

[0022] Figure 4 It is a schematic diagram of the verification of the time series frequency spectrum analysis provided by the present invention.

[0023] Figure 5 It is a schematic diagram of the Mann-Kendall mutation test provided by the present invention.

[0024] Figure 6 It is a schematic structural diagram of the anomaly detection device for soil moisture monitoring data based on time series provided by the present invention.

[0025] Figure 7 It is a schematic physical structure diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0026] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] Soil moisture, as an important part of water resources, is one of the most important factors in terrestrial ecosystems. Therefore, it is crucial to monitor the status of soil moisture. All along, there have been many methods for monitoring soil moisture, such as the most commonly used oven-drying method, resistance method, γ-ray method, and time domain reflectometry, etc. These methods can accurately measure the soil moisture value at a single point, but the measurement accuracy is easily affected by human activities, weather conditions, and the monitoring range. The single-point measurement results cannot meet the application requirements. Therefore, accurate, timely, and long-term large-scale ground observations of soil moisture can reveal the water cycle trends related to climate or land cover changes.

[0028] In recent years, with the development and progress of satellite remote sensing technology, soil moisture monitoring technologies based on satellite visible-light near-infrared and thermal infrared remote sensing data, active microwave, and passive microwave have also been developed, greatly improving the ability of soil moisture monitoring. However, the satellite remote sensing inversion of surface soil moisture has not achieved the expected accuracy. The product accuracy is low, and the standard specifications among different products are inconsistent, unable to meet the user requirements. Therefore, site-based soil moisture measurement that can represent the true value at the pixel scale and characterize the time dynamic characteristics is the key to calibrating and validating satellite-based soil moisture inversion.

[0029] Due to the high cost of building a long-term soil moisture measurement network and the insufficient understanding of the importance of soil moisture in climate simulation and regional weather forecasting, long-term soil moisture measurement networks are very scarce. The International Soil Moisture Network uses a variety of methods to detect suspicious soil moisture measurement values, which can be subdivided into geophysical dynamic range verification, geophysical consistency methods, and time series spectrum-based methods. Geophysical consistency and spectrum-based methods are only applicable to soil moisture observation data, while geophysical dynamic range verification is applicable to all dynamic variables in the database.

[0030] With the development of the soil moisture network, many progresses have been made in basic theory and applied research. However, the existing observation data still have deficiencies such as inconsistent sensors and soil depth, uneven sampling distribution, and only at the regional scale.

[0031] In the embodiments of the present invention, by using the site network resources of existing industries and research institutions, ground observation equipment can be efficiently deployed to ensure the high-precision acquisition of common product data of soil moisture at each site. To achieve the precise monitoring of soil moisture changes. Considering factors such as soil type, terrain, and vegetation coverage, representative and different sites are selected as test station nodes. This ensures the wide coverage and in-depth representativeness of the data, which helps to comprehensively reflect the soil moisture conditions of different ecosystems.

[0032] For example, it covers various ecological types such as grasslands, farmlands, deserts, and forests. This wide coverage of ecological types enables a more accurate assessment of the responses of different ecosystems to soil moisture changes. By using high-precision soil moisture and soil temperature sensors to monitor the changes in soil temperature and humidity in real time, it not only provides accurate in-situ data but also supports pixel-scale observations, which helps to deeply reveal the spatial distribution and dynamic changes of soil moisture.

[0033] Optionally, the anomaly detection method for soil moisture monitoring data based on time series in the embodiments of the present application can be executed by a server, or by a terminal device, or jointly by a server and a terminal device. Taking the execution of the anomaly detection method for soil moisture monitoring data based on time series in this embodiment by a server as an example.

[0034] Figure 1 is a schematic flowchart of the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention. As Figure 1 shown, the method includes the following steps.

[0035] Step 101, obtain soil moisture values at multiple time steps collected by a soil moisture sensor.

[0036] In the embodiments of the present invention, a soil automatic observation network for typical ecosystems (including grasslands, farmlands, deserts, and forests) is established in advance. The network has unified sensors (i.e., soil moisture sensors), unified sample designs, unified collection depths, unified calibration methods, and unified data processing procedures. The soil automatic observation network includes multiple soil moisture sensors, which can automatically obtain in-situ soil humidity data (i.e., soil moisture values) in real time, as well as a spatial scale up to the pixel level.

[0037] For example, at the detection site corresponding to each soil moisture sensor, sample points are arranged in a verification sample area with the local typical vegetation type and relatively flat terrain. The area of the sample area at each site is about 1 ha, and the measurement interval of the soil moisture and soil temperature sensors is 30 minutes. Each test station consists of 10 observation nodes (i.e., soil moisture sensors), and each observation node includes 4 observation soil layer depths (5 cm, 10 cm, 20 cm, and 40 cm).

[0038] Step 102: Perform moisture change detection based on soil moisture values at multiple time steps to determine the soil moisture change result at the target time step.

[0039] To ensure the data reliability of the soil moisture sensor, it is necessary to complete the quality marking of the original data. In the embodiments of the present invention, an automatic quality control method is adopted, mainly selecting methods such as threshold dynamic range verification, auxiliary data indirect verification, time series spectrum analysis verification, and MK test trend analysis.

[0040] In some embodiments, the threshold dynamic range verification detects soil moisture observation values that exceed the physically reasonable range through the direct threshold method. The direct threshold method detects the observed data through the preset maximum and minimum values of the soil moisture content. If the observed value exceeds this range, these data (observed values exceeding the threshold) are considered outliers, which may be caused by measurement errors, instrument failures, or other reasons. For example, the soil moisture content between 0.0 (m³ / m³) and 0.6 (m³ / m³) is considered reasonable.

[0041] Reference Figure 2 , Figure 2 is a schematic diagram of the threshold dynamic range verification of the target site provided by the present invention, where the vertical coordinate is the soil moisture (value) and the horizontal coordinate is the time.

[0042] Soil moisture content less than 0 m³ / m³ or greater than 0.6 m³ / m³ is marked as invalid data. As Figure 2 shown, only a small part of the time periods exceed the saturated moisture content, and the time periods exceeding the saturated moisture content are mainly concentrated around September 2018 and 2021, mainly because there are obvious rainfall events around these time periods.

[0043] According to the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, moisture change detection is performed based on soil moisture values at multiple time steps to determine the soil moisture change result at the target time step, including: Determine the first soil moisture value at the target time step, the second soil moisture value at the previous time step of the target time step, and the third soil moisture value at the time step 24 hours before the target time step; When the first soil moisture value is greater than the second soil moisture value, determine that the soil moisture change result at the target time step meets the first precipitation condition, where the first precipitation condition is used to indicate that there is an increase in the target soil moisture value; Determine the difference between the first soil moisture value and the third soil moisture value; When the difference is greater than the moisture deviation threshold, it is determined that the soil moisture change result at the target time step meets the second precipitation condition, where the second precipitation condition is used to indicate that the target soil moisture value exceeds the daily value.

[0044] In the embodiments of the present invention, the relationship between precipitation and soil moisture response is used as the basis for identifying false soil moisture observations. If the soil moisture rises but there is no significant rainfall within the previous 24 hours, the observed value is marked as suspicious.

[0045] In some embodiments, the moisture deviation threshold is determined based on the standard deviation of the soil moisture value within 24 hours.

[0046] If the following conditions for time step t are met, a significant increase in soil moisture (i.e., exceeding the daily variation caused by temperature and noise) can be identified: where is the first soil moisture value at the target time step t (in hours), represents the second soil moisture value at the previous time step of the target time step, represents the third soil moisture value at the time step 24 hours before the target time step, is the standard deviation of the soil moisture value (x) within the previous 24 hours.

[0047] If , it is determined that the soil moisture at the target time step has an upward trend and meets the first precipitation condition.

[0048] If , it is determined that the soil moisture change result at the target time step meets the second precipitation condition, that is, the soil moisture value exceeds the daily fluctuation range.

[0049] Here, the first precipitation condition ensures that only the increase in soil moisture is marked. The second precipitation condition ensures that the identified increase in soil moisture exceeds the daily variation of soil moisture, such as the variation caused by temperature changes.

[0050] According to the above conditions, it can be comprehensively judged whether the soil moisture change at the target time step meets the precipitation condition. For example, if both the first precipitation condition and the second precipitation condition are met, it may indicate a significant precipitation event that causes a significant increase in soil moisture. If only the first precipitation condition is met, it may indicate an upward trend in soil moisture, but not necessarily a significant precipitation event. If neither is met, it may indicate that the soil moisture change is within the normal range.

[0051] Refer to Figure 3 ,Figure 3 This is a schematic diagram for indirectly verifying auxiliary data provided by the present invention. Here, the left vertical coordinate represents soil moisture, the right vertical coordinate represents precipitation, and the horizontal coordinate represents time; Indirect verification of the increase in soil moisture using rainfall data is applicable when the sensor depth is less than or equal to 10 cm. When the soil water content and rainfall meet the above precipitation conditions, they will be quality marked. Figure 3 As shown, during periods with fewer precipitation events, soil moisture decreases over time. During periods with larger changes in soil moisture, the marked points are more dispersed, while during periods with slower changes in soil moisture, the marked points are more concentrated, indicating that this method is more sensitive to changes in soil moisture within a certain range. However, this also leads to some mislabeling phenomena. But for most cases of soil moisture increase without rainfall events, this method has captured them well.

[0052] Through the embodiments of the present invention, by setting the first precipitation condition (soil moisture increase) and the second precipitation condition (soil moisture exceeding the daily value), this method can more accurately identify precipitation events. It is more accurate than a single soil moisture threshold judgment because it takes into account the change trend and daily fluctuation range of soil moisture.

[0053] Step 103: Determine the minimum rainfall for the response of the soil moisture sensor based on the measurement depth and measurement accuracy of the soil moisture sensor.

[0054] In the embodiments of the present invention, for the measured values (soil moisture values) that meet the above first precipitation condition and second precipitation condition, check the occurrence and amount of precipitation within 24 hours before the measurement; if an increase in soil moisture is identified within 24 hours before the measurement but there is no rainfall, mark this measured value as suspicious. Whether the soil moisture sensor can detect a rainfall event depends on the rainfall amount, the installation depth of the sensor, and the accuracy of the sensor. Therefore, the minimum rainfall required for the sensor response (in meters) is determined by the following factors: where D is the measurement depth (m) of the soil moisture sensor, A is the accuracy of the soil moisture sensor, and p is the soil porosity. Here, we use an average sensor accuracy of 0.05 m 3 / m -3 , and an average p value of 0.5. If the total rainfall (P) within 24 hours before the increase in soil moisture is less than , then mark the soil moisture measurement value as suspicious.

[0055] Step 104: Determine the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall for the response.

[0056] Among them, the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

[0057] According to the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, based on the soil moisture change result and the response minimum rainfall, determining the data detection result of the first soil moisture value at the target time step includes: Obtain the actual rainfall at the target time step; When the actual rainfall is less than the response minimum rainfall and the soil moisture change result at the target time step satisfies the first precipitation condition and the second precipitation condition, determine that the data detection result of the first soil moisture value at the target time step is data anomaly.

[0058] In the embodiment of the present invention, the first precipitation condition ( ) ensures that the soil moisture increases at time step t relative to the previous time step t - 1. The second precipitation condition ( ) ensures that the increase in soil moisture at time step t relative to the moisture value 24 hours ago exceeds twice the standard deviation of the soil moisture within the previous 24 hours.

[0059] For the measured values that meet the above soil moisture increase conditions, check the rainfall situation within the previous 24 hours. If there is no rainfall record within the previous 24 hours, or the rainfall is less than the minimum rainfall required for the sensor response ( ), then mark the soil moisture measurement value (soil moisture value) as suspicious.

[0060] It can be understood that the embodiment of the present invention is applicable to non-irrigated areas.

[0061] Through the embodiment of the present invention, by setting clear soil moisture increase conditions (the first precipitation condition and the second precipitation condition), the significant increase in soil moisture can be accurately identified, avoiding the interference of daily fluctuations and noise; combined with the inspection of rainfall events, it further ensures the correlation between the identified soil moisture increase and rainfall events, improving the accuracy of monitoring.

[0062] In some embodiments, an anomaly detection method for soil moisture monitoring data based on time series provided by the present invention can also be implemented by means of deep learning, specifically including the following steps: Step 1, collect soil moisture values at multiple time steps from the soil moisture sensor.

[0063] The soil moisture values should contain sufficient historical data so that the deep learning model can learn the natural change patterns of soil moisture and perform data cleaning (removing outliers or missing values in the data to ensure data quality) and standardization (standardizing or normalizing the soil moisture values and other features) on the collected soil moisture values.

[0064] Step 2: Select a suitable deep learning model. Such as recurrent neural network (RNN), long short-term memory network (LSTM), gated recurrent unit (GRU), or convolutional neural network (CNN), etc. In particular, the long short-term memory network performs excellently in processing time series data.

[0065] Among them, the model architecture of the deep learning model includes an input layer, a hidden layer, and an output layer. The input layer receives the soil moisture values and other features; the hidden layer is responsible for learning the internal laws and patterns of the data; the output layer predicts whether the change in soil moisture at the target time step is caused by rainfall.

[0066] Train the model using the historical data of soil moisture values, and optimize the performance of the model by adjusting model parameters (such as learning rate, batch size, number of iterations, etc.); use independent validation sets and test sets to evaluate the performance of the model to ensure that the model has good generalization ability.

[0067] Step 3: Take the response minimum rainfall as an input feature or constraint condition of the model.

[0068] According to the measurement depth and measurement accuracy of the soil moisture sensor, combined with the historical data of rainfall events, determine the minimum rainfall that the sensor can respond to, which can be achieved by analyzing the correlation between rainfall events and soil moisture changes. Take the response minimum rainfall as an input feature or constraint condition of the model to help the model more accurately judge whether the change in soil moisture is caused by rainfall.

[0069] Step 4: Use the trained deep learning model to predict the soil moisture values at the target time step, obtain the data detection results, and judge whether they are caused by rainfall.

[0070] According to the output of the model, provide the data detection results, including information such as whether soil moisture changes caused by rainfall are detected, the magnitude and duration of the changes, etc.

[0071] Through the embodiments of the present invention, by means of the deep learning method, deploy the trained deep learning model into the actual environment to realize real-time monitoring and early warning of soil moisture changes, and can more accurately and efficiently detect whether soil moisture changes are caused by rainfall.

[0072] According to the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, after determining the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the response to the minimum rainfall, the above method further includes: When the data detection result is data anomaly, perform time series frequency spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step.

[0073] In the embodiments of the present invention, by the shape of the time series of the soil moisture value, detect outliers (peaks), positive and negative interruptions, signal saturation, and non-responsive sensors.

[0074] According to the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, performing time series frequency spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step includes: Determine the first ratio between the first soil moisture value at the target time step and the second soil moisture value at the previous time step of the target time step; When the first ratio is greater than the rising threshold or less than the falling threshold, determine that the first soil moisture value meets the first peak condition; Determine the second ratio between the second soil moisture value and the fourth soil moisture value at the next time step of the target time step; When the absolute value of the second derivative of the second ratio is within a preset numerical range, determine that the first soil moisture value meets the second peak condition; Determine the variance and average value of the soil moisture values between the time step 12 hours before the target time step and the time step 12 hours after the target time step; When the absolute value of the third ratio of the variance to the average value is less than 1, determine that the first soil moisture value meets the third peak condition; When the first soil moisture value meets the first peak condition, the second peak condition, and the third peak condition, determine that the first soil moisture value at the target time step is an abnormal peak.

[0075] Traditional peak detection methods may misjudge the sudden increase in soil moisture caused by precipitation as an error. Therefore, the embodiments of the present invention propose a new method based on a series of conditions applied to measure the humidity value (soil moisture value) and its second derivative.

[0076] In some embodiments, the derivative is calculated by using a Savitzky-Golay filter with a second-order polynomial fitting and a window size of 3 hours.

[0077] An observed data (i.e., the first soil moisture value) will be marked as a spike (abnormal peak) only when all of the following 3 conditions (peak conditions) are met.

[0078] First, there must be a large change in soil moisture over 2 consecutive time steps (i.e., the first peak condition). That is, a minimum rise or fall of 15% compared to the previous value in time step t (the target time step) is recognized as a potential spike. The first peak condition can be expressed by the following formula: where, represents the first soil moisture value at the target time step, represents the second soil moisture value at the previous time step of the target time step.

[0079] The above formula cannot distinguish whether the peak is due to precipitation activity. Based on the typical behavior of the second derivative near the peak, the second peak condition is added in the embodiments of the present invention.

[0080] A positive peak (negative peak) will cause a strong negative peak (positive peak) in the second derivative at time step t, and be accompanied by two lower positive peaks (negative peaks) around the t - 1 and t + 1 moments. Assuming that there is no drastic change in soil moisture content between t - 1 and t + 1, the ratio between these time steps is close to 1 (whether it is a positive peak or a negative peak).

[0081] In some embodiments, it is observed from the calibration dataset that the natural variation range of this ratio is between 0.8 and 1.2. The second peak condition can be expressed by the following formula: where, represents the second derivative of the second soil moisture value at the previous time step of the target time step, represents the second derivative of the fourth soil moisture value at the next time step of the target time step.

[0082] Since the second derivative cannot handle noisy data well, we add a third criterion. Based on the coefficient of variation centered on t but not including t itself within 24 hours, the third peak condition can be expressed by the following formula: where, represents the soil moisture values between the time step 12 hours before the target time step and the time step 12 hours after the target time step, is the variance, is the average value, and the threshold is generated from the characteristics of the coefficient of variation, where values greater than 1 indicate data with a large amount of noise. It should be noted that since a 24-hour time window covering "future" observations is required, this check cannot be performed in near real-time.

[0083] Through the embodiments of the present invention, the first peak condition is used to identify significant increases or decreases in soil moisture, that is, potential spikes. By setting a threshold of 15%, those soil moisture data points that change significantly in a short period of time can be screened out, providing a basis for subsequent spike detection.

[0084] The second peak condition utilizes the typical characteristics of the second derivative near the peak, that is, the second derivative is zero at the peak and the signs on both sides are opposite. By checking the characteristics of the second derivative at times t - 1 and t + 1, the existence of spikes can be further confirmed, and those soil moisture changes that do not conform to the spike characteristics can be excluded. At the same time, by setting a threshold for the ratio (between 0.8 and 1.2), those outliers that may be caused by noise or other factors can be further screened out.

[0085] The third peak condition is used to suppress the influence of noise on spike detection. By calculating the coefficient of variation (i.e., the ratio of variance to the average value) within a 24-hour time window centered on t but not including t itself, the fluctuation of soil moisture within this time period can be evaluated. When the coefficient of variation is less than 1, it is considered that the data noise within this time period is small, which is conducive to the accurate detection of spikes.

[0086] According to the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, perform time series spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step, including: Determine the relative change and absolute change between the first soil moisture value at the target time step and the second soil moisture value at the previous time step of the target time step; When the relative change is greater than the first change threshold and the absolute change is greater than the second change threshold, determine that the first soil moisture value satisfies the first interruption condition; Determine the average value of the first derivatives of the soil moisture values at all time steps within 24 hours centered on the target time step; Determine the product of the average value and the preset interruption weight; When the first derivative of the first soil moisture value is greater than the product, determine that the first soil moisture value satisfies the second interruption condition; Based on the second derivative of the first soil moisture value, determine whether the first soil moisture value satisfies the third interruption condition; When the first soil moisture value satisfies the first interruption condition, the second interruption condition, and the third interruption condition, determine that the first soil moisture value at the target time step is an abnormal interruption.

[0087] Abrupt interruptions are characterized by a sudden increase (jump) or decrease (drop) in soil moisture. Jumps and drops usually result in a persistent deviation from the previous period. In an embodiment of the present invention, this typical behavior is captured by observing the time series and its first and second derivatives. The derivatives are calculated in a similar manner to spike detection. To be marked as an interruption (abrupt interruption), the following 3 conditions (interruption conditions) need to be met.

[0088] In the first interruption condition, the relative change in soil moisture needs to reach at least 10%. Additionally, to prevent over - marking of low absolute humidity values, the absolute change in soil moisture needs to reach at least 0.01 m³ / m -3 , and the first interruption condition can be referred to the following formula: and where, represents the first soil moisture value at the target time step, represents the second soil moisture value at the previous time step of the target time step.

[0089] In an embodiment of the present invention, the relative change is used to represent the percentage change of the soil moisture value relative to the previous time step; the absolute change is used to represent the actual change amount of the soil moisture value between two time steps.

[0090] Negative (positive) mutations are manifested as a strong negative (positive) change in the first derivative. In an embodiment of the present invention, it is assumed that the value should be at least 10 times smaller (larger). To make this criterion more robust, the first derivative is compared with the average value of all first derivative values within 24 hours centered on t, so the second interruption condition can be referred to the following formula: where, represents the first derivative of the first soil moisture value, represents the total number of time steps, k represents the index of the time step (from the time step 12 hours before to the time step 12 hours after), represents the target time step.

[0091] In an embodiment of the present invention, it is necessary to calculate the first derivatives of the soil moisture values of all time steps within 24 hours centered on the target time step t, and find the average value of these first derivatives. The first derivative reflects the rate of change of the soil moisture value over time; then, we calculate the product of this average value and the preset interruption weight (10 in the above formula); finally, we compare the first derivative of the soil moisture value at the target time step t with the above - mentioned product. In some embodiments, the second derivative reflects the change in the rate of change of the soil moisture value, i.e., acceleration.

[0092] According to the anomaly detection method for soil moisture monitoring data based on time series provided by the present invention, the third interruption condition is expressed by the following formula: and where x represents the second derivative of the first soil moisture value, x represents the fourth soil moisture value at the next time step of the target time step, x represents the fifth soil moisture value at the second next time step of the target time step.

[0093] Negative (positive) mutations result in large negative (positive) second derivatives at time t, followed by large positive (negative) values at time t + 1: the peak sizes in the second derivative are approximately the same (but of opposite signs), resulting in a ratio close to 1. At time t + 2, the second derivative returns to a value close to zero; thus, the ratio of the absolute second derivative between t + 1 and t + 2 is very large. This leads to the following conditions: and Since it is not possible to make a reliable statement about the reasonableness of the measured values before and after the jump, only the jump (interruption) itself is marked. Similarly, this check cannot be applied to near-real-time data.

[0094] Reference Figure 4 , Figure 4 is a schematic diagram of the time series spectrum analysis verification provided by the present invention, where the vertical coordinate is soil moisture and the horizontal coordinate is time.

[0095] As Figure 4 shown, Figure 4 the red dots in Figure 4 are the points of detected positive breaks. Obviously Figure 4 there are no points of negative breaks in Figure 4 , largely because there is no obvious negative break phenomenon. For positive breaks, although the method has well identified some points of positive breaks, there are still some points of positive breaks that are not definitely identified, which is attributed to the difficulty of distinguishing artificial events from the natural increase in soil moisture. Similar to the spike detection test, when values are missing before the jump, the interruption detection does not work.

[0096] Through the embodiments of the present invention, by combining relative change and absolute change, first derivative, second derivative, and preset thresholds and weights, possible abnormal interruptions can be more comprehensively identified.

[0097] Through the above steps of the embodiments of the present invention, by obtaining the soil moisture values collected by the soil moisture sensor at multiple time steps, the soil moisture condition can be monitored in real time; based on the measurement depth and measurement accuracy of the soil moisture sensor to determine its response to the minimum rainfall, which helps to determine the performance limitations and applicable ranges of the soil moisture sensor, thereby improving the accuracy of data interpretation; by combining the soil moisture change results with the response to the minimum rainfall to determine the data detection results, the rationality of the data can be further verified, and the misjudgment caused by sensor errors or environmental factors can be reduced; when the data detection result is data anomaly, the anomaly type is further determined through time series spectrum analysis. The application of time series spectrum analysis can more accurately identify the anomaly types in the soil moisture values, improving the accuracy and reliability of the soil moisture monitoring data.

[0098] In some embodiments, constant plateau (regions where soil moisture values remain relatively constant over a period of time) soil moisture values (or plateaus) are identified through three conditions. First, in order to distinguish them from regular wet events and peaks, these values need to remain unchanged for at least 12 hours.

[0099] Therefore, throughout the data cycle, time intervals with a minimum length of 12 hours are searched, within which the variance of the soil moisture shall not exceed 1% of the average uncertainty of the sensor, which is 0.05 m³ / m. -3 This small variation is allowed because the soil moisture readings of plateaus are not always completely stable. Specifically, it can refer to the following formula: where represents that the time interval is a symmetric interval centered on a certain time step t, represents the variance of the soil moisture within the time interval, n represents the interval duration (greater than 6 hours), and interval represents the continuous time range.

[0100] To distinguish plateaus from other constant soil moisture values (for example, soil moisture values after a long period of no precipitation), a second criterion based on the first derivative is introduced. Since plateaus usually occur after intense precipitation events, there will be a local maximum of the first derivative before the plateau. The end of the plateau is represented by a local minimum of the first derivative. The intervals generated by the first condition provide a basis for checking the local extrema that must satisfy these thresholds. The minimum and maximum thresholds of the first derivative are determined based on the calibration dataset.

[0101] where, t plateau_start represents the time point when the plateau starts, t plateau_end represents the time point when the plateau ends, Indicates the existence of a local maximum of the first derivative within the interval , indicating the existence of a local maximum of the first derivative within the interval Indicates the existence of a local minimum of the first derivative within the interval .

[0102] From a physics perspective, plateaus always occur at the highest soil moisture values in the time series. Therefore, the embodiments of the present invention add a third condition, assuming that plateaus only occur at at least 95% of the maximum soil moisture level measured during the entire observation period: µ(x[t plateau_start , t plateau_end ) > max(x[t0,t end ) * 0.95 where µ(x[t plateau_start , t plateau_end ) represents the average value of the soil moisture values between the plateau start time t plateau_start and the plateau end time t plateau_end , and max(x[t0,tend]) represents the maximum soil moisture level during the entire observation period, that is, the maximum value of the soil moisture values from the observation start time t0 to the observation end time tend.

[0103] Low constant values are essentially different from plateaus, and they are usually the result of frozen soil or sensor failures. Although frost events can be captured by negative soil temperatures (see "Using Soil Temperature"), there will be a sharp negative mutation before the low-level plateaus caused by sensor failures. Therefore, the negative mutation detection algorithm described in "Mutation Detection" is first used to identify potential offsets of low constant values. Starting from this offset, all subsequent measurements are marked as "low constant values" as long as they meet the following conditions: where represents the variance of the soil moisture values from time t to t + n, represents the average value of the soil moisture values from time t to t + n, t represents the detected negative mutation, and the minimum length of n is 12, which is consistent with plateaus.

[0104] In some embodiments, the Mann-Kendall method is used to analyze the long-term evolution trend of the basin hydro-meteorological sequence and analyze the precipitation evolution trend.

[0105] Mutation test: Let the element time series be X1, X2,... X n , and S k represents the jth sample X j > X iThe cumulative number of (1 ≤ i ≤ j), define the statistic S k : where, represents the statistic (i.e., the number of samples X j > X i ), k represents the number of samples that have been counted, n represents the total number of samples, equals 1 (when ) or 0 (when ).

[0106] Under the assumption of random independence of the time series, the mean and variance of S k are respectively: where, represents 's mean, represents 's variance, k represents the number of samples that have been counted, n represents the total number of samples.

[0107] Standardize S k : where UF1 = 0, given the significance level α, if |UF k | > Uα, it indicates that there is an obvious trend change in the sequence. All UF k can form a curve. Apply this method to the reverse sequence, and represent the reverse sequence X n , X n-1 ,...., X1 as X'1, X'2,...., X'n. When j' = n + 1 - j, then the UB k of the reverse sequence is: where UB1 = 0. UB k is not simply equal to the negative value of UF k , but is inverted and then negated. Here, UF k is calculated according to the reverse sequence.

[0108] Given the significance level, if α = 0.05, then the critical value is ±1.96. Draw the UF k and UB k curves and the two straight lines of ±1.96 on one graph. Analyze the drawn UF k and UB k curves. If UF k or UB kIf the value is greater than 0, it indicates that the sequence shows an upward trend; if it is less than 0, it indicates a downward trend. When they exceed the critical line, it indicates a significant upward or downward trend. The range exceeding the critical line is determined as the time region where mutation occurs. If UF k and UB k have an intersection point, and the intersection point is between the critical lines, then the corresponding moment of the intersection point is the time when the mutation starts.

[0109] The MK mutation test is a non-parametric statistical method used to evaluate whether there are obvious trends in time series data. It does not require assumptions about the distribution of the data, so it is applicable to various types of time series data, including data that does not follow a normal distribution, and is not affected by a few outliers. It can effectively identify the mutation points in the data, that is, the points where the data trend changes significantly.

[0110] The MK mutation test technology can accurately identify the upward or downward trend in time series data by calculating the statistics UF k and UB k . When the value of UFk or UBk exceeds the critical value, it indicates that there is a significant trend change in the data.

[0111] By plotting the curves of UF k and UB k and combining with the critical line, the mutation point can be accurately located. The corresponding moment of the mutation point is the moment when the data trend changes significantly, which is of great significance for prediction and early warning.

[0112] Reference Figure 5 , Figure 5 is a schematic diagram of the Mann-Kendall mutation test provided by the present invention. Among them, the abscissa is the date (t (year)), and the ordinate is the statistic.

[0113] Perform the Mann-Kendall mutation test on the daily soil moisture data at different depths from July 2019 to December 2023 at the target site. Given a significance level of 0.05, the analysis results are shown in Figure 5 . In the figure, UF is the standard normal distribution, UB is the inverse sequence of UF. If there is an intersection point between these two curves of UF and UB, and the intersection point is between the critical lines of the significance level, then the corresponding moment of the intersection point is the time when the mutation starts.

[0114] The following describes the anomaly detection device for soil moisture monitoring data based on time series provided by the present invention. The anomaly detection device for soil moisture monitoring data based on time series described below can be mutually corresponding and referenced with the anomaly detection method for soil moisture monitoring data based on time series described above.

[0115] ReferenceFigure 6 , Figure 6 is a schematic structural diagram of an anomaly detection device for soil moisture monitoring data based on time series provided by the present invention.

[0116] An acquisition module 601, configured to acquire soil moisture values at multiple time steps collected by a soil moisture sensor; A moisture detection module 602, configured to perform moisture change detection based on the soil moisture values at the multiple time steps, and determine a soil moisture change result at a target time step; A determination module 603, configured to determine a minimum rainfall response of the soil moisture sensor based on a measurement depth and a measurement accuracy of the soil moisture sensor; A precipitation detection module 604, configured to determine a data detection result of a first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

[0117] Specifically, the above-mentioned anomaly detection device for soil moisture monitoring data based on time series provided by the present invention can implement all method steps implemented by the above-mentioned anomaly detection method embodiment for soil moisture monitoring data based on time series, and can achieve the same technical effects. The same parts and beneficial effects as those in the method embodiment will not be specifically described herein.

[0118] Figure 7 is a schematic physical structure diagram of an electronic device provided by the present invention. As Figure 7 shown, the electronic device may include: a processor 710, a communication interface 720, a memory 730, and a communication bus 740. Among them, the processor 710, the communication interface 720, and the memory 730 communicate with each other through the communication bus 740. The processor 710 can call logical instructions in the memory 730 to execute an anomaly detection method for soil moisture monitoring data based on time series. The method includes: acquiring soil moisture values at multiple time steps collected by a soil moisture sensor; performing moisture change detection based on the soil moisture values at the multiple time steps to determine a soil moisture change result at a target time step; determining a minimum rainfall response of the soil moisture sensor based on a measurement depth and a measurement accuracy of the soil moisture sensor; and determining a data detection result of a first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

[0119] In addition, when the logical instructions in the above-mentioned memory 730 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0120] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the anomaly detection method for soil moisture monitoring data based on time series provided by the above-mentioned various methods. The method includes: obtaining soil moisture values at multiple time steps collected by a soil moisture sensor; performing moisture change detection based on the soil moisture values at multiple time steps to determine the soil moisture change result at a target time step; determining the minimum rainfall response of the soil moisture sensor based on the measurement depth and measurement accuracy of the soil moisture sensor; and determining the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

[0121] In yet another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the anomaly detection method for soil moisture monitoring data based on time series provided by the above-mentioned various methods. The method includes: obtaining soil moisture values at multiple time steps collected by a soil moisture sensor; performing moisture change detection based on the soil moisture values at multiple time steps to determine the soil moisture change result at a target time step; determining the minimum rainfall response of the soil moisture sensor based on the measurement depth and measurement accuracy of the soil moisture sensor; and determining the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

[0122] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative effort.

[0123] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. An anomaly detection method for soil moisture monitoring data based on time series, characterized in that, including: Obtaining soil moisture values at multiple time steps collected by a soil moisture sensor; Performing moisture change detection based on the soil moisture values at the multiple time steps to determine the soil moisture change result at a target time step; Determining the minimum rainfall response of the soil moisture sensor based on the measurement depth and measurement accuracy of the soil moisture sensor; Determining a data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

2. The anomaly detection method for soil moisture monitoring data based on time series according to claim 1, wherein The performing moisture change detection based on the soil moisture values at the multiple time steps to determine the soil moisture change result at a target time step includes: Determining a first soil moisture value at the target time step, a second soil moisture value at the previous time step of the target time step, and a third soil moisture value at the time step 24 hours before the target time step; When the first soil moisture value is greater than the second soil moisture value, determining that the soil moisture change result at the target time step meets a first precipitation condition, where the first precipitation condition is used to indicate an increase in the target soil moisture value; Determining the difference between the first soil moisture value and the third soil moisture value; When the difference is greater than a moisture deviation threshold, determining that the soil moisture change result at the target time step meets a second precipitation condition, where the second precipitation condition is used to indicate that the target soil moisture value exceeds the daily value.

3. The anomaly detection method for soil moisture monitoring data based on time series according to claim 2, characterized in that, The determining a data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response includes: Obtaining the actual rainfall at the target time step; When the actual rainfall is less than the minimum rainfall response and the soil moisture change result at the target time step meets the first precipitation condition and the second precipitation condition, determining that the data detection result of the first soil moisture value at the target time step is data anomaly.

4. The anomaly detection method for soil moisture monitoring data based on time series according to claim 1, wherein, After the determining a data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, the method further includes: When the data detection result is data anomaly, performing time series spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step; The performing time series spectrum analysis on the first soil moisture value at the target time step to determine the anomaly type of the first soil moisture value at the target time step includes: Determining a first ratio between the first soil moisture value at the target time step and the second soil moisture value at the previous time step of the target time step; When the first ratio is greater than an increase threshold or less than a decrease threshold, determining that the first soil moisture value meets a first peak condition; Determining a second ratio between the second soil moisture value and a fourth soil moisture value at the next time step of the target time step; When the absolute value of the second derivative of the second ratio is within a preset numerical range, it is determined that the first soil moisture value satisfies the second peak condition; Determine the variance and average value of the soil moisture values between the 12-hour time step before the target time step and the 12-hour time step after the target time step; When the absolute value of the third ratio of the variance to the average value is less than 1, it is determined that the first soil moisture value satisfies the third peak condition; When the first soil moisture value satisfies the first peak condition, the second peak condition, and the third peak condition, it is determined that the first soil moisture value at the target time step is an abnormal peak.

5. The abnormal detection method for soil moisture monitoring data based on time series according to claim 4, characterized in that, The time series frequency spectrum analysis of the first soil moisture value at the target time step to determine the abnormal type of the first soil moisture value at the target time step includes: Determine the relative change and absolute change between the first soil moisture value at the target time step and the second soil moisture value at the previous time step of the target time step; When the relative change is greater than the first change threshold and the absolute change is greater than the second change threshold, it is determined that the first soil moisture value satisfies the first interruption condition; Determine the average value of the first derivatives of the soil moisture values of all time steps within 24 hours centered on the target time step; Determine the product of the average value and the preset interruption weight; When the first derivative of the first soil moisture value is greater than the product, it is determined that the first soil moisture value satisfies the second interruption condition; Based on the second derivative of the first soil moisture value, determine whether the first soil moisture value satisfies the third interruption condition; When the first soil moisture value satisfies the first interruption condition, the second interruption condition, and the third interruption condition, it is determined that the first soil moisture value at the target time step is an abnormal interruption.

6. The anomaly detection method for soil moisture monitoring data based on time series according to claim 5, characterized in that, The third interruption condition is represented by the following formula: and Wherein, represents the second derivative of the first soil moisture value, represents the fourth soil moisture value at the next time step of the target time step, represents the fifth soil moisture value at the second next time step of the target time step.

7. An abnormal detection device for soil moisture monitoring data based on time series, characterized in that, including: An acquisition module for acquiring the soil moisture values of multiple time steps collected by a soil moisture sensor; A moisture detection module for performing moisture change detection based on the soil moisture values of the multiple time steps to determine the soil moisture change result at the target time step; A determination module for determining the minimum rainfall response of the soil moisture sensor based on the measurement depth and measurement accuracy of the soil moisture sensor; A precipitation detection module for determining the data detection result of the first soil moisture value at the target time step based on the soil moisture change result and the minimum rainfall response, where the data detection result is used to indicate whether the soil moisture change result at the target time step is caused by rainfall.

8. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the abnormal detection method for soil moisture monitoring data based on time series according to any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the abnormal detection method for soil moisture monitoring data based on time series according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the abnormal detection method for soil moisture monitoring data based on time series according to any one of claims 1 to 6.

Images