A groundwater pollution monitoring method and system based on edge computing

Abstract

This invention relates to the field of edge computing technology, specifically to a groundwater pollution monitoring method and system based on edge computing. The method includes the following steps: acquiring groundwater sequence data and calculating unit-time fluctuation intensity parameters; comparing threshold intervals to generate sampling frequencies; collecting physicochemical indicators to generate deviation difference sets and extracting urgency weight parameters; combining the remaining computing power of edge nodes to generate task scheduling strategies; collecting environmental indicators and writing them into a deviation accumulation matrix and accumulating them row by row; and comparing with set risk thresholds to generate pollution risk levels. In this invention, by adaptively matching the sampling frequency based on the fluctuation intensity state, the fixed interval blind zone is effectively eliminated and abrupt signals are captured. Combining multi-dimensional deviation features and node computing power to construct an allocation index to generate a scheduling strategy, dynamic task allocation and computing power balancing are completed at the edge, avoiding link congestion caused by redundant baseline transmission. Based on the accumulation matrix, pollution trends are tracked, achieving proactive assessment.

Description

Technical Field

[0001] This invention relates to the field of edge computing technology, and in particular to a groundwater pollution monitoring method and system based on edge computing. Background Technology

[0002] Edge computing technology involves computing architectures that collect, process, and analyze data close to the data source. Its core aspects include distributed node deployment, edge data preprocessing, field device access protocol management, low-latency data transmission mechanisms, and layered collaboration mechanisms with cloud platforms. This technology typically involves setting up edge nodes with computing capabilities between sensing terminals and central servers. Raw data is filtered, integrated, and preliminarily calculated locally before being uploaded, thereby constructing a layered architecture that includes a sensing layer, an edge processing layer, and a central management layer. It is widely used in scenarios such as environmental monitoring, industrial control, and energy management.

[0003] Traditional groundwater pollution monitoring methods refer to technical solutions that continuously observe and record changes in pollutant concentrations in groundwater. The main technical aspects involved are the collection and analysis of indicators such as groundwater level, temperature, conductivity, dissolved oxygen, and heavy metal ion concentration. Traditional methods typically involve setting up sampling tubes and submersible pumps inside monitoring wells to periodically extract water samples. Various parameter data are obtained through chemical reagent colorimetric detection, atomic absorption spectrophotometry, or electrode methods. The data is then transmitted to a remote monitoring center via wired communication lines or mobile communication terminals. The central server summarizes the data at preset time intervals, compares it with historical records, and generates trend curves. The entire process relies on a centralized data processing structure and manual inspection and maintenance to determine the groundwater pollution status.

[0004] Traditional monitoring methods rely on submersible pumps installed inside monitoring wells to periodically extract water samples and transmit them to a remote central server for aggregation and comparison. The fixed centralized data processing and periodic sampling operation mode makes it impossible to respond in time to sudden water quality deterioration. Long-distance transmission of raw parameters is prone to communication link congestion and generates a large amount of redundant baseline information. The operation mode that relies on manual inspection and fixed time intervals lacks the ability to adaptively adjust to dynamic changes in hydrological conditions. It is very easy to miss key high-frequency sudden change signals in the early stage of pollution spread and cause delayed judgment and false alarms. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a groundwater pollution monitoring method and system based on edge computing.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a groundwater pollution monitoring method based on edge computing, comprising the following steps: S1: Obtain groundwater sequence data of edge nodes within the time window, accumulate the absolute change in water level to generate the fluctuation energy integral value, accumulate the sampling interval to generate the standardized time base, calculate the ratio of the fluctuation energy integral value to the standardized time base, and generate the fluctuation intensity parameter per unit time. S2: Compare the unit time fluctuation intensity parameter with the preset graded frequency mapping table, select the corresponding sampling frequency, and generate the target sampling frequency; S3: Collect changes in liquid level, conductivity, and heavy metals at the target sampling frequency, calculate the difference with the benchmark threshold, generate a deviation difference set, and calculate the ratio between the maximum value in the deviation difference set and the upper limit of the benchmark threshold to generate an urgency weight parameter. S4: Read the remaining computing power cycle of the edge node and the expected execution time of the task to be executed, calculate the computing power matching coefficient, and calculate the task allocation index in combination with the urgency weight parameter, and sort the task allocation index to generate a task scheduling strategy. S5: Based on the task scheduling strategy, trigger the edge nodes to calculate the conductivity difference, heavy metal ratio, and dissolved oxygen reverse value to generate three deviations, write them into the deviation accumulation matrix, accumulate the deviation accumulation matrix and compare it with the set risk threshold to generate the pollution risk level.

[0007] As a further aspect of the present invention, the unit time fluctuation intensity parameter includes instantaneous disturbance characterization value, trend activity index and anomaly sensitivity coefficient; the target sampling frequency includes high frequency response level, low frequency energy saving level and adaptive adjustment level; the urgency weight parameter includes pollution diffusion risk level, severity of exceeding standard indicator and priority disposal level factor; the task scheduling strategy includes core task priority sequence, delayed task buffer sequence and resource reallocation scheme; and the pollution risk level includes safety status level, early warning status level and control status level.

[0008] As a further aspect of the present invention, the step of obtaining the unit time fluctuation intensity parameter specifically includes: S101: Obtain the groundwater sequence data of the edge node within the time window, perform time boundary parsing on the groundwater sequence data, locate each continuous sampling period included within it, extract the water level change corresponding to each sampling period, extract the time span corresponding to each sampling period as the sampling interval, map the water level change to the sampling interval according to the triggering order, and establish a sampling feature mapping set. S102: Call the water level change recorded in the sampling feature mapping set, perform scalar sign stripping on the water level change at the corresponding position of each sampling period, extract the absolute value of each water level change, perform arithmetic accumulation on the absolute value, obtain the arithmetic accumulation output result, and obtain the wave energy integral value. S103: Based on the sampling feature mapping set, read the sampling interval matching each sampling period, perform continuous numerical accumulation on each sampling interval, designate the accumulated output value as the standardized time base, assign the fluctuation energy integral value as the numerator operation parameter, assign the standardized time base as the denominator operation parameter, perform division arithmetic operation between the numerator operation parameter and the denominator operation parameter, extract the quotient data, and generate the unit time fluctuation intensity parameter.

[0009] As a further aspect of the present invention, the step of obtaining the target sampling frequency specifically includes: S201: Obtain the local node's preset hierarchical frequency mapping table, read the threshold interval corresponding to the continuous mapping level configured in the hierarchical frequency mapping table, perform numerical boundary extraction for the threshold interval, extract the upper and lower bound parameters of the threshold interval, combine the extracted upper and lower bound parameters into key-value pairs, and establish a frequency judgment benchmark set. S202: Call the unit time fluctuation intensity parameter, and synchronously perform size comparison judgment with the upper and lower bound parameters of the frequency judgment benchmark set. When the judgment parameter value is greater than the upper bound parameter, record the upper boundary trigger state. When the judgment parameter value is less than the lower bound parameter, record the lower boundary trigger state. Perform data integration on the upper boundary trigger state and the lower boundary trigger state to generate a frequency mapping matching state. S203: For the frequency mapping matching state, parse the recorded state type. When an upper boundary trigger state is identified, retrieve the corresponding first frequency in the hierarchical frequency mapping table. When a lower boundary trigger state is identified, retrieve the corresponding second frequency in the hierarchical frequency mapping table. Perform parameter value extraction on the retrieved first or second frequency and assign parameter values ​​to the hardware node register to obtain the target sampling frequency.

[0010] As a further aspect of the present invention, the process of reading the threshold interval corresponding to the continuous mapping level configured in the graded frequency mapping table and performing numerical boundary extraction for the threshold interval specifically involves: retrieving the historical water level fluctuation energy statistical sequence in the local storage state; sorting the historical water level fluctuation energy statistical sequence according to the ascending order to generate a cumulative energy distribution sequence; extracting the peak value of environmental noise energy fluctuation and configuring it as the initial reference value; performing a truncation and segmentation operation on the cumulative energy distribution sequence according to the set percentile step size to generate multiple energy distribution sub-intervals; configuring each energy distribution sub-interval as a continuous mapping level; reading the lower boundary value and upper boundary value of each energy distribution sub-interval; performing a summation operation on the initial reference value and the lower boundary value to generate a lower bound parameter; performing a summation operation on the initial reference value and the upper boundary value to generate an upper bound parameter; combining the lower bound parameter and the upper bound parameter to generate a threshold interval; locating the threshold interval matched with each continuous mapping level; performing data bit field parsing for the threshold interval to separate and extract the upper bound parameter and lower bound parameter contained within the threshold interval.

[0011] As a further aspect of the present invention, the step of obtaining the urgency weight parameter specifically includes: S301: Call the target sampling frequency to trigger the edge node sensor, collect the current cycle liquid level change, conductivity change and heavy metal concentration change, read the reference threshold corresponding to each variable in the device storage matrix, perform arithmetic subtraction between the liquid level change, conductivity change and heavy metal concentration change and the reference threshold respectively, extract the deviation value of each variable, perform key value aggregation mapping on each deviation value, and establish a deviation difference set; S302: Call the deviation values ​​recorded in the deviation difference set and extract the absolute value parameters. Input the absolute value parameters into the bubble sort algorithm. Perform descending sort by comparing the size of adjacent absolute value parameters and swapping array positions to construct a descending deviation sequence. Perform first-order index address positioning on the descending deviation sequence, extract the recorded parameter values ​​in the first-order address, and generate the first-order deviation difference. S303: Obtain the original indicator type associated with the first deviation difference, retrieve and extract the corresponding upper limit value parameter in the limit configuration table according to the original indicator type, configure the first deviation difference as the numerator parameter, configure the upper limit value parameter as the denominator parameter, perform the division arithmetic operation between the numerator parameter and the denominator parameter, and extract the quotient data to obtain the urgency weight parameter.

[0012] As a further aspect of the present invention, the step of obtaining the task scheduling strategy specifically includes: S401: Obtain the remaining computing power cycle of the edge node, extract the estimated execution time of the task to be executed, perform a division operation between the remaining computing power cycle and the estimated execution time, extract the quotient, and generate the computing power matching coefficient; S402: Call the urgency weight parameter and computing power matching coefficient, collect the node's remaining energy value and the task's rated energy consumption benchmark value, monitor the link packet loss rate, and calculate and obtain the task allocation index; S403: For each task to be executed, a one-dimensional feature sequence is established for the task allocation index associated with it. The one-dimensional feature sequence is loaded into a hardware comparator to perform numerical descending sorting, and the allocation identifier of the first and second data items is extracted to obtain the task scheduling strategy.

[0013] As a further aspect of the present invention, the step of obtaining the pollution risk level specifically includes: S501: Invoke the task scheduling strategy to trigger the control node to collect the current conductivity, current heavy metal concentration and current dissolved oxygen, read the preset benchmark limit, perform subtraction calculation between the current conductivity and the benchmark limit to extract the difference value, define it as the first deviation, perform division operation between the current heavy metal concentration and the control limit to extract the quotient value data, define it as the second deviation, perform subtraction calculation between the current dissolved oxygen and the minimum control value to extract the difference, perform sign reversal on the difference to extract the negative vector value, and generate the third deviation. S502: Read the deviation cumulative matrix, write the first deviation, the second deviation and the third deviation into the monitoring column of the deviation cumulative matrix, extract the recorded values ​​of three consecutive periods for the deviation cumulative matrix, perform horizontal row-by-row accumulation on the recorded values, extract the summation output value associated with each data row, and establish the matrix accumulation result; S503: Based on the matrix accumulation result, call the summation output value, extract the preset risk threshold of the environmental monitoring component, compare the summation output value with the set risk threshold execution value, count the total number of the summation output value that crosses the set risk threshold, and use the total number of the crossovers to perform key-value retrieval in the preset status level mapping table to extract the classification label and obtain the pollution risk level.

[0014] As a further aspect of the present invention, the method for setting the risk threshold specifically involves reading a locally stored dataset of historical water quality samples without abnormalities, extracting the extreme values ​​of normal conductivity fluctuations, the extreme values ​​of heavy metal concentration baselines, and the lower limit of dissolved oxygen from the historical sample dataset, substituting the extreme values ​​of normal conductivity fluctuations, the extreme values ​​of heavy metal concentration baselines, and the lower limit of dissolved oxygen into the deviation calculation and the continuous three-item cycle-by-line accumulation operation logic for synchronous processing, obtaining the limit accumulation output parameters reached by each environmental indicator under normal water quality conditions, performing arithmetic addition on the extracted multiple limit accumulation output parameters, extracting the scalar value obtained by the addition, and configuring it as the set risk threshold.

[0015] An edge computing-based groundwater pollution monitoring system, wherein the edge computing-based groundwater pollution monitoring system is used to implement the aforementioned edge computing-based groundwater pollution monitoring method, the system comprising: The fluctuation intensity analysis module acquires groundwater sequence data of edge nodes within a time window, accumulates the absolute change in water level to generate the fluctuation energy integral value, accumulates the sampling interval to generate a standardized time base, calculates the ratio of the fluctuation energy integral value to the standardized time base, and generates the fluctuation intensity parameter per unit time. The sampling frequency adjustment module compares the unit time fluctuation intensity parameter with a preset graded frequency mapping table, selects the corresponding sampling frequency, and generates the target sampling frequency. The weight dynamic adjustment module collects changes in liquid level, conductivity, and heavy metals at the target sampling frequency, calculates the difference between the data and the benchmark threshold, generates a deviation difference set, and calculates the ratio between the maximum value in the deviation difference set and the upper limit of the benchmark threshold to generate an urgency weight parameter. The real-time task allocation module reads the remaining computing power cycle of the edge node and the estimated execution time of the task to be executed, calculates and generates a computing power matching coefficient, and calculates a task allocation index in combination with the urgency weight parameter, and sorts the task allocation index to generate a task scheduling strategy. The pollution risk assessment module, based on the task scheduling strategy, triggers edge nodes to calculate the conductivity difference, heavy metal ratio, and dissolved oxygen reverse value to generate three deviations, writes them into the deviation accumulation matrix, accumulates the deviation accumulation matrix and compares it with the set risk threshold to generate a pollution risk level.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the fluctuation energy integral of groundwater sequence data is extracted and time reference is used to generate a unit time fluctuation intensity parameter. The sampling frequency is adaptively matched according to the state of intensity crossing the threshold interval, effectively eliminating the blind spots caused by fixed intervals and accurately capturing abrupt change signals. Simultaneously, the deviation characteristics of multi-dimensional indicators and the remaining computing power cycle of nodes are combined to construct a task allocation index and generate a scheduling strategy. High-frequency task dynamic allocation and computing power balance are completed at the edge of the data source, avoiding link congestion caused by redundant baseline data transmission. Based on the cumulative deviation matrix of continuous multi-cycle deviation, the pollution deterioration trend is tracked horizontally and the risk threshold is compared to realize the autonomous judgment and risk warning of environmental status in advance. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0018] Figure 1 This is a schematic diagram of the workflow of the present invention; Figure 2 This is a detailed flowchart of S1 of the present invention; Figure 3 This is a detailed flowchart of the S2 process of the present invention; Figure 4 This is a detailed flowchart of the S3 process of the present invention; Figure 5 This is a detailed flowchart of the S4 process of the present invention; Figure 6 This is a detailed flowchart of S5 of the present invention; Figure 7 This is a system flowchart of the present invention. Detailed Implementation

[0019] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0020] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0021] Please see Figure 1 This invention provides a technical solution: a groundwater pollution monitoring method based on edge computing, comprising the following steps: S1: Obtain groundwater sequence data of edge nodes within the time window and extract water level changes and sampling intervals for each sampling period. Take the absolute value of water level changes for each sampling period and perform cumulative calculation to generate the fluctuation energy integral value. Perform cumulative calculation on the sampling intervals for each sampling period to generate a standardized time reference. Calculate the ratio between the fluctuation energy integral value and the standardized time reference to generate the fluctuation intensity parameter per unit time. S2: Compare the unit time fluctuation intensity parameter with the threshold range configured in the graded frequency mapping table. When the unit time fluctuation intensity parameter is greater than the upper limit parameter of the threshold range, select the first frequency. When the unit time fluctuation intensity parameter is less than the lower limit parameter of the threshold range, select the second frequency. Extract the first frequency or the second frequency to generate the target sampling frequency. S3: Collect the changes in liquid level, conductivity, and heavy metal concentration in the current period at the edge nodes according to the target sampling frequency. Calculate the difference between the changes in liquid level, conductivity, and heavy metal concentration and the baseline threshold to generate a deviation difference set. Input the deviation difference set into the bubble sort algorithm for descending sorting and extract the first deviation difference. Calculate the ratio between the first deviation difference and the upper limit of the baseline threshold to generate an urgency weight parameter. S4: Read the remaining computing power cycle of the edge node and the expected execution time of the task to be executed. Calculate the ratio of the remaining computing power cycle to the expected execution time of the task to be executed to generate a computing power matching coefficient. Perform a product operation on the urgency weight parameter and the computing power matching coefficient to generate a task allocation index. Perform a descending sort operation on the task allocation index to generate a task scheduling strategy. S5: Based on the task scheduling strategy, trigger edge nodes to collect the current conductivity, current heavy metal concentration, and current dissolved oxygen. Calculate the difference between the current conductivity and the baseline limit to generate the first deviation. Calculate the ratio between the current heavy metal concentration and the control limit to generate the second deviation. Calculate the difference between the current dissolved oxygen and the minimum control value and take the negative vector value to generate the third deviation. Write the first, second, and third deviations into the deviation accumulation matrix. Perform a row-by-row accumulation operation on the recorded values ​​of three consecutive periods in the deviation accumulation matrix to generate the matrix accumulation result. Compare the matrix accumulation result with the set risk threshold to generate the pollution risk level.

[0022] The unit time fluctuation intensity parameter includes instantaneous disturbance characterization value, trend activity index and anomaly sensitivity coefficient; the target sampling frequency includes high frequency response level, low frequency energy saving level and adaptive adjustment level; the urgency weight parameter includes pollution diffusion risk level, severity of exceeding standard indicator and priority disposal level factor; the task scheduling strategy includes core task priority sequence, delayed task buffer sequence and resource reallocation scheme; and the pollution risk level includes safety status level, early warning status level and control status level.

[0023] Please see Figure 2 The specific steps for obtaining the fluctuation intensity parameter per unit time are as follows: S101: Obtain groundwater sequence data of edge nodes within the time window, perform time boundary parsing on the groundwater sequence data, locate each continuous sampling period included within it, extract the water level change corresponding to each sampling period, extract the time span corresponding to each sampling period as the sampling interval, map the water level change to the sampling interval according to the triggering order, and establish a sampling feature mapping set. Obtain the groundwater sequence data file from the edge nodes and read the raw pressure sensor readings within the current 24-hour time window recorded in the file. Perform null value cleaning on the raw reading sequence, traversing all data rows within the sequence. When an outlier field with empty values ​​is identified, extract the two immediately preceding and following valid readings, calculate the average by adding these two valid readings and dividing by 2, and then apply the average to the outlier field. Perform time boundary parsing on the cleaned groundwater sequence data, read the timestamp associated with each data record, extract the timestamps of two adjacent data records, perform arithmetic subtraction, locate continuous data segments with a time difference strictly within 5 minutes, and define each continuous data segment as a single continuous sampling period. Extract the ending water level height value corresponding to the last data record within each continuous sampling period, extract the beginning water level height value corresponding to the first data record within the period, and perform a mathematical calculation: water level change = ending water level height value - beginning water level height value. Selecting a specific continuous sampling period, the final water level is calculated as 15.6 meters, and the initial water level is 14.1 meters. Substituting these values ​​into the formula, the water level change is calculated as 15.6 - 14.1 = 1.5 meters. The end and start timestamps of this continuous sampling period are extracted, and a mathematical subtraction operation is performed to obtain the time span as the sampling interval. If the end timestamp is 10:45 and the start timestamp is 10:15, the extracted sampling interval is 30 minutes. Following the increasing trigger order of the timestamps, the 1.5-meter water level change is used as the key, and the 30-minute sampling interval is used as the key value to establish a sampling feature mapping set.

[0024] S102: Call the water level change recorded in the sampling feature mapping set, perform scalar sign stripping on the water level change at the corresponding position of each sampling period, extract the absolute value of each water level change, perform arithmetic accumulation on the absolute value, obtain the arithmetic accumulation output result, and obtain the wave energy integral value. The system retrieves all water level change data items recorded in the sampling feature map set, iterating through the water level changes at the corresponding positions in each consecutive sampling period. For water level changes with positive or negative numerical signs, a scalar sign stripping operation is performed to remove leading plus or minus signs, extracting the pure absolute value of each water level change. The register accumulator is then used to perform mathematical calculations on the extracted absolute values, resulting in an arithmetic sum: the sum of the first, second, and third absolute values, plus the sum of all subsequent absolute values. For example, water level changes of 1.5 meters, -0.8 meters, and 2.1 meters are selected for three consecutive sampling periods. After scalar sign stripping, the absolute values ​​are extracted as 1.5 meters, 0.8 meters, and 2.1 meters, respectively. Substituting these absolute values ​​into the arithmetic summation formula, the result is 1.5 + 0.8 + 2.1 = 4.4 meters. The calculated 4.4 meters is directly configured as the integral value of the fluctuation energy within this time window and stored in the local cache.

[0025] S103: Based on the sampling feature mapping set, read the sampling interval matched by each sampling period, perform continuous numerical accumulation on each sampling interval, specify the accumulated output value as the standardized time base, assign the fluctuation energy integral value as the numerator operation parameter, assign the standardized time base as the denominator operation parameter, perform the division arithmetic operation of the numerator operation parameter and the denominator operation parameter, extract the quotient data, and generate the unit time fluctuation intensity parameter. Based on the sampling feature mapping set residing in the buffer, the sampling interval value matching each continuous sampling period is traversed and read. The hardware accumulation controller is called to perform a continuous mathematical accumulation formula on each extracted sampling interval value. The accumulated output value = first sampling interval + second sampling interval + subsequent sampling intervals. The extracted three sampling interval values ​​are set to 30 minutes, 45 minutes, and 25 minutes respectively. Substituting these values ​​into the formula, the accumulated output value = 30 + 45 + 25 = 100 minutes. This 100 minutes is designated as the standardized time base for this time period. The 4.4-meter fluctuation energy integral value is retrieved from the local buffer and assigned as the numerator operation parameter. The 100-minute standardized time base is retrieved and assigned as the denominator operation parameter. The mathematical formula is executed, and the fluctuation intensity parameter per unit time = numerator operation parameter / denominator operation parameter. Substituting the numerator operation parameter 4.4 and the denominator operation parameter 100 into the formula, the fluctuation intensity parameter per unit time = 4.4 / 100 = 0.044 meters per minute. The extracted quotient data of 0.044 is directly generated and written into the global variable table of the main control chip as a parameter of fluctuation intensity per unit time.

[0026] Please see Figure 3 The specific steps for obtaining the target sampling frequency are as follows: S201: Obtain the local node's preset hierarchical frequency mapping table, read the threshold interval corresponding to the continuous mapping level configured in the hierarchical frequency mapping table, perform numerical boundary extraction for the threshold interval, extract the upper and lower bound parameters of the threshold interval, combine the extracted upper and lower bound parameters into key-value pairs, and establish a frequency judgment benchmark set. The process of reading the threshold intervals corresponding to the continuous mapping levels configured in the graded frequency mapping table and extracting the numerical boundaries of the threshold intervals is as follows: First, retrieve the historical water level fluctuation energy statistical sequence from local storage. Sort the historical water level fluctuation energy statistical sequence according to ascending order to generate a cumulative energy distribution sequence. Then, extract the peak value of environmental noise energy fluctuations and configure it as the initial baseline value. Next, perform a truncation and segmentation operation on the cumulative energy distribution sequence according to the set percentile step size to generate multiple energy distribution sub-intervals. Configure each energy distribution sub-interval as a continuous mapping level. Then, read the lower and upper boundary values ​​of each energy distribution sub-interval. Summate the initial baseline value and the lower boundary value to generate a lower bound parameter. Summate the initial baseline value and the upper boundary value to generate an upper bound parameter. Combine the lower and upper bound parameters to generate a threshold interval. Locate the threshold intervals that match each continuous mapping level. Finally, perform data bit field parsing on the threshold intervals to separate and extract the upper and lower bound parameters contained within the threshold intervals. Access the local node's preset hierarchical frequency mapping table file. The hierarchical frequency mapping table file is derived and set based on the cumulative distribution sequence of historical water level fluctuation energy and combined with the environmental noise benchmark value. Its mapping rules and internal technical logic are based on the comparison of the unit time fluctuation intensity parameter with the set threshold range and the out-of-bounds state, and dynamically and adaptively match the sampling frequency of high frequency or low frequency. Specifically, the derivation and setting method for the threshold interval corresponding to the continuous mapping level is as follows: Retrieve the historical water level fluctuation energy statistical sequence from local storage, sort the historical water level fluctuation energy statistical sequence according to ascending order, and generate a cumulative energy distribution sequence; extract the peak value of environmental noise energy fluctuation and configure it as the initial benchmark value; perform a truncation and segmentation operation on the cumulative energy distribution sequence according to the set percentile step size to generate multiple energy distribution sub-intervals, and configure each energy distribution sub-interval as a continuous mapping level; read the lower boundary value and upper boundary value of each energy distribution sub-interval, perform a summation operation between the initial benchmark value and the lower boundary value to generate a lower bound parameter, perform a summation operation between the initial benchmark value and the upper boundary value to generate an upper bound parameter, and combine the lower bound parameter and the upper bound parameter to generate a threshold interval; read the threshold interval data segment corresponding to the continuous mapping level configured in the graded frequency mapping table. Numerical boundary extraction is performed for the threshold range. Historical water level fluctuation energy statistics for the past 30 days, recorded on the local storage disk, are retrieved. A sorting algorithm is called to compare and determine the 500 energy statistics values ​​contained within the historical water level fluctuation energy statistics sequence. Position swapping and sorting are performed according to the ascending order of values ​​from smallest to largest, generating a cumulative energy distribution sequence containing 500 nodes. An environmental microphone sensor is driven to collect ambient sound wave frequencies, extracting the peak value of environmental noise energy fluctuation (1.2 dB) and setting it as the initial baseline value. The cumulative energy distribution sequence is truncated and segmented according to a set percentile step size of 25%, with truncation performed at the 125th, 250th, and 375th data nodes, generating four energy distribution sub-intervals. Each energy distribution sub-interval is directly configured as a continuous mapping level. Read the lower and upper bound values ​​of the four energy distribution sub-intervals. For the second energy distribution sub-interval, read its lower bound value as 3.5 and upper bound value as 6.8. Perform mathematical calculations: lower bound parameter = initial reference value + lower bound value; upper bound parameter = initial reference value + upper bound value. Substitute the initial reference value 1.2 with the lower bound value 3.5 and the upper bound value 6.8 into the calculations: lower bound parameter = 1.2 + 3.5 = 4.7; upper bound parameter = 1.2 + 6.8 = 8.0. Combine 4.7 and 8.0 to generate a threshold interval covering the range of 4.7 to 8.0. Locate the specific threshold interval associated with each continuous mapping level matching. Perform data bit field parsing operations on this threshold interval to separate and extract the upper bound parameter 8.0 and the lower bound parameter 4.7 contained within the threshold interval. Combine and package the extracted upper and lower bound parameters as key-value pairs to establish a frequency judgment reference set.

[0027] Table 1 Frequency Judgment Reference Set Mapping Table: ; As shown in Table 1, this process details the results of the combination of lower and upper bound parameters extracted from the four consecutive mapping levels.

[0028] S202: Call the unit time fluctuation intensity parameter, and synchronously compare the unit time fluctuation intensity parameter with the upper and lower bound parameters in the frequency judgment benchmark set. When the judgment parameter value is greater than the upper bound parameter, record the upper over-boundary trigger state. When the judgment parameter value is less than the lower bound parameter, record the lower over-boundary trigger state. Perform data integration on the upper and lower over-boundary trigger states to generate a frequency mapping matching state. The unit-time fluctuation intensity parameter, calculated as 0.044, is retrieved from the global variable table of the main control chip. This parameter is then synchronously compared with the upper bound parameter 8.0 and the lower bound parameter 4.7 recorded within the frequency judgment benchmark set in the hardware numerical comparator. A mathematical formula is executed: if the unit-time fluctuation intensity parameter is greater than the upper bound parameter, a 1 is written to the first flag bit of the status register to record the upper boundary trigger state; if it is less than the lower bound parameter, a 1 is written to the second flag bit of the status register to record the lower boundary trigger state. Since 0.044 is less than the lower bound parameter 4.7, the trigger condition is met, and the hardware comparator writes a 1 to the second flag bit to record the lower boundary trigger state. The upper and lower boundary trigger state values ​​recorded in the status register are then integrated and concatenated to generate a frequency mapping matching status bitmap representing the current intensity of hydrological fluctuations.

[0029] S203: For frequency mapping matching status, parse the recorded status type. When an upper out-of-bounds trigger status is identified, retrieve the corresponding first frequency in the hierarchical frequency mapping table. When a lower out-of-bounds trigger status is identified, retrieve the corresponding second frequency in the hierarchical frequency mapping table. Perform parameter value extraction on the retrieved first or second frequency and assign parameter values ​​to the hardware node register to obtain the target sampling frequency. For the generated frequency mapping matching state bitmap, a bitwise AND operation is performed to parse the recorded state type identifier. When the parsing identifies that the first flag bit of the state register is 1 (i.e., an upper out-of-bounds trigger state), a search command is sent to the hierarchical frequency mapping table to extract a preset sampling frequency of 100 times per second as the first frequency. When the parsing identifies that the second flag bit of the state register is 1 (i.e., a lower out-of-bounds trigger state), a search command is sent to the hierarchical frequency mapping table to extract a preset sampling frequency of 5 times per second as the second frequency. Based on the aforementioned identification results, the corresponding second frequency of 5 times per second is retrieved and extracted. The parameter value of the retrieved second frequency is extracted to obtain the pure number 5, and the number 5 is sent via the internal bus to the internal timer control register of the hardware node to perform a parameter assignment overwrite operation, completing the hardware configuration update and obtaining the final target sampling frequency.

[0030] Please see Figure 4 The specific steps for obtaining the urgency weight parameter are as follows: S301: Call the target sampling frequency to trigger the edge node sensor, collect the current cycle liquid level change, conductivity change and heavy metal concentration change, read the reference threshold corresponding to each variable in the device storage matrix, perform arithmetic subtraction between the liquid level change, conductivity change and heavy metal concentration change and the reference threshold respectively, extract the deviation value of each variable, perform key value aggregation mapping on each deviation value, and establish a deviation difference set; The target sampling frequency (5 times per second) written to the timer control register is invoked, generating a pulse width modulation square wave signal of the corresponding frequency and sending it to the edge node sensor array. The changes in liquid level, conductivity, and heavy metal concentration for the current monitoring period are collected. The baseline thresholds for each variable in the device's storage matrix are read, where the baseline threshold for liquid level is 0.5 meters, the baseline threshold for conductivity is 20 microsiemens per centimeter, and the baseline threshold for heavy metals is 0.02 milligrams per liter. A mathematical calculation is performed: deviation = actual collected change - baseline threshold. For example, if the collected liquid level change is 1.8 meters, the conductivity change is 55 microsiemens per centimeter, and the heavy metal concentration change is 0.08 milligrams per liter, the deviations are calculated as follows: liquid level deviation = 1.8 - 0.5 = 1.3 meters, conductivity deviation = 55 - 20 = 35 microsiemens per centimeter, and heavy metal deviation = 0.08 - 0.02 = 0.06 milligrams per liter. The three calculated deviation values ​​are prefixed with the original indicator names and then subjected to key-value aggregation mapping and packaging operations to generate a data sequence with a uniform format and establish a deviation difference set.

[0031] S302: Call the deviation values ​​recorded in the deviation difference set and extract the absolute value parameters. Input the absolute value parameters into the bubble sort algorithm. Perform descending sort by comparing the size of adjacent absolute value parameters and swapping array positions to construct a descending deviation sequence. Perform first-order index address location for the descending deviation sequence, extract the parameter values ​​recorded in the first-order address, and generate the first-order deviation difference. Access the deviation difference set, retrieve the internal records for liquid level deviation (1.3), conductivity deviation (35), and heavy metal deviation (0.06), strip away the positive and negative data formatting, and extract the absolute values ​​of these three values ​​as 1.3, 35, and 0.06, respectively. Input these three absolute value parameters into the bubble sort algorithm's operation logic, setting the outer loop count to the total number of parameters minus 1 (2 times), and the inner loop to compare the values ​​of adjacent parameters. In the first round of comparison, the first parameter 1.3 is compared with the second parameter 35. Since 1.3 is less than 35, an array position swap is triggered, changing the sequence to 35, 1.3, 0.06. The second parameter 1.3 is then compared with the third parameter 0.06. Since 1.3 is greater than 0.06, no swap is performed, and the first round of comparison ends. In the second round of comparison, the first parameter 35 is extracted and compared with the second parameter 1.3. Since 35 is greater than 1.3, no swap is performed, the algorithm terminates, and a descending deviation sequence is constructed with the order 35, 1.3, 0.06. A pointer positioning instruction is issued for the descending deviation sequence, performing first-order index address positioning, i.e., locating to the memory address with array index 0. The parameter value 35 recorded at this first-order address is extracted, defined, and the first-order deviation difference is generated.

[0032] S303: Obtain the original indicator type associated with the first deviation difference, retrieve and extract the corresponding upper limit value parameter in the limit configuration table according to the original indicator type, configure the first deviation difference as the numerator parameter, configure the upper limit value parameter as the denominator parameter, perform the division arithmetic operation between the numerator parameter and the denominator parameter, and extract the quotient value data to obtain the urgency weight parameter. The original index type associated with the first deviation difference of 35 is obtained, namely conductivity. Using this conductivity index type string as the primary key, an exact match search is performed within the limit configuration table to locate and extract the corresponding upper limit parameter of conductivity, which is preset to 100 microSiemens per centimeter. The first deviation difference of 35 is configured as the numerator parameter of the division operation, and the extracted upper limit parameter 100 is configured as the denominator parameter. The hardware arithmetic logic unit is called to perform the mathematical expression operation, where the urgency weight parameter = numerator parameter / denominator parameter. Substituting the numerator parameter 35 and the denominator parameter 100 into the expression operation, the urgency weight parameter = 35 / 100 = 0.35. The quotient value of 0.35 is extracted and stored in the scheduling cache register to obtain the urgency weight parameter.

[0033] Please see Figure 5 The specific steps for obtaining the task scheduling strategy are as follows: S401: Obtain the remaining computing power cycle of the edge node, extract the estimated execution time of the task to be executed, perform a division operation between the remaining computing power cycle and the estimated execution time, extract the quotient, and generate the computing power matching coefficient; The system status register of the edge node's central processing unit is read to obtain the current remaining computing power cycles, which is calculated to be 4500 idle clock ticks. The task description file is retrieved to extract the estimated execution time of the task, which is marked as 1500 clock units. The arithmetic logic unit is invoked to perform a mathematical calculation: computing power matching coefficient = remaining computing power cycles / estimated execution time. Substituting the remaining computing power cycles of 4500 and the estimated execution time of 1500 into the calculation, the computing power matching coefficient = 4500 / 1500 = 3.0. The resulting quotient value of 3.0 is extracted and written into the global variable scheduling table to generate the computing power matching coefficient.

[0034] S402: Invoke the urgency weight parameter and computing power matching coefficient, collect the node's remaining energy value and the task's rated energy consumption baseline value, monitor the link packet loss rate, and use the formula: ; Calculate and obtain the task allocation index; in, Represents the task allocation index. The urgency weight parameter is obtained by calling the ratio of the first deviation difference to the upper limit value. The computing power matching coefficient is obtained by calling the quotient of the remaining computing power cycle and the expected execution time. The normalized value representing the remaining energy of the node is obtained by reading the power data from the hardware controller and performing an extreme value division transformation. The normalized value representing the baseline rated power consumption of the task is obtained by extracting the power consumption field from the configuration file and performing an extreme value division transformation. It represents the packet loss rate of the link, and is obtained by calculating the ratio of the number of lost data packets to the total number of data packets sent within the communication interface. The urgency weight parameter of 0.35 stored in the register is matched with the computing power coefficient of 3.0. The power management chip interface of the hardware controller is read to extract the power data as 800 mAh, and the battery's full-load rated power data is read as 1000 mAh. An extreme value division conversion operation is performed, and the normalized value of the node's remaining energy value is 800 / 1000 = 0.8. The power consumption field reading in the task configuration file is extracted to 120 mW, and the hardware maximum power consumption limit value is read as 300 mW. An extreme value division conversion operation is performed, and the normalized value of the task's rated energy consumption baseline is 120 / 300 = 0.4. The network communication interface status table is accessed, and the number of lost data packets in the past 10 minutes is 50, and the total number of sent data packets is 1000. A ratio calculation is performed to obtain the link packet loss rate: 50 / 1000 = 0.05. Mathematical calculations are then performed. First, an internal division operation is performed on the absolute value. The normalized value of the node's remaining energy (0.8) is divided by the normalized value of the task's rated energy consumption baseline (0.4), resulting in a quotient of 2.0. An internal subtraction operation is then performed, subtracting the link packet loss rate (0.05) from 2.0, resulting in a difference of 1.95. The absolute value of 1.95 remains 1.95. A hardware square root operation is then performed on 1.95, yielding 1.396. A pre-multiplication operation is then performed, multiplying the urgency weight parameter (0.35) by the computing power matching coefficient (3.0), resulting in a product of 1.05. Finally, a final addition operation is performed: the task allocation index = 1.05 + 1.396 = 2.446. The final task allocation index of 2.446 is obtained and saved.

[0035] S403: For each task to be executed, a one-dimensional feature sequence is established based on the task allocation index associated with it. The one-dimensional feature sequence is then loaded into a hardware comparator to perform numerical sorting in descending order. The allocation identifier of the first and second data items is extracted to obtain the task scheduling strategy. For the four task allocation indices associated with the tasks to be executed stored in the queue, which are 2.446, 1.855, 3.102, and 0.958 respectively, a one-dimensional feature sequence is established. This one-dimensional feature sequence is loaded into a hardware comparator via a direct memory access channel. The hardware comparator uses internal hard-wired logic to perform numerical sorting in descending order, and the reassembled sequence order is 3.102, 2.446, 1.855, and 0.958. The allocation identifier associated with the first data item 3.102 after reassembly, i.e., task number 3, is extracted and written into the scheduling instruction delivery channel to obtain the task scheduling policy.

[0036] Table 2: Ranking Table of Task Allocation Index ; As shown in Table 2, this process lists the task allocation index and corresponding sorting position of the four tasks to be executed after being sorted in descending order by the hardware comparator.

[0037] Please see Figure 6 The specific steps for obtaining the pollution risk level are as follows: S501: Invoke the task scheduling strategy, trigger the control node to collect the current conductivity, current heavy metal concentration and current dissolved oxygen, read the preset benchmark limit, perform subtraction calculation between the current conductivity and the benchmark limit to extract the difference value, define it as the first deviation, perform division operation between the current heavy metal concentration and the control limit to extract the quotient data, define it as the second deviation, perform subtraction calculation between the current dissolved oxygen and the minimum control value to extract the difference, perform sign reversal on the difference to extract the negative vector value, and generate the third deviation. The task scheduling strategy is invoked to trigger the working pins of the environmental sensors at the control node's underlying layer, collecting data on the current water conductivity (150 μSiemens per centimeter), heavy metal concentration (0.05 mg / L), and dissolved oxygen (4.5 mg / L). Preset baseline limits are read, with a conductivity baseline limit of 100 μSiemens per centimeter, a heavy metal control limit of 0.02 mg / L, and a minimum dissolved oxygen control value of 5.0 mg / L. Mathematical calculations are performed: first deviation = current conductivity - baseline limit; second deviation = current heavy metal concentration / control limit; third deviation = current dissolved oxygen - minimum control value. The difference in the third deviation is then inverted to extract the negative vector value. Substitute the actual collected values ​​into the formula for calculation. The first deviation is 150-100=50 microSiemens per centimeter, the second deviation is 0.05 / 0.02=2.5, and the third deviation preprocessing difference is 4.5-5.0=-0.5. Perform a sign reversal operation on the negative 0.5 to extract the positive number 0.5 and generate the final third deviation.

[0038] S502: Read the deviation cumulative matrix, write the first deviation, the second deviation and the third deviation into the monitoring column of the deviation cumulative matrix, extract the recorded values ​​of three consecutive periods for the deviation cumulative matrix, perform horizontal row-by-row accumulation on the recorded values, extract the summation output value associated with each data row, and establish the matrix accumulation result; The system reads a 3x3 deviation cumulative matrix space allocated in system memory. The newly generated first deviation (50), second deviation (2.5), and third deviation (0.5) are written as new data columns into the rightmost monitoring column of the deviation cumulative matrix, while the leftmost expired data column is removed. Historical data covering three consecutive periods are extracted from the deviation cumulative matrix, with data rows corresponding to conductivity, heavy metal, and dissolved oxygen deviations, respectively. A horizontal, row-by-row summation mathematical formula is performed on the recorded values. The summation output value for each data row is equal to the sum of the first, second, and third period values. For the first deviation data row, the three period values ​​are 40, 45, and 50, and the summation output value is 40 + 45 + 50 = 135. For the second deviation data row, the three period values ​​are 1.5, 2.0, and 2.5, and the summation output value is 1.5 + 2.0 + 2.5 = 6.0. For the third deviation data row, the three period values ​​are 0.2, 0.3, and 0.5 respectively. Substituting these values ​​into the formula and calculating the sum, the output value is 0.2 + 0.3 + 0.5 = 1.0. The sums of these three values ​​(135, 6.0, and 1.0) are combined to construct a one-dimensional array containing three elements, and the matrix is ​​then accumulated.

[0039] S503: Based on the matrix accumulation result, call the summation output value, extract the preset risk threshold of the environmental monitoring component, compare the summation output value with the execution value of the preset risk threshold, count the total number of the summation output value that crosses the set risk threshold, and use the total number of the crossover to perform key-value retrieval in the preset state level mapping table to extract the classification label and obtain the pollution risk level. The specific method for setting risk thresholds is as follows: read the local storage of historical water quality sample datasets without abnormalities, extract the extreme values ​​of normal conductivity fluctuations, the extreme values ​​of heavy metal concentration baselines, and the lower limit of dissolved oxygen from the historical sample datasets, substitute the extreme values ​​of normal conductivity fluctuations, the extreme values ​​of heavy metal concentration baselines, and the lower limit of dissolved oxygen into the deviation calculation and the continuous three-item cycle row-by-row accumulation operation logic to perform synchronous processing, obtain the limit accumulation output parameters reached by each environmental indicator under normal water quality conditions, perform arithmetic addition on the extracted multiple limit accumulation output parameters, extract the scalar value obtained by addition and configure it as the set risk threshold; Based on the matrix accumulation result array, the internal summation output values ​​for the three terms are 135, 6.0, and 1.0. The set risk threshold array stored in the environmental monitoring component is extracted. This threshold array contains three fixed indicator limits: 120, 5.0, and 0.8. The summation output value is compared with the corresponding element of the set risk threshold using a comparator. Since 135 is greater than 120, 6.0 is greater than 5.0, and 1.0 is greater than 0.8, the total number of times the summation output value crosses the set risk threshold is counted as 3. Using the total number of crosses of 3, a key-value search is performed in the preset status level mapping table to extract the classification label. The classification label corresponding to the total number of crosses of 3 is extracted as a red alert, thus obtaining the specific pollution risk level. The derivation logic for the set risk threshold is as follows: read the locally stored historical water quality sample dataset without abnormalities, and iterate through all normal sample data. The extreme values ​​of conductivity fluctuations were extracted from historical sample datasets: 25 μSiemens per centimeter, the extreme value of heavy metal concentration baseline was 0.01 mg / L, and the lower limit of dissolved oxygen was 0.2 mg / L. These three extreme parameters were substituted into the same deviation calculation and continuous three-cycle cumulative summation formula as in S501 to S502, simulating the maintenance of this extreme state for three cycles. The cumulative output parameters of conductivity limits were 25 + 25 + 25 = 75, heavy metal limits were 0.01 + 0.01 + 0.01 = 0.03, and dissolved oxygen limits were 0.2 + 0.2 + 0.2 = 0.6. These three cumulative output parameters were then mathematically added together, and the risk threshold scalar value was set as: first parameter + second parameter + third parameter. Substituting the extreme value calculation results, the risk threshold scalar value was set as: 75 + 0.03 + 0.6 = 75.63. Configure and save the scalar value of 75.63 and the limit summation value of each individual parameter directly as the risk threshold parameter for global use in the system.

[0040] Table 3: Status Level Mapping Table ; As shown in Table 3, this table clearly shows the pollution risk level labels corresponding to the total number of boundary violations and the control strategies for subsequent hardware responses, verifying the logical accuracy of retrieving the red alert classification label by the total number of boundary violations (3).

[0041] Please see Figure 7 An edge computing-based groundwater pollution monitoring system is used to execute the aforementioned edge computing-based groundwater pollution monitoring method. The system includes: The fluctuation intensity analysis module acquires groundwater sequence data of edge nodes within a time window, accumulates the absolute change in water level to generate the fluctuation energy integral value, accumulates the sampling interval to generate a standardized time base, calculates the ratio of the fluctuation energy integral value to the standardized time base, and generates the fluctuation intensity parameter per unit time. The sampling frequency adjustment module compares the unit time fluctuation intensity parameter with the preset graded frequency mapping table, selects the corresponding sampling frequency, and generates the target sampling frequency. The weight dynamic adjustment module collects changes in liquid level, conductivity, and heavy metals at the target sampling frequency, calculates the difference between the data and the benchmark threshold, generates a deviation difference set, and calculates the ratio between the maximum value in the deviation difference set and the upper limit of the benchmark threshold to generate an urgency weight parameter. The real-time task allocation module reads the remaining computing power cycle of the edge nodes and the estimated execution time of the tasks to be executed, calculates and generates a computing power matching coefficient, and calculates a task allocation index in combination with the urgency weight parameter, and sorts the task allocation indexes to generate a task scheduling strategy. The pollution risk assessment module, based on the task scheduling strategy, triggers edge nodes to calculate the conductivity difference, heavy metal ratio, and dissolved oxygen reverse value to generate three deviations, writes them into the deviation accumulation matrix, accumulates the deviation accumulation matrix and compares it with the set risk threshold to generate the pollution risk level.

[0042] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0043] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0044] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0045] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0046] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0047] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0048] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0049] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0050] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0051] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of protection of the described technical solutions.

Claims

1. A groundwater pollution monitoring method based on edge computing, characterized in that, Includes the following steps: S1: Obtain groundwater sequence data of edge nodes within the time window, accumulate the absolute change in water level to generate the fluctuation energy integral value, accumulate the sampling interval to generate the standardized time base, calculate the ratio of the fluctuation energy integral value to the standardized time base, and generate the fluctuation intensity parameter per unit time. S2: Compare the unit time fluctuation intensity parameter with the preset graded frequency mapping table, select the corresponding sampling frequency, and generate the target sampling frequency; S3: Collect changes in liquid level, conductivity, and heavy metals at the target sampling frequency, calculate the difference with the benchmark threshold, generate a deviation difference set, and calculate the ratio between the maximum value in the deviation difference set and the upper limit of the benchmark threshold to generate an urgency weight parameter. S4: Read the remaining computing power cycle of the edge node and the expected execution time of the task to be executed, calculate the computing power matching coefficient, and calculate the task allocation index in combination with the urgency weight parameter, and sort the task allocation index to generate a task scheduling strategy. S5: Based on the task scheduling strategy, trigger the edge nodes to calculate the conductivity difference, heavy metal ratio, and dissolved oxygen reverse value to generate three deviations, write them into the deviation accumulation matrix, accumulate the deviation accumulation matrix and compare it with the set risk threshold to generate the pollution risk level.

2. The groundwater pollution monitoring method based on edge computing according to claim 1, characterized in that, The unit time fluctuation intensity parameter includes instantaneous disturbance characterization value, trend activity index and anomaly sensitivity coefficient; the target sampling frequency includes high frequency response level, low frequency energy saving level and adaptive adjustment level; the urgency weight parameter includes pollution diffusion risk level, severity of exceeding standard indicator and priority disposal level factor; the task scheduling strategy includes core task priority sequence, delayed task buffer sequence and resource reallocation scheme; the pollution risk level includes safety status level, early warning status level and control status level.

3. The groundwater pollution monitoring method based on edge computing according to claim 1, characterized in that, The specific steps for obtaining the fluctuation intensity parameter per unit time are as follows: S101: Obtain the groundwater sequence data of the edge node within the time window, perform time boundary parsing on the groundwater sequence data, locate each continuous sampling period included within it, extract the water level change corresponding to each sampling period, extract the time span corresponding to each sampling period as the sampling interval, map the water level change to the sampling interval according to the triggering order, and establish a sampling feature mapping set. S102: Call the water level change recorded in the sampling feature mapping set, perform scalar sign stripping on the water level change at the corresponding position of each sampling period, extract the absolute value of each water level change, perform arithmetic accumulation on the absolute value, obtain the arithmetic accumulation output result, and obtain the wave energy integral value. S103: Based on the sampling feature mapping set, read the sampling interval matching each sampling period, perform continuous numerical accumulation on each sampling interval, designate the accumulated output value as the standardized time base, assign the fluctuation energy integral value as the numerator operation parameter, assign the standardized time base as the denominator operation parameter, perform division arithmetic operation between the numerator operation parameter and the denominator operation parameter, extract the quotient data, and generate the unit time fluctuation intensity parameter.

4. The groundwater pollution monitoring method based on edge computing according to claim 3, characterized in that, The specific steps for obtaining the target sampling frequency are as follows: S201: Obtain the local node's preset hierarchical frequency mapping table, read the threshold interval corresponding to the continuous mapping level configured in the hierarchical frequency mapping table, perform numerical boundary extraction for the threshold interval, extract the upper and lower bound parameters of the threshold interval, combine the extracted upper and lower bound parameters into key-value pairs, and establish a frequency judgment benchmark set. S202: Call the unit time fluctuation intensity parameter, and synchronously perform size comparison judgment with the upper and lower bound parameters of the frequency judgment benchmark set. When the judgment parameter value is greater than the upper bound parameter, record the upper boundary trigger state. When the judgment parameter value is less than the lower bound parameter, record the lower boundary trigger state. Perform data integration on the upper boundary trigger state and the lower boundary trigger state to generate a frequency mapping matching state. S203: For the frequency mapping matching state, parse the recorded state type. When an upper boundary trigger state is identified, retrieve the corresponding first frequency in the hierarchical frequency mapping table. When a lower boundary trigger state is identified, retrieve the corresponding second frequency in the hierarchical frequency mapping table. Perform parameter value extraction on the retrieved first or second frequency and assign parameter values ​​to the hardware node register to obtain the target sampling frequency.

5. The groundwater pollution monitoring method based on edge computing according to claim 4, characterized in that, The process of reading the threshold interval corresponding to the continuous mapping level configured in the graded frequency mapping table and extracting the numerical boundary of the threshold interval specifically involves: retrieving the historical water level fluctuation energy statistical sequence in local storage, sorting the historical water level fluctuation energy statistical sequence according to ascending order to generate a cumulative energy distribution sequence; extracting the peak value of environmental noise energy fluctuation and configuring it as the initial reference value; performing a truncation and segmentation operation on the cumulative energy distribution sequence according to a set percentile step size to generate multiple energy distribution sub-intervals, configuring each energy distribution sub-interval as a continuous mapping level; reading the lower boundary value and upper boundary value of each energy distribution sub-interval, performing a summation operation between the initial reference value and the lower boundary value to generate a lower bound parameter, performing a summation operation between the initial reference value and the upper boundary value to generate an upper bound parameter, and combining the lower bound parameter and the upper bound parameter to generate a threshold interval; locating the threshold interval matched with each continuous mapping level; and performing data bit field parsing on the threshold interval to separate and extract the upper bound parameter and lower bound parameter contained within the threshold interval.

6. The groundwater pollution monitoring method based on edge computing according to claim 4, characterized in that, The specific steps for obtaining the urgency weight parameter are as follows: S301: Call the target sampling frequency to trigger the edge node sensor, collect the current cycle liquid level change, conductivity change and heavy metal concentration change, read the reference threshold corresponding to each variable in the device storage matrix, perform arithmetic subtraction between the liquid level change, conductivity change and heavy metal concentration change and the reference threshold respectively, extract the deviation value of each variable, perform key value aggregation mapping on each deviation value, and establish a deviation difference set; S302: Call the deviation values ​​recorded in the deviation difference set and extract the absolute value parameters. Input the absolute value parameters into the bubble sort algorithm. Perform descending sort by comparing the size of adjacent absolute value parameters and swapping array positions to construct a descending deviation sequence. Perform first-order index address positioning on the descending deviation sequence, extract the recorded parameter values ​​in the first-order address, and generate the first-order deviation difference. S303: Obtain the original indicator type associated with the first deviation difference, retrieve and extract the corresponding upper limit value parameter in the limit configuration table according to the original indicator type, configure the first deviation difference as the numerator parameter, configure the upper limit value parameter as the denominator parameter, perform the division arithmetic operation between the numerator parameter and the denominator parameter, and extract the quotient data to obtain the urgency weight parameter.

7. The groundwater pollution monitoring method based on edge computing according to claim 6, characterized in that, The steps for obtaining the task scheduling strategy are as follows: S401: Obtain the remaining computing power cycle of the edge node, extract the estimated execution time of the task to be executed, perform a division operation between the remaining computing power cycle and the estimated execution time, extract the quotient, and generate the computing power matching coefficient; S402: Call the urgency weight parameter and computing power matching coefficient, collect the node's remaining energy value and the task's rated energy consumption benchmark value, monitor the link packet loss rate, and calculate and obtain the task allocation index; S403: For each task to be executed, a one-dimensional feature sequence is established for the task allocation index associated with it. The one-dimensional feature sequence is loaded into a hardware comparator to perform numerical descending sorting, and the allocation identifier of the first and second data items is extracted to obtain the task scheduling strategy.

8. The groundwater pollution monitoring method based on edge computing according to claim 7, characterized in that, The specific steps for obtaining the pollution risk level are as follows: S501: Invoke the task scheduling strategy to trigger the control node to collect the current conductivity, current heavy metal concentration and current dissolved oxygen, read the preset benchmark limit, perform subtraction calculation between the current conductivity and the benchmark limit to extract the difference value, define it as the first deviation, perform division operation between the current heavy metal concentration and the control limit to extract the quotient value data, define it as the second deviation, perform subtraction calculation between the current dissolved oxygen and the minimum control value to extract the difference, perform sign reversal on the difference to extract the negative vector value, and generate the third deviation. S502: Read the deviation cumulative matrix, write the first deviation, the second deviation and the third deviation into the monitoring column of the deviation cumulative matrix, extract the recorded values ​​of three consecutive periods for the deviation cumulative matrix, perform horizontal row-by-row accumulation on the recorded values, extract the summation output value associated with each data row, and establish the matrix accumulation result; S503: Based on the matrix accumulation result, call the summation output value, extract the preset risk threshold of the environmental monitoring component, compare the summation output value with the set risk threshold execution value, count the total number of the summation output value that crosses the set risk threshold, and use the total number of the crossovers to perform key-value retrieval in the preset status level mapping table to extract the classification label and obtain the pollution risk level.

9. The groundwater pollution monitoring method based on edge computing according to claim 8, characterized in that, The specific method for setting the risk threshold is as follows: read the locally stored historical sample dataset of water quality without abnormalities, extract the extreme values ​​of normal fluctuations in conductivity, the extreme values ​​of baseline heavy metal concentration, and the lower limit of allowable dissolved oxygen from the historical sample dataset, substitute the extreme values ​​of normal fluctuations in conductivity, the extreme values ​​of baseline heavy metal concentration, and the lower limit of allowable dissolved oxygen into the deviation calculation and the continuous three-item cycle row-by-row accumulation operation logic to perform synchronous processing, obtain the limit accumulation output parameters reached by each environmental indicator under normal water quality conditions, perform arithmetic addition on the extracted multiple limit accumulation output parameters, extract the scalar value obtained by addition and configure it as the set risk threshold.

10. A groundwater pollution monitoring system based on edge computing, characterized in that, The system is used to implement the groundwater pollution monitoring method based on edge computing as described in any one of claims 1-9, and the system comprises: The fluctuation intensity analysis module acquires groundwater sequence data of edge nodes within a time window, accumulates the absolute change in water level to generate the fluctuation energy integral value, accumulates the sampling interval to generate a standardized time base, calculates the ratio of the fluctuation energy integral value to the standardized time base, and generates the fluctuation intensity parameter per unit time. The sampling frequency adjustment module compares the unit time fluctuation intensity parameter with a preset graded frequency mapping table, selects the corresponding sampling frequency, and generates the target sampling frequency. The weight dynamic adjustment module collects changes in liquid level, conductivity, and heavy metals at the target sampling frequency, calculates the difference between the data and the benchmark threshold, generates a deviation difference set, and calculates the ratio between the maximum value in the deviation difference set and the upper limit of the benchmark threshold to generate an urgency weight parameter. The real-time task allocation module reads the remaining computing power cycle of the edge node and the estimated execution time of the task to be executed, calculates and generates a computing power matching coefficient, and calculates a task allocation index in combination with the urgency weight parameter, and sorts the task allocation indexes to generate a task scheduling strategy. The pollution risk assessment module, based on the task scheduling strategy, triggers edge nodes to calculate the conductivity difference, heavy metal ratio, and dissolved oxygen reverse value to generate three deviations, writes them into the deviation accumulation matrix, accumulates the deviation accumulation matrix and compares it with the set risk threshold to generate a pollution risk level.

Images