A data analysis-based zanthoxylum bungeanum planting environment anomaly detection method

Abstract

This invention relates to the field of planting environment monitoring technology, specifically to a data analysis-based method for detecting anomalies in Sichuan pepper planting environments. The method includes: obtaining soil root coupling characterization values ​​based on the time span differences between the same-order subsequences in the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence, as well as the time span differences between the same-order subsequences in the Sichuan pepper root cadmium ion concentration fluctuation sequence and the Sichuan pepper root pH fluctuation sequence; obtaining soil root propagation direction characterization values ​​based on the differences in start time and range between the soil cadmium ion concentration fluctuation sequence and the Sichuan pepper root cadmium ion concentration fluctuation sequence; adjusting a fixed empirical lag order based on the soil root coupling characterization values ​​and the soil root propagation direction characterization values ​​to obtain an optimized lag order; and using an ARMA model with the optimized lag order to detect anomalies in the soil cadmium content of the Sichuan pepper planting area. Furthermore, this invention can improve the accuracy of detecting anomalies in Sichuan pepper planting environments.

Description

Technical Field

[0001] This invention relates to the field of planting environment detection technology, specifically to a method for detecting abnormalities in the planting environment of Sichuan pepper based on data analysis. Background Technology

[0002] Cadmium (Cd), a highly toxic and cumulative heavy metal, not only damages the physical and chemical structure of soil and inhibits soil microbial activity, but also accumulates continuously in the roots and fruits of Sichuan pepper through soil-plant migration and bioaccumulation. This interferes with the plant's key physiological metabolic processes, leading to a decrease in yield and deterioration in quality. Therefore, in order to ensure the quality and safety of Sichuan pepper fruits, it is crucial to conduct abnormal detection of cadmium content in the soil where Sichuan pepper is grown.

[0003] Existing methods for detecting abnormal cadmium content in soil used for Sichuan pepper cultivation, based on an autoregressive moving average (ARMA) model, rely on the statistical stationarity of time-series data. They use residual thresholds to determine abnormal fluctuations in cadmium content, and further rely on empirically determined fixed lag orders to model historical data. However, at specific growth stages of Sichuan pepper, the roots secrete large amounts of organic acids, leading to a significant decrease in rhizosphere soil pH. This disrupts the adsorption-desorption balance of cadmium, inducing short-term, non-polluting, physiological fluctuations in soil cadmium ion concentration. These fluctuations, caused by the physiological activities of the Sichuan pepper plant, are not due to exogenous pollution, but rather to the Sichuan pepper plant itself. Fluctuations caused by plant physiological activities are not cadmium pollution caused by real environmental pollution sources. However, ARMA models with fixed lag orders cannot effectively distinguish between these physiological fluctuations and real pollution fluctuations. Consequently, ARMA models with fixed lag orders may overfit when faced with physiological fluctuations and underfit and miss detections when faced with real pollution fluctuations. This seriously weakens the accuracy and reliability of detecting abnormal cadmium content in soil used for Sichuan pepper cultivation. Therefore, how to adaptively adjust the lag order of the ARMA model to improve the accuracy of detecting abnormal cadmium content in soil used for Sichuan pepper cultivation has become an urgent problem to be solved. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides a data analysis-based method for detecting anomalies in the Sichuan pepper growing environment. The specific technical solution employed is as follows:

[0005] One embodiment of the present invention provides a method for detecting anomalies in the Sichuan pepper planting environment based on data analysis, comprising the following steps:

[0006] Obtain the fluctuation sequence of the pepper planting detection area, the fluctuation sequence including the soil cadmium ion concentration fluctuation sequence, the soil pH value fluctuation sequence, the pepper root cadmium ion concentration fluctuation sequence, and the pepper root pH value fluctuation sequence;

[0007] Subsequences are obtained by dividing the corresponding fluctuation sequence based on the extreme values ​​in the fluctuation sequence. Soil root coupling characterization values ​​are obtained based on the time span difference between the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence and the time span difference between the same-order subsequences in the pepper root cadmium ion concentration fluctuation sequence and the pepper root pH fluctuation sequence. Soil root propagation direction characterization values ​​are obtained based on the difference in starting time and range between the soil cadmium ion concentration fluctuation sequence and the pepper root cadmium ion concentration fluctuation sequence. The optimized lag order is obtained by adjusting the fixed empirical lag order based on the soil root coupling characterization value and the soil root propagation direction characterization value.

[0008] Anomaly detection and early warning of soil cadmium content in pepper planting areas is carried out using an ARMA model with optimized lag order.

[0009] Beneficial Effects: This invention first obtains the fluctuation sequences of the pepper planting detection area, including soil cadmium ion concentration fluctuation sequences, soil pH fluctuation sequences, pepper root cadmium ion concentration fluctuation sequences, and pepper root pH fluctuation sequences. Then, based on the extreme values ​​in the fluctuation sequences, subsequences are obtained from the corresponding fluctuation sequences. Based on the time span differences between the soil cadmium ion concentration fluctuation sequences and the soil pH fluctuation sequences, as well as the time span differences between the pepper root cadmium ion concentration fluctuation sequences and the pepper root pH fluctuation sequences, soil root coupling characterization values ​​are obtained. Based on the differences in starting time and range between the soil cadmium ion concentration fluctuation sequences and the pepper root cadmium ion concentration fluctuation sequences, soil root propagation direction characterization values ​​are obtained. Based on the soil root coupling characterization values ​​and the soil root propagation direction characterization values, the fixed empirical lag order is adjusted to obtain an optimized lag order. Finally, based on the ARMA model with the optimized lag order, anomaly detection and early warning of soil cadmium content in the pepper planting detection area are performed. Furthermore, this invention dynamically adjusts the lag order of the model based on the soil root coupling characterization value and the soil root propagation direction characterization value, and makes predictions and identifies anomalies based on the model after the lag order adjustment. This can effectively reduce the interference of data fluctuations caused by short-term physiological activities on the long-term predictive performance of the model, improve the sensitivity and accuracy of identifying real soil cadmium pollution events, and thus improve the accuracy and reliability of abnormal detection and early warning of cadmium ion content in soil for pepper planting. Attached Figure Description

[0010] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart of a data analysis-based method for detecting anomalies in the Sichuan pepper planting environment according to the present invention. Detailed Implementation

[0012] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the protection scope of the embodiments of the present invention.

[0013] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art.

[0014] This embodiment provides a method for detecting anomalies in the Sichuan pepper planting environment based on data analysis, which is described in detail below:

[0015] like Figure 1 As shown, this data analysis-based method for detecting anomalies in the Sichuan pepper growing environment includes the following steps:

[0016] Step S001: Obtain the fluctuation sequence of the pepper planting detection area. The fluctuation sequence includes the soil cadmium ion concentration fluctuation sequence, the soil pH value fluctuation sequence, the pepper root cadmium ion concentration fluctuation sequence, and the pepper root pH value fluctuation sequence.

[0017] When Sichuan pepper plants are in specific growth cycles (such as the budding and flowering stage or the fruit enlargement stage), their roots secrete large amounts of organic acids and other metabolic products, thereby lowering the pH value of the root zone soil. This decrease in soil pH significantly alters the adsorption-desorption balance of cadmium ions in the soil, increasing their bioavailability and mobility. This leads to abnormal fluctuations in the concentration of cadmium ions in the soil solution in the short term. These fluctuations are essentially due to the plant's own physiological activities, not actual external environmental pollution or genuine abnormal fluctuations in soil cadmium ion concentration. However, physiological fluctuations caused by root exudates and pollution fluctuations caused by external pollution inputs may exhibit similar time-series patterns in statistical characteristics, but their physical generation mechanisms and model structure requirements are fundamentally different. Furthermore, ARMA models with fixed lag orders cannot effectively distinguish between these physiological fluctuations and genuine pollution fluctuations, thus... The fixed lag order of the ARMA model may lead to overfitting when faced with physiological fluctuations, and underfitting and missing detections when faced with real pollution fluctuations. This results in low accuracy and reliability of detecting abnormal cadmium content in Sichuan pepper planting soil, or in other words, it will seriously weaken the sensitivity of identifying and warning of real soil cadmium pollution events. In order to ensure high sensitivity to real pollution events while effectively suppressing the interference of physiological fluctuations on anomaly detection and warning, that is, to improve the accuracy and reliability of anomaly detection and warning of cadmium content in Sichuan pepper planting soil, this embodiment will subsequently dynamically adjust the lag order of autoregression and moving average to distinguish and suppress non-polluting cadmium concentration fluctuations caused by root secretion of organic acids, while retaining high sensitivity to true anomalies caused by exogenous pollution, thereby achieving the goal of improving the accuracy and reliability of anomaly detection and warning of cadmium content in Sichuan pepper planting soil.

[0018] First, this embodiment selects any Sichuan pepper planting area that needs to be tested for abnormal soil cadmium content, and divides the Sichuan pepper planting area into sub-regions by grid. Then, the sub-region containing Sichuan pepper plants is selected as the Sichuan pepper planting test area. The grid spacing needs to be set according to the actual situation such as monitoring accuracy and the distance between adjacent Sichuan pepper plants. For example, in this embodiment, the grid spacing can be set to 50 meters, but the grid area is required not to be larger than the area of ​​the Sichuan pepper planting test area. In this embodiment, the soil cadmium content in the Sichuan pepper planting test area will be tested for abnormality. Since the method for abnormal soil cadmium content detection and early warning in each Sichuan pepper planting test area is the same, for ease of understanding and description, this embodiment will describe the abnormal soil cadmium content detection process of any Sichuan pepper planting test area as an example. That is, the Sichuan pepper planting test areas that appear later are the same Sichuan pepper planting test area. In this embodiment, to improve the accuracy and reliability of anomaly detection of soil cadmium content in the pepper planting area, the following steps will first obtain the extraction sequences for the current pepper planting area. These extraction sequences include sequences for soil cadmium ion concentration, soil pH value, pepper root cadmium ion concentration, and pepper root pH value. These extraction sequences form the basis for subsequent dynamic adjustment of the lag order. The specific process for obtaining the extraction sequences for the current pepper planting area is as follows:

[0019] First, based on cadmium ion sensors and pH sensors deployed in the soil and at the root system of the Sichuan pepper plantation testing area, the cadmium ion concentration and pH value in the soil and the root system of the Sichuan pepper plantation testing area are collected. The cadmium ion concentration and pH value in the soil and the root system of the Sichuan pepper plantation testing area collected by the sensors at different collection times are recorded as the cadmium ion concentration of the untreated soil, the pH value of the untreated soil, the cadmium ion concentration of the untreated Sichuan pepper root system, and the pH value of the untreated Sichuan pepper root system at the corresponding collection time. The soil cadmium ion concentration, soil pH value, Sichuan pepper root cadmium ion concentration, and Sichuan pepper root pH value are collected synchronously. In specific applications, the implementer needs to set the interval between adjacent collection times or the data collection frequency according to the actual situation. For example, in this embodiment, the interval between adjacent collection times can be set to 1 hour.

[0020] Next, a preset current reference time period is obtained. This preset current reference time period needs to be set according to actual conditions such as the interval between adjacent collection times and prediction accuracy. For example, in this embodiment, the preset current reference time period is required to consist of the current time and the continuous historical time period before the current time. Furthermore, the length of the preset current reference time period is required to be much greater than the interval between adjacent collection times. For example, in this embodiment, the time period consisting of the current time and the continuous 200 hours before the current time can be selected as the preset current reference time period to capture short-term changes caused by root activity or environmental disturbances, while ensuring that the model training can obtain sufficient time-series samples and taking into account the overall efficiency of data processing. Then, the following data are obtained within the preset current reference time period: cadmium ion concentration of all untreated soils, pH value of all untreated soils, cadmium ion concentration of all untreated pepper roots, and cadmium ion concentration of all untreated pepper roots in the pepper planting detection area. The time series of soil cadmium ion concentrations in the pepper planting test area, collected within the preset current reference time period, is recorded as the soil cadmium ion concentration untreated sequence for the current pepper planting test area. Similarly, the time series of soil pH values ​​in the pepper planting test area, collected within the preset current reference time period, is recorded as the soil pH value untreated sequence for the current pepper planting test area. The time series of soil cadmium ion concentrations in the pepper roots in the pepper planting test area, collected within the preset current reference time period, is recorded as the pepper root cadmium ion concentration untreated sequence for the current pepper planting test area. Finally, the time series of soil pH values ​​in the pepper roots in the pepper planting test area, collected within the preset current reference time period, is recorded as the pepper root pH value untreated sequence for the current pepper planting test area.

[0021] Next, all the obtained sequences to be processed are preprocessed, and the preprocessed sequences are recorded as the sequences to be extracted in the current pepper planting detection area. The sequences to be processed include the soil cadmium ion concentration sequence, the soil pH value sequence, the pepper root cadmium ion concentration sequence, and the pepper root pH value sequence. The sequences to be extracted obtained by preprocessing the sequences to be processed include the soil cadmium ion concentration sequence, the soil pH value sequence, the pepper root cadmium ion concentration sequence, and the pepper root pH value sequence. The data preprocessing in this embodiment includes, but is not limited to, filtering and denoising the sequences to be processed. Filtering and denoising can remove outliers caused by obvious physical interference. In this embodiment, existing denoising methods are selected to denoise the sequences, such as median filtering.

[0022] The cadmium ion sensor in this embodiment is an anodic stripping voltammetry cadmium ion sensor, which is resistant to soil organic matter interference. The pH sensor in this embodiment is a composite glass electrode sensor. In this embodiment, the sensor used to collect the cadmium ion concentration and pH value of the pepper roots in the pepper planting detection area can be fixed to the surface of the newly formed root system of any pepper plant in the pepper planting detection area by a micro-invasive or close-fitting method. In this embodiment, the sensor used to collect the cadmium ion concentration and pH value in the soil of the pepper planting detection area is buried in the soil at a distance of 5 to 10 cm from the sensor used to collect the cadmium ion concentration and pH value of the pepper roots in the pepper planting detection area. In this embodiment, all sensors are required to be at the same depth to eliminate vertical concentration gradient interference.

[0023] In this embodiment, after obtaining the sequence to be extracted from the current pepper planting detection area, the fluctuation or unstable data in the sequence is extracted to obtain the fluctuation sequence of the current pepper planting detection area. The fluctuation sequence also includes the soil cadmium ion concentration fluctuation sequence, soil pH fluctuation sequence, pepper root cadmium ion concentration fluctuation sequence, and pepper root pH fluctuation sequence. The purpose of extracting the fluctuation or unstable data is to analyze the characteristics of soil root coupling and soil root propagation direction. In other words, the extracted fluctuation or unstable data is the basis for subsequent identification of fluctuation attributes. The characteristics of soil root coupling and soil root propagation direction can reflect whether the fluctuation of soil cadmium ion concentration is mainly caused by plant physiological activities or by actual cadmium pollution, which is the key to adjusting the lag order. The specific process of obtaining the fluctuation sequence of the current pepper planting detection area is as follows:

[0024] For any sequence to be extracted: Based on the difference between each data point in the sequence and the mean of the sequence, a stable characterization value is obtained for each data point in the sequence. All data points with stable characterization values ​​less than a preset fluctuation threshold are acquired from the sequence. These data points are then arranged in chronological order of acquisition time and denoised, and the result is recorded as the fluctuation sequence of the sequence to be extracted. Existing denoising methods are used here to eliminate isolated data caused by instantaneous sensor drift or external physical disturbances, retaining data that are relatively dense in time and reflect continuous fluctuations. Data with stable characterization values ​​less than the preset fluctuation threshold are classified as fluctuating or unstable data, while data with stable characterization values ​​not less than the preset fluctuation threshold are classified as stable data. In practical applications, implementers need to set a preset fluctuation judgment threshold based on the actual situation such as the range of stable characterization values, experimental statistics, and planting scenarios. In this embodiment, the preset fluctuation judgment threshold is required to be relatively small. For example, the preset fluctuation judgment threshold can be set to 0.4. The purpose of requiring a relatively small preset fluctuation judgment threshold is to improve the sensitivity to the extraction of unstable data.

[0025] The stable representation value of any data point in any sequence to be extracted is the result of a negative correlation mapping between the absolute difference between that data point and the mean of the sequence to be extracted. Here, a negative exponential function with a base e is used to achieve this negative correlation mapping. The expression for the stable representation value of the b-th data point in the sequence to be extracted is: Where exp() is an exponential function with base e. For the b-th data in the sequence to be extracted, The mean of the sequence to be extracted is the mean of the data in the sequence to be extracted; the smaller the difference between the data in the sequence and the mean of the sequence, the better. The smaller the value, the less significant the fluctuation in the corresponding data, and the more likely it is to be in a stable state, or the greater the probability that the corresponding data is stable. A larger value indicates greater fluctuation in the corresponding data, suggesting a more likely unstable state, or a higher probability that the data is either unstable or fluctuating. The larger the value, the smaller the stable representation value of the corresponding data. Therefore, the smaller the stable representation value of the data, the greater the possibility that the corresponding data is unstable or fluctuating. Conversely, the larger the stable representation value, the greater the possibility that the corresponding data is stable.

[0026] Therefore, this embodiment can obtain the fluctuation sequences of all sequences to be extracted through the above process. Since the sequences to be extracted in this embodiment include the soil cadmium ion concentration sequence, the soil pH value sequence, the Sichuan pepper root cadmium ion concentration sequence, and the Sichuan pepper root pH value sequence, this embodiment can obtain the fluctuation sequences of the soil cadmium ion concentration sequence, the soil pH value sequence, the Sichuan pepper root cadmium ion concentration sequence, and the Sichuan pepper root pH value sequence according to the above fluctuation sequence extraction process. The fluctuation sequences of soil cadmium ion concentration, soil pH value, Sichuan pepper root cadmium ion concentration, and Sichuan pepper root pH value to be extracted are respectively denoted as the current Sichuan pepper planting and detection area's soil cadmium ion concentration fluctuation sequence, soil pH value fluctuation sequence, Sichuan pepper root cadmium ion concentration fluctuation sequence, and Sichuan pepper root pH value fluctuation sequence. That is, the fluctuation sequence includes the soil cadmium ion concentration fluctuation sequence, soil pH value fluctuation sequence, Sichuan pepper root cadmium ion concentration fluctuation sequence, and Sichuan pepper root pH value fluctuation sequence.

[0027] Step S002: Based on the extreme values ​​in the fluctuation sequence, subsequences are obtained by dividing the corresponding fluctuation sequence. Based on the time span difference between the same-order subsequences in the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence, as well as the time span difference between the same-order subsequences in the pepper root cadmium ion concentration fluctuation sequence and the pepper root pH fluctuation sequence, soil root coupling characterization values ​​are obtained. Based on the difference in starting time and range between the soil cadmium ion concentration fluctuation sequence and the pepper root cadmium ion concentration fluctuation sequence, soil root propagation direction characterization values ​​are obtained. Based on the soil root coupling characterization values ​​and the soil root propagation direction characterization values, the fixed empirical lag order is adjusted to obtain the optimized lag order.

[0028] In this embodiment, after obtaining the fluctuation sequence, the fluctuation sequence is analyzed to quantify the coupling correlation characteristics between cadmium ion concentration fluctuation and pH value fluctuation, as well as the migration direction characteristics of cadmium ions between soil and roots, namely, the soil root coupling characterization value and the soil root propagation direction characterization value. Based on the soil root coupling characterization value and the soil root propagation direction characterization value, it is possible to distinguish between the increased cadmium activity caused by plant physiological activities (physiological fluctuation) and the increased cadmium concentration caused by external environmental pollution input (pollution fluctuation), that is, to distinguish between the physiological fluctuation of pepper plants and the actual pollution fluctuation, which is the decision basis for subsequent differential adjustment of model parameters. Soil root coupling characteristics and soil root propagation direction characteristics can reflect the causes of physiological fluctuations and actual pollution fluctuations in Sichuan pepper plants. Physiological fluctuations are usually caused by a local pH decrease due to root exudates; therefore, cadmium ion concentration fluctuations caused by physiological fluctuations should be highly synchronous and strongly correlated with pH fluctuations in time, and the migration direction of cadmium ions is mainly from roots to soil. Pollution fluctuations, on the other hand, are usually caused by external inputs; therefore, the correlation between concentration changes and soil pH changes is weaker. Furthermore, an increase in cadmium ion concentration may first appear in the soil and then migrate to the roots. In other words, when cadmium ion concentration fluctuations are highly synchronous or strongly correlated with pH fluctuations in time, and the migration direction of cadmium ions is mainly from roots to soil, it indicates that the cadmium ion concentration fluctuations are physiological fluctuations caused primarily by plant physiological activities. Conversely, when cadmium ion concentration fluctuations are weakly correlated with pH fluctuations in time, and the migration direction of cadmium ions is mainly from soil to roots, it indicates that the cadmium ion concentration fluctuations are pollution fluctuations caused primarily by actual pollution inputs.

[0029] Therefore, based on the above analysis, this embodiment needs to first obtain soil root coupling characterization values ​​and soil root propagation direction characterization values. Before obtaining soil root coupling characterization values, it is necessary to divide the fluctuation sequence. The purpose of the division is to analyze the temporal correlation characteristics between the pH value fluctuation sequence and the cadmium ion concentration fluctuation sequence. The specific process of dividing the fluctuation sequence to obtain subsequences is as follows: For any fluctuation sequence: First, identify and obtain the extreme values ​​in the fluctuation sequence, and then use the extreme values ​​in the fluctuation sequence to divide the sequence, and divide the results... This is denoted as a subsequence on the fluctuation sequence; the process of identifying extreme values ​​in the sequence is a known technique; for example: if the 3rd and 5th data points in the fluctuation sequence are extreme values, then the number of subsequences on the fluctuation sequence is 3, and the data segment from the 1st to the 3rd data point in the fluctuation sequence is the 1st subsequence on the fluctuation sequence, the data segment from the 3rd to the 5th data point in the fluctuation sequence is the 2nd subsequence on the fluctuation sequence, and the data segment from the 5th data point to the last data point in the fluctuation sequence is the 3rd subsequence on the fluctuation sequence.

[0030] Since the time span difference between subsequences at the same position in the cadmium ion concentration fluctuation sequence and the pH value fluctuation sequence can reflect temporal synchronicity, this embodiment, after dividing the fluctuation sequence, obtains the soil root coupling characterization value based on the time span difference between the same-order subsequences in the soil cadmium ion concentration fluctuation sequence and the soil pH value fluctuation sequence, as well as the time span difference between the same-order subsequences in the Sichuan pepper root cadmium ion concentration fluctuation sequence and the Sichuan pepper root pH value fluctuation sequence. The specific process of obtaining the soil root coupling characterization value based on the time span difference between the same-order subsequences in the soil cadmium ion concentration fluctuation sequence and the soil pH value fluctuation sequence, as well as the time span difference between the same-order subsequences in the Sichuan pepper root cadmium ion concentration fluctuation sequence and the Sichuan pepper root pH value fluctuation sequence, is as follows:

[0031] First, based on the time span difference between the same-order subsequences of the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence, a soil coupling characterization value is obtained. Then, based on the time span difference between the same-order subsequences of the Sichuan pepper root cadmium ion concentration fluctuation sequence and the Sichuan pepper root pH fluctuation sequence, a root coupling characterization value is obtained. Both the soil coupling characterization value and the root coupling characterization value can reflect whether the cadmium ion concentration fluctuation is mainly caused by plant physiological activities or by actual pollution. Finally, the soil coupling characterization value and the root coupling characterization value are fused to obtain the soil root coupling characterization value, which also reflects whether the cadmium ion concentration fluctuation is mainly caused by plant physiological activities or by actual pollution. Furthermore, in this embodiment, the soil root coupling characterization value is the result of multiplying the soil coupling characterization value and the root coupling characterization value. , These are soil coupling characterization values. This is a characterization value for root coupling. and A higher concentration indicates a stronger correlation between cadmium ion concentrations in both the soil and root system and pH value. It also suggests a high degree of temporal synchronicity and strong correlation between fluctuations in cadmium ion concentrations in both the soil and root system and pH value fluctuations. and The larger the value, the greater the soil-root coupling characterization value. Therefore, a larger soil-root coupling characterization value indicates a stronger correlation between cadmium ion concentration and pH value in both the soil and root systems. It also indicates a high degree of synchronicity and strong correlation between cadmium ion concentration fluctuations and pH value fluctuations in both the soil and root systems over time. In this case, the probability of the fluctuations being physiological fluctuations is higher, or the probability that the cadmium ion concentration fluctuations are caused by plant physiological activities is higher. Conversely, a smaller soil-root coupling characterization value indicates a weaker correlation between cadmium ion concentration and pH value in both the soil and root systems. It also indicates a weaker synchronicity and correlation between cadmium ion concentration fluctuations and pH value fluctuations in both the soil and root systems over time. In this case, the probability of the fluctuations being true pollution fluctuations is higher, or the probability that the cadmium ion concentration fluctuations are caused by true cadmium pollution is higher.

[0032] The specific process for obtaining soil coupling characterization values ​​based on the time span difference between the same-order subsequences in the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence is as follows:

[0033] The soil cadmium ion concentration fluctuation sequence is matched with the subsequences of the same order in the soil pH fluctuation sequence to obtain subsequence pairs; min(M1,M2) is the number of subsequence pairs, min() is the minimum value function, M1 is the number of subsequences in the soil cadmium ion concentration fluctuation sequence, M2 is the number of subsequences in the soil pH fluctuation sequence, and the c-th subsequence in the soil cadmium ion concentration fluctuation sequence and the c-th subsequence in the soil pH fluctuation sequence form a subsequence pair, where c is not greater than min(M1,M2); the time span value of each subsequence is obtained, and the time of any subsequence is... The span value is the time interval between the starting and ending data in the subsequence, or the length of time elapsed from the starting to the ending data in the subsequence. The absolute value of the difference between the time span values ​​of the two subsequences in each subsequence pair is calculated and denoted as the time span difference of the corresponding subsequence pair. A negative correlation mapping is performed on the time span difference of the subsequence pair using a negative exponential function with a base of constant e, and the mapping result is denoted as the time span characteristic value of the corresponding subsequence pair. The mean of the time span characteristic values ​​of all subsequence pairs is calculated and denoted as the soil coupling characterization value. The expression for the soil coupling characterization value is:

[0034]

[0035] in, These are soil coupling characterization values. The number of subsequence pairs is represented by exp(), which is an exponential function with base e. Let be the time span value of the first subsequence in the j-th subsequence pair. This represents the time span value of the second subsequence in the j-th subsequence pair; The temporal synchronicity or coupling correlation between the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence was quantified; The smaller, The larger the value, the more synchronous the changes in the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence are in time, and the more tightly coupled the fluctuation patterns are. In other words, the stronger the coupling correlation between the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence, or the stronger the coupling correlation between cadmium ion concentration and acidity / alkalinity, the higher the probability that the fluctuation is physiological. The smaller the value, the less synchronous the changes in the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence are in time, and the less tightly coupled the fluctuation patterns are. In other words, the weaker the coupling relationship between the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence is, or the weaker the coupling relationship between cadmium ion concentration and acidity / alkalinity is. The weaker the coupling relationship, the higher the probability that the fluctuation is a true pollution fluctuation.

[0036] Since the process of obtaining root coupling characterization values ​​based on the time span difference of the same order subsequences on the cadmium ion concentration fluctuation sequence and the pH value fluctuation sequence of Sichuan pepper roots is the same as the process of obtaining soil coupling characterization values ​​based on the time span difference of the same order subsequences on the cadmium ion concentration fluctuation sequence and the pH value fluctuation sequence of soil, this embodiment will not describe the process of obtaining root coupling characterization values ​​in detail.

[0037] In this embodiment, after obtaining the soil root coupling characterization value, the soil root propagation direction characterization value is further obtained. This value reflects the cadmium ion migration direction, which in turn reflects whether the cadmium ion concentration fluctuation is primarily caused by plant physiological activities or by actual pollution. Furthermore, the differences in start time and range between the soil cadmium ion concentration fluctuation sequence and the Sichuan pepper root cadmium ion concentration fluctuation sequence reflect whether cadmium ions diffuse from the roots to the soil or whether the cadmium ion concentration increases and first appears in the soil before migrating to the roots. Therefore, the differences in start time and range between the soil cadmium ion concentration fluctuation sequence and the Sichuan pepper root cadmium ion concentration fluctuation sequence characterize the cadmium ion migration direction. Consequently, this embodiment then obtains the soil root propagation direction characterization value based on the differences in start time and range between the soil cadmium ion concentration fluctuation sequence and the Sichuan pepper root cadmium ion concentration fluctuation sequence. The specific process for obtaining the soil root propagation direction characterization value based on the differences in start time and range between the soil cadmium ion concentration fluctuation sequence and the Sichuan pepper root cadmium ion concentration fluctuation sequence is as follows:

[0038] The normalized result of the difference between the starting time of the soil cadmium ion concentration fluctuation sequence and the starting time of the pepper root cadmium ion concentration fluctuation sequence is denoted as the time difference index value. If the starting time of the soil cadmium ion concentration fluctuation sequence is later than the starting time of the pepper root cadmium ion concentration fluctuation sequence, then the time interval between these two sequences, plus a negative sign, is the difference between the starting times of the soil cadmium ion concentration fluctuation sequence and the pepper root cadmium ion concentration fluctuation sequence. The normalized result of the difference between the range of the pepper root cadmium ion concentration fluctuation sequence and the range of the soil cadmium ion concentration fluctuation sequence is denoted as the change index value. Here, the hyperbolic tangent function is used for normalization. The product of the time difference index value and the change index value is denoted as the soil root propagation direction characterization value. The expression for the soil root propagation direction characterization value is:

[0039]

[0040] Where U2 represents the direction of root propagation in the soil, tanh() is the hyperbolic tangent function, and the mean of tanh()+1 is used for normalization. This represents the time corresponding to the initial data in the soil cadmium ion concentration fluctuation sequence. The time corresponding to the initial data in the cadmium ion concentration fluctuation sequence of Sichuan pepper roots refers to the time when the data was collected or the original data was collected. The range of the cadmium ion concentration fluctuation sequence in the roots of Sichuan pepper. The range of soil cadmium ion concentration fluctuation series is the difference between the maximum and minimum values ​​in any series. and The onset time of cadmium ion fluctuations in soil and the onset time of cadmium ion fluctuations in the roots of Sichuan pepper were characterized, respectively. It can characterize whether the onset time of cadmium ion fluctuations in Sichuan pepper roots precedes the onset time of cadmium ion fluctuations in the soil, and the significance of whether the onset time of cadmium ion fluctuations in Sichuan pepper roots precedes or follows soil cadmium ion fluctuations; and When the value is greater than 0 and the larger it is, it not only indicates that the start time of cadmium ion fluctuation in the root system of Sichuan pepper is earlier than the start time of cadmium ion fluctuation in the soil, but also indicates that the start time of cadmium ion fluctuation in the root system of Sichuan pepper is significantly earlier than the start time of cadmium ion fluctuation in the soil. At this time, the fluctuation is more strongly caused by root activity and spreads outward, which is more consistent with the propagation characteristics of physiological fluctuations. In other words, the probability that the cadmium ion fluctuation is caused by the physiological activity of the plant is greater at this time. The smaller the value (less than 0) and the smaller the value (the smaller the value), the more it indicates that the start time of cadmium ion fluctuations in the root system of Sichuan pepper is later than that of cadmium ion fluctuations in the soil. It also indicates that the start time of cadmium ion fluctuations in the soil is significantly earlier than that of cadmium ion fluctuations in the root system of Sichuan pepper. At this time, the fluctuations are more likely to be caused by soil environment input and may migrate to the root system, which is more consistent with the propagation characteristics of pollution fluctuations. In other words, the probability that the cadmium ion fluctuations at this time are caused by real pollution is greater. It reflects the intensity of cadmium ion fluctuations in the root system. Reflecting the severity of fluctuations in cadmium ions in the soil, The difference between the degree of fluctuation in cadmium ion concentration in roots and the degree of fluctuation in cadmium ion concentration in soil was quantified. A higher root density indicates more dramatic fluctuations in cadmium ion concentration compared to soil cadmium ion concentration. This suggests that the fluctuations are more strongly driven by root activity and diffuse outwards, aligning better with the propagation characteristics of physiological fluctuations. In other words, the probability that these cadmium ion fluctuations are caused by plant physiological activities is higher. Conversely, a lower root density indicates more dramatic fluctuations. The smaller the value, the less drastic the fluctuations in cadmium ion concentration in the roots are compared to those in the soil. This indicates that the fluctuations are more likely to be input from the soil environment and migrate to the roots, which is more consistent with the propagation characteristics of pollution fluctuations, or in other words, the cadmium ion fluctuations at this time are more likely to be caused by actual pollution. The greater the sum When U2 is larger, and tends towards 1, a larger U2 indicates that the fluctuation is more strongly caused by root activity and spreads outward, which is more consistent with the propagation characteristics of physiological fluctuations. In other words, the probability that the cadmium ion fluctuation is caused by plant physiological activity is higher. Conversely, a smaller U2, and tends towards 0, indicates that the fluctuation is more strongly caused by soil environmental input and may migrate to the root system, which is more consistent with the propagation characteristics of pollution fluctuations. In other words, the probability that the cadmium ion fluctuation is caused by actual pollution is higher. Furthermore, and When the value equals 0, it indicates that the fluctuations in cadmium ion concentration in the root system and soil are completely synchronized in time and the intensity of the fluctuations is also consistent. This is a fuzzy range of fluctuation attributes. That is, it is impossible to determine whether net cadmium ion migration has occurred, or whether root-dominated outward diffusion and soil-dominated inward migration have occurred. In order to ensure that the model maintains a neutral, stable and reliable baseline response in the fuzzy region, it is determined that U2 is not calculated and there is no need to adjust the fixed empirical lag order.

[0041] Therefore, this embodiment obtains soil root coupling characterization values ​​and soil root propagation direction characterization values ​​through the above process. Since these values ​​reflect the attributes causing fluctuations, and to ensure that the ARMA model can effectively distinguish between physiological fluctuations and true pollution fluctuations, thus avoiding overfitting when dealing with physiological fluctuations and underfitting when dealing with true pollution fluctuations, this embodiment reduces the lag order for physiological cadmium ion concentration fluctuations caused by plant physiological activities to weaken their interference with model fitting and reduce the probability of overfitting when dealing with physiological fluctuations. For pollution-related cadmium ion concentration fluctuations caused by external pollution, the lag order is increased to highlight abnormal characteristics and reduce the probability of underfitting when dealing with true pollution fluctuations. Therefore, this embodiment will next adjust the fixed empirical lag order based on the soil root coupling characterization values ​​and soil root propagation direction characterization values ​​to obtain the optimized lag order. The specific process is as follows:

[0042] First, based on the soil root coupling characterization value and the soil root propagation direction characterization value, a scaling factor is obtained. The scaling factor characterizes the direction and degree of subsequent adjustment to the fixed empirical lag order. The specific process for obtaining the scaling factor is as follows: The product of the soil root coupling characterization value and the soil root propagation direction characterization value is recorded as the comprehensive characterization value. The result of subtracting the comprehensive characterization value from the preset category attribute discrimination threshold and adding an adjustment constant is recorded as the scaling factor. In this embodiment, the adjustment constant is required to be the maximum value within the range of the comprehensive characterization value, i.e., a constant of 1. The integer value of the product of the scaling factor and the fixed empirical lag order is used as the optimized lag order. Here, the round() function is used for rounding. round() is a built-in function for rounding numbers. In this embodiment, the rounded result is required to be an integer. The expression for obtaining the optimized lag order is:

[0043]

[0044] in, To optimize the lag order, `round()` is a built-in function that rounds numbers. As a preset threshold for class attribute discrimination, U1 is the soil root coupling characterization value, and U2 is the soil root propagation direction characterization value. The fixed empirical lag order is a fixed lag order determined based on traditional experience. For example, the current fixed empirical lag order is often empirically set to between 1 and 2. Greater than When this occurs, it indicates a higher probability that the current cadmium ion concentration fluctuation is caused by plant physiological activities, or it can be determined that the current cadmium ion concentration fluctuation is caused by plant physiological activities. Therefore, [the following should be considered]. Turn it down, and Greater than The greater the degree, The smaller the adjustment; Less than This indicates that the current cadmium ion concentration fluctuation is more likely to be caused by actual pollution input, or it can be determined that the current cadmium ion concentration fluctuation is caused by actual pollution. In this case, [the following should be done]. Increase it, and Less than The greater the degree, The greater the increase; equal No adjustments are made at this point to avoid introducing misclassification errors due to subjective intervention at the discrimination boundary, ensuring that the model maintains a neutral, stable, and reliable baseline response in the fuzzy region. In this embodiment, the above formula... This is to control the direction and degree of adjustment, that is Less than 0 and the smaller the value, Adding 1 can Reduce it, and reduce it even more. When it is greater than 0 and the larger it is, Adding 1 can Increase it, and increase it even more, that is to say The smaller, The larger, When it is larger, The smaller.

[0045] Therefore, this embodiment completes the dynamic adjustment of the fixed empirical lag order. Moreover, the adjustment method provided in this embodiment can not only effectively reduce the interference of data fluctuations caused by short-term physiological activities on the long-term predictive performance of the model and improve the sensitivity and accuracy of identifying real soil cadmium pollution events, but also enable the model to better adapt to the data generation process under different types of fluctuations, thereby improving the prediction accuracy and the residual's ability to distinguish real anomalies. In addition, in specific applications, implementers need to set a preset category attribute discrimination threshold according to the actual situation such as the range of comprehensive characterization values. For example, in this embodiment, the preset category attribute discrimination threshold can be set to the median of the range of comprehensive characterization values, which is 0.5. Setting the preset category attribute discrimination threshold to the median of the range of comprehensive characterization values ​​can take into account the discrimination accuracy of physiological fluctuations and pollution fluctuations, ensuring that the model achieves a balanced effect of differentiated adjustment of the two types of fluctuations, and at the same time providing a reasonable benchmark value for the fine-tuning of thresholds under different planting scenarios.

[0046] Step S003: Based on the ARMA model with the order of optimization lag order, anomaly detection and early warning of soil cadmium content in the pepper planting detection area is carried out.

[0047] In this embodiment, after adjusting the fixed empirical lag order to obtain the optimized lag order, the ARMA model with the optimized lag order is trained to obtain the trained ARMA model. Based on this trained ARMA model, the soil cadmium ion concentration in the pepper planting detection area is predicted in the future, and the predicted soil cadmium ion concentration in the pepper planting detection area at the future time is output. The training process of the ARMA model is well known; this embodiment only changes the process of determining the lag order, without changing the process of prediction and anomaly detection and early warning based on the ARMA model. The input for prediction is the soil cadmium ion concentration sequence of the pepper planting detection area collected and denoised within a historical time period, or the soil cadmium ion concentration sequence to be extracted can be input. Then, based on the obtained predicted soil cadmium ion concentration in the pepper planting detection area at the future time, the soil cadmium ion concentration in the pepper planting detection area is calculated. The process of detecting and warning of abnormal cadmium content in the soil of the pepper planting area based on the predicted soil cadmium ion concentration at a future time is well-known. For example, the residuals at different times can be calculated based on the predicted soil cadmium ion concentration and the actual soil cadmium ion concentration. Then, based on the confirmation that the model residuals basically meet the characteristics of white noise, if the residuals exceed the normal fluctuation range set based on historical statistical characteristics, an early warning signal is immediately triggered and the time point is marked as a potential cadmium pollution anomaly, prompting on-site verification and intervention. This achieves accurate detection and early warning of cadmium pollution in the pepper planting environment. That is, if 95% of the residuals at all historical times are in the interval W, but the residual at a certain time is not in the interval W, an early warning signal is immediately triggered and the time point is marked as a potential cadmium pollution anomaly.

[0048] Thus, this embodiment completes the anomaly detection of the Sichuan pepper planting environment. Furthermore, this embodiment's method of dynamically adjusting the model's lag order based on soil root coupling characterization values ​​and soil root propagation direction characterization values, and then using the adjusted model for prediction and anomaly identification, effectively reduces the interference of data fluctuations caused by short-term physiological activities on the model's long-term predictive performance. This improves the sensitivity and accuracy of identifying real soil cadmium pollution events and allows the model to better adapt to data generation processes under different types of fluctuations, thereby enhancing prediction accuracy and the residual's ability to discriminate real anomalies. In other words, this embodiment can improve the accuracy and reliability of anomaly detection and early warning of cadmium content anomalies in Sichuan pepper planting soil. Additionally, in this embodiment, all parameters involved in mathematical operations must ignore their dimensions; that is, all operational parameters in this embodiment involve only numerical values.

[0049] In summary, this embodiment first obtains the fluctuation sequences of the pepper planting detection area, including soil cadmium ion concentration fluctuation sequences, soil pH fluctuation sequences, pepper root cadmium ion concentration fluctuation sequences, and pepper root pH fluctuation sequences. Then, based on the extreme values ​​in the fluctuation sequences, subsequences are obtained from the corresponding fluctuation sequences. Based on the time span differences between the soil cadmium ion concentration fluctuation sequences and the soil pH fluctuation sequences, as well as the time span differences between the pepper root cadmium ion concentration fluctuation sequences and the pepper root pH fluctuation sequences, soil root coupling characterization values ​​are obtained. Based on the differences in start time and range between the soil cadmium ion concentration fluctuation sequences and the pepper root cadmium ion concentration fluctuation sequences, soil root propagation direction characterization values ​​are obtained. Based on the soil root coupling characterization values ​​and the soil root propagation direction characterization values, the fixed empirical lag order is adjusted to obtain an optimized lag order. Finally, based on the ARMA model with the optimized lag order, anomaly detection and early warning of soil cadmium content in the pepper planting detection area are performed. Furthermore, this embodiment dynamically adjusts the lag order of the model based on the soil root coupling characterization value and the soil root propagation direction characterization value, and makes predictions and identifies anomalies based on the model after the lag order adjustment. This can effectively reduce the interference of data fluctuations caused by short-term physiological activities on the long-term predictive performance of the model, improve the sensitivity and accuracy of identifying real soil cadmium pollution events, and thus improve the accuracy and reliability of abnormal detection and early warning of cadmium ion content in soil for pepper planting.

[0050] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for detecting anomalies in the Sichuan pepper planting environment based on data analysis, characterized in that, The method includes the following steps: Obtain the fluctuation sequence of the pepper planting detection area, the fluctuation sequence including the soil cadmium ion concentration fluctuation sequence, the soil pH value fluctuation sequence, the pepper root cadmium ion concentration fluctuation sequence, and the pepper root pH value fluctuation sequence; Subsequences are obtained by dividing the corresponding fluctuation sequence based on the extreme values ​​in the fluctuation sequence. Soil root coupling characterization values ​​are obtained based on the time span difference between the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence and the time span difference between the same-order subsequences in the pepper root cadmium ion concentration fluctuation sequence and the pepper root pH fluctuation sequence. Soil root propagation direction characterization values ​​are obtained based on the difference in starting time and range between the soil cadmium ion concentration fluctuation sequence and the pepper root cadmium ion concentration fluctuation sequence. The optimized lag order is obtained by adjusting the fixed empirical lag order based on the soil root coupling characterization value and the soil root propagation direction characterization value. Anomaly detection and early warning of soil cadmium content in pepper planting areas is carried out based on an ARMA model with optimized lag order. The method for obtaining soil root coupling characterization values ​​includes: obtaining soil coupling characterization values ​​based on the time span difference between the same order subsequences on the soil cadmium ion concentration fluctuation sequence and the soil pH fluctuation sequence; obtaining root coupling characterization values ​​based on the time span difference between the same order subsequences on the Sichuan pepper root cadmium ion concentration fluctuation sequence and the Sichuan pepper root pH fluctuation sequence; the methods for obtaining soil coupling characterization values ​​and root coupling characterization values ​​are the same; and recording the product of the soil coupling characterization value and the root coupling characterization value as the soil root coupling characterization value. The method for obtaining soil coupling characterization values ​​includes: matching the soil cadmium ion concentration fluctuation sequence with the subsequences of the same order in the soil pH fluctuation sequence to obtain subsequence pairs; recording the negative correlation mapping result of the absolute value of the difference between the time span values ​​of the two subsequences in each subsequence pair as the time span feature value of the corresponding subsequence pair; and recording the mean of the time span feature values ​​of all subsequence pairs as the soil coupling characterization value. The method for obtaining the soil root propagation direction characterization value includes: recording the normalized result of the difference between the time corresponding to the initial data in the soil cadmium ion concentration fluctuation sequence and the time corresponding to the initial data in the Sichuan pepper root cadmium ion concentration fluctuation sequence as the time difference index value; recording the normalized result of the difference between the range of the Sichuan pepper root cadmium ion concentration fluctuation sequence and the range of the soil cadmium ion concentration fluctuation sequence as the change index value; and recording the product of the time difference index value and the change index value as the soil root propagation direction characterization value. The method for obtaining the optimized lag order includes: taking the rounded value of the product of the scaling factor and the fixed empirical lag order as the optimized lag order; the scaling factor is the result of subtracting the comprehensive characterization value from the preset category attribute discrimination threshold and adding an adjustment constant; and the comprehensive characterization value is the product of the soil root coupling characterization value and the soil root propagation direction characterization value.

2. The method for detecting abnormalities in the Sichuan pepper planting environment based on data analysis as described in claim 1, characterized in that, Methods for obtaining fluctuation sequences include: The sequence to be extracted from the pepper planting detection area is obtained. Based on the difference between each data in the sequence to be extracted and the mean of the corresponding sequence, the stable characterization value corresponding to each data in the sequence to be extracted is obtained. The data in the sequence to be extracted whose stable characterization value is less than the preset fluctuation judgment threshold are arranged in chronological order and then denoised. The result is recorded as the fluctuation sequence.

3. The method for detecting abnormalities in the Sichuan pepper planting environment based on data analysis as described in claim 2, characterized in that, The stable representation value of any data in any sequence to be extracted is the negative correlation mapping result of the absolute value of the difference between the corresponding data and the mean of the corresponding sequence to be extracted.

4. The method for detecting abnormalities in the Sichuan pepper planting environment based on data analysis as described in claim 2, characterized in that, The sequences to be extracted include soil cadmium ion concentration, soil pH value, pepper root cadmium ion concentration, and pepper root pH value. The sequences to be extracted are obtained by collecting, arranging, and preprocessing data from the pepper planting and testing area, including soil cadmium ion concentration, soil pH value, pepper root cadmium ion concentration, and pepper root pH value.

5. The method for detecting abnormalities in the Sichuan pepper planting environment based on data analysis as described in claim 1, characterized in that, The time span of any subsequence is the length of time elapsed from the start data to the end data in the corresponding subsequence.

Images