A Method and System for Predicting Soil Pollution Remediation Parameters Based on Multi-Source Data

Abstract

This invention relates to the field of soil pollution remediation technology, specifically to a method and system for predicting soil pollution remediation parameters based on multi-source data. It addresses the technical problem that existing technologies using single parameter values ​​lack necessary engineering tolerance, and the prediction results and engineering stability are insufficient to meet practical needs. The method includes: determining a first sensitivity between the target heavy metal pollutant concentration and iron loading, and a second sensitivity between the target heavy metal pollutant concentration and the relative abundance of sulfate-reducing bacteria; determining a core offset based on the first synergistic strength between iron loading and the target heavy metal pollutant concentration, and the second synergistic strength between the relative abundance of sulfate-reducing bacteria and the target heavy metal pollutant concentration; and determining the prediction range for iron loading and the prediction range for sulfate-reducing bacteria dosage based on the corrected first and second sensitivities. This invention is applicable to soil pollution remediation scenarios.

Description

Technical Field

[0001] This invention relates to the field of soil pollution remediation technology, specifically to a method and system for predicting soil pollution remediation parameters based on multi-source data. Background Technology

[0002] In the field of soil pollution remediation, especially for sites contaminated with heavy metals such as antimony and arsenic, a synergistic remediation strategy combining chemical fixation with iron-based materials and microbial remediation using sulfate-reducing bacteria is commonly employed. However, the actual soil environment exhibits high spatial heterogeneity and dynamic fluctuations. The physicochemical properties (such as pH and microbial community structure) and pollutant concentration distribution of soils at different depths and in different regions vary significantly. This necessitates the comprehensive consideration of multi-source heterogeneous data in selecting remediation parameters to achieve precise and effective remediation.

[0003] Currently, most existing methods for predicting remediation parameters rely on static fitting of historical or small-scale experimental data, failing to fully explore the deep correlations between multiple data sources such as soil pH, microbial abundance, heavy metal concentration, and iron loading. Furthermore, they typically only output recommended values ​​for a single remediation parameter. Given the complex and ever-changing environmental disturbances faced by field remediation projects, single parameter values ​​lack the necessary engineering tolerance. If actual conditions deviate from the predicted conditions, it may lead to substandard remediation results or uncontrolled remediation costs, making it difficult for existing methods to meet practical needs in terms of predictive effectiveness and engineering stability. Summary of the Invention

[0004] To address the technical problem that existing methods using single parameter values ​​lack necessary engineering tolerance, and that deviations between actual working conditions and predicted conditions can lead to substandard remediation results or runaway remediation costs, making it difficult to meet practical needs in terms of prediction effectiveness and engineering stability, this invention aims to provide a method and system for predicting soil pollution remediation parameters based on multi-source data. The specific technical solution adopted is as follows: In a first aspect, the present invention provides a method for predicting soil pollution remediation parameters based on multi-source data. The method includes: acquiring time-series detection data of multiple samples of soil to be remediated; the time-series detection data includes the relative abundance of sulfate-reducing bacteria, the concentration of a target heavy metal pollutant, and iron loading; the target heavy metal pollutant is antimony or arsenic; determining a first sensitivity between the target heavy metal pollutant concentration and iron loading, and determining a second sensitivity between the target heavy metal pollutant concentration and the relative abundance of sulfate-reducing bacteria; determining a core offset based on a first synergistic strength between iron loading and the target heavy metal pollutant concentration, and a second synergistic strength between the relative abundance of sulfate-reducing bacteria and the target heavy metal pollutant concentration; correcting the first and second sensitivities based on a comparison of the core offset with a preset offset threshold; and determining the prediction range for iron loading and the prediction range for sulfate-reducing bacteria dosage based on the corrected first and second sensitivities.

[0005] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining a first action delay coefficient based on the moment when the time-series data of the target heavy metal pollutant concentration and iron loading first show a synchronous decrease in time series; and determining a first sensitivity based on the synergistic strength between the degree of change in iron loading and the degree of change in the target heavy metal pollutant concentration in the subsequent period starting from the moment of synchronous decrease, and the first action delay coefficient.

[0006] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining the moment when the relative change rate of sulfate-reducing bacteria abundance first exceeds zero and the relative change rate of the target heavy metal pollutant concentration is less than zero, based on time-series data of sulfate-reducing bacteria relative abundance, as the starting point of bacterial community remediation; determining a second action delay coefficient based on the time required from the starting point of bacterial community remediation until the degree of change in bacterial abundance converges to within a preset fluctuation threshold; and determining a second sensitivity based on the synergistic strength between the relative abundance of sulfate-reducing bacteria and the concentration of the target heavy metal pollutant after the starting point of bacterial community remediation, as well as the second action delay coefficient.

[0007] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: generating a first synergistic strength sequence based on the synergistic strength between iron loading and the target heavy metal pollutant concentration; generating a second synergistic strength sequence based on the synergistic strength between the relative abundance of sulfate-reducing bacteria and the target heavy metal pollutant concentration; generating time-series data of the synergistic strength difference characterizing the transfer of remediation cores between physisorption and bioremediation based on the first and second synergistic strength sequences; determining the remediation core transfer time based on the time-series data of the synergistic strength difference; and determining the core offset as the time-series average of the absolute values ​​of the synergistic strength difference from the remediation core transfer time to the end of remediation.

[0008] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: determining the relative dispersion of the core offset among multiple samples; the relative dispersion is used to characterize the stability of the core offset; determining a weighted core offset based on the mean of the core offset and the relative dispersion; when the weighted core offset is less than or equal to a preset offset threshold, enhancing the first sensitivity while keeping the second sensitivity unchanged; when the weighted core offset is greater than the preset offset threshold, enhancing the second sensitivity while keeping the first sensitivity unchanged.

[0009] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: scaling a preset iron loading baseline range according to a modified first sensitivity to generate a predicted range for iron loading; and scaling a preset sulfate-reducing bacteria dosage baseline range according to a modified second sensitivity to generate a predicted range for sulfate-reducing bacteria dosage.

[0010] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: calculating the first rate of change of adjacent moments in the time series data of the target heavy metal pollutant concentration, and the second rate of change of adjacent moments in the time series data of iron load; determining the moment when both the first rate of change and the second rate of change are negative for the first time in the time series as the moment of synchronous decline; and determining the time difference between the moment of synchronous decline and the moment of remediation start as the first action delay coefficient.

[0011] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: taking the difference between the absolute values ​​of the first and second coordination strength sequences at the same moment as the coordination strength difference at that moment, and generating coordination strength difference time series data; determining the difference trend of adjacent moments in the coordination strength difference time series data; and determining the moment when the difference trend changes from negative to positive and the corresponding coordination strength difference is a local minimum as the core transfer repair moment.

[0012] In conjunction with the first aspect mentioned above, in one possible implementation, the method specifically includes: collecting multiple samples from the soil area to be remediated and recording the soil at multiple depths for each sample; detecting the initial target heavy metal pollutant concentration, initial soil pH, and initial relative abundance of sulfate-reducing bacteria at each depth; setting different iron loading and sulfate-reducing bacteria dosage for each sample based on the initial target heavy metal pollutant concentration; recording the changes in soil pH, relative abundance of sulfate-reducing bacteria, target heavy metal pollutant concentration, and iron loading for each sample according to a preset time series during the remediation process, and normalizing the detected data.

[0013] Secondly, the present invention provides a soil pollution remediation parameter prediction system based on multi-source data. The system includes: a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the methods described above.

[0014] The present invention has the following beneficial effects: This invention acquires time-series detection data from multiple soil samples to be remediated, including the relative abundance of sulfate-reducing bacteria, the concentration of target heavy metal pollutants, and iron loading. It directly determines the first sensitivity between the target heavy metal pollutant concentration and iron loading, and the second sensitivity between the target heavy metal pollutant concentration and the relative abundance of sulfate-reducing bacteria. Then, it uses the first synergistic strength between iron loading and the target heavy metal concentration, and the second synergistic strength between the relative abundance of sulfate-reducing bacteria and the target heavy metal concentration, to quantify the degree of core shift in remediation. Based on the comparison between the core shift degree and a preset shift threshold, the two sensitivities are corrected, and finally, the prediction range for iron loading and sulfate-reducing bacteria dosage is determined. Abandoning the existing method of statically fitting a single parameter value, this method achieves prediction of remediation parameter ranges through the synergistic strength and core shift among multi-source data. It can not only respond to fluctuations in soil environmental data in real time, but also provide a parameter range with tolerance for errors in actual engineering. This significantly improves the accuracy and operability of remediation strategies for complex contaminated sites. It overcomes the technical defects of existing methods that rely on single prediction results and are difficult to adapt to uncertain field conditions. This solves the technical problem that existing methods lack the necessary engineering tolerance when using single parameter values. Once the actual working conditions deviate from the prediction conditions, it may lead to substandard remediation results or uncontrolled remediation costs, making it difficult for the prediction results and engineering stability to meet actual needs. Attached Figure Description

[0015] 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.

[0016] Figure 1 This is a flowchart illustrating a method for predicting soil pollution remediation parameters based on multi-source data, provided in one embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the soil pollution remediation parameter prediction method and system based on multi-source data proposed in this invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] 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 to which this invention pertains.

[0019] The specific scheme of the soil pollution remediation parameter prediction method and system based on multi-source data provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0020] Please see Figure 1 This document illustrates a flowchart of a method for predicting soil pollution remediation parameters based on multi-source data, according to an embodiment of the present invention. The method includes: S101. Obtain time-series detection data of multiple samples of the soil to be remediated.

[0021] The time-series detection data includes the relative abundance of sulfate-reducing bacteria, the concentration of the target heavy metal pollutant, and the iron loading. The target heavy metal pollutant is antimony or arsenic.

[0022] Optionally, multiple representative soil samples are pre-collected for the target soil area to be remediated, and remediation simulation experiments are conducted on each sample separately. During the experiment, various parameters of each sample are continuously detected at preset time intervals (e.g., every 12 or 24 hours, which is not limited in this invention), and the relative abundance of sulfate-reducing bacteria, the concentration of target heavy metal pollutants, and the real-time monitored iron loading are recorded at each moment, thereby forming multi-dimensional time-series detection data.

[0023] Furthermore, in some embodiments of the present invention, obtaining time-series detection data of multiple samples of soil to be remediated includes: collecting multiple samples in the area of ​​soil to be remediated and recording the soil at multiple depth levels for each sample; detecting the initial target heavy metal pollutant concentration, initial soil pH, and initial relative abundance of sulfate-reducing bacteria at each depth level. Based on the initial target heavy metal pollutant concentration for each sample, different iron loading and sulfate-reducing bacteria dosages were set for each sample. During the remediation process of each sample, changes in soil pH, relative abundance of sulfate-reducing bacteria, target heavy metal pollutant concentration, and iron loading were recorded according to a preset time series.

[0024] Optionally, for the target soil area to be remediated, an S-shaped sampling path is planned using a stratified controlled sampling method. Multiple sampling points are selected, and soil samples are collected from the top 0-20cm, middle 20-50cm, and lower than 50cm layers at each sampling point, resulting in multiple samples, denoted as the set. After collection, soil samples were immediately cleaned of stones, root debris, and other foreign objects, then placed in a refrigerated truck and transported to the laboratory within 24 hours. For each sample, high-throughput sequencing was used to detect soil microbiota, and the relative abundance of sulfate-reducing bacteria was calculated. Inductively coupled plasma mass spectrometry was used to detect the initial concentration of target heavy metal pollutants. The initial value of iron loading in the remediation material was determined by weighing. For each sample at each depth, the tests were repeated at least three times. The Raida criterion was used to identify and remove outliers caused by operational contamination or instrument deviation. The arithmetic mean of the valid test results was calculated as the initial data for that parameter of that sample.

[0025] Understandably, heavy metal pollutants in soil exhibit significant migration differences and uneven distribution across the vertical profile, meaning that samples from a single depth cannot represent the overall pollution status of the entire remediation area. By simultaneously collecting soil samples from different depths—top, middle, and deep—it is possible to comprehensively capture the vertical concentration gradient changes of pollutants and the heterogeneity of soil physicochemical properties. This provides more representative and spatially comprehensive time-series monitoring data for subsequent construction of topological correlation models and determination of sensitivity and core offset.

[0026] Optionally, different combinations of repair parameters can be set for different samples. First, for the sample... To set benchmark remediation parameters, the selection of iron loading benchmark remediation parameters involves first calculating the amount of the target pollutant per unit weight of soil based on the detected initial antimony and / or arsenic concentrations. Then, based on the reaction mechanism between the remediation material and the pollutant, the corresponding chemical reaction equation is selected. Using the stoichiometric coefficients of the reaction equation, the theoretical number of moles of remediation agent required for complete conversion or passivation of the target pollutant is calculated and converted to mass. Since soil is not a pure aqueous solution, it contains competing ions such as phosphate, mass transfer resistance, and effectiveness loss. A redundancy factor (usually 2-10) is multiplied by the calculated theoretical mass to compensate for agent loss under complex conditions, ensuring that the initial iron loading benchmark value has sufficient reaction margin. This ultimately yields the iron loading benchmark value. .

[0027] It is understandable that theoretical calculations of iron demand are based on a pure water system. However, in actual soil, competing ions, mass transfer resistance, side reaction consumption, and spatial heterogeneity exist, causing some iron to be unable to effectively act on the target pollutants. Therefore, a redundancy coefficient is used to multiply and amplify the theoretical value to ensure sufficient effective iron loading, thereby guaranteeing remediation effectiveness and long-term stability. The specific value is determined through small-scale experiments or empirical selection based on pollutant concentration, soil clay content, and competing ion levels; this invention does not impose any limitations on this.

[0028] For selecting the baseline value for sulfate-reducing bacteria (SRB) dosage, based on prior biological knowledge and soil microbial community analysis, a target competitive advantage ratio is set after SRB dosage. Typically, exogenous bacteria are required to account for 1%-10% of the total community; taking 5% as an example, the baseline SRB dosage = initial total soil microbial concentration (which can be obtained through high-throughput sequencing or plate counting) × competitive advantage ratio of 5%. This ensures that SRB dosage makes the bacteria the dominant species, preventing reduced soil remediation capacity due to competition among microorganisms. The final baseline SRB dosage is thus determined. Obtain the sample Repair parameter combination Then, adjust the preset parameter step size. Construct a basic repair parameter matrix with a fixed percentage of the baseline value, such as 25%: , , , Assign the parameter combinations in the basic repair parameter matrix to all but one of the following: The remaining samples were then used for remediation in parallel for each soil sample, and the data were recorded over time. The curve of change, in which To repair the starting point, all collected time-series data were normalized and mapped to the interval [0,1] to obtain standardized time-series detection data, including the relative abundance of sulfate-reducing bacteria, the concentration of target heavy metal pollutants, and the iron loading.

[0029] It should be noted that, in this embodiment of the invention, the normalization processing of the time-series detection data can use maximum-minimum value normalization processing to obtain the maximum and minimum values ​​of each parameter across all samples and all times. Subtracting the minimum value from each parameter value and then dividing by the difference between the maximum and minimum values ​​maps the data to the [0,1] interval. Of course, in other embodiments of the invention, other dimensionless processing methods such as decimal scaling normalization can also be used, as long as parameters with different dimensions can be compared on the same scale. This invention does not impose any limitations on this.

[0030] S102. Determine the first sensitivity between the target heavy metal pollutant concentration and iron loading, and determine the second sensitivity between the target heavy metal pollutant concentration and the relative abundance of sulfate-reducing bacteria.

[0031] Optionally, based on time-series detection data from multiple samples, for each sample, a first sensitivity between the target heavy metal pollutant concentration and iron loading, and a second sensitivity between the target heavy metal pollutant concentration and the relative abundance of sulfate-reducing bacteria are calculated. The first sensitivity characterizes the extent to which changes in iron loading affect the decrease in the target heavy metal pollutant concentration, and the second sensitivity characterizes the extent to which changes in the relative abundance of sulfate-reducing bacteria affect the decrease in the target heavy metal pollutant concentration.

[0032] Optionally, the calculation of the first sensitivity is based on the target heavy metal pollutant concentration sequence and iron loading sequence, and is quantified by analyzing the synergistic relationship between the two over time (e.g., the start time of synchronous decline, the consistency of change intensity in subsequent periods, and response delay). The calculation of the second sensitivity is based on the target heavy metal pollutant concentration sequence and sulfate-reducing bacteria relative abundance sequence, and is quantified by analyzing the synergistic relationship between the two over time (e.g., the correlation time between the start of bacterial growth and the start of concentration decline, the adaptation time required for the bacterial community to reach stable metabolism, and the synchronization intensity in subsequent periods). Both sensitivity calculations adopt a pre-defined evaluation framework, using a time delay factor as a penalty term to weight the mean of synergistic intensity to ensure the comparability of the sensitivities of different remediation factors. Finally, the calculated first and second sensitivities are used as the initial edge weights between the target heavy metal pollutant concentration and iron loading, and between the target heavy metal pollutant concentration and sulfate-reducing bacteria relative abundance, respectively, in the topological correlation model.

[0033] Furthermore, in some embodiments of the present invention, determining the first sensitivity between the target heavy metal pollutant concentration and the iron loading includes: determining a first action delay coefficient based on the moment when the time-series data of the target heavy metal pollutant concentration and the iron loading first synchronously decrease in time series; and determining the first sensitivity based on the synergistic strength between the degree of change in the iron loading and the degree of change in the target heavy metal pollutant concentration in subsequent periods starting from the moment of synchronous decrease, and the first action delay coefficient.

[0034] Specifically, the time-series data of the target heavy metal pollutant concentration and the iron loading are time-series aligned to identify the moment when they first show a synchronous decrease in time. The time difference between this moment and the remediation start moment is used as the first action delay coefficient. Then, from the moment of synchronous decrease to the end of remediation, the synergistic strength between the degree of change in iron loading and the degree of change in the target heavy metal pollutant concentration is calculated at each moment (e.g., the absolute value of the product of their rates of change is used as the local synergistic strength at that moment), and the time-series average of the local synergistic strength is calculated for all moments. Finally, this time-series average is weighted using the first action delay coefficient (e.g., using a penalty factor in the form of a negative exponential function) to obtain the first sensitivity.

[0035] It should be noted that the above method of setting the local synergistic intensity only applies to scenarios where both decrease synchronously (both rates of change are negative). For asynchronous situations such as a decrease in iron loading but an increase in pollutant concentration, or an increase in iron loading but a decrease in pollutant concentration, the local synergistic intensity is set to zero because it does not meet the physical expectations of synergistic remediation. This ensures that the first sensitivity truly reflects the positive contribution of iron loading to pollutant removal.

[0036] For example, the first sensitivity satisfies the following formula: in, The first sensitivity (the sensitivity of the target heavy metal pollutant concentration to the iron loading). The normalized iron load application delay coefficient is obtained by normalizing the first application delay coefficient through the maximum and minimum values. The larger the value, the slower the response. is a natural exponential function with base e, used to map the time delay penalty factor to the interval (0,1] to achieve negative correlation penalty; This represents the total number of moments in the time series. The index position in the time series where the time series curves of target heavy metal pollutant concentration and iron loading begin to decrease synchronously. The summation index variable is from b to n; Let c be the relative rate of change of the iron load at time c; Let be the relative change rate of the target heavy metal pollutant concentration at time c; A preset non-zero constant (e.g., 0.0001) is used to avoid the denominator being zero. From the start of the synchronous descent to the end of the repair, calculate the absolute value of the product of the rate of change of iron load and the rate of change of concentration at each moment. The physical meaning of the product is that when both decrease simultaneously, the product is positive. The larger the sum, the stronger the coordinated descent of the two. The average value is calculated to reflect the average level of synergy intensity over the entire subsequent period; This is an exponentially decaying penalty factor. Time delay coefficient. The larger the value, the smaller the exponent (because of the negative exponent), and the heavier the penalty on the average cooperating strength. That is, the slower the response (the longer the time), the lower the sensitivity will be, even if the subsequent cooperating strength is high.

[0037] It should be noted that the local synergy intensity setting method is only for scenarios where both iron loading and pollutant concentration decrease simultaneously (both rates of change are negative). For asynchronous situations such as a decrease in iron loading but an increase in pollutant concentration, or an increase in iron loading but a decrease in pollutant concentration, the local synergy intensity is set to zero because it does not meet the physical expectations of synergistic remediation. This ensures that the first sensitivity truly reflects the positive contribution of iron loading to pollutant removal.

[0038] It should be noted that if no synchronous decrease moment is identified during the entire monitoring period (i.e., there is no moment when both change rates are negative at the same time), it is determined that the change in iron load has no observable inhibitory effect on the concentration of the target heavy metal pollutant. In this case, the first sensitivity is set to zero, and subsequent weighted calculations involving the first sensitivity of this sample are skipped.

[0039] Understandably, this step overcomes the shortcomings of traditional sensitivity analysis, which only focuses on the intensity of change and ignores the timeliness of response, by introducing a first-action delay coefficient as a penalty factor. Specifically, even if the iron loading and pollutant concentration show a strong synergistic decreasing trend in subsequent periods, if no concentration response is observed for a long time after the iron loading is added (i.e., a large delay), the actual sensitivity of this pathway will be reasonably suppressed.

[0040] Furthermore, in some embodiments of the present invention, determining the second sensitivity between the target heavy metal pollutant concentration and the relative abundance of sulfate-reducing bacteria includes: determining, based on time-series data of the relative abundance of sulfate-reducing bacteria, the moment when the relative change rate of bacterial abundance first exceeds zero and the relative change rate of the target heavy metal pollutant concentration is less than zero as the bacterial community remediation start point; determining a second action delay coefficient based on the time required from the bacterial community remediation start point until the degree of change in bacterial abundance converges to within a preset fluctuation threshold; and determining the second sensitivity based on the synergistic strength between the relative abundance of sulfate-reducing bacteria and the target heavy metal pollutant concentration after the bacterial community remediation start point and the second action delay coefficient.

[0041] Specifically, based on the time-series data of the relative abundance of sulfate-reducing bacteria, the relative rate of change between adjacent time points is calculated. The moment when the relative rate of change of bacterial abundance first exceeds zero and the relative rate of change of the target heavy metal pollutant concentration is less than zero is determined as the initiation point of bacterial community remediation. Then, starting from the initiation point, the fluctuation of the degree of change in bacterial abundance is monitored. When the standard deviation of the rate of change over multiple consecutive time points (e.g., three consecutive time points) is less than a preset fluctuation threshold, the bacterial community is determined to have entered a stable phase. The first time point in this consecutive phase is taken as the steady-state point, and the time difference between the steady-state point and the initiation point is used as the second action delay coefficient. Finally, from the initiation point to the end of remediation, the synergistic strength between the rate of change of bacterial abundance and the rate of change of the target heavy metal pollutant concentration at each time point is calculated (e.g., the absolute value of the product of the two rates of change is taken as the synergistic strength at that time point). The time-series average of the synergistic strength at all time points is then calculated, and the second action delay coefficient is used to weight this average (e.g., using a natural exponential function with base e). The second sensitivity is derived by considering the relationship between the concentration of the target heavy metal pollutant and the relative abundance of sulfate-reducing bacteria. A mathematical evaluation framework consistent with iron loading is adopted to ensure the comparability of remediation factors across different dimensions within the topological network. A second action delay coefficient is introduced as a penalty term. If the bacterial community can quickly establish a steady state under pH shift interference (i.e., the relative rate of change converges rapidly), it indicates strong adaptability to the soil environment, and a smaller penalty coefficient is assigned, thus retaining a higher edge weight. Conversely, if the bacterial community remains in a fluctuating phase for a long time due to slow mitigation of heavy metal toxicity or excessive environmental stress, the second action delay coefficient increases, and the value of the negative exponential penalty factor decreases accordingly, meaning a greater discount on synergistic strength (increased penalty intensity), thereby reducing the second sensitivity. The specific calculation logic is the same as for the first sensitivity and will not be repeated here.

[0042] Furthermore, in some embodiments of the present invention, determining a first action delay coefficient based on the moment when the time-series data of the target heavy metal pollutant concentration and iron loading first show a synchronous decrease in time series includes: calculating a first rate of change of adjacent moments in the time-series data of the target heavy metal pollutant concentration and a second rate of change of adjacent moments in the time-series data of the iron loading; determining the moment when both the first rate of change and the second rate of change are negative for the first time series as the moment of synchronous decrease; and determining the time difference between the moment of synchronous decrease and the moment of remediation initiation as the first action delay coefficient.

[0043] After obtaining the raw time-series detection data (without normalization) for each sample, the first rate of change of the target heavy metal pollutant concentration at adjacent time points is directly calculated. This is done by subtracting the concentration value of the previous time point from the current concentration value, and then dividing the difference by the sum of the previous concentration value and a preset non-zero minimum constant (e.g., 0.0001) to obtain the relative rate of change at that time point. The second rate of change is calculated using the same method for the raw iron loading data. If the raw detection data already contains zero values ​​(e.g., extremely low concentrations), adding a non-zero minimum constant ensures that the denominator is always greater than zero.

[0044] Based on the first and second rate of change sequences calculated above, the analysis is performed sequentially, starting from the second moment after the repair begins. At each moment, it is checked whether both the first and second rates of change are simultaneously negative. When both are negative for the first time, that moment is recorded as the synchronous descent moment.

[0045] After determining the synchronous decline time, the remediation start time is obtained (i.e., the time corresponding to the first data point in the time series detection data, usually the time point at which the remediation experiment begins). The time difference between the synchronous decline time and the remediation start time is calculated; this difference is the first action delay coefficient, and its dimension is consistent with the sampling interval unit of the time series (e.g., hours or days). This delay coefficient quantitatively describes the length of time from the start of remediation to the actual reduction of pollutant concentration caused by the iron loading. The larger the delay coefficient, the more severely the iron-based material is affected by factors such as mass transfer resistance and competing ion consumption, and the slower the response; the smaller the delay coefficient, the faster the iron-based material can act on the target pollutant.

[0046] S103. Determine the core offset based on the first synergistic strength between iron loading and target heavy metal pollutant concentration, and the second synergistic strength between sulfate-reducing bacteria relative abundance and target heavy metal pollutant concentration.

[0047] Optionally, based on time-series detection data from multiple samples, a first co-intensity between iron loading and the target heavy metal pollutant concentration, and a second co-intensity between the relative abundance of sulfate-reducing bacteria and the target heavy metal pollutant concentration are calculated for each sample. The first co-intensity characterizes the temporal synchronicity and correlation between changes in iron loading and changes in target heavy metal pollutant concentration, while the second co-intensity characterizes the temporal synchronicity and correlation between changes in the relative abundance of sulfate-reducing bacteria and changes in target heavy metal pollutant concentration. Then, the first and second co-intensities at the same time point are compared to obtain the degree of difference between them (e.g., calculating the absolute value of the difference).

[0048] Understandably, in the initial stages of remediation, iron-based materials rapidly adsorb pollutants, with the first synergistic strength significantly higher than the second. As remediation progresses, the iron materials gradually passivate, while sulfate-reducing bacteria adapt to the environment and enhance metabolism, causing the second synergistic strength to gradually increase and potentially surpass the first. By calculating the absolute difference between the first and second synergistic strengths at each time step, a curve can be constructed showing how this difference changes over time. When the trend of this curve changes from decreasing to increasing (the difference reaches its minimum point), it represents the moment when the remediation core shifts from physisorption to bioremediation. From this point onward, if the difference continues to increase, it indicates that bioremediation has become dominant, with a significant shift in the core; if the difference remains consistently small, it indicates that the two mechanisms work synergistically or alternately, with a smaller shift in the core.

[0049] Further, in some embodiments of the present invention, determining the core offset based on a first synergistic strength between iron loading and target heavy metal pollutant concentration, and a second synergistic strength between sulfate-reducing bacteria relative abundance and target heavy metal pollutant concentration, includes: generating a first synergistic strength sequence based on the synergistic strength between iron loading and target heavy metal pollutant concentration; generating a second synergistic strength sequence based on the synergistic strength between sulfate-reducing bacteria relative abundance and target heavy metal pollutant concentration; generating time-series data of the synergistic strength difference characterizing the transfer of the remediation core between physical adsorption and bioremediation based on the first and second synergistic strength sequences; determining the remediation core transfer time based on the time-series data of the synergistic strength difference; and determining the core offset as the time-series average of the absolute values ​​of the synergistic strength difference from the remediation core transfer time to the end of remediation.

[0050] Specifically, for each time point, the degree of change in iron loading is integrated with the degree of change in the target heavy metal pollutant concentration; for example, the absolute value of the product of their rates of change is taken as the synergistic strength at that time. This synergistic strength quantifies the degree of synchronization between the decrease in iron loading and the decrease in pollutant concentration at the same time point; a larger value indicates a closer linkage between the two. Arranging the synergistic strengths of all time points in chronological order constitutes the first synergistic strength sequence.

[0051] For each time point, the change in the relative abundance of sulfate-reducing bacteria is integrated with the change in the concentration of the target heavy metal pollutant. For example, the absolute value of the product of their rates of change is taken as the synergistic strength at that time point. This synergistic strength quantifies the degree of synchronization between the increase in bacterial abundance and the decrease in pollutant concentration at the same time point; a larger value indicates a stronger correlation between bacterial activity and pollutant removal. Arranging the synergistic strengths of all time points in chronological order constitutes the second synergistic strength sequence.

[0052] For each time point, the absolute value of the first coordination strength is subtracted from the absolute value of the second coordination strength, and the absolute value of the difference is taken to obtain the coordination strength difference for that time point. The coordination strength differences for all time points are then arranged in chronological order to generate time-series data of coordination strength differences.

[0053] The difference sequence is obtained by calculating the first-order difference between adjacent time points in the difference sequence (i.e., the difference between the next time point and the difference between the previous time point). In the early stages of remediation, physisorption dominates, and the difference may be large and show a decreasing trend (bioremediation gradually strengthens). When bioremediation begins to surpass physisorption, the difference drops to a minimum and then begins to rise. Therefore, the moment when the value in the difference sequence changes from negative to positive and the corresponding synergistic strength difference reaches a local minimum is determined as the core transfer moment of remediation. After determining the core transfer moment, the synergistic strength difference of all time points within this period is extracted from that moment until the end of remediation. The arithmetic mean of these differences (i.e., the sum of the differences divided by the number of time points) is calculated, and this average value is determined as the core shift. This average value quantifies the average level of the difference between the two synergistic strengths after the transfer moment. The larger the core shift, the more obvious and stable the advantage of bioremediation over physisorption; the smaller the core shift, the more similar or alternating the contributions of the two mechanisms throughout the post-transfer period.

[0054] For example, the core offset repair satisfies the following formula: in, The term "repair core offset" indicates the degree of drastic transfer of the repair core from physical adsorption to biorepair. This represents the total number of moments in the time series. To repair the index position of the core transfer moment in the time series (the index where the valley point is located); The summation index variable is from e to n; For the first The synergistic strength between the iron loading at any given moment and the target heavy metal pollutant concentration can be the absolute value of the product after synchronous decrease or other measures. For the first The synergistic strength between the relative abundance of sulfate-reducing bacteria and the concentration of the target heavy metal pollutant at any given time; To prevent the denominator from being zero, a non-zero constant (e.g., 0.0001) is preset.

[0055] The absolute difference between the synergistic strength of physical adsorption and bioremediation at the same time point is calculated to reflect the difference in their contributions at that moment. The larger the difference, the more dominant one mechanism is; the smaller the difference, the more similar their effects are. Determine the time from the start of the core transfer during repair to the end of the repair, and sum the differences at all points in time; Calculate the average value (number of time points is...) The time-series average difference is obtained; R is the average value: the larger the value, the greater the difference in the synergistic strength of the two mechanisms after the transfer time, that is, the repair core quickly and stably biases towards one side; the smaller the value, the smaller the difference between the two after the transfer or the alternating fluctuations.

[0056] Furthermore, in some embodiments of the present invention, determining the repair core transfer time based on the time series data of the coordination strength difference includes: taking the difference between the absolute values ​​of the first coordination strength sequence and the second coordination strength sequence at the same time as the coordination strength difference at that time, and generating coordination strength difference time series data; determining the difference trend of adjacent times in the coordination strength difference time series data; and determining the time when the difference trend changes from negative to positive and the corresponding coordination strength difference is a local minimum as the repair core transfer time.

[0057] Specifically, after obtaining the first and second synergistic intensity sequences, the two sequences are aligned using the same time index. For each time point, the difference between the absolute values ​​of the first and second synergistic intensities is calculated, and the absolute value of this difference is taken to obtain the synergistic intensity difference for that time point. This difference reflects the magnitude of the difference in contribution of the physical adsorption mechanism and the bioremediation mechanism to pollutant removal at that time point: the closer the difference is to zero, the more similar the strengths of the two mechanisms; the larger the difference, the more dominant one mechanism is. The synergistic intensity differences for all time points are arranged sequentially in chronological order to generate the synergistic intensity difference time-series data. For two adjacent time points in the difference sequence, the difference between the later and earlier time points is calculated to obtain the first-order difference value of that adjacent interval. The sign of this difference value reflects whether the difference sequence is increasing or decreasing: a positive difference value indicates that the synergistic intensity difference increases over time, and the gap between physical adsorption and bioremediation is widening; a negative difference value indicates that the synergistic intensity difference decreases over time, and the contributions of the two mechanisms tend to be similar. By performing the above calculations sequentially on the entire difference sequence, the changing pattern of the difference value sign can be determined, thereby capturing the turning point of the difference sequence.

[0058] Based on the obtained difference trend sequence, the signs of the difference values ​​at each time point are iterated in chronological order. When the difference value changes from negative to positive, it indicates that the synergistic strength difference sequence changes from decreasing to increasing at this point, meaning that the synergistic strength difference at that time is a local minimum. The time corresponding to this local minimum is determined as the time of repair core transfer. The physical meaning of this judgment logic is that before the transfer of the repair core, the physical adsorption mechanism gradually weakens while the biological repair mechanism gradually strengthens, and the strength difference between the two mechanisms continues to narrow; when the difference narrows to a minimum, the biological repair mechanism begins to surpass physical adsorption and become dominant, after which the difference widens. Therefore, this minimum difference point precisely corresponds to the critical moment when the repair core transfers from physical adsorption to biological repair.

[0059] S104. Based on the comparison results between the core offset and the preset offset threshold, the first sensitivity and the second sensitivity are corrected.

[0060] After determining the core shift, it is compared with a preset shift threshold. This preset shift threshold is used to determine the significance of the succession from physioremediation to bioremediation in the remediation core. If the core shift is greater than the preset shift threshold, it indicates that the bioremediation pathway has become dominant and has a significant advantage in the remediation process. In this case, the secondary sensitivity should be enhanced while the primary sensitivity remains unchanged to reflect the higher reliability of the bioremediation sensitivity. If the core shift is less than or equal to the preset shift threshold, it indicates that the physioremediation pathway is still dominant or the difference between the two is not significant. In this case, the primary sensitivity should be enhanced while the secondary sensitivity remains unchanged. In actual processing of multi-sample data, the mean of the core shift is first calculated based on the sample groups, and then the mean is weighted using its relative dispersion to obtain the weighted core shift. Then, the weighted core shift is compared with the preset shift threshold. The larger the weighted core shift, the more stable and dominant the bioremediation succession is, and the magnitude of the enhancement of the secondary sensitivity should be increased accordingly. Conversely, if the weighted core shift is less than or equal to the threshold, the primary sensitivity should be enhanced. It should be noted that the enhancement operation can linearly amplify the original sensitivity using a gain factor. The magnitude of the gain factor can be determined based on the deviation between the weighted core offset and a preset offset threshold. Specifically, firstly, the absolute difference between the weighted core offset and the preset offset threshold is calculated as the deviation. Then, based on this deviation, a monotonically increasing mapping relationship is used to determine the gain factor; the larger the deviation, the larger the gain factor. When the deviation is zero (i.e., the weighted core offset equals the preset offset threshold), the gain factor is 1, indicating no enhancement. As a specific implementation, the deviation can be divided by a preset normalization constant, and then the result can be incremented by one to obtain the gain factor; this normalization constant can be set according to the preset offset threshold or the maximum expected deviation value.

[0061] Furthermore, in some embodiments of the present invention, the first sensitivity and the second sensitivity are modified based on the comparison result of the core offset and the preset offset threshold, including: determining the relative dispersion of the core offset among multiple samples; the relative dispersion is used to characterize the stability of the core offset; determining the weighted core offset based on the mean of the core offset and the relative dispersion; when the weighted core offset is less than or equal to the preset offset threshold, the first sensitivity is enhanced while the second sensitivity remains unchanged; when the weighted core offset is greater than the preset offset threshold, the second sensitivity is enhanced while the first sensitivity remains unchanged.

[0062] Specifically, after obtaining the core offsets of multiple samples, the relative dispersion of these core offsets is calculated to characterize their stability. Relative dispersion is a dimensionless indicator that eliminates the influence of the absolute magnitude of the data. One specific implementation is to calculate the standard deviation of the core offsets and divide it by the sum of the mean and a preset non-zero minimum constant (e.g., 0.0001, to prevent the denominator from being 0) to obtain the coefficient of variation, or to calculate the ratio of the variance to the square of the mean (i.e., the relative variance). For example, first, the arithmetic mean and standard deviation of the core offsets of all samples are calculated, and then the standard deviation is divided by the mean to obtain the coefficient of variation; or the variance is calculated first and then divided by the square of the mean to obtain the relative variance. The smaller the relative dispersion, the more consistent the core offsets are among different samples, and the higher the stability; the larger the relative dispersion, the greater the fluctuation of the core offsets, and the worse the stability.

[0063] Based on the mean and relative dispersion of the core offset across multiple samples, a stability-weighted processing is applied to the mean. Specifically, the relative dispersion is first exponentially negative to obtain a stability coefficient between zero and one. The smaller the relative dispersion, the closer the stability coefficient is to one; the larger the relative dispersion, the closer the stability coefficient is to zero. Then, the stability coefficient is incremented by one to obtain a gain factor (with a value between one and two). Finally, this gain factor is multiplied by the mean of the core offset to obtain the weighted core offset.

[0064] The weighted core shift is compared with a pre-set shift threshold. This threshold is used to determine whether the succession from physisorption to bioremediation of remediation cores has reached a significant level; its specific value is set empirically or experimentally (e.g., 0.2). If the weighted core shift is less than or equal to this threshold, it indicates that the physisorption mechanism still dominates throughout the remediation process, or that bioremediation, although occurring, has not been significant enough to be a major contributing pathway. In this case, the first sensitivity (i.e., the sensitivity of iron loading to the effect of pollutant concentration) should be enhanced while the second sensitivity remains unchanged. This enhancement can be achieved by multiplying a gain factor by the first sensitivity; the value of the gain factor can be determined based on how close the weighted core shift is to the threshold.

[0065] If the weighted core offset exceeds a preset offset threshold, it indicates that the remediation core has clearly shifted from physical adsorption to bioremediation, and this succession exhibits high consistency and stability across multiple samples. At this point, sulfate-reducing bacteria have become, or are about to become, the dominant factor in pollutant removal. Therefore, the second sensitivity (i.e., the sensitivity of the relative abundance of sulfate-reducing bacteria to the effect of pollutant concentration) should be enhanced, while maintaining the first sensitivity unchanged. The magnitude of the enhancement operation can be positively correlated with the degree to which the weighted core offset exceeds the threshold: the greater the exceedance, the greater the gain, fully reflecting the dominant position and sensitivity weight of the bioremediation pathway in the current soil environment.

[0066] It is understandable that the preset offset threshold is determined based on control experiments or empirical values. For control samples with only iron-based materials added (without bacteria), the core offset is usually close to zero; while for samples with successful iron-bacterial co-remediation and biological succession, the offset is often greater than 0.2. Therefore, the preset offset threshold is generally set to an empirical value between 0.1 and 0.3. In specific implementation, it can also be adaptively calibrated according to the results. This invention does not limit this.

[0067] S105. Based on the corrected first and second sensitivities, determine the prediction range for iron loading and the prediction range for sulfate-reducing bacteria dosage.

[0068] Optionally, for iron loading, the theoretical dosage range calculated using stoichiometry based on the initial contamination concentration is used as the starting point, and this baseline range is scaled according to the corrected first sensitivity. The scaling principle is: the higher the first sensitivity, the more sensitive the iron loading is to changes in contaminant concentration; in this case, the prediction range should be narrowed to provide more accurate dosage guidance. Conversely, the lower the first sensitivity, the smaller the impact of iron loading fluctuations on the remediation effect; in this case, the prediction range can be widened to increase the engineering tolerance. Similarly, for sulfate-reducing bacteria dosage, the preset sulfate-reducing bacteria dosage baseline range is used as the starting point, and the range is scaled according to the corrected second sensitivity.

[0069] For example, assuming the baseline range for iron loading has been experimentally determined to be A to B grams per kilogram of soil, the corrected first sensitivity is a normalized value between 0 and 1. When this sensitivity is close to 1, the baseline range is shrunk to a smaller proportion of its original width (e.g., 50% of its original width), resulting in a narrower prediction range; when the sensitivity is close to 0, the baseline range is expanded to a larger proportion of its original width (e.g., 150% of its original width), resulting in a wider prediction range. The specific mapping relationship between sensitivity and scaling ratio can be achieved using a negative exponential function or an inverse proportional function. The prediction range for sulfate-reducing bacteria dosage is generated in the same way, using the corrected second sensitivity as the scaling basis.

[0070] Furthermore, in some embodiments of the present invention, determining the prediction range for iron loading and the prediction range for sulfate-reducing bacteria dosage based on the modified first sensitivity and second sensitivity includes: Specifically, a preset iron loading baseline range is first obtained. This baseline range can be calculated using chemometrics methods based on the initial target heavy metal pollutant concentration in the soil to be remediated. For example, it can be a range defined by a minimum effective dosage and a maximum safe dosage. Then, this baseline range is multiplicatively scaled according to the corrected first sensitivity. The basic principle of scaling is: the higher the corrected first sensitivity, the more sensitive the iron loading is to changes in pollutant concentration. In this case, the baseline range should be narrowed towards the center to generate a narrower prediction range, providing more accurate dosage guidance; conversely, the lower the corrected first sensitivity, the less the fluctuation in iron loading has an impact on the remediation effect. In this case, the baseline range should be appropriately widened to generate a wider prediction range, increasing the engineering tolerance margin. The mapping relationship between the scaling ratio and sensitivity can be achieved using a preset linear function.

[0071] Preferably, a linear scaling method can be used to determine the prediction range for iron loading and sulfate-reducing bacteria dosage. Specifically, the corrected first and second sensitivities are used as sensitivity-driven search range adjustment factors, and the scaling ratio is calculated using a linear function. For example, the linear mapping function satisfies the following formula: in, The scaling factor is used to multiplicatively scale the preset iron load baseline range or sulfate-reducing bacteria dosage baseline range, and output the final predicted range width. μ>1 indicates range expansion, μ=1 indicates no change, and μ<1 indicates contraction. This is the preset maximum scaling ratio, with a value greater than 1 (e.g., 1.5). It corresponds to the scaling ratio when the sensitivity W′=0 after correction, and is used to provide the widest engineering fault tolerance range when the sensitivity is low. This is the preset minimum scaling ratio, less than 1 (e.g., 0.5), corresponding to the scaling ratio when the sensitivity W′=1 after correction, used to provide the most accurate constraint range when highly sensitive; The sensitivity is the corrected value (the first corrected sensitivity is for iron loading, and the second corrected sensitivity is for sulfate-reducing bacteria dosage).

[0072] Understandably, the maximum scaling ratio corresponds to the case of lowest sensitivity. In this case, the accuracy requirement for the repair factor is not high, and appropriately widening the prediction interval (e.g., widening it to 150% of the baseline interval) can provide higher fault tolerance redundancy for the project and cope with on-site uncertainties. The minimum scaling ratio corresponds to the case of highest sensitivity. In this case, the parameter range must be strictly constrained to ensure the repair effect. Therefore, the interval is narrowed (e.g., narrowed to 50% of the baseline interval) to improve determinism. Both are usually set based on engineering experience or experimental data, ensuring that the upper limit after widening does not exceed the safety limit and the lower limit after narrowing is not lower than the effective threshold. This invention does not impose any limitations on this.

[0073] Specifically, firstly, a preset baseline range for sulfate-reducing bacteria (SRB) dosage is obtained. This baseline range can be determined by multiplying the initial total bacterial concentration of the soil to be remediated by a preset competitive advantage ratio (this ratio represents the expected proportion of exogenously added SRB to the total bacterial population in the later stages of remediation, for example, exogenous bacteria accounting for 5% of the total bacterial population), thus establishing a basic dosage range. Then, this baseline range is multiplicatively scaled based on the corrected second sensitivity. The basic principle of scaling is similar to that of iron loading: the higher the corrected second sensitivity, the more sensitive the abundance of SRB to changes in pollutant concentration. In this case, the baseline range should be narrowed towards the center to generate a narrower prediction range, ensuring the dosage accuracy of this highly sensitive factor during engineering implementation; conversely, the lower the corrected second sensitivity, the less the fluctuation in SRB dosage affects the remediation effect. In this case, the baseline range should be appropriately widened to generate a wider prediction range, increasing redundancy during construction. This scaling operation can also use the same nonlinear mapping function as the first sensitivity to ensure consistency in the generation logic of the prediction ranges for the two parameters. The final output of the predicted range for sulfate-reducing bacteria dosage is the recommended range for sulfate-reducing bacteria dosage in engineering.

[0074] It is understandable that the preset competitive advantage ratio is set based on prior knowledge of microbial ecology: if the ratio is too low (e.g., below 1%), the exogenous bacteria will have difficulty surviving in the competition and will not be able to effectively initiate bioremediation; if the ratio is too high (e.g., above 10%), the cost of the bacterial agent will increase significantly and may lead to waste of resources. Therefore, an empirical value between 1% and 10% (e.g., 5%) is usually taken, and the specific value can be calibrated according to the colonization effect and remediation efficiency of the bacterial community in the experiment. This invention does not limit this value.

[0075] On the other hand, the present invention also provides a soil pollution remediation parameter prediction system based on multi-source data. The system includes: a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the methods described above.

[0076] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0077] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for predicting soil pollution remediation parameters based on multi-source data, characterized in that, The method includes: Time-series detection data were obtained from multiple samples of the soil to be remediated; the time-series detection data included the relative abundance of sulfate-reducing bacteria, the concentration of the target heavy metal pollutant, and the iron loading; the target heavy metal pollutant was antimony or arsenic. The first sensitivity was determined between the concentration of the target heavy metal pollutant and the iron loading, and the second sensitivity was determined between the concentration of the target heavy metal pollutant and the relative abundance of sulfate-reducing bacteria. The core offset is determined based on the first synergistic strength between iron loading and target heavy metal pollutant concentration, and the second synergistic strength between sulfate-reducing bacteria relative abundance and target heavy metal pollutant concentration. Based on the comparison between the core offset and the preset offset threshold, the first sensitivity and the second sensitivity are corrected. Based on the corrected first and second sensitivities, the prediction ranges for iron loading and sulfate-reducing bacteria dosage were determined.

2. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 1, characterized in that, The determination of the first sensitivity between the target heavy metal pollutant concentration and iron loading includes: The first action delay coefficient is determined based on the moment when the time series data of the target heavy metal pollutant concentration and the iron loading first show a synchronous decrease in time series. The first sensitivity is determined based on the synergistic strength between the degree of change in the iron load and the degree of change in the target heavy metal pollutant concentration during the subsequent period starting from the moment of the synchronous descent, and the first action delay coefficient.

3. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 1, characterized in that, The second sensitivity for determining the relationship between the concentration of the target heavy metal pollutant and the relative abundance of sulfate-reducing bacteria includes: Based on the time-series data of the relative abundance of sulfate-reducing bacteria, the moment when the relative change rate of bacterial abundance first exceeds zero and the relative change rate of the target heavy metal pollutant concentration is less than zero is determined as the starting point for bacterial remediation. The second action delay coefficient is determined based on the time required from the start of the microbial community repair until the degree of change in microbial community abundance converges to within a preset fluctuation threshold. The second sensitivity is determined based on the synergistic strength between the relative abundance of sulfate-reducing bacteria after the initiation point of the microbial community repair and the concentration of the target heavy metal pollutant, as well as the second action delay coefficient.

4. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 1, characterized in that, The determination of core offset based on the first synergistic strength between iron loading and target heavy metal pollutant concentration, and the second synergistic strength between sulfate-reducing bacteria relative abundance and target heavy metal pollutant concentration, includes: A first synergy strength sequence is generated based on the synergy strength between the iron loading and the target heavy metal pollutant concentration. A second synergistic strength sequence is generated based on the synergistic strength between the relative abundance of sulfate-reducing bacteria and the concentration of the target heavy metal pollutant. Based on the first synergistic strength sequence and the second synergistic strength sequence, time-series data of synergistic strength difference characterizing the transfer of the remediation core between physical adsorption and bioremediation are generated; The timing of the repair core transfer is determined based on the time series data of the collaborative strength difference. The time-series average of the absolute values ​​of the collaborative strength difference during the period from the core transfer time to the end of the repair is determined as the core offset.

5. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 1, characterized in that, The step of correcting the first sensitivity and the second sensitivity based on the comparison result between the core offset and the preset offset threshold includes: Determine the relative dispersion of the core offset across multiple samples; the relative dispersion is used to characterize the stability of the core offset. The weighted core offset is determined based on the mean of the core offset and the relative dispersion. When the weighted core offset is less than or equal to the preset offset threshold, the first sensitivity is enhanced while the second sensitivity remains unchanged. When the weighted core offset is greater than the preset offset threshold, the second sensitivity is enhanced while the first sensitivity remains unchanged.

6. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 1, characterized in that, The determination of the prediction range for iron loading and the prediction range for sulfate-reducing bacteria dosage based on the corrected first and second sensitivities includes: The preset iron load reference range is scaled according to the modified first sensitivity to generate the prediction range of the iron load. Based on the modified second sensitivity, the preset reference range for sulfate-reducing bacteria dosage is scaled to generate a predicted range for the sulfate-reducing bacteria dosage.

7. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 2, characterized in that, The step of determining the first action delay coefficient based on the moment when the time-series data of the target heavy metal pollutant concentration and the iron loading first show a synchronous decrease in time includes: Calculate the first rate of change of the target heavy metal pollutant concentration at adjacent time points in the time series data, and the second rate of change of the iron load at adjacent time points in the time series data; The moment when both the first rate of change and the second rate of change first appear simultaneously in the time sequence is determined as the moment of synchronous decline; The time difference between the synchronous descent time and the repair start time is determined as the first action delay coefficient.

8. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 4, characterized in that, The step of determining the repair core transfer time based on the time series data of the collaborative strength difference includes: The difference between the absolute values ​​of the first and second coordination strength sequences at the same time is used as the coordination strength difference at that time to generate the coordination strength difference time series data. Determine the difference trend between adjacent time points in the time series data of the collaborative strength difference; The moment when the differential trend changes from negative to positive and the corresponding difference in synergy strength is a local minimum is determined as the moment of core transfer for repair.

9. The method for predicting soil pollution remediation parameters based on multi-source data according to claim 1, characterized in that, The acquisition of time-series detection data from multiple samples of the soil to be remediated includes: Multiple samples were collected from the soil area to be remediated, and the soil at multiple depths was recorded for each sample. The initial concentrations of target heavy metal pollutants, initial soil pH, and initial relative abundance of sulfate-reducing bacteria were detected at various soil depths. Based on the initial target heavy metal pollutant concentration for each sample, different iron loading and sulfate-reducing bacteria dosages were set for each sample. During the remediation process of each sample, changes in soil pH, relative abundance of sulfate-reducing bacteria, concentration of target heavy metal pollutants, and iron load were recorded according to a preset time series, and the detected data were normalized.

10. A soil pollution remediation parameter prediction system based on multi-source data, the system comprising: A memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that the processor, when executing the computer program, implements the steps of the method as claimed in any one of claims 1-9.

Images