Method and device for evaluating performance of modular fillers of constructed wetlands and making precise replacement decisions

Abstract

The embodiment of the specification provides a method and device for evaluating the performance of a modular filler of a constructed wetland and making a precise replacement decision, wherein the method comprises the following steps: deploying a multi-parameter sensor in a filler unit, monitoring the microenvironment parameters in real time, and spatiotemporally fusing the system water inlet load to calculate a dynamic characteristic response vector representing the physiological state of the unit; establishing a dynamic model containing inherent attenuation, load impact accumulation and temperature effect based on the performance attenuation mechanism of the filler, and using data assimilation algorithms such as unscented Kalman filtering to estimate the performance index and model parameters of each unit in real time; using Monte Carlo simulation to predict the remaining effective time and its probability distribution of each unit; finally, taking the minimization of system performance risk and operation and maintenance cost as the target, taking inventory and process continuity as the constraint, constructing a multi-objective mixed integer programming model, and generating a globally optimal precise replacement plan. The scientificity, economy and system stability of operation and maintenance are significantly improved.

Description

Technical Field

[0001] The embodiments in this specification relate to the field of environmental engineering technology, and in particular to a method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions. Background Technology

[0002] Constructed wetlands, as an ecological and low-cost advanced wastewater treatment and purification technology, are widely used globally for the treatment of domestic sewage, agricultural drainage, river replenishment, and industrial effluent. Their purification mechanism primarily relies on the physical interception, chemical transformation, and biodegradation of pollutants by the composite ecosystem of packing material, microorganisms, and plants. Among these, the packing material, serving as the carrier for microbial attachment, the reaction interface for pollutant transformation, and the medium for hydraulic conduction, is crucial; its long-term stability is the material basis for ensuring the sustained and effective treatment efficiency of constructed wetlands.

[0003] However, during long-term operation, constructed wetland fillers generally face the problem of performance degradation. Therefore, a better solution is urgently needed. Summary of the Invention

[0004] In view of this, embodiments of this specification provide a method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions. One or more embodiments of this specification also relate to a device for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions, a computing device, a computer-readable storage medium, and a computer program, to address the technical deficiencies existing in the prior art.

[0005] According to a first aspect of the embodiments of this specification, a method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions is provided, including: Collect microenvironmental parameters inside multiple modular filler units in constructed wetlands and influent load parameters of constructed wetland systems; For each modular packing unit, the microenvironment parameters are fused with the spatiotemporally aligned influent load parameters to calculate the characteristic response vector that characterizes the response state of the modular packing unit. Based on a pre-defined packing performance decay mechanism model and using a data assimilation algorithm, the feature response vector is used as the input of the observation value to dynamically estimate the real-time performance index and model parameters of each modular packing unit. Based on the estimated real-time performance index, model parameters, and future load assumptions, the performance degradation trajectory of each modular packing unit is predicted through numerical simulation, and the remaining effective time and probability distribution of the remaining effective time when the modular packing unit reaches the preset failure threshold are calculated. Using the probability distribution of the remaining effective time of all modular packing units, the process criticality weight, replacement cost, and preset constraints as inputs, a multi-objective optimization model is constructed with the goal of minimizing system performance risk and total operation and maintenance cost. The multi-objective optimization model is solved to generate a replacement plan. Output and display the replacement plan and information related to the health status of the modular packing unit.

[0006] In one possible implementation, the microenvironment parameters include redox potential, dissolved oxygen concentration, pH value, and temperature; the influent load parameters include influent flow rate, chemical oxygen demand concentration, ammonia nitrogen concentration, total nitrogen concentration, and total phosphorus concentration.

[0007] In one possible implementation, the characteristic response vector includes the redox potential dynamic response index per unit pollutant load, the dissolved oxygen consumption rate index per unit pollutant load, the nitrate load response index based on pH change, and the relative decay index of the hydraulic conductivity coefficient of the packing unit.

[0008] In one possible implementation, the mechanistic model includes the rate of change of the real-time performance index over time, which is negatively correlated with the temperature-related intrinsic decay rate, an acceleration function based on the cumulative effect of historical inflow load shocks, and a power function of the real-time performance index.

[0009] In one possible implementation, unscented Kalman filtering or extended Kalman filtering is used as a data assimilation algorithm to dynamically estimate the real-time performance index and model parameters.

[0010] In one possible implementation, the probability distribution of the remaining effective time is calculated through Monte Carlo simulation, which includes: drawing multiple sets of parameter samples from the posterior distribution of the state estimate, performing multiple forward numerical integrations in conjunction with future load scenarios, and statistically obtaining the probability distribution.

[0011] In one possible implementation, the multi-objective optimization model is a mixed-integer linear programming model, the decision variables are binary variables representing whether to replace a specific packing unit in a specific time window, the objective function is the weighted sum of system performance risk cost and total operation and maintenance cost, and the constraints include inventory constraints, the constraint that each unit can be replaced at most once, and the constraint of the maximum replacement ratio within the same process partition.

[0012] According to a second aspect of the embodiments of this specification, a device for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions is provided, comprising: The data acquisition module is configured to collect microenvironmental parameters inside multiple modular filler units in the constructed wetland and influent load parameters of the constructed wetland system. The feature determination module is configured to fuse microenvironmental parameters with spatiotemporally aligned influent load parameters for each modular packing unit, and calculate a feature response vector characterizing the response state of the modular packing unit. The dynamic estimation module is configured to dynamically estimate the real-time performance index and model parameters of each modular packing unit based on a preset packing performance decay mechanism model and using a data assimilation algorithm, with the feature response vector as the input of the observation value. The probability distribution module is configured to predict the performance degradation trajectory of each modular packing unit through numerical simulation based on the estimated real-time performance index, model parameters, and future load assumptions, and to calculate the remaining effective time and probability distribution of the modular packing unit when it reaches the preset failure threshold. The objective solution module is configured to take the probability distribution of the remaining effective time of all modular packing units, the process criticality weight, the replacement cost, and the preset constraints as inputs, construct a multi-objective optimization model with the goal of minimizing system performance risk and total operation and maintenance cost, and solve the multi-objective optimization model to generate a replacement plan. The information display module is configured to output and display replacement plans and information related to the health status of modular packing units.

[0013] According to a third aspect of the embodiments of this specification, a computing device is provided, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the above-mentioned modular filler performance evaluation and precise replacement decision-making method for artificial wetlands are implemented.

[0014] According to a fourth aspect of the embodiments of this specification, a computer-readable storage medium is provided that stores computer-executable instructions, which, when executed by a processor, implement the steps of the above-described method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions.

[0015] According to a fifth aspect of the embodiments of this specification, a computer program is provided, wherein when the computer program is executed in a computer, the computer is instructed to perform the steps of the above-described method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions.

[0016] This specification provides a method and apparatus for evaluating the performance of modular packing materials in constructed wetlands and for making precise replacement decisions. The method includes: deploying multi-parameter sensors within the packing unit to monitor its microenvironmental parameters in real time, and performing spatiotemporal fusion with the system's influent load to calculate a dynamic characteristic response vector representing the unit's "physiological state"; establishing a dynamic model based on the packing performance decay mechanism, including inherent decay, cumulative load impact, and temperature effects, and using data assimilation algorithms such as unscented Kalman filtering to estimate the performance index and model parameters of each unit in real time; using Monte Carlo simulation to predict the remaining effective time and probability distribution of each unit; and finally, with the goal of minimizing system performance risk and operation and maintenance costs, and constrained by inventory and process continuity, constructing a multi-objective mixed integer programming model to generate a globally optimal precise replacement plan. This significantly improves the scientific nature, economy, and system stability of operation and maintenance. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions, provided in one embodiment of this specification. Figure 2 This is a schematic diagram of a system for evaluating the performance of modular filler material in constructed wetlands and making precise replacement decisions, provided in one embodiment of this specification. Figure 3 This is a schematic diagram of a modular filler material performance evaluation and precise replacement decision-making device for constructed wetlands, provided in one embodiment of this specification. Figure 4 This is a structural block diagram of a computing device provided in one embodiment of this specification. Detailed Implementation

[0018] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0019] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of this specification. The singular forms “a” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0020] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0021] This specification provides a method for evaluating the performance of modular filler material in constructed wetlands and making precise replacement decisions. This specification also relates to a device for evaluating the performance of modular filler material in constructed wetlands and making precise replacement decisions, a computing device, and a computer-readable storage medium, which will be described in detail in the following embodiments.

[0022] See Figure 1 , Figure 1 A flowchart is shown of a method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions according to an embodiment of this specification, which specifically includes the following steps.

[0023] Step 101: Collect microenvironmental parameters inside multiple modular filler units in the constructed wetland and influent load parameters of the constructed wetland system.

[0024] Step 102: For each modular packing unit, the microenvironment parameters are fused with the spatiotemporally aligned influent load parameters to calculate the characteristic response vector that characterizes the response state of the modular packing unit.

[0025] Step 103: Based on the preset packing performance decay mechanism model and using the data assimilation algorithm, the feature response vector is used as the input of the observation value to dynamically estimate the real-time performance index and model parameters of each modular packing unit.

[0026] Step 104: Based on the estimated real-time performance index and model parameters, as well as future load assumptions, predict the performance degradation trajectory of each modular packing unit through numerical simulation, and calculate the remaining effective time and probability distribution of the remaining effective time when the modular packing unit reaches the preset failure threshold.

[0027] Step 105: Using the probability distribution of the remaining effective time of all modular packing units, the process criticality weight, replacement cost, and preset constraints as input, construct a multi-objective optimization model with the goal of minimizing system performance risk and total operation and maintenance cost, and solve the multi-objective optimization model to generate a replacement plan.

[0028] Step 106: Output and display the replacement plan and information related to the health status of the modular packing unit.

[0029] Modular packing units refer to standardized, independently installable and replaceable physical modules of packing material in constructed wetlands, serving as the fundamental carriers of purification functions. Microenvironmental parameters refer to environmental variables directly measured at specific locations within the packing unit, reflecting its local physicochemical and biological state. Influent load parameters refer to the total amount of pollutants carried by wastewater entering the constructed wetland per unit time and its hydraulic conditions, serving as external inputs driving packing performance degradation. Characteristic response vectors refer to a set of dimensionless or normalized indices calculated to correlate microenvironmental changes with influent load, used to eliminate the impact of load fluctuations and directly characterize the "physiological" state of the packing unit. Packing performance degradation mechanism models refer to mathematical relationships established based on physical, chemical, and biological processes, describing the degradation of packing performance over time. Data assimilation algorithms refer to a class of algorithms that combine observed data with mechanism models to optimally estimate the internal state and parameters of the model. Real-time performance index refers to a value varying between zero and one, used to quantify the percentage of the packing unit's current performance relative to its new state. Model parameters can refer to the coefficients in the mechanistic model used to characterize the specific characteristics of the packing unit, such as decay rate and sensitivity to load. Future load assumptions can refer to the pollutant load patterns entering the wetland in the future, predicted based on historical data or operational plans. Performance decay trajectory can refer to the predicted change curve of the real-time performance index over future time. Remaining effective time can refer to the length of time from the current moment when the predicted real-time performance index first drops to the preset failure threshold. Probability distribution can refer to the statistical regularity of possible values ​​for the remaining effective time obtained through multiple simulations. Process criticality weights can refer to the values ​​assigned to the packing unit based on its position and functional importance in the wetland process, used to differentiate between different units in decision-making. Replacement cost can refer to the expenses incurred in purchasing new packing modules and carrying out replacement work. Preset constraints can refer to the restrictions that must be followed in actual operation and maintenance, such as the upper limit on the number of modules that can be replaced per week and the limit on the proportion of simultaneous replacements in the same area. System performance risk can refer to the potential costs such as environmental penalties and reputational damage caused by the failure of packing units leading to excessive effluent quality in the constructed wetland. Total maintenance cost can refer to the total cost incurred for all planned replacement activities within the planning period. A replacement plan can refer to an operational procedure that specifies which particular packing units will be replaced at which future points in time.

[0030] The present invention will be further described below through a detailed embodiment: In one embodiment of the present invention, a method for evaluating the effectiveness of modular packing material and making precise replacement decisions in constructed wetlands is used in a composite vertical flow constructed wetland for treating effluent from a municipal wastewater treatment plant. The wetland is divided into thousands of standard modular packing material units, with the packing materials categorized by different functions such as phosphorus removal, nitrification, and denitrification.

[0031] After the system starts operating, the first step is data acquisition. Miniature sensor nodes deployed within each representative packing unit continuously measure the redox potential, dissolved oxygen concentration, pH value, and temperature of the core reaction zone at set time intervals (e.g., every 5 minutes). Simultaneously, an online water quality analyzer and flow meter installed at the wetland's main inlet monitor the influent flow rate, chemical oxygen demand (COD), ammonia nitrogen, total nitrogen, and total phosphorus concentrations. All data is synchronized via an IoT gateway and then transmitted to the central data processing platform.

[0032] Subsequently, the system performs a feature extraction step for each packing unit. The platform cleans and filters the received raw data. Since wastewater takes time to flow through the wetland, the system estimates the hydraulic retention time delay for each unit, aligning the influent load data on the time axis to the moment it affects that unit. Then, within a set time window (e.g., the past 6 hours), the system calculates multiple characteristic response indicators for that unit. For example, for phosphorus removal units, it calculates the rate of decrease in redox potential per unit of total phosphorus load; for nitrification units, it calculates the rate of dissolved oxygen consumption per unit of ammonia nitrogen load; for denitrification units, it calculates the pH fluctuation amplitude per unit of nitrate load; in addition, it estimates the rate of decrease in hydraulic conductivity relative to the initial value based on the head difference between the upstream and downstream of the unit. These temperature-corrected indicators together constitute the current characteristic response vector of the unit.

[0033] Next, the system performs a state estimation step. The platform maintains an estimator based on an unscented Kalman filter for each packing unit. At the core of this estimator is a pre-defined packing performance decay mechanism model. This model posits that the rate of decline of a packing unit's real-time performance index depends on a temperature-dependent intrinsic decay rate, an acceleration function reflecting the cumulative impact of all historical influent pollutant loads, and the current performance index itself. The system uses the characteristic response vector calculated in the previous step as observations and inputs it into the unscented Kalman filter. The filter runs its prediction and update steps, dynamically and optimally estimating the current, unmeasurable real-time performance index of each packing unit (e.g., a value between 0 and 1, where 1 represents brand new), and simultaneously updating key parameters in the mechanism model (such as the intrinsic decay rate and sensitivity to load shocks). These estimates are displayed on the management platform's dashboard.

[0034] Next, the system performs a lifetime prediction step. Based on the real-time performance index and model parameters of each unit estimated in the previous step, and assuming that the influent load will remain at the recent average level or a certain typical pattern for a future period (e.g., the next 60 days), the system extrapolates the performance decay trajectory of each unit using numerical integration methods (e.g., the Runge-Kutta method). To quantify the uncertainty of the prediction, the system employs Monte Carlo simulation: a large number of samples are drawn from the parameter probability distribution provided by the unscented Kalman filter, and the above trajectory extrapolation is performed for each sample, thereby obtaining a large number of possible future trajectories. By statistically analyzing the time points at which these trajectories first cross a preset failure threshold (e.g., when the performance index drops to 0.4), the probability distribution of the remaining effective time of each unit can be obtained, for example, "the probability of unit A failing within 30 days is 10%, and the probability of failing within 45 days is 50%."

[0035] Afterward, the system executes optimization decision-making steps. The decision engine automatically starts weekly. It reads the probability distribution of the remaining effective time of all packing units, combining this with pre-configured process criticality weights for various units (e.g., higher weight for descaling units near the inlet), replacement costs for various units, and operational constraints (e.g., maximum replacement quantity supported by weekly inventory, and a single replacement ratio not exceeding 30% within the same hydraulic channel). The engine constructs a multi-objective mixed-integer linear programming model, with the decision variable being "whether to replace the I-th unit in the W-th week." Its objective is to balance minimizing the system performance risk cost due to premature unit failure with minimizing the total packing procurement and replacement labor costs. By calling the optimization solver, the system solves the model, generating an optimal replacement plan for the next few weeks (e.g., the next 4 weeks), clearly indicating the recommended unit numbers and their time windows for replacement.

[0036] Finally, the system outputs and displays the steps. The generated replacement plan is displayed in Gantt chart format on the web management interface. Simultaneously, a heat map of the wetland visually displays the current health status of each packing unit (coded with green, yellow, orange, and red colors based on real-time performance indices) and its predicted remaining lifespan. The system also automatically generates early warning work orders, pushing them to the mobile app of on-site maintenance personnel to guide them in performing precise replacement operations. All historical status, prediction, and decision data are stored in the database, forming a digital full lifecycle archive of the packing units.

[0037] Through iterative execution of the above steps, the system achieves continuous perception, evaluation, prediction, and proactive maintenance decision-making regarding the performance of constructed wetland fillers.

[0038] The beneficial effect of this embodiment is that it transforms the operation and maintenance management of constructed wetland filler from a passive response mode that relies on fixed cycles or excessive effluent quality to a predictive proactive maintenance mode based on the actual health status of each filler unit. Through multi-source data fusion and advanced feature engineering, the implicit performance degradation process becomes visible and measurable. Using a mechanism- and data-driven model, accurate and probabilistic predictions of the remaining lifespan of the filler are achieved, providing a scientific time window for maintenance actions. Finally, through system-level optimization decision-making, a precise replacement plan that optimally balances cost and risk is automatically generated while meeting process requirements and resource constraints. This avoids resource waste caused by "over-maintenance" and system risks caused by "under-maintenance," significantly improving the reliability, stability, and economy of constructed wetland operation.

[0039] In one possible implementation, the microenvironment parameters include redox potential, dissolved oxygen concentration, pH value, and temperature; the influent load parameters include influent flow rate, chemical oxygen demand concentration, ammonia nitrogen concentration, total nitrogen concentration, and total phosphorus concentration.

[0040] Among these, oxidation-reduction potential (ORP) refers to the electrochemical measurement of the relative strength of oxidants and reductants in a solution, measured in millivolts (mV). It indicates whether the interior of the packing unit is in an aerobic, anoxic, or anaerobic state, and is associated with key biochemical processes such as nitrification, denitrification, and phosphorus release / uptake. Dissolved oxygen concentration (DO) refers to the content of molecular oxygen dissolved in water, measured in milligrams per liter (mg / L), directly reflecting the metabolic activity of aerobic microorganisms and oxygen mass transfer efficiency. pH value refers to the acidity or alkalinity of the solution, affecting microbial enzyme activity, pollutant speciation, and certain chemical precipitation reactions. Temperature refers to the water's temperature in degrees Celsius (°C), a key factor influencing the rates of all biological and chemical reactions. Influent flow rate refers to the volume of wastewater entering the constructed wetland per unit time, serving as the basis for calculating pollutant load. Chemical oxygen demand (COD) concentration refers to the amount of oxygen consumed by organic matter in water that can be oxidized by strong oxidants, reflecting the organic pollution load. Ammonia nitrogen concentration refers to the content of nitrogen in water in the form of ammonia or ammonium ions, a major substrate in the nitrification process. Total nitrogen concentration can refer to the sum of all forms of nitrogen in water (ammonia nitrogen, nitrate nitrogen, nitrite nitrogen, organic nitrogen, etc.). Total phosphorus concentration can refer to the sum of all forms of phosphorus in water (orthophosphate, polyphosphate, organic phosphorus, etc.).

[0041] In practical applications, this step constructs the system's "sensory nervous system." A miniature multi-parameter sensor probe is permanently integrated or embedded within the core reaction zone of each modular packing unit (numbered i, i=1,2,...,N). This probe continuously monitors at least the following microenvironmental parameters (Class A data): Oxidation-reduction potential (ORPi(t)): millivolt-level accuracy, used to indicate the oxidation-reduction state within the unit, a key indicator for judging the intensity of biochemical processes such as nitrification, denitrification, and phosphorus release / uptake. Dissolved oxygen concentration (DOi(t)): mg / L accuracy, directly reflecting the intensity of aerobic biochemical reactions and oxygen transport efficiency. pH (pHi(t)): used to monitor processes such as nitrification (alkali consumption, pH decrease), denitrification (alkali production, pH increase), and is also an important parameter for some chemical precipitation reactions (such as phosphorus precipitation). Temperature (Ti(t)): Celsius accuracy, a significant factor affecting the rates of all biological and chemical reactions. Simultaneously, online water quality analyzers and flow meters are installed at the inlet and key process nodes of the constructed wetland system to continuously monitor system-level operating parameters (Class B data): influent water quality and quantity: flow rate (Qin(t)), chemical oxygen demand (CODin(t)), and ammonia nitrogen (NH4). + -Nin(t)), nitrate nitrogen (NO3) - -Nin(t)), Total Nitrogen (TNin(t)), Total Phosphorus (TPin(t)), pHin(t), and Temperature (Tin(t)). Water Level / Differential Pressure: Water level gauges or differential pressure sensors are installed at key locations in the packing layer to estimate hydraulic retention time and detect blockage. Meteorological data may also be included: ambient temperature, light intensity, rainfall, wind speed, etc., to correct for the effects of natural reoxygenation and evaporation in open water bodies. All Class A and Class B data are collected synchronously via an IoT gateway, with the sampling frequency adjustable as needed (e.g., 1-15 minutes / time) and transmitted to the central data processing platform.

[0042] In practical implementation, the monitoring of microenvironmental parameters and influent load parameters were clearly defined. Within each modular packing unit equipped with sensors, an integrated microprobe continuously measures at least four core parameters: oxidation-reduction potential (ORP), used to determine whether the unit favors nitrification (higher positive value) or denitrification / phosphorus removal (lower negative value); dissolved oxygen concentration, directly characterizing the intensity of aerobic biochemical reactions; pH value, used to monitor the decrease caused by alkali consumption during nitrification or the increase caused by alkali production during denitrification; and temperature, used to correct for biochemical reaction rates. Simultaneously, at the main inlet of the constructed wetland system, an online water quality analyzer and flow meter must be installed to continuously monitor the concentrations of five key pollutants: influent flow rate, chemical oxygen demand (COD), ammonia nitrogen (AM), total nitrogen (TNO), and total phosphorus (TP). These parameters collectively constitute the fundamental data source necessary for evaluating the performance response of the packing unit and for attenuation modeling. For example, characterizing a nitrifying packing unit requires simultaneously using internally measured dissolved oxygen and pH values, as well as ammonia nitrogen concentration and flow rate data at the inlet, to calculate a meaningful oxygen consumption index per unit ammonia nitrogen load.

[0043] The beneficial effect of this embodiment lies in clarifying the core components of the monitoring parameter system. The four microenvironmental parameters—oxidation-reduction potential, dissolved oxygen, pH, and temperature—can effectively capture the key biogeochemical processes within the packing material; while the five influent load parameters—flow rate, chemical oxygen demand, ammonia nitrogen, total nitrogen, and total phosphorus—comprehensively cover the main pollutant targets in constructed wetland treatment. This set of parameters ensures that the system can accurately assess the performance of different functional packing materials (such as phosphorus removal, nitrification, and denitrification) while also considering the feasibility and cost of monitoring technology, providing reliable and necessary data input for the entire intelligent assessment and decision-making process.

[0044] In one possible implementation, the characteristic response vector is calculated and includes the dynamic response index of redox potential per unit pollutant load, the dissolved oxygen consumption rate index per unit pollutant load, the response index of nitrate load per unit pH change, and the relative decay index of the hydraulic conductivity coefficient of the packing unit.

[0045] The dynamic response index of redox potential per unit pollutant load refers to the ratio of the cumulative decrease in redox potential within a packing unit to the cumulative load of total phosphorus (or other reducible pollutants) acting on that unit during a specific time window. A decrease in this index indicates a weakening of the packing material's phosphorus adsorption / reduction capacity. The dissolved oxygen consumption rate index per unit pollutant load refers to the ratio of the cumulative rate of decrease in dissolved oxygen concentration within a packing unit to the cumulative load of chemical oxygen demand (COD) or ammonia nitrogen (the latter multiplied by the theoretical oxygen demand coefficient) acting on that unit during a specific time window. A decrease in this index reflects a decline in the activity of aerobic microorganisms (heterotrophic or nitrifying bacteria). The unit nitrate load response index based on pH changes refers to the ratio of the fluctuation range of pH value within a packing unit (e.g., the difference between the maximum and minimum values) to the cumulative load of nitrate nitrogen acting on that unit during a specific time window. This index utilizes the characteristic that denitrification consumes hydrogen ions, causing pH to rise; a decrease in this index may indicate reduced denitrifying bacteria activity or insufficient carbon source. The relative decay index of the hydraulic conductivity of a packing unit can be defined as the ratio of the hydraulic conductivity of the packing unit at the current moment, estimated according to Darcy's law, to its initial (or reference) hydraulic conductivity. A decrease in this index indicates a deterioration in the permeability of the packing pores due to blockage.

[0046] In the feature extraction step, the system calculates a set of core characteristic response indices for each packing unit. For units whose primary function is phosphorus removal or that operate in a strongly reducing environment, the system calculates the dynamic response index of redox potential per unit total phosphorus load. This index quantifies the severity of the change in the internal redox state of the packing when subjected to phosphorus load shocks. For aerobic units responsible for carbon oxidation and nitrification, the system calculates the dissolved oxygen consumption rate index per unit chemical oxygen demand load and the dissolved oxygen consumption rate index per unit ammonia nitrogen load, respectively. These two indices are directly related to the metabolic activity of heterotrophic microorganisms and nitrifying bacteria. For anoxic / anaerobic units responsible for denitrification, the system calculates the unit nitrate load response index based on pH changes, assessing its effectiveness by capturing the characteristic pH rise caused by denitrification. Furthermore, regardless of the functional unit, the system estimates its hydraulic conductivity based on monitored head loss and calculates the relative decay index relative to the initial state to independently assess the physical decay process of pore blockage. These calculated and temperature-normalized indices together form a multidimensional characteristic response vector. They transform the raw, load-fluctuation-affected sensor readings into characteristic signals that can stably characterize the "physiological" state and performance response of the packing unit, providing high-quality observational input for subsequent state estimation.

[0047] In practical applications, quality control of raw data is performed, including outlier removal (based on the 3σ principle or interquartile range method), missing value imputation (such as linear interpolation or Kalman smoothing), and signal smoothing filtering (such as moving average or low-pass filtering). Due to the hydraulic retention time (HRT) of wastewater flowing through the wetland, the influent load data must be spatiotemporally aligned with the internal response data of each unit. A simplified hydraulic transport model is established for each unit i, estimating its time delay τi relative to the inlet. This can be initially determined through tracer experiments or computational fluid dynamics (CFD) simulations, and fine-tuned during operation through data correlation analysis. The "effective influent load vector" Li(t) acting on unit i at time t is then:

[0048] Furthermore, for each unit i, within a sliding time window Δt (e.g., 6 hours or 1 day), a set of dimensionless or normalized dynamic characteristic response indices are calculated to form its eigenvector Ri(t): ORP dynamic response index per unit pollutant load : Targeted for phosphorus removal or strong reducing units.

[0049]

[0050] In the formula, ε in the denominator is a very small positive number to prevent division by zero errors. A decrease in this exponent indicates a weakening of the filler's adsorption / reduction capacity for phosphorus.

[0051] Carbon oxidation and nitration activity index , : Targeting the aerobic unit.

[0052]

[0053]

[0054] In the formula, 4.57 is the theoretical oxygen demand coefficient for complete nitrification of ammonia nitrogen. The decrease in these indices reflects the decline in the activity of aerobic microorganisms.

[0055] Denitrification Characteristic Index : For anoxic / anaerobic denitrification units. The denitrification process consumes H₂... + The property that causes pH to rise.

[0056]

[0057] A decrease in this index may indicate reduced denitrifying bacteria activity or insufficient carbon source.

[0058] Hydraulic property attenuation index Based on Darcy's law, the head loss upstream and downstream of the monitoring unit is monitored. The relative decay of the hydraulic conductivity K is evaluated based on the estimated flow rate.

[0059]

[0060] The decline in this index indicates that pore blockage is worsening.

[0061] All the above-mentioned indices involving reaction rates need to be normalized to temperature using the Arrhenius equation to eliminate the influence of temperature fluctuations on eigenvalues ​​and facilitate long-term trend analysis.

[0062]

[0063] Where θ is the temperature coefficient (usually taken as 1.02~1.06), and Tref is the reference temperature (e.g., 20℃).

[0064] Finally, the state of each unit i at time t can be represented by its eigenvector:

[0065] The beneficial effect of this embodiment lies in defining a set of characteristic response indicators with clear physicochemical significance. These indicators, by directly correlating microenvironmental changes with influent pollutant load and performing dimensionless processing, effectively eliminate the influence of influent load fluctuations on the observed signals, allowing the extracted features to more purely reflect the performance status of the packing unit itself. Simultaneously, differentiated features were designed for packing materials with different functions (e.g., ORP response for phosphorus removal, DO response for aerobic processes, and pH response for denitrification), enhancing the specificity and accuracy of the condition assessment. The evaluation of the hydraulic conductivity coefficient independently reflects the physical blockage status. This set of features collectively constitutes a comprehensive "health check" indicator set, laying a solid foundation for subsequent accurate modeling and prediction.

[0066] In one possible implementation, the mechanistic model includes the rate of change of the real-time performance index over time, which is negatively correlated with the temperature-related intrinsic decay rate, an acceleration function based on the cumulative effect of historical inflow load shocks, and a power function of the real-time performance index. The acceleration function is constructed as a weighted cumulative function of historical load shock intensities, indicating that the current decay acceleration effect is determined by the weighted sum of all historical load shocks, with recent historical loads having a higher weight than long-term historical loads.

[0067] In this context, the acceleration function refers to a mathematical function used in the packing material performance decay mechanism model to characterize the accelerating effect of influent pollutant load on the performance decay process. Historical load impact intensity refers to the total amount or concentration of pollutants entering the packing unit at various past time points, usually represented by the norm of a load vector. The weighted cumulative function refers to a mathematical operation that performs a weighted summation of historical sequence data. The decay acceleration effect refers to the increase in the rate of packing material performance decay caused by the pollutant load. Recent historical load refers to the load within a relatively recent time period (e.g., within a few days or weeks). Long-term historical load refers to the load within a relatively distant time period (e.g., several months ago).

[0068] When constructing the packing performance decay mechanism model, the acceleration function used to describe the load acceleration effect adopts a specific construction form. This function is not a simple linear term directly proportional to the current load, but rather designed as a "weighted cumulative function." Its core idea is that the accelerated decay of packing performance is not only the result of the current load impact, but also the cumulative result of all load impacts it has experienced throughout its service history. However, not all historical loads have the same impact. This function achieves the physical intuition of "greater impact in the near term, smaller impact in the long term" through an exponentially decaying weighting function. Specifically, when calculating the acceleration effect at the current moment, the system integrates (sums) the entire historical load sequence from the start of operation to the current moment, but multiplies each historical load impact by a weighting coefficient that decays exponentially with time distance (the difference between the current moment and a historical moment). This means that the impact of a load impact from a few days ago on the current decay rate is far greater than that of a load impact from several months ago. This model structure better reflects the cumulative effect of "wear and tear" on packing performance during long-term operation, i.e., the "fatigue" phenomenon, making the prediction model more consistent with engineering reality than a simpler model.

[0069] In practical applications, this embodiment can establish a mapping relationship from the observable feature vector Ri(t) to the internal performance state Ei(t) that cannot be directly measured, and estimate Ei(t) in real time.

[0070] Specifically, the real-time performance index Ei(t) ∈ [0, 1] is defined for filler unit i, where 1 represents a brand-new state (optimal performance) and 0 represents complete failure. Its decay kinetics can be described as follows:

[0071] Among them, the inherent decay term: -βi(T) · [Ei(t)]γβi(T): the temperature-dependent inherent decay rate constant. According to van der Hoff's rule or Arrhenius's formula: βi(T) = βi,20 · θi(T(t)-20). βi,20 is the rate constant at 20℃, which is the parameter to be estimated. γ: the decay kinetic order. When γ=1, it is first-order kinetics (exponential decay); when γ>1, the decay is slow in the early stage and accelerates in the later stage; when γ<1, the opposite is true. γ reflects the decay mode of the packing performance and is a key hyperparameter of the model. The load acceleration term: g(Li(t), Θi), this function describes the accelerating effect of the influent pollutant load on the decay process.

[0072] This invention proposes a nonlinear impact accumulation model:

[0073] Where |Li(τ)| is a norm of the load vector (such as the Euclidean norm), representing the total pollution load impact intensity. a: Impact sensitivity coefficient, indicating the degree to which unit cumulative load accelerates the decay rate. Th: "Memory time constant", characterizing how quickly the influence of historical loads on the current decay rate decays. The larger Th is, the longer the system "memory" lasts.

[0074] The physical meaning of this model is clear: the performance degradation of the packing material is not only a function of the current load, but also a weighted cumulative result of all historical load impacts, with recent impacts having a greater impact. This reflects the actual aging process better than the simple linear relationship g=1+k·L(t). Since Ei(t) cannot be directly measured, we need to use the observable eigenvector Ri(t) to estimate it. To this end, a state-space model is constructed, and data assimilation techniques are used for real-time state estimation.

[0075] State variables: Define the extended state vector Xi(t) = [Ei(t), Si(t), βi,20, a, Th]T. Where Si(t) ∈ [0,1] is the "instantaneous saturation," used to describe the temporary performance degradation caused by load shocks in the short term (such as temporary saturation of adsorption sites). Its dynamic equation can be described as:

[0076] Where α is the saturation rate and λ is the recovery (desorption / regeneration) rate.

[0077] Observation equation: Establishes the relationship between state variables and observed eigenvectors. This is a nonlinear mapping.

[0078] For example, the ORP response characteristics can be modeled as: IORP,i(t) = η · Ei(t) · (1 - Si(t)) + v(t), where η is the scaling factor and v(t) is the observation noise.

[0079] The beneficial effect of this embodiment lies in using a weighted cumulative function to model the accelerating effect of load, giving the decay model a "memory" capability and better reflecting the actual physical process of packing performance aging. This model recognizes that packing decay is not instantaneous but rather an integral result of historical service conditions, and that recent high-load operation has a more significant impact on its current health status. This improves the model's ability to describe performance evolution under complex load histories, making remaining life predictions based on this model more accurate and reliable, and providing a more solid model foundation for developing proactive maintenance plans.

[0080] In one possible implementation, unscented Kalman filtering or extended Kalman filtering is used as a data assimilation algorithm to dynamically estimate the real-time performance index and model parameters.

[0081] Unscented Kalman Filtering (UKF) can refer to an algorithm for state estimation of nonlinear systems. It approximates the probability distribution of the state by carefully selecting a set of sample points (Sigma points) without linearizing the nonlinear model, and has high estimation accuracy and numerical stability. Extended Kalman Filtering (EKF) can refer to another algorithm for state estimation of nonlinear systems. It handles nonlinear problems by performing a first-order Taylor expansion (linearization) on the nonlinear model and observation equations at the current estimate, and is a widely used nonlinear filtering method. Dynamic estimation can refer to the algorithm continuously and online updating the optimal estimates of the system's internal state (such as the real-time performance index) and unknown parameters (such as the decay rate in the model) as new observation data arrives.

[0082] In the state estimation step, the core problem the system needs to solve is how to use observable but noisy feature response vectors to estimate the implicit real-time performance index and unknown parameters in the mechanistic model that cannot be directly measured. This invention preferably employs data assimilation algorithms such as unscented Kalman filtering or extended Kalman filtering to accomplish this task. The system instantiates a filter (e.g., a UKF filter) for each packing unit. The filter's "state equation" is the packing performance decay mechanism model described in claim 1 (possibly discretized), while its "observation equation" describes the mathematical relationship between the feature response vector and state variables (including the real-time performance index and model parameters). Whenever a new feature response vector is calculated, it is input to the filter as an observation. The filter executes its standard procedure: first, a "prediction step" is performed, predicting the current state based on the previous state estimate and mechanistic model; then, an "update step" is performed, comparing the predicted state with the new observation, and optimally fusing the prediction and observation information using Kalman gain to obtain the optimal estimate of the current state and parameters, and providing the uncertainty of the estimate. Through continuous operation, the filter enables real-time, dynamic tracking of the health status and individual characteristics of each packing unit.

[0083] In practical applications, due to the strong nonlinearity of the observation equation h(·), this invention preferably employs the Unscented Kalman Filter (UKF) for state estimation. The UKF approximates the probability distribution of the state using a carefully selected set of "Sigma points," avoiding the complexity and linearization error inherent in the Extended Kalman Filter (EKF) which requires calculating the Jacobian matrix. Details are as follows.

[0084] Initialization: Set initial values ​​for the state vector. 0 and its error covariance matrix P0.

[0085] Prediction step (time update): Based on the state equation (mechanistic model) and process noise, predict the mean state at the next time step. k|k-1 and covariance Pk|k-1. Specifically, this is achieved by passing a set of Sigma points through the state equation f(·).

[0086] Update step (measurement update): The predicted Sigma point is used to obtain the predicted observations through the observation equation h(·). The mean and covariance of the predicted observations are calculated. The Kalman gain Kk is calculated. When new observation data Ri(k) arrives, the state estimate is updated: k= k|k-1Kk[Ri(k) - k|k-1].

[0087] Update error covariance: Pk = Pk|k-1KkPyyKk T Through continuous operation of UKF, the system can output the optimal estimate Êi(t) of the efficiency index Ei(t) of each packing unit in real time, as well as its estimation uncertainty (the diagonal elements of the covariance matrix).

[0088] The beneficial effect of this embodiment lies in employing mature and efficient data assimilation algorithms (such as UKF or EKF) to solve the state estimation problem. These algorithms can optimally fuse noisy observation data and mechanistic model predictions based on physical knowledge, estimating key state variables (efficiency indices) and model parameters that cannot be directly measured with high accuracy and robustness. Compared to simple regression or threshold judgment, this method can handle the nonlinearity, dynamics, and uncertainty of the system, providing a reliable and real-time state awareness foundation for the entire predictive maintenance chain, and serving as a key bridge connecting "data" and "model".

[0089] In one possible implementation, the probability distribution of the remaining effective time is calculated through Monte Carlo simulation. Specifically, this involves: drawing multiple sets of parameter samples from the posterior distribution of the state estimate, performing multiple forward numerical integrations in conjunction with future load scenarios, and statistically obtaining the probability distribution.

[0090] Monte Carlo simulation can refer to a statistical simulation method that obtains numerical results through repeated random sampling. The posterior distribution of state estimation refers to the probability distribution of the model state and parameters obtained after updating with a data assimilation algorithm (such as UKF), containing the estimated values ​​and their uncertainties. Parameter samples refer to a set of specific parameter values ​​(such as decay rate, sensitivity coefficient, etc.) randomly selected from the above posterior probability distribution according to its statistical characteristics. Future load scenarios refer to predictions or assumptions about the influent pollutant load during the planning period; these can be a deterministic sequence or a sequence containing random fluctuations. Forward numerical integration refers to using the current state as a starting point, a mechanistic model, and a set of specific parameters, and using numerical calculation methods (such as the Euler method or Runge-Kutta method) to progressively calculate the state values ​​at various future time points. Statistically obtained probability distributions refer to frequency statistics or kernel density estimation of results obtained from a large number of simulations (such as failure time), thereby forming its empirical probability distribution.

[0091] In the lifetime prediction step, to quantify the uncertainty of the prediction and provide probabilistic remaining lifetime information, the system employs a Monte Carlo simulation method. The specific implementation process is as follows: First, from the posterior distribution generated by the aforementioned unscented Kalman filtering and other state estimation steps, a large number of groups (e.g., 1000 groups) of packing unit model parameters and state samples are randomly selected. Each group of samples represents a possible true state of the unit. Then, for each group of samples, the system assumes a future influent load scenario (e.g., maintaining the average load of the most recent week, or adding random fluctuations), and uses the packing effectiveness decay mechanism model to perform forward numerical integration, simulating the trajectory of the effectiveness index change from the current moment to a sufficiently long future time. Next, for each simulated trajectory, the moment when its effectiveness index first falls below a preset failure threshold (e.g., 0.4) is found; the time difference between this moment and the current moment is the remaining effective time under that simulation. Finally, after simulating all samples, a large number of predicted values ​​for the remaining effective time are obtained. The system performs statistical analysis on these values, such as calculating their mean, median, and standard deviation, and plotting histograms or cumulative distribution functions, thereby obtaining the complete probability distribution of the remaining effective time. This allows decision-makers not only to know "approximately how much longer it can be used," but also "how likely it is to fail before a certain point in time."

[0092] In practical applications, this embodiment uses the current state estimate Êi(t) and its uncertainty, as well as model parameters, to make probabilistic predictions of future performance.

[0093] Given a future inflow load scenario Lifuture(τ) (usually assumed to be the recent average load or a typical load pattern), the state equation is numerically integrated (e.g., by the fourth-order Runge-Kutta method) to obtain the predicted trajectory Êipred(τ) of the efficiency index from the current time t to the future time t+T, where τ∈[t, t+T].

[0094] Furthermore, a failure threshold Eth is defined (e.g., 0.3 or 0.4, which can be set according to process requirements). It is defined as the time from the current moment until the performance index first falls below the threshold.

[0095] Considering the uncertainties of model parameters (β, a, Th, etc.) and future loads, a single deterministic prediction is unreliable. This invention employs Monte Carlo simulation for probabilistic prediction, as detailed below.

[0096] From the posterior state distribution N provided by UKF ( In a dataset (k, Pk), M groups (e.g., 1000 groups) of state and parameter samples {Xi(m)} are extracted. For each sample group, considering future load random fluctuations (e.g., following a normal distribution), a deterministic prediction using S4.1 is performed, resulting in M ​​future efficiency index trajectories. The TRUL,i(m) corresponding to each trajectory is calculated. These M RUL values ​​are statistically analyzed to obtain their empirical probability distribution function (PDF) and cumulative distribution function (CDF). This allows for probabilistic descriptions such as "the probability of this unit failing after 30 days is 80%" or "the remaining effective time of this unit has a 90% confidence level between 25 and 38 days."

[0097] The beneficial effect of this embodiment is that it achieves probabilistic remaining lifetime prediction through Monte Carlo simulation. This method fully considers the uncertainty of model parameter estimation and the randomness of future loads, expanding the prediction result from a single numerical value to a probability distribution. This provides richer and more reliable information support for operation and maintenance decisions, enabling decision-makers to formulate maintenance plans based on risk probabilities (e.g., "the probability of this unit failing next month is greater than 80%)," thereby making a more scientific and refined trade-off between risk control and cost savings.

[0098] In one possible implementation, the multi-objective optimization model is a mixed-integer linear programming model, the decision variables are binary variables representing whether to replace a specific packing unit in a specific time window, the objective function is the weighted sum of system performance risk cost and total operation and maintenance cost, and the constraints include inventory constraints, the constraint that each unit can be replaced at most once, and the constraint of the maximum replacement ratio within the same process partition.

[0099] Mixed-integer linear programming models refer to a class of mathematical optimization models where some or all decision variables are restricted to integers (especially binary 0 or 1), and the objective function and constraints are linear functions of the decision variables. Decision variables refer to the unknowns that need to be determined in the optimization problem. Binary variables refer to decision variables that can only take the values ​​0 or 1, often used to represent binary choices such as "yes / no" or "do / don't do." A specific time window refers to a discrete time period (e.g., in weeks) that divides the future planning period (e.g., 4 weeks). System performance risk cost refers to the quantified value of the economic losses (e.g., environmental fines, operational interruption losses, etc.) that may result from the failure of the packing material due to untimely replacement, leading to excessive effluent from the constructed wetland. Total operation and maintenance cost refers to all direct costs incurred in executing the replacement plan, including the cost of purchasing new packing material and the labor cost of replacement construction. Weighted sum refers to a method of transforming a multi-objective problem into a single-objective problem by multiplying two or more objective functions by specific weights and then adding them together. Inventory constraints refer to the limitation on the number of filler modules available for replacement within any given time window, which is restricted by available stock or supply capacity. The constraint of "at most one replacement per unit" refers to a logical constraint in the optimization model that ensures any specific filler unit is replaced at most once during the entire planning period. The constraint of "maximum replacement ratio within the same process zone" refers to a restriction imposed to maintain the process continuity and treatment stability of constructed wetlands, preventing hydraulic short-circuiting or a sudden drop in treatment capacity due to excessive unit replacement in localized areas. This restricts the number of units allowed to be replaced within a single time window in the same hydraulic channel or functional zone to no more than a certain percentage of the total number of units in that zone.

[0100] In the optimization decision-making process, the system needs to generate a globally optimal replacement plan based on the predicted information of all units. To this end, the system constructs a specific mixed-integer linear programming model. The decision variables in this model are two-dimensional binary variables, for example, variable X_{i,w}, where a value of 1 indicates a plan to replace the i-th packing unit in week w, and a value of 0 indicates no replacement. The model has two objectives that need to be minimized simultaneously: first, the system performance risk cost, which is proportional to the probability of each unit failing in each week without replacement (calculated from the probability distribution obtained from Monte Carlo simulation) and its process criticality weight; second, the total maintenance cost, i.e., the sum of the replacement costs of all units planned for replacement. To solve this, the system typically uses a weighted sum method, assigning preference weights to the two objectives and merging them into a comprehensive single-objective linear function.

[0101] Simultaneously, the model must satisfy a series of realistic constraints: inventory constraints limit the maximum number of different types (A / B / C types) of packing modules that can be replaced each week; logical constraints ensure that each unit is not repeatedly scheduled for replacement during the planning period; and process continuity constraints require that, for each process zone (such as an independent treatment corridor) in the wetland, the number of units planned for replacement in the same week cannot exceed a certain set percentage (such as 25%) of the total number of units in that zone, to prevent excessive disturbance. The system inputs the above model into a professional mathematical optimization solver for solution, and finally outputs a set of binary decision variables that satisfies all constraints and optimizes the overall objective. This set directly corresponds to an executable and detailed future replacement plan.

[0102] In practical applications, based on the predicted RUL and its probability distribution of all N packing units, as well as operation and maintenance constraints, the optimal maintenance (replacement) plan for the future planning period (such as the next 4 weeks) is generated.

[0103] First, we model the optimization problem. For the decision variables, we define binary decision variables xi,w∈{0, 1}, where i=1,...,N represent the unit numbers, and w=1,...,W represent the future maintenance time windows (e.g., in weeks). xi,w=1 indicates that unit i is planned to be replaced in week w.

[0104] For the objective function: it is necessary to minimize two conflicting objectives simultaneously. Objective F1: System performance risk cost. Replacing too late will lead to unit failure, increasing the risk of effluent exceeding standards; replacing too early will waste the remaining value of the packing material. Risk cost is proportional to the probability of unit failure and the weight of its impact on the system.

[0105]

[0106] Where Pfail,i(w) is the probability that unit i will fail in week w (which can be calculated from CDF in S4.3), Crisk is the average cost of a single failure event (including environmental penalties, reputational damage, etc.), and Wi is the process criticality weight of unit i (for example, units located at the inlet and undertaking the main phosphorus removal task have a higher weight).

[0107] Target F2: Total Operation and Maintenance Costs. This includes the cost of purchasing new packing material and the labor costs for replacement work.

[0108]

[0109] The specific constraints are as follows: Inventory constraints: The number of filler modules available for replacement each week is limited.

[0110]

[0111] Logical constraint: Each unit can be replaced at most once.

[0112]

[0113] Process continuity constraints: To prevent hydraulic short circuits or sudden drops in treatment capacity due to excessive local replacements, the proportion of units allowed to be replaced each week within the same hydraulic flow channel or functional zone shall not exceed ρ (e.g., 30%).

[0114]

[0115] Where Gk represents the set of cells in the k-th partition.

[0116] Non-negativity and binary constraints: xi,w ≥ 0 and are integers (actually 0 or 1).

[0117] The above problem is a multi-objective mixed-integer linear programming (MILP) problem. This invention uses the ε-constraint method or the weighted sum method to transform it into a single-objective problem for solution.

[0118] Weighted sum method: Construct a comprehensive objective function F = ω1F1ω2F2, where ω1, ω2 ≥ 0 and ω1+ω2=1, and the weights reflect the decision-maker's preference for risk and cost.

[0119] The transformed single-objective MILP problem is solved using commercial optimization solvers (such as Gurobi, CPLEX) or heuristic algorithms (such as genetic algorithms) to obtain a Pareto optimal solution set or an optimal solution with specific weights. The solution results are converted into an executable maintenance work order plan and displayed through a human-computer interaction interface, including: Replacement Plan Gantt Chart: Visually displays the units planned for replacement in the coming weeks and their timing.

[0120] Unit health status panorama: The current performance index Êi(t) and predicted RUL of each unit are displayed on the wetland plan in the form of a heat map, and the health level is encoded with color (green / yellow / orange / red).

[0121] Warning list: Lists units with a RUL less than a predetermined threshold (e.g., 2 weeks) and reminds users to pay priority attention.

[0122] Inventory and cost forecasting: Forecasting the consumption of packing inventory and cumulative operation and maintenance costs at various future time points.

[0123] The beneficial effect of this embodiment lies in formalizing the complex operation and maintenance decision-making problem into a rigorous mixed-integer linear programming model. By introducing binary decision variables, the decision space of "which to replace and when to replace" is clearly defined. By simultaneously incorporating risk (the product of failure probability and consequences) and cost into the objective function, a comprehensive optimization of economy and reliability is achieved. The designed inventory, logic, and process continuity constraints ensure that the generated plan is feasible in terms of materials, logic, and process operations. This method completely changes the fuzzy decision-making model that relies on manual experience for scheduling, making the replacement plan formulation process automated, scientific, and optimized. It maximizes the lifespan of the packing material and minimizes the total life-cycle operation and maintenance cost while ensuring the safe and stable operation of the system.

[0124] In one overall embodiment, see Figure 2 This solution is used for the performance evaluation and precise replacement decision-making system of modular filler in constructed wetlands. The system is an integrated hardware and software platform that combines IoT, edge computing, cloud computing, and artificial intelligence technologies, and adopts a layered distributed architecture. 1. On-site perception and execution layer: Intelligent sensing module: Encapsulates miniaturized, low-power, corrosion-resistant probes for ORP, DO, pH, and temperature sensors, with built-in signal conditioning and analog-to-digital conversion circuitry. Supports industry-standard protocols (such as Modbus RTU / TCP) or low-power wide-area network protocols (such as LoRaWAN, NB-IoT).

[0125] Unit identification: Each filler module is embedded with an RFID tag or printed with a unique QR code, which is bound to the built-in sensor ID.

[0126] Automatic actuators: In large or specific designs, lightweight robotic arms or gripping devices can be equipped to automatically locate and replace failed filler modules upon receiving instructions.

[0127] 2. Edge computing and network transport layer: IoT Gateway / Edge Server: Deployed within the wetland field control cabinet. Responsibilities include: 1) Aggregating all sensor data; 2) Performing preliminary data cleaning, caching, and protocol conversion; 3) Performing lightweight feature computation (S2) and state estimation (a simplified version of S3), providing local decision support during network outages; 4) Uploading data to the cloud platform via wired (fiber optic, industrial Ethernet) or wireless (4G / 5G) networks.

[0128] 3. Cloud Platform Core Service Layer: This is the system's "intelligent brain," built on a microservices architecture and containing the following key services: Data access and storage services: Receive and persistently store massive amounts of time-series data from the edge to build a historical database.

[0129] Stream processing and feature computation service: Real-time computation of the high-level feature vector Ri(t) for each unit.

[0130] Mechanism-Data Hybrid Model Service: Instantiates a UKF state estimator for each cell, runs continuously, and outputs Êi(t) and state covariance in real time.

[0131] Lifetime prediction service: Regularly (e.g., daily) start Monte Carlo simulations to update the RUL probability distribution of all cells.

[0132] Optimize decision engine service: Call the optimization solver at a set period (e.g., weekly) to solve MILP problems in S5 and generate maintenance plans.

[0133] Model training and adaptive services: Regularly use historical data to train and calibrate key parameters (such as γ, θ, α, λ) in the mechanistic model offline, enabling the model to evolve itself.

[0134] 4. Application Interaction and Presentation Layer: Web management platform: Provides administrators with a rich set of visual dashboards, including real-time data monitoring, health maps, forecasting and early warning, maintenance plans, report statistics and other functions.

[0135] Mobile App: Provides on-site maintenance personnel with mobile operation support such as work order push, navigation and positioning, QR code confirmation, and replacement record uploading.

[0136] API Interface: Provides standardized data interfaces that can be integrated with higher-level smart water management platforms, asset management systems (EAM), or enterprise resource planning (ERP) systems.

[0137] 5. Security and Operations Management Team: Ensure system communication security (encryption, authentication), data security (backup, disaster recovery), and stable operation (monitoring, alarms).

[0138] Corresponding to the above method embodiments, this specification also provides an embodiment of a modular filler material performance evaluation and precise replacement decision-making device for constructed wetlands. Figure 3 This specification illustrates a schematic diagram of a modular constructed wetland filler performance evaluation and precise replacement decision-making device according to one embodiment. Figure 3 As shown, the device includes: The data acquisition module 201 is configured to collect microenvironmental parameters inside multiple modular filler units in the constructed wetland and influent load parameters of the constructed wetland system. The feature determination module 202 is configured to fuse microenvironmental parameters with spatiotemporally aligned influent load parameters for each modular packing unit, and calculate a feature response vector characterizing the response state of the modular packing unit. The dynamic estimation module 203 is configured to dynamically estimate the real-time performance index and model parameters of each modular packing unit based on a preset packing performance decay mechanism model and using a data assimilation algorithm, with the feature response vector as the input of the observation value. The probability distribution module 204 is configured to predict the performance degradation trajectory of each modular packing unit through numerical simulation based on the estimated real-time performance index, model parameters, and future load assumptions, and to calculate the remaining effective time and probability distribution of the remaining effective time when the modular packing unit reaches the preset failure threshold. The objective solver module 205 is configured to take the probability distribution of the remaining effective time of all modular packing units, the process criticality weight, the replacement cost and the preset constraints as inputs, construct a multi-objective optimization model with the goal of minimizing system performance risk and total operation and maintenance cost, and solve the multi-objective optimization model to generate a replacement plan. Information display module 206 is configured to output and display replacement plans and information related to the health status of modular packing units.

[0139] In one possible implementation, the microenvironment parameters include redox potential, dissolved oxygen concentration, pH value, and temperature; the influent load parameters include influent flow rate, chemical oxygen demand concentration, ammonia nitrogen concentration, total nitrogen concentration, and total phosphorus concentration.

[0140] In one possible implementation, the characteristic response vector includes the redox potential dynamic response index per unit pollutant load, the dissolved oxygen consumption rate index per unit pollutant load, the nitrate load response index based on pH change, and the relative decay index of the hydraulic conductivity coefficient of the packing unit.

[0141] In one possible implementation, the mechanistic model includes the rate of change of the real-time performance index over time, which is negatively correlated with the temperature-related intrinsic decay rate, an acceleration function based on the cumulative effect of historical inflow load shocks, and a power function of the real-time performance index.

[0142] In one possible implementation, unscented Kalman filtering or extended Kalman filtering is used as a data assimilation algorithm to dynamically estimate the real-time performance index and model parameters.

[0143] In one possible implementation, the probability distribution of the remaining effective time is calculated through Monte Carlo simulation, which includes: drawing multiple sets of parameter samples from the posterior distribution of the state estimate, performing multiple forward numerical integrations in conjunction with future load scenarios, and statistically obtaining the probability distribution.

[0144] In one possible implementation, the multi-objective optimization model is a mixed-integer linear programming model, the decision variables are binary variables representing whether to replace a specific packing unit in a specific time window, the objective function is the weighted sum of system performance risk cost and total operation and maintenance cost, and the constraints include inventory constraints, the constraint that each unit can be replaced at most once, and the constraint of the maximum replacement ratio within the same process partition.

[0145] The above is a schematic scheme of a modular wetland packing performance evaluation and precise replacement decision-making device according to this embodiment. It should be noted that the technical solution of this modular wetland packing performance evaluation and precise replacement decision-making device and the technical solution of the aforementioned modular wetland packing performance evaluation and precise replacement decision-making method belong to the same concept. Details not described in detail in the technical solution of the modular wetland packing performance evaluation and precise replacement decision-making device can be found in the description of the aforementioned modular wetland packing performance evaluation and precise replacement decision-making method.

[0146] Figure 4 A structural block diagram of a computing device 300 according to one embodiment of this specification is shown. The components of the computing device 300 include, but are not limited to, a memory 310 and a processor 320. The processor 320 is connected to the memory 310 via a bus 330, and a database 350 is used to store data.

[0147] The computing device 300 also includes an access device 340, which enables the computing device 300 to communicate via one or more networks 360. Examples of these networks include Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or combinations of communication networks such as the Internet. The access device 340 may include one or more of any type of wired or wireless network interface (e.g., a network interface card (NIC)), such as an IEEE 802.11 Wireless Local Area Network (WLAN) wireless interface, a Wi-MAX (Worldwide Interoperability for Microwave Access) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, or a Near Field Communication (NFC) interface.

[0148] In one embodiment of this specification, the aforementioned components of the computing device 300 and Figure 3 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 3 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0149] The computing device 300 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 300 can also be a mobile or stationary server.

[0150] The processor 320 executes computer-executable instructions, which, when executed by the processor, implement the steps of the aforementioned method for evaluating and precisely replacing modular filler material in constructed wetlands. The above is a schematic representation of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the aforementioned method for evaluating and precisely replacing modular filler material in constructed wetlands belong to the same concept. Details not described in detail in the technical solution of the computing device can be found in the description of the technical solution of the aforementioned method for evaluating and precisely replacing modular filler material in constructed wetlands.

[0151] An embodiment of this specification also provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the above-described method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions.

[0152] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium belongs to the same concept as the technical solution of the above-mentioned method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the above-mentioned method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions.

[0153] An embodiment of this specification also provides a computer program, wherein when the computer program is executed in a computer, the computer is instructed to perform the steps of the above-described method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions.

[0154] The above is an illustrative example of a computer program in this embodiment. It should be noted that the technical solution of this computer program belongs to the same concept as the technical solution of the aforementioned method for evaluating and precisely replacing modular filler material in constructed wetlands. Details not described in detail in the computer program's technical solution can be found in the description of the aforementioned method for evaluating and precisely replacing modular filler material in constructed wetlands.

[0155] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0156] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or certain intermediate forms. The computer-readable medium may include any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium may be appropriately added to or subtracted according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media may not include electrical carrier signals and telecommunication signals.

[0157] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments in this specification are not limited to the described order of actions, because according to the embodiments in this specification, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments in this specification.

[0158] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0159] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments described herein. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions, characterized in that, Includes the following steps: The microenvironmental parameters inside multiple modular filler units in the constructed wetland and the influent load parameters of the constructed wetland system were collected. For each modular packing unit, the microenvironment parameters are fused with the influent load parameters that have been spatiotemporally aligned to calculate the feature response vector characterizing the response state of the modular packing unit. Based on a preset packing performance decay mechanism model and using a data assimilation algorithm, the feature response vector is used as the input of the observation value to dynamically estimate the real-time performance index and model parameters of each modular packing unit. Based on the estimated real-time performance index, the model parameters, and future load assumptions, the performance degradation trajectory of each modular packing unit is predicted through numerical simulation, and the remaining effective time of the modular packing unit reaching the preset failure threshold and the probability distribution of the remaining effective time are calculated. Using the probability distribution of the remaining effective time of all the modular packing units, the process criticality weight, the replacement cost, and the preset constraints as input, a multi-objective optimization model is constructed with the goal of minimizing system performance risk and total operation and maintenance cost. The multi-objective optimization model is then solved to generate a replacement plan. Output and display the replacement plan and information related to the health status of the modular packing unit.

2. The method according to claim 1, characterized in that, The microenvironment parameters include redox potential, dissolved oxygen concentration, pH value, and temperature; the influent load parameters include influent flow rate, chemical oxygen demand concentration, ammonia nitrogen concentration, total nitrogen concentration, and total phosphorus concentration.

3. The method according to claim 1, characterized in that, The characteristic response vector includes the dynamic response index of redox potential under unit pollutant load, the dissolved oxygen consumption rate index under unit pollutant load, the response index of unit nitrate load based on pH change, and the relative decay index of hydraulic conductivity coefficient of packing unit.

4. The method according to claim 1, characterized in that, The mechanistic model includes the rate of change of the real-time performance index over time, which is negatively correlated with the temperature-related inherent decay rate, the acceleration function based on the cumulative effect of historical inflow load shocks, and the power function of the real-time performance index.

5. The method according to claim 1, characterized in that, The real-time performance index and the model parameters are dynamically estimated using unscented Kalman filtering or extended Kalman filtering as the data assimilation algorithm.

6. The method according to claim 1, characterized in that, The probability distribution of the remaining effective time is calculated by Monte Carlo simulation, including: drawing multiple sets of parameter samples from the posterior distribution of the state estimate, performing multiple forward numerical integrations in combination with future load scenarios, and statistically obtaining the probability distribution.

7. The method according to claim 1, characterized in that, The multi-objective optimization model is a mixed-integer linear programming model. The decision variables are binary variables representing whether to replace a specific packing unit in a specific time window. The objective function is the weighted sum of system performance risk cost and total operation and maintenance cost. The constraints include inventory constraints, a constraint that each unit can be replaced at most once, and a constraint that the maximum replacement ratio within the same process partition is limited.

8. A device for evaluating the effectiveness of modular filler material in constructed wetlands and for making precise replacement decisions, characterized in that, include: The data acquisition module is configured to collect microenvironmental parameters inside multiple modular filler units in the constructed wetland and the influent load parameters of the constructed wetland system. The feature determination module is configured to, for each of the modular packing units, fuse the microenvironment parameters with the spatiotemporally aligned influent load parameters to calculate a feature response vector characterizing the response state of the modular packing unit. The dynamic estimation module is configured to dynamically estimate the real-time performance index and model parameters of each modular packing unit based on a preset packing performance decay mechanism model and using a data assimilation algorithm, with the feature response vector as the input of the observation value. The probability distribution module is configured to predict the performance degradation trajectory of each modular packing unit through numerical simulation based on the estimated real-time performance index, the model parameters, and future load assumptions, and to calculate the remaining effective time of the modular packing unit when it reaches a preset failure threshold and the probability distribution of the remaining effective time. The objective solution module is configured to take the probability distribution of the remaining effective time of all the modular packing units, the process criticality weight, the replacement cost and the preset constraints as input, construct a multi-objective optimization model with the objective of minimizing system performance risk and total operation and maintenance cost, and solve the multi-objective optimization model to generate a replacement plan; The information display module is configured to output and display the replacement plan and information related to the health status of the modular packing unit.

9. A computing device, characterized in that, include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the modular filler performance evaluation and precise replacement decision method for artificial wetlands according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the method for evaluating the effectiveness of modular filler material in constructed wetlands and making precise replacement decisions as described in any one of claims 1 to 7.

Images