A pollution control method and system driven by environmental variables

Abstract

The present invention relates to the technical field of environmental monitoring data processing, and discloses a pollution control method and system driven by environmental variables. The method comprises: converting the numerical changes of environmental variables into single-bit fluctuation symbols at edge nodes, calculating the temporal collaborative correlation of multi-parameter fluctuation symbol sequences at regional nodes, and generating an instability index based on the weighted correlation and triggering a predictive control instruction on a central platform. The present invention realizes the extraction of system instability trends from the fluctuation direction of environmental variables through the collaborative mechanism of fluctuation symbolization and bit operation, breaking through the lag of traditional threshold monitoring in responding to complex pollution. At the same time, the system is combined with disturbance marking and inertia correction algorithm to enable the system to have both slow-changing risk prediction and sudden disturbance response capabilities in seconds, significantly improving the accuracy and timeliness of pollution control.

Description

Technical Field

[0001] The present invention relates to a pollution control method and system driven by environmental variables, belonging to the technical field of environmental monitoring data processing. Background Art

[0002] In the field of environmental monitoring and pollution control, data processing systems usually rely on fixed thresholds to trigger early warning mechanisms. The core logic is to determine whether a single environmental variable (such as PM2.5 concentration, pH value) exceeds the preset limit. Although such methods can respond to significant pollution events, they are difficult to capture hidden risks caused by the coordinated fluctuations of multiple variables. Taking industrial parks as an example, when the air humidity continues to be higher than 85%, the fan speed is reduced to 70% due to power-saving strategies, and the groundwater level rises to a high level due to rainfall, although each parameter is within the independent threshold range, their synergistic effect may cause complex pollution such as accelerated acid mist deposition and heavy metal infiltration into groundwater.

[0003] Specifically, existing technologies have the following fundamental limitations: 1. They rely on single-parameter threshold monitoring and ignore the dynamic synergistic effects of multiple variables, resulting in a complete blindness to compliant but unstable systemic risks; 2. They can only passively respond to excessive pollution incidents that have already occurred and lack the ability to predict the precursors of pollution, which doubles the cost of governance; 3. If the accuracy is improved by increasing the number of sensors or increasing the sampling frequency, it will lead to a surge in data transmission and computing power requirements, making it difficult to deploy on a large scale on low-cost edge nodes.

[0004] Some existing improvements attempt to incorporate multivariate correlation analysis, but these still require transmitting raw data to the cloud for processing. The communication latency and computational complexity of these solutions make them inadequate for the second-by-second response required for sudden pollution incidents. Therefore, the technical challenge addressed by this invention is how to achieve early identification and predictive intervention of coordinated fluctuations in multiple environmental variables. Summary of the Invention

[0005] The present invention provides an environmental variable-driven pollution control method, the main purpose of which is to solve the problem that traditional monitoring systems cannot capture the hidden pollution risks caused by the coordinated fluctuations of multiple environmental variables due to their reliance on single-parameter threshold judgment, and the existing improvement schemes have the problems of delayed response and high deployment costs.

[0006] To achieve the above objectives, the present invention provides an environmental variable driven pollution control method, comprising the following steps:

[0007] Step a: Periodically monitor the values ​​of environmental variables at multiple edge computing nodes, and compare the current values ​​of the environmental variables with the values ​​of the previous cycle to generate a single-bit fluctuation symbol that represents the fluctuation direction of the environmental variables, wherein the fluctuation symbol includes a first bit state and a second bit state. When the current value is greater than the value of the previous cycle, the first bit state is generated; when the current value is less than or equal to the value of the previous cycle, the second bit state is generated.

[0008] Step b: At the regional aggregation node, the fluctuation symbols sent by multiple edge computing nodes are received, and for a determined pair of environmental parameters, the temporal coordination correlation between the corresponding fluctuation symbol sequences is calculated. The temporal coordination correlation reflects the coordination of the fluctuation direction of the parameter pair within a fixed-length time window.

[0009] Step c: On the central platform, the time series coordination correlation reported by the regional aggregation nodes is received. Based on the weights determined to reflect the degree of influence of different parameters on the pollution formation process, an instability index representing the overall state of the environmental system is calculated. The instability index is the weighted sum of the time series coordination correlations.

[0010] In step d, when the value of the instability index exceeds the set warning threshold, the system triggers a pollution risk warning and automatically issues pollution control instructions based on the control strategy configured by the system. The pollution control instructions include changing the operating state of at least one environmental variable to block the fluctuation synergy conditions that cause the instability index to exceed the warning threshold.

[0011] Preferably, the calculation of the timing synergy correlation degree in step b includes performing a bitwise XOR operation on the fluctuation symbol bit sequence corresponding to the two parameters in the environmental parameter pair, and determining the timing synergy correlation degree based on the percentage of the number of bits with a bit value of zero to the total length of the sequence in the XOR operation result. The higher the percentage, the greater the synergy correlation degree. The edge computing node additionally calculates the absolute value of the difference between the current value of the environmental variable and the value of the previous cycle. If the absolute value of the difference exceeds a certain mutation threshold, a burst disturbance mark is generated and sent to the regional convergence node together with the fluctuation symbol; and, when the regional convergence node receives any burst disturbance mark, it immediately shortens the calculation period of the instability index to less than one second, and switches the calculation method of the instability index to ,in, is the risk value calculated instantaneously based on the latest data, The inertia coefficient is greater than or equal to 0.9 and less than 1.

[0012] Preferably, when the regional convergence node does not receive any sudden disturbance mark within one minute, the system restores the calculation period of the instability index to more than one minute, and switches the calculation method of the instability index to a conventional prediction mode.

[0013] Preferably, the edge computing node utilizes the network time protocol or beacon timing function to obtain a high-precision synchronized clock, and when generating a wave symbol, appends an event occurrence timestamp provided by the synchronized clock to the wave symbol.

[0014] Preferably, the regional aggregation node caches the fluctuation symbol flow in a short period of time, and when it detects that two or more spatially adjacent nodes monitoring the same environmental variable have successively reported the same fluctuation symbol, it calculates the corresponding timestamp difference; and, based on the timestamp difference and the known spatial positions between adjacent nodes, determines the migration direction of the environmental variable change in space.

[0015] Preferably, the determined migration direction is used as a weighting factor and input into the instability index calculation model of the central platform, thereby increasing the risk weight of the downstream area of ​​the migration path.

[0016] Preferably, when the regional aggregation node collects and processes the fluctuation symbol sequence, it calculates the average consistency ratio of the fluctuation symbol sequence of any node in a group of spatially adjacent edge computing nodes monitoring the same environmental variable and the fluctuation symbol sequences of all other nodes in the group; and when the average consistency ratio is lower than 0.5, the regional aggregation node generates diagnostic information indicating a fault of any node and reports it to the central platform.

[0017] Preferably, the diagnostic information is used to drive targeted maintenance of the faulty edge computing node.

[0018] Preferably, the environmental variables include relative humidity of the air, fan speed of the production equipment, and underground water level in the park, and the pollution control instructions include forcibly increasing the fan power according to the instability index.

[0019] A pollution control system driven by environmental variables, the system comprising:

[0020] Multiple edge computing units are used to periodically monitor the values ​​of environmental variables and compare the current values ​​of the environmental variables with the values ​​of the previous cycle to generate a single-bit fluctuation symbol that represents the fluctuation direction of the environmental variables, wherein the fluctuation symbol includes a first bit state and a second bit state. When the current value is greater than the value of the previous cycle, the first bit state is generated; when the current value is less than or equal to the value of the previous cycle, the second bit state is generated; at least one regional convergence unit is used to receive the fluctuation symbols sent by the multiple edge computing units, and for a determined pair of environmental parameters, calculate the time series coordination correlation between the corresponding fluctuation symbol sequences, wherein the time series coordination correlation reflects the coordination of the fluctuation direction of the parameter pair within a time window of a fixed length;

[0021] A central processing unit is used to receive the time series synergy correlation reported by each regional convergence unit, and calculate the instability index representing the overall state of the environmental system based on the determined weights reflecting the degree of influence of different parameters on the pollution formation process, where the instability index is the weighted sum of the time series synergy correlation;

[0022] A governance instruction issuing unit is used to trigger a pollution risk warning when the value of the instability index exceeds the set warning threshold, and automatically issue pollution control instructions according to the governance strategy configured by the system. The pollution control instructions include changing the operating state of at least one environmental variable to block the fluctuation coordination conditions that cause the instability index to exceed the warning threshold.

[0023] Compared with the prior art, the present invention has the following beneficial effects:

[0024] 1. Compress the numerical changes of environmental variables into single-bit fluctuation symbols at the edge nodes. Combined with the bitwise XOR operation of the symbol sequence by the regional nodes, the system only needs extremely low computing resources to capture the temporal coordination of multi-parameter fluctuations. This dimensionality reduction processing of information dimensions and the coordination of lightweight computing mechanisms make it possible to identify the instability trend of the environmental system no longer rely on breaking through the physical quantity threshold. Instead, it can perceive the risk formation conditions in advance through the coordinated correlation of the fluctuation direction, thus avoiding the delayed response of traditional monitoring methods to hidden complex pollution from the root.

[0025] 2. The detection of sudden changes in amplitude by edge nodes triggers disturbance marking, which works in synergy with the mechanism of regional nodes using the historical attenuation inertia algorithm to dynamically correct the instability index, enabling the system to autonomously switch between conventional prediction and emergency response modes. When a sudden disturbance occurs, the system uses the inertia coefficient to smoothly transition the risk value calculation, maintaining the ability to respond to impact events in seconds while avoiding command jitter caused by transient noise. This allows a single system to simultaneously cope with two completely different operating scenarios: slowly changing risk evolution and sudden pollution events.

[0026] 3. The precise timestamp generated by the timing function of the communication protocol is attached to the fluctuation symbol. Combined with the phase analysis of the time difference of the same-direction symbols of adjacent nodes by the regional node, the system can reconstruct the migration vector of environmental variables in space without adding dedicated sensors. This mechanism expands the risk perception dimension from intensity scalar to direction vector by mining the spatiotemporal correlation information in the existing bit stream, providing a precise spatial blocking path for governance instructions, and significantly improving the efficiency of intervention in responding to the spread of cross-media pollution.

[0027] 4. When calculating the collaborative correlation degree, regional nodes simultaneously perform cross-correlation analysis of the fluctuation symbols of similar nodes. By comparing the symbol consistency ratio of spatially adjacent nodes for the same environmental parameter, the abnormal data source is automatically located. This mechanism reuses the existing data processing flow to realize equipment status monitoring, forming the active diagnosis capability of faulty nodes at a low cost, blocking the decision distortion caused by sensor failure at the source of the data, and ensuring the reliability and self-maintenance of the system in long-term operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a schematic diagram of the three-level distributed structure flow of the pollution control method and system driven by environmental variables of the present invention;

[0029] Figure 2 Schematic diagram comparing the dynamic change of the instability index of the system of the present invention and the warning timeliness of the traditional monitoring system;

[0030] Figure 3 This is a timing diagram of the process of determining the spatial migration direction of environmental variable changes and prioritizing the triggering of pollution control instructions in the present invention.

[0031] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0032] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0033] The embodiment of the present application discloses a pollution control method and system driven by environmental variables, the overall architecture of which is designed as a three-level distributed intelligent system, including multiple edge computing units deployed at the monitoring site to perform real-time perception and preliminary processing of data sources; at least one regional aggregation unit to complete the fusion and correlation analysis of multi-dimensional information within the jurisdiction; and a central processing unit to conduct global situation analysis and final decision-making. Under this architecture, the complete operation process of the method begins with the edge computing node's periodic sampling of the physical world environmental variables, and generates single-bit fluctuation symbols representing the fluctuation direction through its internal procedures. These symbol streams are then transmitted to the regional aggregation node, where the multi-parameter time series collaborative correlation is calculated. Finally, the correlation data calculated by each regional aggregation node is uniformly reported to the central platform for calculating the instability index representing the overall state of the environmental system, and based on the comparison result of the index with the preset warning threshold, the control instruction issuing unit triggers the predictive pollution control instruction.

[0034] In the practice of complex industrial environment monitoring, a key technical bottleneck is how to effectively extract early signals that indicate systemic risks from massive amounts of noisy raw sensor data at extremely low computational and transmission costs. To meet this challenge, the method disclosed in the present invention configures a set of efficient data dimensionality reduction procedures at the front end of data acquisition, namely the edge computing node. The node samples the monitored environmental variables, such as relative humidity of the air, fan speed of production equipment, and groundwater level in the park, at a certain period, for example, once per second, and compares the value obtained in the current period with the value of the previous period stored in the node memory. If the current value is greater than the value of the previous period, the node samples the value of the monitored environmental variables, such as relative humidity of the air, fan speed of production equipment, and groundwater level in the park, and compares the value obtained in the current period with the value of the previous period stored in the node memory. The processor in the node generates and outputs a first bit state representing an upward trend, such as a binary value of 1; conversely, if the current value is less than or equal to the value of the previous cycle, a second bit state representing a downward or stable trend is generated, such as a binary value of 0. In this way, continuous floating-point environmental variable values ​​that usually occupy multiple bytes of storage space are converted in real time into a single-bit fluctuation symbol sequence that only occupies a single bit but accurately retains the core information of its change direction. This processing not only sharply reduces the data transmission volume by several orders of magnitude, but also fundamentally unifies the analysis benchmarks of parameters of different physical dimensions, laying a solid foundation for subsequent cross-parameter collaborative analysis.

[0035] Traditional monitoring methods, due to their adherence to the judgment of the absolute value of the physical quantity of a single parameter, are often completely blind to the complex pollution risks formed by the coordinated evolution of multiple parameters within their respective safety thresholds. In view of this, the core of this method is reflected in the temporal collaborative correlation calculation logic performed by the regional aggregation node. The node receives and caches the fluctuation symbol bit sequence sent by multiple edge computing nodes under its jurisdiction, and for the predetermined environmental parameter pairs with potential physical or chemical correlation in the system, such as air relative humidity and fan speed, within a fixed-length time window, for example, a sequence containing 60 consecutive symbols, a bit-wise XOR operation is performed on their corresponding fluctuation symbol bit sequences. Given the mathematical property of the XOR operation that the result is 0 when the two bits are the same and 1 when they are different, the proportion of 0 in the operation result sequence can directly and quantitatively reflect the coordination of the fluctuation directions of the two parameters within the time window. The regional aggregation node calculates the ratio of the number of bits with zero value in the XOR operation result to the percentage of the total sequence length. By calculating the time series synergy correlation ratio, a time series synergy correlation between 0 and 1 can be obtained without any complex floating-point operations. The higher the percentage, the more consistent or completely synchronously opposite the fluctuation directions of the two environmental variables, and the greater the synergy correlation. This reveals the system instability trend hidden behind independent compliance data with extremely high computational efficiency. The final assessment and intervention of systemic risk requires integrating the correlation trends of various local parameter pairs into a unified indicator that can macro-represent the global stability state and trigger precise governance actions accordingly. To this end, after aggregating multiple time series synergy correlations reported by regional aggregation nodes, the central platform calculates the final instability index based on a preset weight model. The weight coefficients in this model are obtained through offline regression analysis or mechanism modeling of historical data of specific pollution scenarios. They are used to reflect the degree of influence of different parameters on the specific pollution formation process. The instability index is determined as the weighted sum of each time series synergy correlation and its corresponding weight: ,in For the The temporal co-correlation of parameter pairs, For its corresponding weight, when the real-time calculated value of the instability index continues for a period of time, such as 30 seconds, and exceeds a warning threshold set based on risk tolerance, the system will automatically trigger a pollution risk warning and issue one or more pollution control instructions based on the preset control strategies in the configuration library. For example, if the main reason for the instability index exceeding the limit is the high coordination between the increase in humidity and the speed reduction of the fan, the instruction will forcibly change the operating state of at least one environmental variable, such as increasing the fan power to a predetermined level, thereby directly blocking or weakening the fluctuation coordination conditions that cause instability from the root, and realizing predictive control of pollution risks.

[0036] In the specific deployment of the present invention, the generation and configuration of its core parameters follow a complete set of data-driven deterministic procedures. First, for the historical monitoring data set containing multi-dimensional environmental variables in a specific application scenario, by calculating the Pearson correlation coefficient between each variable and generating a correlation matrix, the parameter pairs with an absolute value of the correlation coefficient higher than the preset statistical significance level, such as 0.6, are screened out to form the collaborative correlation target set to be analyzed; secondly, for each parameter pair in this set, its instability index calculation model is The corresponding weight coefficient , is achieved by taking an objective function The output obtained by optimizing the solution is defined as ,in is the weight vector to be found, and are the mean values ​​of the instability index calculated based on the quasi-pollution condition and normal condition time windows marked in the historical data, is the standard deviation of the instability index under normal working conditions, and In order to prevent the denominator from being zero, the optimization process uses the simulated annealing algorithm to find the optimal solution within the preset number of iterations. The weight vector with the largest value , thereby transforming the determination of the weight coefficient into a standardized operating process driven by raw data, with a unique optimization goal and a deterministic optimization path; the warning threshold and inertia coefficient in the system It is also uniquely determined by its preceding, self-consistent calculation procedure, where the warning threshold is the weight coefficient After being determined by the above procedures, it is set by performing receiver operating characteristic analysis on the historical instability index curve. During the analysis, the corresponding true positive rate TPR and false positive rate FPR are calculated by traversing all possible thresholds, and the ROC curve is drawn. Finally, the instability index value corresponding to the point where the Youden index J=TPR-FPR reaches the maximum value is selected as the warning threshold; and the emergency response model The inertia coefficient in , whose value comes from fitting an ideal response model, which is defined as the unit step response of a critically damped second-order system , where the natural frequency It is a performance indicator preset according to the response speed requirements of specific scenarios, such as 1rad / s. The value is in the range of 0.9 to 1, and the unique solution is searched by minimizing the root mean square error between the actual response curve and the ideal response curve, thereby ensuring that all core parameters of the system are derived from the inevitable deduction of internal logic rather than relying on external experience settings. The weight coefficient The least squares regression operation is performed on the feature matrix generated by the historical monitoring data set containing environmental variables such as relative air humidity, fan speed, groundwater level and park air pressure. Each column of the feature matrix corresponds to a time series vector formed by sampling a single variable 24 hours a day for 30 consecutive days. The regression operation aims to maximize the difference between the mean of the instability index in the marked quasi-pollution condition window and the mean of the instability index in the normal condition window and minimize the standard deviation. The regression coefficient is solved as the weight by QR decomposition. The warning threshold is determined by calculating the 90th percentile value of the empirical distribution function of the historical instability index series and taking this value as the threshold. The inertia coefficient The value range is between 0.92 and 0.97. By injecting emergency simulation data into the instability index calculation process, the instability index is able to change from the initial value to the 90th percentile value within 3 seconds after the event is triggered without overshoot fluctuations, and the optimal value is taken from the monotonic curve fitting result. The starting point of this process is the original physical quantity sequence monitored by the database and is completed in a closed loop through executable calculation steps such as matrix generation, regression solution, percentile value selection and dynamic response curve fitting.

[0037] A robust industrial-grade environmental governance system must have the ability to predict slowly changing cumulative risks and respond to sudden disturbance events in seconds. In order to achieve this dual-modal response characteristic within a unified framework, this method adds a disturbance detection function to the edge computing node. When performing periodic comparisons, the node additionally calculates the absolute value of the difference between the current value and the value of the previous period, and compares it with a mutation threshold determined for the variable. The calibration process of this mutation threshold has a deterministic procedure: at the initial stage of system deployment, collect at least 24 hours of continuous historical data of the environmental variable and calculate the standard deviation of its value change rate , and set the mutation threshold to a specific multiple of the standard deviation, i.e. ,in This coefficient can be adjusted between 3 and 5 according to the sensitivity requirements of the application scenario. When the absolute value of the difference exceeds this threshold, the edge node generates a sudden disturbance flag and sends it to the regional aggregation node together with the fluctuation symbol of the period. Once the regional aggregation node receives the sudden disturbance flag from any node under its jurisdiction, it immediately switches the calculation period of the instability index from the conventional one minute or more to the emergency one second. At the same time, its calculation method is also switched to an emergency model based on inertia correction: ,in, It is the risk value calculated instantaneously based on the data of the latest cycle. is the instability index value before switching, and As an inertia coefficient greater than or equal to 0.9 and less than 1, its specific value is determined through fitting analysis of historical emergency data, aiming to ensure that the risk index can smoothly transition while quickly responding to shocks, and avoid violent jitters in governance instructions caused by instantaneous data noise. Accordingly, when the regional aggregation node does not receive any sudden disturbance mark within one consecutive minute, the system will automatically return to the normal prediction mode, that is, the calculation period of the instability index will be restored to more than one minute, and it will switch back to the conventional weighted calculation method based on the time window. This deterministic, data-marker-driven mode switching mechanism allows the system to switch autonomously and smoothly between the two working conditions without human intervention, ensuring adaptability and reliability in all scenarios.

[0038] In order to further deal with spatial migration problems such as the spread of pollution clusters, this method reuses the time synchronization function of the existing communication network. The edge computing node is configured to use the network time protocol or beacon timing function to obtain a globally unified high-precision synchronized clock, and when generating each fluctuation symbol, it adds an event occurrence timestamp provided by the synchronized clock with millisecond accuracy. When the regional aggregation node processes the symbol stream, it will continuously monitor a group of nodes that are spatially adjacent and monitor the same environmental variables. Once it detects that two or more such nodes have reported the same fluctuation symbol in a very short period of time, it will immediately calculate the timestamp difference corresponding to these symbols, and then combine it with the timestamp difference of these adjacent nodes pre-stored in the system. The precise spatial position coordinates between nodes can be used to determine the migration direction and speed of the environmental variable changes in space in real time through vector operations. This determined migration direction information is then reported as a dynamic weighting factor and input into the instability index calculation model of the central platform. Its role is to systematically increase the risk weights of other relevant parameters in the downstream area of ​​the migration path, so that governance instructions can be more predictably and preferentially directed to the areas that are about to be affected, realizing a strategic upgrade from passive interception to active blocking. The setting of the mutation threshold is to generate a vector based on the absolute value of the difference between the values ​​of each two adjacent sampling points in the time series of the corresponding environmental variable with a continuous period of 1 second for 24 hours, and the standard deviation of the vector is calculated. and multiply by a fixed coefficient Then the threshold coefficient The value is 4 and remains unchanged. The humidity variable, fan speed, and groundwater level correspond to thresholds of 4 respectively. Humidity, 4 Speed, 4 Water level, the threshold is used by the edge computing node to calculate the absolute value of the difference between the current sampling value and the previous sampling value in each cycle. The disturbance mark is compared to generate a disturbance mark. The disturbance mark indicates a mutation trigger with a binary flag bit 1 and is uploaded to the regional aggregation node at the same time as the periodic fluctuation symbol sequence. After receiving the disturbance mark for the first time, the regional aggregation node immediately switches the instability index calculation period to 1 second and executes the inertia correction mode. When no disturbance mark is received for 60 consecutive seconds, it automatically restores the normal calculation mode with a period of 60 seconds. This process is implemented in the form of a fixed sampling period, standard deviation operation and multiplier amplification, and direct comparison and judgment to generate a binary mark, without any external parameter dependence and can be executed completely in a closed loop within the system.

[0039] In order to ensure the data credibility and decision reliability of the entire system during long-term operation, an endogenous device health status self-diagnosis mechanism is integrated into the data processing flow of the regional aggregation node. While collecting and processing the fluctuation symbol sequence, the node will periodically perform consistency analysis on a group of spatially adjacent edge computing nodes that monitor the same environmental variables. The specific procedure is: in each analysis cycle, the fluctuation symbol sequence of any node in the group is used as the benchmark, and its consistency ratio with the fluctuation symbol sequence of all other nodes in the group is calculated respectively. The calculation method of this ratio is the same as the aforementioned time series collaborative correlation, that is, it is obtained by counting the proportion of zero values ​​after bitwise XOR. Subsequently, the arithmetic mean of these consistency ratios is calculated. If the average consistency ratio is constant over multiple consecutive cycles, for example, Within 5 cycles, it was stably below the logical threshold of 0.5, which statistically means that the fluctuation behavior of this node is significantly and non-randomly opposite or unrelated to the vast majority of similar nodes in the surrounding area. The regional aggregation node then generates a diagnostic information indicating that the benchmark node may have a fault or data anomaly, and reports this information together with evidence such as the node ID and anomaly timestamp to the central platform. After receiving this diagnostic information, the central platform can automatically trigger the work order system to drive maintenance personnel to conduct targeted inspections and maintenance on the faulty edge computing node. This design reuses the existing data processing process and builds a self-purification and repair barrier for the system at a low incremental cost, ensuring that the decision-making basis of the entire predictive governance system is solid and reliable from the source of the data.

[0040] Example 1: In a continuously operating large-scale chemical park, the method of the present invention is deployed to monitor and prevent complex pollution risks caused by the synergistic effects of multiple environmental factors. The challenge of this application scenario does not come from the drastic exceedance of a single parameter, but rather from the fact that multiple environmental variables evolve together in a hidden manner within their own independent safety thresholds, thereby accumulating to form a systemic instability risk when the traditional monitoring system is completely oblivious. In the early summer of the park, a district-wide energy-saving and consumption-reduction policy led to a general reduction in the fan speed of the exhaust systems of hundreds of production equipment. At the same time, seasonal rainfall caused the park's groundwater level to reach its annual low. At a high level, although the current values ​​of various environmental variables such as relative humidity, fan speed and groundwater level in the park monitored and reported by each edge computing node do not exceed the safety thresholds set independently, the traditional threshold-based monitoring system therefore determines that the environmental status is normal. However, a hidden risk driven by the collaboration of multiple variables is quietly forming. Under this working condition, the method of the present invention does not judge whether the absolute value of any physical quantity is compliant in isolation, but compresses the continuous environmental variables into a single-bit fluctuation symbol representing the trend in the information dimension, thereby transforming the core monitoring issue from static numerical limit violations to a quantitative assessment of the dynamic collaborative relationship of multiple variables.

[0041] In specific operations, edge computing nodes throughout the park continuously convert the numerical changes of various environmental variables into single-bit fluctuating symbol streams. Among them, the symbol sequence of relative humidity generally presents the first bit state representing an upward trend, while the symbol sequence of fan speed continues to show the second bit state representing a downward trend. Here, the efficient data dimension reduction procedures of edge nodes and the time series collaborative correlation calculation logic of regional nodes form a significant synergy. The former provides the latter with unified, refined and high-frequency analysis materials, while the latter converts the original information of the former into risk insights with clear physical meanings by performing bit-wise XOR operations on these symbol streams. The regional convergence nodes Based on this, it was calculated that the temporal synergistic correlation between the two sets of fluctuating symbol sequences, air relative humidity and fan speed, was significantly high. This quantitative result clearly revealed that the two key variables were co-evolving in a direction that was conducive to the accumulation and transformation of pollutants. After receiving the above-mentioned high-synergistic correlation data, the central platform calculated the instability index representing the overall state of the environmental system based on the preset weights and concluded that it had exceeded the set warning threshold. The system immediately triggered a pollution risk warning and automatically issued a pollution control instruction with a clear target. This instruction bypassed the conventional energy-saving operation strategy and forcibly increased the fan power operation status in a specific area to the preset risk blocking mode.

[0042] Furthermore, when responding to a short thunderstorm, the system demonstrated its ability to resolve core technical contradictions within a single architecture. In the field of environmental monitoring, there has long been an inherent conflict between sensitivity and reliability. That is, high sensitivity can easily lead to misjudgment of instantaneous noise, while high reliability may sacrifice the speed of response to early risks. During this thunderstorm, the relative humidity of the air monitored by an edge computing node increased sharply due to direct splashing of rain. The absolute value of the difference between its current value and the value of the previous cycle exceeded the mutation threshold set according to the deterministic procedure, thereby generating a sudden disturbance mark. The system immediately switched to the sudden disturbance mark determined by the formula The emergency response mode defined here is a complementary mechanism between the generation mechanism of the sudden disturbance marker and the emergency algorithm based on the inertia coefficient. The former ensures the system's ability to respond to emergencies in seconds and achieves sensitivity; while the latter uses an inertia coefficient greater than or equal to 0.9. Giving historical data a very high weight makes the final instability index Although there was an increase in response, this brief and unrepresentative data shock did not immediately trigger unnecessary global governance instructions, thus ensuring the robustness and reliability of decision-making and achieving the unity of seemingly opposing goals. This intervention broke the synergistic conditions of high humidity and low wind speed fluctuations that led to the instability index exceeding the limit. As the fan power increased, local air circulation was enhanced, and accumulated moisture and potential acid gases were effectively dissipated. Correspondingly, the synergistic correlation between relative humidity and fan speed decreased rapidly, and the instability index also fell back to a safe range. The risk evolution path was effectively blocked before a complex pollution event actually occurred and caused equipment corrosion or soil contamination. Its environmental safety protection does not come from setting increasingly stringent static thresholds for a single physical quantity, but depends on a deep understanding and active management of the dynamic relationships between the multiple elements that constitute the environmental system. Through a closed-loop governance system that predicts and blocks the conditions for risk formation, the present invention achieves a significant improvement in the timeliness and accuracy of pollution control.

[0043] Example 2: In order to objectively quantify the advantage of the method of the present invention in terms of timeliness of early warning over the traditional threshold monitoring mechanism in dealing with slowly changing composite pollution risks, this verification test is specially designed and executed. The test relies on a high-fidelity industrial environment simulation platform, which can accurately simulate the dynamic evolution of multiple key environmental variables including relative humidity of air, fan speed of production equipment and groundwater level in the park, and deploys two monitoring systems in parallel on the platform: System A, as the control group, adopts the traditional single-parameter independent threshold alarm mechanism, and the alarm threshold of each parameter is set to 95% of the industry-standard safety limit; System B, as the experimental group, fully realizes the disclosed method of the present invention, based on edge computing, regional aggregation and center To deal with the pollution control method driven by environmental variables in the three-level architecture, in order to ensure the engineering authenticity of the experiment, the core parameter settings in the experiment follow a rigorous decision-making logic. Among them, the sampling period of the edge node is set to 1 second. This decision aims to balance the real-time nature of the data and the system processing load to ensure that the second-level variable fluctuations can be captured without excessively increasing the consumption of computing resources; the time window length used by the regional aggregation node to calculate the time series collaborative correlation is determined to be 60 sampling points, that is, one minute. The fundamental consideration of this setting is the technical trade-off between statistical reliability and response sensitivity, aiming to smooth out the interference of high-frequency random noise while ensuring sufficient recognition sensitivity for continuous collaborative trends; the weight coefficient used by the central platform to calculate the instability index It is calibrated through regression analysis of thousands of historical pollution event data in the simulation platform to ensure that it can reflect the actual contribution of different parameters to the specific pollution formation process to the greatest extent.

[0044] After the test process was initiated, the simulation platform, following a preset script, drove the values ​​of three key environmental variables to evolve slowly and collaboratively from an initial stable state to a critically unstable state within 600 seconds. During this process, the absolute values ​​of each parameter were always controlled below the independent alarm threshold set by System A. During the test, the state outputs and key intermediate variables of the two systems were synchronously recorded. Table 1 shows data snapshots at some key moments during the test process.

[0045] Table 1: Snapshot of key data during the test.

[0046]

[0047] The test results are shown in Table 1. During the first 480 seconds of the test, System A remained silent and alert-free, failing to reveal potential system risks, as no single environmental variable reached its limit. In stark contrast, System B's instability index continued to rise steadily as the synergistic trend among the variables strengthened. The underlying mechanism lies in the fact that the regional aggregation node, through bitwise XOR operations on the fluctuating symbol sequences, quantified the dynamic relationship between the coevolution of environmental variables into a continuously increasing temporal synergistic correlation. This metric, serving as the core input, caused the central platform's instability index to exceed the set warning threshold of 0.80 at the 480th second, 120 seconds (two minutes) before any physical quantity transgression. This triggered a predictive pollution control command. Following the command, the fan speed was controlled to increase, disrupting the original synergistic conditions and causing the instability index to rapidly decline, effectively preventing further risk development. It was not until the 600th second that System A issued an alarm due to the relative humidity finally exceeding the 90% threshold.

[0048] Example 3: This example combines Figures 1 to 3 , an environmental variable driven pollution control method and system are described, such as Figure 1 As shown, it includes an edge computing layer, a regional convergence layer, and a central platform layer. The edge computing layer is equipped with edge node 1 (air humidity monitoring), edge node 2 (fan speed monitoring), and edge node 3 (groundwater level monitoring). Each edge node periodically monitors environmental variables and generates a fluctuation symbol (0 / 1). When the current value is greater than the previous cycle value, 1 is generated, and when the current value is less than or equal to the previous cycle value, 0 is generated. The fluctuation symbol (0 / 1) is sent to the regional convergence node of the regional convergence layer. The regional convergence node performs time series collaborative correlation calculation (exclusive OR operation) and simultaneously performs sudden disturbance detection to determine whether to enter the emergency mode: The regional convergence node sends the calculation results to the central processing unit of the core platform layer, which calculates the instability index. The calculation formula is: If the instability index is greater than the warning threshold, the pollution control instruction issuing unit will be triggered to execute the pollution control instruction to block the spread of instability risk caused by coordinated fluctuations.

[0049] like Figure 2As shown, the figure contains three curves, representing the upstream unit local threshold, the downstream unit local threshold and the downstream unit neighborhood risk accumulation value respectively. The horizontal axis is time (minutes), ranging from 0 to 50 minutes, indicating the duration of the experiment. The vertical axis is the risk value, ranging from 0 to 1.0, indicating the risk level calculated by the system. It can be seen from the figure that in the first 25 minutes of the experiment, although the risk values ​​of the upstream unit local threshold and the downstream unit local threshold were at a low level and rose slowly, the downstream unit neighborhood risk accumulation value continued to rise and reached a peak of about 0.9 at about 25 minutes; then, as time went on, although the risk values ​​of the upstream unit local threshold and the downstream unit local threshold continued to rise slowly, the downstream unit neighborhood risk accumulation value began to decline, which shows that the system of the present invention can identify potential pollution risks earlier by accumulating neighborhood risks, thereby realizing predictive governance.

[0050] like Figure 3 As shown, Figure 3 The figure shows the processing process between edge node 1 (upstream), edge node 2 (midstream), edge node 3 (downstream), regional aggregation node, central platform, and governance instruction unit when monitoring changes in the same environmental variable at spatially adjacent nodes, including spatially adjacent nodes monitoring the same environmental variable, detecting an increase in environmental variable changes, sending symbol 1 + timestamp T1, detecting interval changes, sending symbol 1 + timestamp T2, detecting interval changes, sending symbol 1 + timestamp T3, regional aggregation node starts monitoring migration patterns, identifies the same symbols of adjacent nodes, calculates time differences ΔT12=T2-T1, ΔT23=T3-T2, combines the spatial position of the nodes, determines the migration direction and speed, reports migration vector information, and the central platform uses the migration direction as a weighting factor to increase the risk weight of the downstream area, recalculates the instability index, triggers the governance instruction unit to execute the downstream priority governance command, prompts downstream pollution interception, prioritizes blocking in the downstream, coordinates midstream scheduling, actively blocks pollution spread, and triggers predictive governance to complete a complete closed-loop governance process for downstream pollution interception.

[0051] Example 4: In a semiconductor manufacturing cleanroom environment that is sensitive to particulate and molecular contaminants, the first step of the procedure is to target the core components of the instability index calculation model in the central platform, that is, the weights reflecting the influence of different parameters on the degree of influence. To perform data-driven calibration, engineers injected historical environmental monitoring data accumulated over a period of time in the cleanroom. This data includes temperature and humidity, pressure difference, specific chemical consumption rate, and product yield fluctuation data at the corresponding time. It covers conditions ranging from completely normal to several recorded quasi-pollution events caused by multiple factors. The data is injected into the central platform. Subsequently, the system launches an iterative optimization algorithm whose objective function is to maximize the instability index within the time window of the quasi-pollution event. The peak and mean values ​​of the instability index during normal production period are minimized. Through this optimization process, the system automatically adjusts the weights of each parameter pair. , until the objective function converges, and finally outputs a set of weight configurations that can best distinguish normal operating conditions from abnormal precursors. In this way, the concept of regression analysis is materialized into an operational process with clear input, optimization goals and deterministic output.

[0052] In weight After being determined, the procedure enters the next step, which is to set the warning threshold of the instability index. The system uses the calibrated weight model to recalculate all historical data and generate a historical instability index curve corresponding to time. Based on this curve, the receiver operating characteristic curve analysis method is used to determine the warning threshold. Specifically, by using different instability index values ​​as potential warning thresholds, the system calculates and draws a two-dimensional curve consisting of the proportion of correct warnings of quasi-contamination events, namely the true positive rate, and the proportion of incorrect warnings of normal working conditions, namely the false positive rate, at each threshold. The inflection point on the receiver operating characteristic curve, that is, the position closest to the coordinate point in the upper left corner of the chart, and its corresponding instability index value statistically represents the best balance between warning sensitivity and specificity. The system sets this value as the warning threshold specifically for the clean room, thereby transforming the consideration of setting based on risk tolerance into a quantifiable, objective decision-making process that pursues optimal classification efficiency.

[0053] Furthermore, for the emergency response model used in the system to deal with emergencies , the inertia coefficient of its core The calibration also follows a deterministic procedure. Engineers inject a series of simulated step-type shock signals with different rise rates and peak values ​​into the simulation environment and set an ideal system response curve as a benchmark. This curve is defined as the step-rate response of a critically damped second-order system that can respond fastest and without overshoot. Subsequently, the system is minimized. The root mean square error between the actual response curve and the ideal response curve is the optimization target, and the algorithm automatically searches and determines This value ensures that when the system encounters sudden disturbances, the process of raising its risk index can not only respond in seconds, but also effectively suppress overshoot and oscillation, ensuring the smoothness and stability of governance instructions.

[0054] Example 5: In an application scenario where the system of the present invention is already running stably, when it is necessary to add or replace any edge computing node, the new node will not be immediately included in the global instability index calculation system, but will first enter a 24-hour observation and calibration mode. During this period, the fluctuation symbol data stream of the new node is received by the regional aggregation node, but is only used for data comparison with a group of calibrated companion nodes that are spatially adjacent to the node and monitor the same environmental variables. The regional aggregation node automatically calculates and generates a set of initial calibration coefficients for the new node by analyzing the average numerical deviation between the new node and the companion node group and the consistency ratio of the fluctuation symbol sequence during the observation period. In subsequent routine operation, the original data reported by the new node will be corrected by this calibration coefficient before generating the fluctuation symbol. This pre-procedure ensures that the data output characteristics of any newly connected hardware can be consistent with the existing network, and eliminates the systematic decision-making deviation introduced by hardware differences from the root.

[0055] Furthermore, in order to deal with the problem of slow performance drift that may occur in sensors during long-term use, the regional aggregation node, while performing its regular time-series collaborative correlation calculation task, also simultaneously performs a continuous and forward-looking health status monitoring task. It not only generates immediate fault diagnosis information when the average consistency ratio of the fluctuation symbol sequence of any node and the rest of the nodes in the same group is lower than 50%, but also tracks and records the changing trend of the consistency ratio of each node in the past few weeks or even months. If the system detects through the trend analysis algorithm that the average consistency ratio of a node shows a continuous, unidirectional downward trend even if it has not yet reached the fault threshold, the system will also generate a preventive maintenance recommendation and report it to the central platform. When a group of spatially adjacent nodes in a certain area are monitored and their consistency ratios show a general and synchronous trend deviation, the system determines that this phenomenon is not caused by a single node failure, but that the overall environmental baseline of the area has changed. At this time, the system will automatically trigger a request to recommend the calculation weights of the environmental variables involving the area in the central platform. and warning thresholds, and re-execute the offline calibration and parameter determination procedures to achieve the adaptive evolution of the entire governance model to environmental changes.

[0056] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0057] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not limiting. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A pollution control method driven by environmental variables, characterized in that: The following steps are involved: Step a: Periodically monitor the values ​​of environmental variables at multiple edge computing nodes, and compare the current values ​​of the environmental variables with the values ​​of the previous cycle to generate a single-bit fluctuation symbol that represents the fluctuation direction of the environmental variables, wherein the fluctuation symbol includes a first bit state and a second bit state. When the current value is greater than the value of the previous cycle, the first bit state is generated; when the current value is less than or equal to the value of the previous cycle, the second bit state is generated. Step b: At the regional aggregation node, the fluctuation symbols sent by multiple edge computing nodes are received, and for a determined pair of environmental parameters, the temporal coordination correlation between the corresponding fluctuation symbol sequences is calculated. The temporal coordination correlation reflects the coordination of the fluctuation direction of the parameter pair within a fixed-length time window. Step c: On the central platform, the time series coordination correlation reported by the regional aggregation nodes is received. Based on the weights determined to reflect the degree of influence of different parameters on the pollution formation process, an instability index representing the overall state of the environmental system is calculated. The instability index is the weighted sum of the time series coordination correlations. Step d: When the value of the instability index exceeds the set warning threshold, the system triggers a pollution risk warning and automatically issues pollution control instructions based on the system's configured control strategy. The pollution control instructions include changing the operating state of at least one environmental variable to block the fluctuation synergy conditions that cause the instability index to exceed the warning threshold; Among them, the calculation of the timing coordination correlation degree in step b includes performing a bit-wise XOR operation on the fluctuation symbol bit sequence corresponding to the two parameters in the environmental parameter pair, and determining the timing coordination correlation degree based on the percentage of the number of bits with zero bit value and the total length of the sequence in the XOR operation result. The higher the percentage, the greater the coordination correlation degree. The edge computing node additionally calculates the absolute value of the difference between the current value of the environmental variable and the value of the previous cycle. If the absolute value of the difference exceeds a certain mutation threshold, a burst disturbance mark is generated and sent to the regional aggregation node together with the fluctuation symbol; and, when the regional aggregation node receives any burst disturbance mark, it immediately shortens the calculation period of the instability index to less than one second and switches the calculation method of the instability index to ,in, is the risk value calculated instantaneously based on the latest data, is the instability index value before switching, The inertia coefficient is greater than or equal to 0.9 and less than 1.

2. The pollution control method driven by environmental variables according to claim 1, characterized in that: When the regional aggregation node does not receive any sudden disturbance mark within one minute, the system restores the calculation period of the instability index to more than one minute and switches the calculation method of the instability index to the conventional prediction mode.

3. The pollution control method driven by environmental variables according to claim 1, characterized in that: The edge computing node uses the network time protocol or beacon timing function to obtain a high-precision synchronized clock, and when generating a wave symbol, it adds an event occurrence timestamp provided by the synchronized clock to the wave symbol.

4. The pollution control method driven by environmental variables according to claim 3, characterized in that: The regional aggregation node caches the fluctuation symbol flow in a short period of time and calculates the corresponding timestamp difference when it detects that two or more spatially adjacent nodes monitoring the same environmental variable have successively reported the same fluctuation symbol; And based on the timestamp difference and the known spatial positions between adjacent nodes, the migration direction of the environmental variable changes in space is determined.

5. The method for pollution control driven by environmental variables according to claim 4, characterized in that: The determined migration direction is used as a weighting factor and input into the instability index calculation model of the central platform.

6. The pollution control method driven by environmental variables according to claim 1, characterized in that: When the regional aggregation node collects and processes the fluctuation symbol sequence, it calculates the average consistency ratio of the fluctuation symbol sequence of any node in a group of spatially adjacent edge computing nodes monitoring the same environmental variable and the fluctuation symbol sequences of all other nodes in the group; and when the average consistency ratio is lower than 0.5, the regional aggregation node generates diagnostic information indicating a fault in any node and reports it to the central platform.

7. The method for pollution control driven by environmental variables according to claim 6, characterized in that: Diagnostic information is used to drive targeted maintenance of faulty edge computing nodes.

8. The pollution control method driven by environmental variables according to claim 1, characterized in that: Environmental variables include relative air humidity, fan speed of production equipment, and groundwater level in the park. Pollution control instructions include forcibly increasing fan power according to the instability index.

9. The method for pollution control driven by environmental variables according to claim 1, characterized in that: The method is implemented by a pollution control system driven by environmental variables, which includes: Multiple edge computing units are used to periodically monitor the values ​​of environmental variables and compare the current values ​​of the environmental variables with the values ​​of the previous cycle to generate a single-bit fluctuation symbol that represents the fluctuation direction of the environmental variables, wherein the fluctuation symbol includes a first bit state and a second bit state. When the current value is greater than the value of the previous cycle, the first bit state is generated; when the current value is less than or equal to the value of the previous cycle, the second bit state is generated; at least one regional convergence unit is used to receive the fluctuation symbols sent by the multiple edge computing units, and for a determined pair of environmental parameters, calculate the time series coordination correlation between the corresponding fluctuation symbol sequences, wherein the time series coordination correlation reflects the coordination of the fluctuation direction of the parameter pair within a time window of a fixed length; A central processing unit is used to receive the time series synergy correlation reported by each regional convergence unit, and calculate the instability index representing the overall state of the environmental system based on the determined weights reflecting the degree of influence of different parameters on the pollution formation process, where the instability index is the weighted sum of the time series synergy correlation; A governance instruction issuing unit is used to trigger a pollution risk warning when the value of the instability index exceeds the set warning threshold, and automatically issue pollution control instructions according to the governance strategy configured by the system. The pollution control instructions include changing the operating state of at least one environmental variable to block the fluctuation coordination conditions that cause the instability index to exceed the warning threshold.

Images