A method for monitoring soil carbon sequestration on the Qinghai-Tibet Plateau based on a capacitive sensor array

Abstract

This invention discloses a method for monitoring soil carbon sequestration on the Qinghai-Tibet Plateau based on a capacitive sensor array, belonging to the field of soil carbon sequestration monitoring. The method includes: deploying a sensor array to collect capacitive signals; combining time-domain filtering and adaptive adjustment techniques to eliminate noise and adapt to the characteristics of the permafrost transition zone; using a support vector machine to identify water and mineral interference types; and using Kalman filtering to fuse multi-point data for signal correction; extracting carbon content change feature vectors based on the corrected signals; integrating time-series information to dynamically capture change trends; and determining carbon storage fluctuation trends through mapping model parameter adjustments, ultimately generating a carbon sequestration monitoring report and carbon storage change indicators with fused deviation correction results. This invention effectively solves the problems of severe signal interference and insufficient measurement accuracy in complex environments, significantly improving the accuracy and reliability of soil carbon sequestration monitoring.

Description

Technical Field

[0001] This invention belongs to the field of soil carbon sequestration monitoring, and in particular relates to a method for monitoring soil carbon sequestration on the Qinghai-Tibet Plateau based on a capacitive sensor array. Background Technology

[0002] Soil carbon sequestration capacity plays a crucial role in the global carbon cycle and the maintenance of ecological balance, especially in climate change-sensitive regions such as the Qinghai-Tibet Plateau, where accurate monitoring of soil carbon storage is of paramount importance. Currently, various monitoring technologies have been developed in this field, among which capacitive sensing-based measurement methods have attracted attention due to their ease of implementation. These technologies indirectly reflect carbon storage by detecting changes in soil dielectric properties and have already established a certain technological foundation in practical applications.

[0003] However, existing capacitive sensing monitoring methods still have significant limitations in geographically complex areas like the Qinghai-Tibet Plateau. Due to the uneven distribution of moisture and minerals in the soil of the plateau region, and the alternating changes between permafrost and non-permafrost areas, capacitive signals are easily affected by coupling interference from various environmental factors. Traditional signal processing methods struggle to effectively distinguish the influence of different interference sources such as moisture and minerals, and lack adaptive adjustment mechanisms for signal characteristics in permafrost areas, resulting in insufficient correction of measurement biases. Furthermore, existing methods are insufficient in multi-point data fusion, dynamic feature capture, and trend judgment, limiting the accuracy and reliability of monitoring results and failing to meet the practical needs of refined monitoring of load changes in complex environments. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method for monitoring soil carbon sequestration on the Qinghai-Tibet Plateau based on a capacitive sensor array, comprising:

[0005] A preliminary signal sequence is obtained based on the capacitance signal data acquired through the sensor array;

[0006] Based on the preliminary signal sequence, a time-domain filtering method is used to remove noise interference, and adaptive adjustments are made according to the characteristics of the permafrost alternation zone to obtain the filtered signal sequence.

[0007] If the deviation value in the filtered signal sequence exceeds a preset threshold, the environmental interference type is classified by support vector machine to obtain a classification label set;

[0008] Based on the classification label set, interference subsequences are extracted from the filtered signal sequence, and Kalman filtering is used to fuse multi-point data to correct the deviation and obtain the corrected signal sequence.

[0009] Based on the corrected signal sequence, a carbon content change-related feature vector is calculated, and time series information is integrated to dynamically capture carbon content changes and obtain a feature vector set.

[0010] If an abnormal pattern appears in the feature vector set, the parameters are adjusted according to the pre-established mapping model to determine the carbon storage fluctuation trend and obtain the trend estimate.

[0011] Based on the estimated trend, a soil carbon sequestration monitoring report is generated, and the signal bias correction results are integrated to obtain the final carbon storage change index.

[0012] Optionally, obtaining the preliminary signal sequence based on the capacitance signal data acquired by the sensor array includes:

[0013] The original signal sequence is obtained by collecting multi-point capacitance signal data in the target area using a sensor array.

[0014] Based on the original signal sequence, signal preprocessing techniques are used to perform denoising operations to obtain a smoothed signal dataset;

[0015] If there are outliers in the smoothed signal dataset, the data points exceeding the threshold are filtered out to obtain the cleaned signal dataset.

[0016] Based on the cleaned signal dataset, the correlation characteristics between the signals at each location and the moisture difference are obtained, and the support vector machine algorithm is used to classify the signal data and determine the main distribution areas of the moisture difference.

[0017] Based on the distribution area of ​​the moisture difference, the influence characteristics of the mineral difference on the capacitance signal are extracted to obtain the distribution pattern of the mineral difference.

[0018] Based on the distribution areas of the moisture differences and the distribution patterns of the mineral differences, and combined with the topographic data of the target area, a spatial mapping relationship between moisture differences and mineral differences is constructed to obtain a comprehensive distribution map of environmental impacts.

[0019] Based on the comprehensive distribution map, signal correction parameters are generated according to the environmental impact characteristics of different regions, and the corrected signal sequence is obtained as the preliminary signal sequence.

[0020] Optionally, based on the preliminary signal sequence, a time-domain filtering method is used to remove noise interference, and adaptive adjustments are made for the characteristics of the permafrost transition zone to obtain a filtered signal sequence, including:

[0021] Based on the preliminary signal sequence, a time-domain filtering method is used for preliminary processing. Noise interference is removed by point-by-point analysis to obtain preliminary clean signal data.

[0022] Based on the preliminary purified signal data, environmental information of the permafrost alternation area is obtained, and the filtering parameters are dynamically adjusted in combination with environmental adaptability characteristics to obtain a filtering configuration suitable for the current area.

[0023] Based on the filtering configuration, an adaptive technique is applied to perform secondary processing on the signal sequence to obtain a filtered signal that conforms to the characteristics of permafrost.

[0024] Based on the time-domain characteristics of the filtered signal, it is determined whether the signal processing meets the preset purity standard. If the standard is not met, the parameters are fine-tuned to obtain the optimized signal result.

[0025] Based on the optimized signal results, signal processing techniques are used to perform smoothing operations to obtain the final filtered signal sequence;

[0026] Based on the final filtered signal sequence, it is compared with the characteristics of the permafrost alternation zone to determine whether the data meets the technical optimization requirements, and the final usable filtered signal sequence is obtained.

[0027] Optionally, if the deviation value in the filtered signal sequence exceeds a preset threshold, a classification label set is obtained by classifying the environmental interference type using a support vector machine, including:

[0028] Based on the filtered signal sequence, signal segments whose deviation values ​​exceed a preset threshold are detected and marked as abnormal signal segments.

[0029] Based on the abnormal signal segments, a support vector machine is used to classify the types of environmental interference, determine whether the source of interference is water or minerals, and obtain a set of classification labels.

[0030] Based on the set of classification labels, extract abnormal signal segments related to water sources, analyze their temporal distribution characteristics, and determine the main time periods of water interference.

[0031] Based on the set of classification labels, abnormal signal segments related to mineral sources are extracted, frequency domain transformation is performed to obtain frequency distribution characteristics, and the frequency range of mineral interference is determined.

[0032] If the main time period of moisture interference overlaps with the frequency range of mineral interference, the signal data in the overlapping area is subjected to secondary filtering to obtain the separated signal segments.

[0033] Based on the separated signal segments and the classification label set, the deviation value is recalculated to determine whether it still exceeds the preset threshold, and the final interference classification result is obtained.

[0034] Based on the final interference classification results, signal adjustment strategies for water and mineral sources are generated to determine the optimization direction for subsequent signal acquisition.

[0035] Optionally, based on the classification label set, interference subsequences are extracted from the filtered signal sequence, and Kalman filtering is used to fuse multi-point data to correct deviations, thereby obtaining a corrected signal sequence, including:

[0036] Based on the classification label set, interference subsequences are extracted from the filtered signal sequence to obtain the distribution range of the interference subsequences;

[0037] Based on the distribution range of the interference subsequences, and by comparing and analyzing data from multiple points, the key deviation points in the interference subsequences are determined.

[0038] Based on the key deviation points, a Kalman filter method is used for data fusion processing to obtain preliminary corrected signal data;

[0039] Based on the pre-corrected signal data, smoothing is performed according to the continuity of the signal sequence to obtain a smoothed signal set;

[0040] If there are abnormal fluctuations in the smoothed signal set, a second correction is performed by comparing the labeling information of the classification labels to determine the stability of the corrected signal.

[0041] Based on the stability of the signal after the second correction, and in conjunction with the technical rules of signal processing, the final corrected signal sequence is obtained;

[0042] If there are still local deviations in the final corrected signal sequence, fine-tuning is performed by supplementing and fusing multi-point data to obtain the final stable corrected signal sequence.

[0043] Optionally, based on the corrected signal sequence, a carbon content change-related feature vector is calculated, and time series information is integrated to dynamically capture carbon content changes, obtaining a feature vector set, including:

[0044] Based on the correction signal sequence, information related to carbon content changes is extracted. If the change exceeds a preset threshold, it is recorded as a key change point, and a set of key points of carbon content change is obtained.

[0045] Based on the set of key points of carbon content change, correlation analysis is performed using time series data, and the trend data of change within each time period is obtained through time window division method;

[0046] Based on the changing trend data, feature information related to dynamic capture is constructed, and principal component analysis is used to reduce the dimensionality of the multidimensional data to obtain a simplified feature matrix.

[0047] Based on the simplified feature matrix, corresponding feature vectors are generated, and feature information from different time periods is uniformly summarized using data integration technology to obtain a set of feature vectors.

[0048] Based on the set of feature vectors, continuous tracking is performed to meet the needs of change monitoring. If a vector in the set of feature vectors deviates from a preset range, an anomaly marker is triggered, and dynamic monitoring results are obtained.

[0049] Optionally, if an abnormal pattern appears in the feature vector set, the parameters are adjusted according to a pre-established mapping model to determine the carbon storage fluctuation trend and obtain a trend estimate, including:

[0050] Based on the feature vector set, the vectors are initially screened using preset rules to obtain a candidate set of abnormal patterns;

[0051] If there are significant abnormal patterns in the candidate set, the abnormal patterns are classified using a pre-built mapping model to determine the specific category of the abnormal patterns.

[0052] Based on the specific category of the abnormal mode, adjust the relevant parameters in the mapping model to obtain the adjusted parameter configuration;

[0053] Based on the adjusted parameter configuration, analyze the fluctuation trend of carbon storage values ​​and determine the stability of the fluctuation trend;

[0054] If the stability of the fluctuation trend is lower than the preset threshold, time series analysis is performed on the carbon storage value to obtain the corrected trend estimate.

[0055] Based on the comparison between the correction results and historical data, the final trend estimate is determined, and reference data for subsequent processing is output.

[0056] If there is a significant deviation in the reference data, a second verification is performed using real-time data from the reserves monitoring to obtain the final trend estimate.

[0057] Optionally, based on the trend estimate, a soil carbon sequestration monitoring report is generated, and the signal bias correction results are integrated to obtain the final carbon storage change index, including:

[0058] Based on the trend estimate, time series decomposition is performed to separate the long-term trend and short-term fluctuation components, and a preliminary trend estimate is obtained.

[0059] Based on the trend estimation results, short-term fluctuation components are extracted, and a signal deviation detection method is used to identify potential deviation intervals and determine data points within the range of deviation influence.

[0060] Based on the data points within the range of the deviation's influence, deviation correction technology is applied, and the signal deviation is adjusted using a preset correction model to obtain the corrected fluctuation data;

[0061] Based on the corrected fluctuation data and long-term trends, the data are recombined and reconstructed using time series methods to generate complete soil carbon sequestration trend data.

[0062] Based on the soil carbon sequestration trend data and combined with the historical records of carbon sequestration monitoring, the dynamic value of carbon storage change is calculated, and the intermediate results of carbon storage change are derived.

[0063] If the intermediate result of the change in carbon reserves exceeds the preset threshold range, the change value is verified a second time. By comparing with historical monitoring and analysis data, it is determined whether the final reserve index is reasonable.

[0064] Based on the carbon storage index after secondary verification, the data from each stage of the analysis process are integrated to generate the final carbon storage change index and determine the comprehensive assessment result of soil carbon sequestration.

[0065] On the other hand, the present invention also provides an electronic device including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0066] On the other hand, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0067] Compared with the prior art, the present invention has the following advantages and technical effects:

[0068] This invention discloses a soil carbon sequestration monitoring method based on multi-point capacitance signal acquisition and processing. Addressing the signal interference problem caused by differences in moisture and mineral content in soil carbon storage monitoring in the permafrost transition zone of the Qinghai-Tibet Plateau, this method constructs a multi-stage signal processing and feature analysis workflow to achieve accurate estimation of carbon storage change trends. First, the invention acquires raw capacitance signal data through a sensor array. Time-domain filtering is used to remove noise and adaptively adjust the signal. Then, a support vector machine is used to classify interference types, and Kalman filtering is combined to correct biases, generating a corrected signal sequence. Based on this, carbon content change feature vectors are extracted, time-series information is dynamically captured, and parameters are adjusted through a mapping model to determine fluctuation trends. Finally, a soil carbon sequestration monitoring report including bias correction results is generated. The core innovation of this invention lies in the multi-point data fusion and interference classification and correction mechanism, ensuring signal accuracy in complex environments, significantly improving the reliability and practicality of carbon storage monitoring, and providing important technical support for ecological protection in the plateau permafrost region. Attached Figure Description

[0069] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0070] Figure 1 This is a schematic diagram of the method flow according to an embodiment of the present invention. Detailed Implementation

[0071] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0072] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0073] Example 1

[0074] like Figure 1 As shown, this embodiment provides a method for monitoring soil carbon sequestration on the Qinghai-Tibet Plateau based on a capacitive sensor array, including:

[0075] A preliminary signal sequence is obtained based on the capacitance signal data acquired through the sensor array;

[0076] Based on the preliminary signal sequence, a time-domain filtering method is used to remove noise interference, and adaptive adjustments are made according to the characteristics of the permafrost alternation zone to obtain the filtered signal sequence.

[0077] If the deviation value in the filtered signal sequence exceeds a preset threshold, the environmental interference type is classified by support vector machine to obtain a classification label set;

[0078] Based on the classification label set, interference subsequences are extracted from the filtered signal sequence, and Kalman filtering is used to fuse multi-point data to correct the deviation and obtain the corrected signal sequence.

[0079] Based on the corrected signal sequence, a carbon content change-related feature vector is calculated, and time series information is integrated to dynamically capture carbon content changes and obtain a feature vector set.

[0080] If an abnormal pattern appears in the feature vector set, the parameters are adjusted according to the pre-established mapping model to determine the carbon storage fluctuation trend and obtain the trend estimate.

[0081] Based on the estimated trend, a soil carbon sequestration monitoring report is generated, and the signal bias correction results are integrated to obtain the final carbon storage change index.

[0082] Step S101: Collect multi-point capacitance signal data from the target area of ​​the Qinghai-Tibet Plateau by deploying a sensor array to obtain a raw dataset containing the effects of moisture differences and mineral differences, and obtain a preliminary signal sequence.

[0083] Data acquisition of capacitive signals at multiple points was completed within the target area of ​​the Qinghai-Tibet Plateau using a sensor array, constructing raw signal sequences containing differences in moisture and mineral content. Based on these raw signal sequences, signal preprocessing techniques were employed to denoise the acquired capacitive signals, resulting in a smoothed signal dataset. If outliers were found in the smoothed dataset, data points exceeding a preset threshold were filtered out, resulting in a cleaned signal dataset. For the cleaned dataset, the correlation characteristics between the signals at each point and moisture differences were obtained. A support vector machine algorithm was used to classify the signal data and determine the main distribution areas of moisture differences. Based on the classification results, the influence characteristics of mineral differences on capacitive signals were extracted, yielding the distribution patterns of mineral differences. Using these distribution patterns and combined with topographic data of the target area, a spatial mapping relationship between moisture and mineral differences was constructed, determining a comprehensive distribution map of environmental impacts. After obtaining the comprehensive distribution map, corresponding signal correction parameters were generated for the environmental impact characteristics of different areas, resulting in a corrected signal sequence.

[0084] Specifically, when deploying a sensor array in the target area of ​​the Qinghai-Tibet Plateau to collect multi-point capacitance signal data, an automated system can evenly distribute 100 sensor nodes within the target area, covering an area of ​​approximately 500 square kilometers. The distance between each node is approximately 2.5 kilometers, ensuring comprehensive data collection. The sensors employ high-precision capacitance measurement equipment with a measurement range of 0.1 to 100 picofarads. The sampling frequency is set to once per minute, and continuous collection lasts for 72 hours, generating an initial dataset of approximately 432,000 data points. Next, when acquiring the raw dataset containing differences in moisture content and the influence of minerals, the system automatically preprocesses the collected capacitance signals, using a median filtering algorithm to remove noise. The window size is set to 5. After filtering, the mean and standard deviation of the signal for each node are calculated. For example, if a node has a 72-hour mean of 25.3 picofarads and a standard deviation of 1.8, it can be preliminarily determined that its moisture content is high or its mineral distribution is abnormal. Subsequently, regarding the generation of the initial signal sequence, the system uses a time-series analysis algorithm to segment the data, with a 24-hour cycle, extracting the signal peak and trough values ​​within each cycle. For example, in a certain cycle, the peak value is 28.6 picofarads and the trough value is 22.1 picofarads. Combined with Fourier transform analysis of the signal frequency components, the dominant frequency is found to be 0.0003 Hz, reflecting the periodic characteristics of environmental changes. To establish a rigorous logical relationship, the system further performs correlation analysis between the signal sequence and meteorological data (such as rainfall). Assuming that on a certain day, when the rainfall is 10 mm, the mean capacitance signal rises to 26.5 picofarads, it is inferred that the influence weight of increased moisture on the signal is approximately 0.7, and the influence weight of minerals is approximately 0.3. The hypothesis is verified through a regression analysis model, with a correlation coefficient of 0.85, indicating a high correlation between signal changes and environmental factors.

[0085] Step S102: Based on the preliminary signal sequence, a time-domain filtering method is used to remove noise interference, and adaptive adjustments are made according to the characteristics of the alternating frozen soil zone to determine the filtered signal sequence.

[0086] For the initial signal sequence, a time-domain filtering method is used for preliminary data processing. Point-by-point analysis removes significant noise interference, resulting in preliminarily clean signal data. Based on this preliminarily clean signal data, environmental information of the permafrost transition zone is obtained. The filtering parameters are dynamically adjusted according to environmental adaptability characteristics to determine a suitable filtering configuration for the current region. For the adjusted filtering configuration, adaptive techniques are applied to perform secondary processing on the signal sequence to obtain a filtered signal that better matches the characteristics of permafrost. By analyzing the time-domain characteristics of the filtered signal, it is determined whether the signal processing meets the preset purity standard. If it does not meet the preset standard, the parameters are fine-tuned to obtain an optimized signal result. Based on the optimized signal result, signal processing techniques are used to smooth the data, obtaining the final filtered signal sequence. The final filtered signal sequence is compared with the permafrost characteristics of the transition zone to determine whether the data meets the requirements of technical optimization, thus determining the final usable signal output.

[0087] Specifically, the time-domain filtering and adaptive adjustment process for signal sequences from permafrost alternation zones can be implemented using the following methods. First, assume we have acquired a temperature signal sequence from a permafrost alternation zone with a sampling frequency of 1Hz and a data length of 1000 points. The original signal contains noise interference, with signal values ​​ranging from -5.0 to 5.0. Noise removal can be achieved through time-domain filtering using a moving average filtering algorithm. The window size is set to 5 points, and the calculation formula is: y[n] = (x[n-2] + x[n-1] + x[n] + x[n+1] + x[n+2]) / 5, where x is the original signal and y is the filtered signal. For boundary points, a mirror-fill method is used to complete the data. After filtering, abrupt noise in the signal is smoothed out. For example, if a point in the original signal has a value of 3.2, and surrounding points have values ​​of 2.8, 3.0, 3.1, and 2.9, the filtered value of that point is 3.0, significantly reducing the noise impact. Next, adaptive adjustments are made to suit the characteristics of the permafrost transition zone. Since temperature changes in permafrost regions exhibit periodicity and regional differences, the filtering parameters need to be dynamically adjusted based on the local variance of the signal. The variance of the signal is calculated for every 100 points. If the variance is greater than 1.5, the signal segment is considered to be changing drastically, and the window size is adaptively reduced to 3 points to preserve details; if the variance is less than 0.5, the window size is expanded to 7 points to enhance the smoothing effect. For example, if the variance of a signal segment is 2.0, adjusting the window size to 3 will result in a filtered signal that more closely resembles the original trend. Finally, the filtered signal sequence is determined, and the denoising effect is evaluated by comparing the mean square error (MSE) of the signals before and after filtering. Assuming an MSE of 0.3, it indicates that the filtered signal retains the original characteristics well while removing noise interference.

[0088] Step S103: If the deviation value in the filtered signal sequence exceeds the preset threshold, the environmental interference type is classified by support vector machine to determine the source of interference, such as water or minerals, and a classification label set is obtained.

[0089] The acquired raw signal sequence is filtered to obtain smoothed signal data. If the deviation value in the filtered signal sequence exceeds a preset threshold, this portion of the signal data is marked as an abnormal signal segment. For the marked abnormal signal segments, a support vector machine is used to classify the type of environmental interference, determining whether the interference is from water or mineral sources, thus obtaining a set of classification labels. Based on the classification label set, abnormal signal segments related to water sources are extracted, and their temporal distribution characteristics are analyzed to determine the main time period of water interference. By performing frequency domain transformation on abnormal signal segments related to mineral sources, their frequency distribution characteristics are obtained to determine the frequency band range of mineral interference. If the main time period of water interference overlaps with the frequency band range of mineral interference, the signal data in the overlapping area is filtered a second time to obtain separated signal segments. Based on the separated signal segments and the classification label set, the deviation value is recalculated to determine whether it still exceeds the preset threshold, obtaining the final interference classification result. Based on the final interference classification result, signal adjustment strategies for water and mineral sources are generated to determine the optimization direction for subsequent signal acquisition.

[0090] Specifically, in the signal processing and environmental interference classification process, the first step is to detect the deviation value of the filtered signal sequence. Assuming the signal sequence has 1000 sampling points, the Kalman filter algorithm is used to smooth noise. The difference between the deviation value and the mean of each sampling point is calculated, with a preset threshold of 0.5. If the deviation value of a point reaches 0.7, it is marked as an outlier. The proportion of outliers is statistically set at 5%. When the proportion exceeds a predetermined 3%, the subsequent classification process is triggered. Next, the support vector machine algorithm is used to classify the abnormal signal into environmental interference types. The feature vectors of the signal are extracted, including frequency components (dominant frequency of 50Hz), amplitude variation rate (average of 0.3), and temporal volatility (standard deviation of 0.2). A training dataset is constructed containing 500 samples of two classes: moisture interference (high feature values) and mineral interference (low feature values). Radial basis function kernels are used for classification. The training model accuracy is 85%. The test set prediction results show that the current signal characteristics are closer to moisture interference, and the classification label is "moisture". Finally, an interference source report is generated based on the classification label set. Combined with historical data analysis, moisture interference is more common in environments with humidity greater than 80%, and it is inferred that the current interference source is the influence of a high humidity environment. The system automatically stores the results in the database and updates the interference distribution map.

[0091] Step S104: Extract interference subsequences from the filtered signal sequence based on the classification label set, and use Kalman filtering to fuse multi-point data to correct the deviation, thereby obtaining the corrected signal sequence.

[0092] Interference subsequences are extracted from the filtered signal and initially divided using classification labels to obtain their distribution range. Based on this range, multi-point data is compared and analyzed to identify key deviation points. Kalman filtering is then used to fuse the data at these key deviation points, yielding preliminarily corrected signal data. This preliminarily corrected signal data is then smoothed based on the signal sequence's continuity, resulting in a smoothed signal set. If abnormal fluctuations are found in the smoothed signal set, secondary correction is performed by comparing the classification label information to assess signal stability. Based on the stability of the secondary corrected signal and signal processing rules, the final corrected signal sequence is obtained. If local deviations still exist in the final corrected signal sequence, fine-tuning is performed through supplementary multi-point data fusion to determine the final stable signal output.

[0093] Step S105: Calculate the carbon content change-related feature vector by correcting the signal sequence, integrate time series information to determine the feature vector set for dynamic capture requirements.

[0094] Initial data is acquired from the original signal sequence. The acquired signal is preprocessed by filtering to remove noise interference, resulting in purified signal data. For the purified signal data, correction processing techniques are used to adjust for deviations. The corrected signal sequence is determined by comparing it with a preset benchmark value. Information related to carbon content changes is extracted from the corrected signal sequence. If the extracted change amplitude exceeds a preset threshold, it is recorded as a key change point, resulting in a set of key points for carbon content changes. Based on this set of key points, correlation analysis is performed using time series data. The change trend data for each time period is obtained through time window segmentation. For the change trend data, feature information related to dynamic capture is constructed. Principal component analysis is used to reduce the dimensionality of the multidimensional data, resulting in a simplified feature matrix. From the simplified feature matrix, corresponding feature vectors are generated. Combined with data integration techniques, the feature information from different time periods is uniformly summarized to determine the final feature vector set. Based on the final feature vector set, continuous tracking is performed to meet change monitoring needs. If any vector in the feature vector set deviates from a preset range, an anomaly marker is triggered, resulting in dynamic monitoring results.

[0095] Step S106: If an abnormal pattern appears in the feature vector set, adjust the parameters according to the pre-established mapping model, determine the carbon storage fluctuation trend, and obtain the trend estimate.

[0096] Feature vectors are extracted from the raw data, and a candidate set of abnormal patterns is obtained by initially screening the vectors using preset rules. If significant abnormal patterns exist in the candidate set, a pre-built mapping model is used to classify the abnormal patterns and determine their specific categories. Based on the classified abnormal patterns, the relevant parameters in the mapping model are adjusted to obtain the adjusted parameter configuration. For the adjusted parameter configuration, the fluctuation trend of carbon reserves is analyzed to determine the stability of the fluctuation trend. If the stability of the fluctuation trend is lower than a preset threshold, time series analysis is performed on the carbon reserves to obtain a corrected trend estimate. By comparing the corrected result with historical data, the final trend estimate is determined, and reference data for subsequent processing is output. If there are significant deviations in the reference data, a secondary verification is performed using real-time data from reserves monitoring to obtain the final trend estimate output.

[0097] Specifically, in the analysis of carbon storage fluctuation trends, the system first extracts abnormal patterns from the feature vector set through a data acquisition system. For example, if the carbon storage data of a certain forest area shows abnormal fluctuations over the past month, with the monitored carbon storage value decreasing from an average of 100 tons / hectare to 90 tons / hectare, exceeding a preset threshold by 10%, the system automatically marks it as an abnormal pattern. Next, based on a pre-established mapping model, the parameters are adjusted using a support vector machine regression (SVR) algorithm. Input features include temperature, rainfall, and vegetation cover. Assuming the current temperature is 25 degrees Celsius, rainfall is 50 millimeters, and vegetation cover is 70%, the model predicts an adjusted baseline carbon storage value of 95 tons / hectare, with an error range controlled within ±2 tons. By comparing with historical data, the analysis concludes that rising temperature is likely the main influencing factor. Then, based on the adjusted parameters and combined with the ARIMA time series analysis algorithm, the system judges the carbon storage fluctuation trend for the next three months. The prediction results show that carbon storage may continue to decline to 88 tons / hectare, with a downward trend slope of -2.5 tons / month. The analysis process considers the continued impact of seasonal fluctuations and anomalous patterns. Finally, the system generates a trend estimate. Combining the above prediction data, the system arrives at an estimated average carbon storage of 90.5 tons / hectare for the next three months and generates a visualization report showing the fluctuation curve and the weights of key influencing factors, with temperature weighted at 0.6, rainfall at 0.3, and vegetation cover at 0.1.

[0098] Step S107: Generate a soil carbon sequestration monitoring report based on the trend estimate, integrate the signal deviation correction results, and determine the final carbon storage change index.

[0099] Step 1: Obtain raw observation data of soil carbon sequestration. Perform time series decomposition processing on the trend estimation module to separate the long-term trend and short-term fluctuation components, obtaining preliminary trend estimation results. Step 2: Extract the short-term fluctuation components from the trend estimation results. Use a signal deviation detection method to identify potential deviation intervals and determine data points within the deviation's influence range. Step 3: For data points within the deviation's influence range, apply deviation correction technology. Adjust the signal deviation using a preset correction model to obtain corrected fluctuation data. Step 4: Recombine the corrected fluctuation data with the long-term trend. Use a time series reconstruction method to generate complete soil carbon sequestration trend data. Step 5: Based on the generated soil carbon sequestration trend data and historical carbon sequestration monitoring data, calculate the dynamic value of carbon storage change and derive intermediate results of carbon storage change. Step 6: If the intermediate results of carbon storage change exceed a preset threshold range, perform a secondary verification of the change value. By comparing with historical monitoring and analysis data, determine whether the final storage index is reasonable. Step 7: Based on the storage index after secondary verification, integrate the data from each stage of the analysis process to generate the final carbon storage change index and determine the comprehensive assessment result of soil carbon sequestration.

[0100] Specifically, in generating the soil carbon sequestration monitoring report, the change in soil carbon storage is first assessed using trend estimates. Assuming we obtain carbon content data for the past five years from soil sampling data of a certain area, with annual averages of 10.5, 11.2, 11.8, 12.3, and 12.9 tons / hectare respectively, we calculate the trend slope using a linear regression algorithm, yielding an annual increase of 0.6 tons / hectare in carbon storage. Combined with standard error analysis, the error range is ±0.1 tons / hectare at a 95% confidence interval, indicating high reliability of the trend. Next, signal bias correction results are integrated to improve data accuracy. Assuming the original data is affected by instrument measurement bias, the data is adjusted using a correction model. Specifically, a calibration coefficient of 1.05 is used to multiply and correct the original data; for example, the corrected value of 12.9 tons / hectare becomes 13.545 tons / hectare. Simultaneously, by comparing with historical calibration datasets, the standard deviation after bias correction is calculated to be 0.08, demonstrating that the correction effectively reduces data uncertainty. Subsequently, the final carbon storage change index was determined. The trend estimate was combined with the corrected data, and a weighted average algorithm was used, with the trend estimate having a weight of 0.6 and the corrected data having a weight of 0.4. The final annual average carbon storage change rate was calculated to be 0.62 tons / hectare. Uncertainty analysis was performed using the Monte Carlo simulation method. After 10,000 simulations, the confidence interval for the change rate was found to be 0.58 to 0.66 tons / hectare.

[0101] On the other hand, this embodiment also provides an electronic device, including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0102] On the other hand, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0103] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for monitoring soil carbon sequestration on the Qinghai-Tibet Plateau based on a capacitive sensor array, characterized in that, include: A preliminary signal sequence is obtained based on the capacitance signal data acquired through the sensor array; Based on the preliminary signal sequence, a time-domain filtering method is used to remove noise interference, and adaptive adjustments are made according to the characteristics of the permafrost alternation zone to obtain the filtered signal sequence. If the deviation value in the filtered signal sequence exceeds a preset threshold, the environmental interference type is classified by support vector machine to obtain a classification label set; Based on the classification label set, interference subsequences are extracted from the filtered signal sequence, and Kalman filtering is used to fuse multi-point data to correct the deviation and obtain the corrected signal sequence. Based on the corrected signal sequence, a carbon content change-related feature vector is calculated, and time series information is integrated to dynamically capture carbon content changes and obtain a feature vector set. If an abnormal pattern appears in the feature vector set, the parameters are adjusted according to the pre-established mapping model to determine the carbon storage fluctuation trend and obtain the trend estimate. Based on the estimated trend, a soil carbon sequestration monitoring report is generated, and the signal bias correction results are integrated to obtain the final carbon storage change index.

2. The method according to claim 1, characterized in that, The step of obtaining a preliminary signal sequence based on capacitance signal data acquired through the sensor array includes: The original signal sequence is obtained by collecting multi-point capacitance signal data in the target area using a sensor array. Based on the original signal sequence, signal preprocessing techniques are used to perform denoising operations to obtain a smoothed signal dataset; If there are outliers in the smoothed signal dataset, the data points exceeding the threshold are filtered out to obtain the cleaned signal dataset. Based on the cleaned signal dataset, the correlation characteristics between the signals at each location and the moisture difference are obtained, and the support vector machine algorithm is used to classify the signal data and determine the main distribution areas of the moisture difference. Based on the distribution area of ​​the moisture difference, the influence characteristics of the mineral difference on the capacitance signal are extracted to obtain the distribution pattern of the mineral difference. Based on the distribution areas of the moisture differences and the distribution patterns of the mineral differences, and combined with the topographic data of the target area, a spatial mapping relationship between moisture differences and mineral differences is constructed to obtain a comprehensive distribution map of environmental impacts. Based on the comprehensive distribution map, signal correction parameters are generated according to the environmental impact characteristics of different regions, and the corrected signal sequence is obtained as the preliminary signal sequence.

3. The method according to claim 1, characterized in that, Based on the preliminary signal sequence, a time-domain filtering method is used to remove noise interference, and adaptive adjustments are made according to the characteristics of the permafrost transition zone to obtain the filtered signal sequence, including: Based on the preliminary signal sequence, a time-domain filtering method is used for preliminary processing. Noise interference is removed by point-by-point analysis to obtain preliminary clean signal data. Based on the preliminary purified signal data, environmental information of the permafrost alternation area is obtained, and the filtering parameters are dynamically adjusted in combination with environmental adaptability characteristics to obtain a filtering configuration suitable for the current area. Based on the filtering configuration, an adaptive technique is applied to perform secondary processing on the signal sequence to obtain a filtered signal that conforms to the characteristics of permafrost. Based on the time-domain characteristics of the filtered signal, it is determined whether the signal processing meets the preset purity standard. If the standard is not met, the parameters are fine-tuned to obtain the optimized signal result. Based on the optimized signal results, signal processing techniques are used to perform smoothing operations to obtain the final filtered signal sequence; Based on the final filtered signal sequence, it is compared with the characteristics of the permafrost alternation zone to determine whether the data meets the technical optimization requirements, and the final usable filtered signal sequence is obtained.

4. The method according to claim 1, characterized in that, If the deviation value in the filtered signal sequence exceeds a preset threshold, then the environmental interference type is classified using a support vector machine to obtain a classification label set, including: Based on the filtered signal sequence, signal segments whose deviation values ​​exceed a preset threshold are detected and marked as abnormal signal segments. Based on the abnormal signal segments, a support vector machine is used to classify the types of environmental interference, determine whether the source of interference is water or minerals, and obtain a set of classification labels. Based on the set of classification labels, extract abnormal signal segments related to water sources, analyze their temporal distribution characteristics, and determine the main time periods of water interference. Based on the set of classification labels, abnormal signal segments related to mineral sources are extracted, frequency domain transformation is performed to obtain frequency distribution characteristics, and the frequency range of mineral interference is determined. If the main time period of moisture interference overlaps with the frequency range of mineral interference, the signal data in the overlapping area is subjected to secondary filtering to obtain the separated signal segments. Based on the separated signal segments and the classification label set, the deviation value is recalculated to determine whether it still exceeds the preset threshold, and the final interference classification result is obtained. Based on the final interference classification results, signal adjustment strategies for water and mineral sources are generated to determine the optimization direction for subsequent signal acquisition.

5. The method according to claim 1, characterized in that, Based on the classification label set, interference subsequences are extracted from the filtered signal sequence, and Kalman filtering is used to fuse multi-point data to correct the deviation, obtaining a corrected signal sequence, including: Based on the classification label set, interference subsequences are extracted from the filtered signal sequence to obtain the distribution range of the interference subsequences; Based on the distribution range of the interference subsequences, and by comparing and analyzing data from multiple points, the key deviation points in the interference subsequences are determined. Based on the key deviation points, a Kalman filter method is used for data fusion processing to obtain preliminary corrected signal data; Based on the pre-corrected signal data, smoothing is performed according to the continuity of the signal sequence to obtain a smoothed signal set; If there are abnormal fluctuations in the smoothed signal set, a second correction is performed by comparing the labeling information of the classification labels to determine the stability of the corrected signal. Based on the stability of the signal after the second correction, and in conjunction with the technical rules of signal processing, the final corrected signal sequence is obtained; If there are still local deviations in the final corrected signal sequence, fine-tuning is performed by supplementing and fusing multi-point data to obtain the final stable corrected signal sequence.

6. The method according to claim 1, characterized in that, Based on the corrected signal sequence, a carbon content change-related feature vector is calculated, and time series information is integrated to dynamically capture carbon content changes, obtaining a feature vector set, including: Based on the correction signal sequence, information related to carbon content changes is extracted. If the change exceeds a preset threshold, it is recorded as a key change point, and a set of key points of carbon content change is obtained. Based on the set of key points of carbon content change, correlation analysis is performed using time series data, and the trend data of change within each time period is obtained through time window division method; Based on the changing trend data, feature information related to dynamic capture is constructed, and principal component analysis is used to reduce the dimensionality of the multidimensional data to obtain a simplified feature matrix. Based on the simplified feature matrix, corresponding feature vectors are generated, and feature information from different time periods is uniformly summarized using data integration technology to obtain a set of feature vectors. Based on the set of feature vectors, continuous tracking is performed to meet the needs of change monitoring. If a vector in the set of feature vectors deviates from a preset range, an anomaly marker is triggered, and dynamic monitoring results are obtained.

7. The method according to claim 1, characterized in that, If an abnormal pattern appears in the feature vector set, the parameters are adjusted according to the pre-established mapping model to determine the carbon storage fluctuation trend and obtain a trend estimate, including: Based on the feature vector set, the vectors are initially screened using preset rules to obtain a candidate set of abnormal patterns; If there are significant abnormal patterns in the candidate set, the abnormal patterns are classified using a pre-built mapping model to determine the specific category of the abnormal patterns. Based on the specific category of the abnormal mode, adjust the relevant parameters in the mapping model to obtain the adjusted parameter configuration; Based on the adjusted parameter configuration, analyze the fluctuation trend of carbon storage values ​​and determine the stability of the fluctuation trend; If the stability of the fluctuation trend is lower than the preset threshold, time series analysis is performed on the carbon storage value to obtain the corrected trend estimate. Based on the comparison between the correction results and historical data, the final trend estimate is determined, and reference data for subsequent processing is output. If there is a significant deviation in the reference data, a second verification is performed using real-time data from the reserves monitoring to obtain the final trend estimate.

8. The method according to claim 1, characterized in that, Based on the estimated trend, a soil carbon sequestration monitoring report is generated, and the signal bias correction results are integrated to obtain the final carbon storage change indicators, including: Based on the trend estimate, time series decomposition is performed to separate the long-term trend and short-term fluctuation components, and a preliminary trend estimate is obtained. Based on the trend estimation results, short-term fluctuation components are extracted, and a signal deviation detection method is used to identify potential deviation intervals and determine data points within the range of deviation influence. Based on the data points within the range of the deviation's influence, deviation correction technology is applied, and the signal deviation is adjusted using a preset correction model to obtain the corrected fluctuation data; Based on the corrected fluctuation data and long-term trends, the data are recombined and reconstructed using time series methods to generate complete soil carbon sequestration trend data. Based on the soil carbon sequestration trend data and combined with the historical records of carbon sequestration monitoring, the dynamic value of carbon storage change is calculated, and the intermediate results of carbon storage change are derived. If the intermediate result of the change in carbon reserves exceeds the preset threshold range, the change value is verified a second time. By comparing with historical monitoring and analysis data, it is determined whether the final reserve index is reasonable. Based on the carbon storage index after secondary verification, the data from each stage of the analysis process are integrated to generate the final carbon storage change index and determine the comprehensive assessment result of soil carbon sequestration.

9. An electronic device comprising a memory, a processor, and a computing program stored in the memory and executable on the processor, characterized in that, When the processor executes the computing program, it implements the method of any one of claims 1-8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-8.

Images