Intelligent sewage multi-stage filtering treatment system based on internet of things

Abstract

This invention relates to the field of Internet of Things (IoT) technology in wastewater treatment, and discloses an intelligent multi-stage wastewater filtration system based on IoT. The system includes modules for data sensing, edge processing, digital twin, intelligent control, and platform monitoring. The data sensing module collects wastewater parameters through a sensor array; the edge processing module cleans the data and calculates the baseline value of the filtration rate and the reference value of the reagent dosage in real time, enabling localized intelligent decision-making; the digital twin module constructs a dynamic simulation model based on the processed data, simulating the mass transfer and reaction kinetics of the filtration process, and outputs a set of optimized operation schemes; the intelligent control module analyzes the schemes to generate control commands; and the platform monitoring module executes the commands and achieves collaborative system management. Through real-time edge computing and digital twin simulation optimization, this system achieves low-latency response and forward-looking precise control of the wastewater treatment process, improving system processing efficiency and operational stability.

Description

Technical Field

[0001] This invention relates to the field of Internet of Things (IoT) technology in wastewater treatment, specifically to an intelligent multi-stage wastewater filtration system based on the Internet of Things. Background Technology

[0002] Traditional multi-stage wastewater filtration systems typically rely on preset programs or human experience for control. These systems generally employ automation schemes based on programmable logic controllers (PLCs), using simple threshold judgments to start / stop equipment or regulate valves. Some remote monitoring systems also exist, capable of uploading sensor data to a central monitoring platform for centralized display and alarms. However, this centralized data processing approach relies entirely on cloud server calculations and feedback for all control logic.

[0003] Existing technical solutions have shortcomings. Because sensor data needs to be transmitted via network to a remote platform for analysis before control commands are sent to field equipment, the entire process involves significant delays. This delay makes the system unable to cope with sudden changes in influent water quality, resulting in lagging feedback control and affecting the stability of effluent water quality. Simple threshold control logic cannot reflect the dynamic relationships between different filtration units and lacks the ability to collaboratively optimize the entire treatment process. Traditional control strategies are based on reactive adjustments using current monitoring data and lack predictive capabilities. The system cannot predict the future trend of filtration performance under current operating parameters, nor can it perform prior evaluation and comparison of multiple control strategies in a virtual environment. This leads to relatively rigid control decisions, making it difficult to achieve an optimal balance between treatment efficiency and energy consumption, and resulting in poor performance when dealing with complex and variable wastewater compositions.

[0004] Current wastewater treatment systems need to address two key issues: improving the real-time performance and local intelligence of system responses, reducing reliance on cloud computing and the impact of network latency; and introducing optimization mechanisms with predictive and simulation capabilities to change the control mode that relies on fixed rules or human experience, thereby achieving a shift from passive response to proactive optimization. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent multi-stage wastewater filtration and treatment system based on the Internet of Things to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides an intelligent multi-stage wastewater filtration and treatment system based on the Internet of Things, the system comprising:

[0007] The data sensing module collects real-time wastewater parameters through a sensor array installed in each stage of the filtration unit. The sensor array includes a pH sensor, a turbidity sensor, and a flow meter. The data sensing module acquires sensor readings at a predetermined sampling frequency and adds device identifiers and timestamps to generate a raw data set.

[0008] The edge processing module receives the raw data set, performs integrity checks and noise filtering on the data, calculates the baseline value of the filtering rate and the reference value of the amount of drug added, and generates a preprocessed data stream.

[0009] Based on the preprocessed data stream, the digital twin module constructs a dynamic simulation model of the virtual filtration system, simulates the material transport and reaction kinetics during the filtration process, adjusts the input of the dynamic simulation model, and outputs an optimized set of solutions.

[0010] The intelligent control module analyzes the control target and generates actuator drive signals based on the optimized scheme set. The drive signals include valve opening adjustment commands and pump power setting commands, forming a control command sequence.

[0011] The platform monitoring module uses the control command sequence to monitor the operating status of the filtering unit through the communication network, execute fault diagnosis routines, and integrate data from multiple units to generate collaborative management commands.

[0012] Preferably, the data sensing module includes:

[0013] The sensor reading acquisition submodule is configured with multiple sensor interfaces to cyclically acquire analog and digital signals from each sensor, convert them into digital readings, and store them in a buffer.

[0014] The data calibration submodule performs deviation correction and linear compensation processing on the digital readings, adjusts the reading error using a calibration curve, and outputs a calibrated data point sequence.

[0015] The data encapsulation submodule packages the calibrated data point sequence according to the communication protocol format, adds the source address and checksum, and forms a transmission data packet;

[0016] The data transmission submodule sends the data packets to the edge processing module via the wireless network, confirms the reception status, and completes the generation of the original data set; the original data set includes an array of sensor readings, a timestamp sequence, and a list of device identifiers.

[0017] Preferably, the edge processing module includes:

[0018] The data receiving submodule parses the transmission data packets in the original data set, extracts the valid data payload, and performs format standardization processing.

[0019] The data cleaning submodule performs outlier detection and missing value imputation on the standardized data, calculates data consistency and integrity scores, and generates a clean dataset.

[0020] The feature extraction submodule calculates time-series and statistical features from the cleaning dataset, including the moving average of the filtration rate and the trend slope of the reagent addition amount;

[0021] The optimization calculation submodule calculates preliminary optimized values ​​of filtration rate and drug dosage based on the time-series and statistical characteristics using a linear regression model, generating a preprocessed data stream; the preprocessed data stream includes the calculated filtration rate value, the estimated drug dosage value, and data quality indicators.

[0022] Preferably, the digital twin module includes:

[0023] The model loading submodule reads the geometric and physical parameters of the filtering system from the database and sets the initial boundary conditions based on the preprocessed data stream.

[0024] The simulation execution submodule runs the differential equation solver of the digital twin model to simulate the flow of sewage through the filter media and the adsorption process of pollutants, and predicts the effluent water quality parameters.

[0025] The optimization analysis submodule compares the simulation output with the target value, and uses the gradient descent algorithm to adjust the operating parameters, including the filtration speed and backwashing frequency, to generate an optimization scheme set. The optimization scheme set includes model parameter adjustment values, simulation output data, and control target descriptions.

[0026] Preferably, the intelligent control module includes:

[0027] The control strategy submodule parses the control objective description in the set of optimization schemes and formulates priority rules and constraints.

[0028] The signal generation submodule calculates the change in valve opening and the increment of pump power according to the control strategy, and generates analog control signals.

[0029] The instruction encoding submodule converts analog control signals into digital instructions, adds device addresses and execution delay times, and assembles them into a control instruction sequence; the control instruction sequence includes valve control code, pump group control code, and execution timing data.

[0030] Preferably, the platform monitoring module includes:

[0031] The status monitoring submodule collects actuator feedback data from the filtering unit in real time, compares the control command sequence with the actual state, and calculates the deviation value.

[0032] The fault diagnosis submodule analyzes the deviation value sequence, matches it with a predefined fault mode library, identifies potential fault types, and generates a diagnostic report.

[0033] The collaborative management submodule integrates diagnostic reports and operational data from multiple units, calculates resource allocation schemes and scheduling plans, and outputs collaborative management instructions; the collaborative management instructions include unit coordination strategies, fault handling codes, and operational mode codes.

[0034] Preferably, the sensor reading acquisition submodule supports configurable sampling frequency and measurement range. The sampling frequency is dynamically adjusted according to the sewage flow rate, and the measurement range is preset according to the sensor type.

[0035] Preferably, the method for constructing the dynamic simulation model in the digital twin module includes:

[0036] The precise geometric structure of each level of the filter unit is obtained through 3D scanning technology, and a 3D mesh model is generated.

[0037] Based on the material property library and chemical reaction kinetic formula of the filter medium, physical properties are assigned to the three-dimensional mesh model and control equations are set.

[0038] The model parameters are calibrated using historical operating data. The model parameters are adjusted by comparing the simulation output values ​​with the actual measured values ​​until the error is lower than the preset threshold.

[0039] Preferably, when the signal generation submodule generates the analog control signal, it adopts the following steps: parsing the priority rules and constraints output by the control strategy submodule; calculating the change in valve opening based on the calculated filtration rate and the estimated amount of reagent added; calculating the increment of pump power based on the pump power setting command and real-time flow data; converting the change and increment into analog signal waveforms, and generating the analog control signal through a digital-to-analog converter.

[0040] Preferably, when the fault diagnosis submodule performs fault diagnosis processing, it adopts the following steps: collecting the actuator feedback data sequence and calculating the deviation value sequence from the control command sequence; matching the deviation value sequence with a predefined fault mode library to identify deviation trend characteristics; when the deviation trend characteristics match a specific fault mode, generating a diagnostic report containing the fault type, location, and severity; and sending the diagnostic report to the collaborative management submodule to generate collaborative management instructions.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] The edge processing module performs real-time calculations at the data acquisition source, generating baseline values ​​for filtration rates and reference values ​​for reagent dosages. By offloading some computational tasks from the cloud to the network edge, the amount of data that needs to be uploaded and its transmission latency are reduced. The data processing flow is optimized, enabling the system to quickly make preliminary adjustments to the filtration rate and reagent dosage based on locally calculated key parameters. This localized intelligence enhances the response speed to fluctuations in influent water quality. Even in extreme cases of network outages, the system can maintain basic stable operation by relying on baseline values ​​calculated at the edge, improving the robustness and autonomy of the entire system.

[0043] The digital twin module constructs a dynamic simulation model of the virtual filtration system, simulating mass transport and reaction kinetics, and outputting a set of optimized solutions. This approach deeply couples the actual physical system with the virtual model, driving the model's operation by importing real-time preprocessed data. The simulation model can reproduce the complex physicochemical reactions within the filtration unit and predict future state changes of the system under different control parameters. This allows control strategy formulation to move beyond being limited to current static data, enabling simulation testing and effect prediction of multiple control schemes in virtual space, thereby selecting the optimal solution. Control behavior shifts from a reaction mode based on fixed rules to an optimization mode based on model predictions and simulation results, achieving forward-looking management of the processing process and refined management of operating parameters, improving processing efficiency and reducing operating costs. Attached Figure Description

[0044] Figure 1 This is a schematic diagram illustrating the working principle of the IoT-based intelligent multi-stage wastewater filtration and treatment system described in this invention.

[0045] Figure 2 A flowchart illustrating how the data perception module works;

[0046] Figure 3 A flowchart illustrating how the edge processing module works;

[0047] Figure 4 This is a graph showing the filtration velocity distribution and density analysis.

[0048] Figure 5 To control for deviation time series trend and moving average analysis chart. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Please see Figure 1 This invention provides an intelligent multi-stage wastewater filtration system based on the Internet of Things (IoT). The system comprises the collaborative operation of a data sensing module, an edge processing module, a digital twin module, an intelligent control module, and a platform monitoring module. The data sensing module collects real-time wastewater parameters through sensor arrays installed in each filtration unit. The sensor arrays include pH sensors, turbidity sensors, and flow meters. The data sensing module acquires sensor readings at a predetermined sampling frequency, adds device identifiers and timestamps, and generates a raw data set. The edge processing module receives the raw data set, performs integrity checks and noise filtering on the data, calculates the baseline value of the filtration rate and the reference value of the reagent dosage, and generates a pre-processed data stream. The digital twin module, based on the pre-processed data stream, constructs a dynamic simulation model of the virtual filtration system, simulates the mass transfer and reaction kinetics during the filtration process, adjusts the input of the dynamic simulation model, and outputs an optimized scheme set. The intelligent control module, based on the optimized scheme set, analyzes the control objectives and generates actuator drive signals, including valve opening adjustment commands and pump power setting commands, forming a control command sequence. The platform monitoring module uses the control command sequence to monitor the operating status of the filtration units through a communication network, executes fault diagnosis routines, and integrates multi-unit data to generate collaborative management commands.

[0051] Example 1: See Figure 2 The implementation of the data sensing module involves the collaborative operation of a sensor reading acquisition submodule, a data calibration submodule, a data encapsulation submodule, and a data transmission submodule. In specific implementation, the sensor reading acquisition submodule is configured with multiple sensor interfaces to connect to a pH sensor, a turbidity sensor, and a flow meter. It cyclically acquires analog and digital signals from each sensor, converts the analog signals into digital readings via an analog-to-digital converter, and stores them in a circular buffer. The digital readings include pH value, turbidity value, and flow rate value. The data calibration submodule performs deviation correction and linear compensation processing on the digital readings, adjusts the reading error using a pre-stored calibration curve, which is generated based on the sensor's factory calibration data, and outputs a calibrated data point sequence. The data encapsulation submodule packages the calibrated data point sequence according to the Modbus communication protocol format, adds a source address field and a cyclic redundancy check code, and forms a transmission data packet. The data transmission submodule sends the transmission data packet to the edge processing module via the LoRa wireless network and waits to receive an acknowledgment frame to confirm the reception status, thus completing the generation of the original data set. The original data set includes a sensor reading array, a timestamp sequence, and a device identifier list. In some embodiments, the sensor reading acquisition submodule supports configurable sampling frequency and measurement range. The sampling frequency is dynamically adjusted according to the sewage flow rate, and the measurement range is preset to a fixed value according to the sensor type. In specific implementations, the dynamic adjustment of the sampling frequency is calculated based on real-time flow data using a linear function, expressed by the following formula:

[0052]

[0053] in: Indicates the sampling frequency. This represents the real-time traffic value. and The preset coefficients and measurement ranges are set to 0-14 for pH sensors, 0-1000 NTU for turbidity sensors, and 0-100 m³ / h for flow meters. Optionally, the sensor reading acquisition submodule supports parallel acquisition of multiple sensor interfaces, switching between different sensor channels via a multiplexer. The deviation correction processing of the data calibration submodule includes zero-point drift compensation and gain adjustment, using a calibration curve to fit the relationship between the sensor output and the actual value. In some embodiments, the data transmission submodule employs a retransmission mechanism to handle transmission failures, automatically retransmitting data packets when no acknowledgment frame is received. Optionally, the data encapsulation submodule supports multiple communication protocol formats, including MQTT and CoAP, to adapt to different network environments. The device identifier list in the original data set uniquely identifies each sensor node for data traceability and processing in subsequent modules.

[0054] Example 2: See Figure 3 The implementation of the edge processing module involves the collaborative operation of a data receiving submodule, a data cleaning submodule, a feature extraction submodule, and an optimization calculation submodule. In specific implementation, the data receiving submodule receives transmission data packets from the data sensing module via an Ethernet interface, parses the protocol header of the transmission data packets, and extracts the valid data payload. The valid data payload includes a sensor reading array, a timestamp sequence, and a device identifier list. The extracted valid data payload is format-standardized, converting data from different sources into standardized data with the same time base and field structure. The data cleaning submodule performs outlier detection and missing value imputation on the standardized data. Outlier detection uses a method based on three standard deviations to identify and remove values ​​that deviate from the normal range. Missing value imputation uses linear interpolation to fill gaps in the data sequence. Data consistency and integrity scores are calculated. The data consistency score is obtained by comparing the logical relationships of data from different sensors at the same time, and the data integrity score is obtained by statistically analyzing the proportion of valid data points to the total sampling points. A clean data set containing valid data points, timestamps, and consistency markers is generated.

[0055] The feature extraction submodule calculates temporal and statistical features from the cleaning dataset. The temporal features include the moving average of the filtration rate, and the statistical features include the trend slope of the pesticide dosage. The moving average of the filtration rate is calculated using the arithmetic mean within a fixed time window, and the trend slope of the pesticide dosage is obtained by fitting a linear trend line using the least squares method. The optimization calculation submodule, based on the temporal and statistical features, applies a linear regression model to calculate the preliminary optimized values ​​for the filtration rate and pesticide dosage. The expression for the linear regression model is:

[0056]

[0057] in: This indicates the estimated amount of medicine to be added. This represents the calculated filtration rate. Represents the regression coefficient. The model intercept is represented, and a preprocessed data stream is generated containing calculated filtering rates, estimated drug dosage, and data quality indicators, which combine data consistency and data integrity scores. In some embodiments, the outlier detection in the data cleaning submodule uses a sliding window mechanism, with the window size dynamically adjusted based on the data sampling frequency. Optionally, the feature extraction submodule supports the calculation of various statistical features, including standard deviation and coefficient of variation. It is understood that the linear regression model parameters in the optimization calculation submodule are obtained through training on historical running data. In some embodiments, the data receiving submodule supports the parsing of various network protocols, including TCP and UDP. Optionally, the missing value imputation processing in the data cleaning submodule supports various interpolation methods, including spline interpolation and nearest neighbor interpolation. In the missing value imputation process of the data cleaning submodule, spline interpolation involves constructing a smooth interpolation function based on time-series data points. It uses piecewise polynomial fitting to adjacent valid data points to ensure the continuity and differentiability of the imputed value. The interpolation process first identifies the data point sequence before and after the missing value, calculates the spline coefficients, generates the interpolation curve, and finally outputs the complete imputed data sequence. Nearest neighbor interpolation, on the other hand, employs a simple and fast filling strategy, directly selecting the valid data point value with the closest timestamp to the missing value as the imputed value. This is suitable for scenarios with high data sampling frequency and gradual changes. During implementation, the nearest neighbor point is determined by calculating the time distance, and its value is copied to the missing position. Both interpolation methods are integrated into the algorithm library of the data cleaning submodule, automatically selected or manually configured according to data characteristics to improve the integrity and consistency of the cleaned data set. It can be understood that the data structure of the preprocessed data stream includes header information and a payload. The header information records the data generation time and processing status.

[0058] Example 3: The implementation of the digital twin module involves the collaborative operation of a model loading submodule, a simulation execution submodule, and an optimization analysis submodule. In specific implementation, the model loading submodule reads the geometric and physical parameters of the filtration system from the system database. The geometric parameters include the structural dimensions and volume data of the filter units, and the physical parameters include the porosity and density data of the filter media. Initial boundary conditions are set based on the preprocessed data stream from the edge processing module. The initial boundary conditions include the influent concentration distribution and the initial flow velocity field. The simulation execution submodule runs the differential equation solver of the digital twin model. The solver is based on the finite volume method to discretize the computational domain and simulate the flow process of wastewater through the filter media and the adsorption process of pollutants. The flow process is described by the Navier-Stokes equations, and the adsorption process is described by the Langmuir isothermal adsorption model. The predicted effluent water quality parameters include turbidity and pollutant concentration. The residual pollutant concentration is analyzed by comparing the simulation output with the preset target value in the optimization analysis submodule. The specific implementation of the Langmuir isotherm adsorption model is integrated in the simulation execution submodule of the digital twin module. Adsorption parameters such as maximum adsorption capacity and adsorption equilibrium constant are obtained from the material property library of the filter medium. These parameters are preset based on the physicochemical properties of the filter medium. The model uses these parameters to calculate the adsorption equilibrium state of pollutants on the surface of the filter medium, simulates the change of pollutant concentration over time during adsorption, and thus predicts the residual pollutant concentration in the effluent water quality parameters. The model parameters are calibrated using historical operating data. The adsorption parameters are adjusted by comparing the simulation output with the actual measured values ​​until the error is lower than the preset threshold, ensuring that the simulated adsorption process is consistent with the actual filtration behavior. The model output is used by the optimization analysis submodule to compare the simulation value with the target value to adjust the filtration rate and backwashing frequency. The gradient descent algorithm is used to adjust the operating parameters, including the filtration rate and backwashing frequency. Through iterative calculation, the simulation output value gradually approaches the target value, generating an optimization scheme set containing the model parameter adjustment values, simulation output data, and a description of the control target.

[0059] The method for constructing the dynamic simulation model in the digital twin module includes obtaining the precise geometric structure of each level of filter unit through 3D laser scanning technology, generating a 3D mesh model containing node coordinates and unit connection relationships, assigning physical properties to the 3D mesh model and setting control equations based on the material property library and chemical reaction kinetic formulas of the filter media, the material property library containing permeability coefficient and adsorption capacity data of different filter materials, the chemical reaction kinetic formulas describing the relationship between pollutant degradation rate and concentration, and the control equations describing mass conservation and momentum conservation. The model parameters are calibrated using historical operating data, and the model parameters are adjusted by comparing the simulation output values ​​with the actual measured values. The actual measured values ​​are obtained from sensor data collected during system operation. The model parameters are adjusted until the average absolute error between the simulation output values ​​and the actual measured values ​​is lower than a preset threshold. In some embodiments, the differential equation solver of the simulation execution submodule uses a semi-implicit algorithm of pressure coupling equations to handle speed-pressure coupling problems. Optionally, the model loading submodule supports the import of multiple geometric file formats, including STL and STEP formats. It is understood that the gradient descent algorithm used in the optimization analysis submodule includes an adaptive learning rate adjustment mechanism. In some embodiments, the chemical reaction kinetic formula used in the construction of the dynamic simulation model is expressed as:

[0060]

[0061] in: Indicates the rate of pollutant degradation. Represents the reaction rate constant. Indicates pollutant concentration. This indicates the reaction order. Optionally, the 3D mesh model generation process includes a mesh quality check step, examining the twist and aspect ratio of the mesh cells. It can be understood that the model parameter calibration process uses the least squares method to optimize the objective function, which is defined as the sum of squared errors between the simulation output values ​​and the actual measured values.

[0062] See Figure 4 This graph is a key component of the digital twin module's quantification of the filtration process characteristics. The model loading submodule of the digital twin module reads the geometric and physical parameters of the filtration system, while the simulation execution submodule simulates the flow of wastewater in the filter media using a differential equation solver, outputting filtration velocity data. A histogram presents the frequency distribution of the filtration velocity, and a kernel density curve fits its probability distribution characteristics. This data provides the foundation for the optimization analysis submodule: it uses a gradient descent algorithm to adjust operating parameters based on the filtration velocity distribution, generating a set of optimized solutions. This graph intuitively reflects the statistical regularity of the filtration velocity and is a core technological manifestation of the digital twin module's ability to achieve dynamic simulation of the filtration process and support the shift of control strategies from passive response to active optimization. It helps the system complete a priori evaluation of operating parameters in a virtual environment, ultimately achieving the optimal balance between treatment efficiency and energy consumption.

[0063] Example 4: The implementation of the intelligent control module involves the collaborative operation of a control strategy submodule, a signal generation submodule, and an instruction encoding submodule. In specific implementation, the control strategy submodule parses the control target description from the optimization scheme set of the digital twin module. The control target description includes the target filtration rate and the target effluent water quality parameters, and formulates priority rules and constraints. The priority rules define the execution order of different control targets, and the constraints limit the operating range of valve opening and pump power. The signal generation submodule calculates the change in valve opening and the increment of pump power according to the control strategy. The change in valve opening is based on the calculated value of the filtration rate. The estimated dosage of the reagent is calculated using a linear function. The increase in pump power is derived from the pump power setting command and real-time flow data through a proportional relationship, generating an analog control signal. The analog control signal is represented as a continuous voltage or current waveform. The instruction encoding submodule converts the analog control signal into a digital instruction. The conversion process uses an analog-to-digital converter to quantize the analog signal into a digital value. The device address and execution delay time are added to the digital instruction. The device address identifies the target actuator, and the execution delay time defines the time offset for sending the instruction. The instructions are then assembled into a control instruction sequence, which includes valve control code, pump control code, and execution timing data.

[0064] The signal generation submodule generates analog control signals using the following steps: parsing the priority rules and constraints output by the control strategy submodule; calculating the change in valve opening based on the calculated filtration rate and estimated reagent dosage; calculating the increment of pump power based on the pump power setting command and real-time flow data; converting the change and increment into analog signal waveforms, and generating analog control signals via a digital-to-analog converter. In specific implementation, the formula for calculating the change in valve opening is:

[0065]

[0066] in: This indicates the change in valve opening. This represents the calculated filtration rate. This indicates the estimated amount of medicine to be added. and The weighting coefficients are used to calculate the incremental pump power based on the deviation ratio between real-time flow data and the reference flow rate. The analog signal waveform is generated using sine or square wave modulation, and the digital-to-analog converter converts the digital values ​​into an analog voltage signal output. In some embodiments, the priority rules of the control strategy submodule include that emergency control commands take precedence over regular adjustment commands. Optionally, the command encoding submodule supports multiple communication protocol encodings, including Modbus RTU and Profibus DP. It is understood that the analog control signal generation process of the signal generation submodule includes a signal filtering step to eliminate high-frequency noise. In some embodiments, referring to Table 1, the execution timing data of the control command sequence is generated synchronously based on the system clock. Optionally, the constraints of the control strategy submodule include minimum and maximum valve opening limits. It is understood that the digital command structure of the command encoding submodule includes a start bit, data bits, and a check bit.

[0067] Table 1: Control Priority Rule Table

[0068]

[0069] Example 5: The implementation of the platform monitoring module involves the collaborative operation of a status monitoring submodule, a fault diagnosis submodule, and a collaborative management submodule. In specific implementation, the status monitoring submodule collects actuator feedback data from the filter unit in real time through a fieldbus interface. The actuator feedback data includes valve position signals and pump current values. It compares the control command sequence from the intelligent control module with the actual state. The control command sequence includes valve control codes and pump control codes. The actual state is parsed from the actuator feedback data, and the deviation value is calculated. The deviation value is the absolute difference between the set value in the control command sequence and the measured value of the actual state. The fault diagnosis submodule analyzes the deviation value sequence. The deviation value sequence is a set of continuous deviation values ​​arranged in chronological order. It is matched with a predefined fault mode library, which stores the deviation mode characteristics of typical faults. Potential fault types are identified and diagnostic reports are generated. The diagnostic reports include fault type codes, occurrence location identifiers, and severity levels. The collaborative management submodule integrates the diagnostic reports and operational data of multiple units. The operational data includes pressure and flow readings of the filter unit. It calculates resource allocation schemes and scheduling plans. The resource allocation scheme defines the processing load allocation of each unit. The scheduling plan arranges the maintenance operation schedule and outputs collaborative management instructions. The collaborative management instructions include unit coordination strategies, fault handling codes, and operating mode codes.

[0070] The fault diagnosis submodule performs the following steps when performing fault diagnosis: It collects the actuator feedback data sequence, calculates the deviation value sequence from the control command sequence (obtained by comparing actual values ​​with command values ​​point-by-point), performs pattern matching between the deviation value sequence and a predefined fault mode library, calculates the similarity between the deviation value sequence and templates in the library, and identifies deviation trend characteristics, including the rising slope or oscillation frequency; when the deviation trend characteristics match a specific fault mode, it generates a diagnostic report containing the fault type, location, and severity; and sends the diagnostic report to the collaborative management submodule to generate collaborative management instructions. In practice, pattern matching uses a formula to calculate the similarity score.

[0071]

[0072] in: This represents the similarity score with the j-th failure mode. This represents the i-th deviation value. This represents the i-th reference deviation value in the j-th fault mode. This represents the weight coefficient of the i-th point. The sequence length is represented by a similarity score, with lower scores indicating higher matching. The fault mode library is generated through training on historical fault data and includes typical patterns of valve jamming and pump overload. In some embodiments, the deviation value calculation of the status monitoring submodule uses a sliding window averaging process to smooth instantaneous noise. Optionally, the resource allocation scheme of the collaborative management submodule is dynamically adjusted based on a load balancing algorithm. The specific implementation of the load balancing algorithm is integrated into the collaborative management submodule. By integrating diagnostic reports and operational data from multiple units, it analyzes the processing load status of each filtration unit in real time, such as flow data and pressure readings, and dynamically calculates the resource allocation scheme. The algorithm adjusts the sewage inflow allocation strategy based on the current load level of the unit, and optimizes task scheduling using a polling mechanism or the minimum load priority principle to ensure balanced workload of each unit in the system, thereby improving overall processing efficiency and outputting collaborative management instructions. It can be understood that the deviation trend feature identification of the fault diagnosis submodule uses a moving average filter to process the data sequence. In some embodiments, the predefined fault mode library supports online updates to adapt to new fault types. Optionally, the actuator feedback data acquisition cycle of the status monitoring submodule is synchronized with the control instruction sequence transmission cycle. It is understandable that the scheduling plan generation of the collaborative management submodule takes into account the equipment maintenance history and processing priority.

[0073] See Figure 5This graph serves as the core data carrier for the platform monitoring module's fault diagnosis and collaborative management. The platform monitoring module's status monitoring submodule collects real-time feedback data from the filter unit actuators, compares it with the control command sequence from the intelligent control module, and calculates the real-time deviation value. A 3-hour moving average smooths the deviation sequence, eliminating instantaneous noise and highlighting trend characteristics. The fault diagnosis submodule analyzes the deviation trend presented in the graph, matches it to a predefined fault mode library, identifies potential fault types such as valve jamming and pump overload, and generates diagnostic reports. The collaborative management submodule integrates this data from multiple units, calculates resource allocation schemes and scheduling plans, and outputs collaborative management commands. Its value lies in upgrading the system from simple status monitoring to predictive fault diagnosis and collaborative optimization capabilities, providing data support for the stable operation of the wastewater treatment system, rapid fault handling, and efficient resource scheduling. It represents a key technological advancement in the platform monitoring module's transition from status awareness to proactive management.

[0074] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0075] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent multi-stage wastewater filtration and treatment system based on the Internet of Things, characterized in that, The system includes: The data sensing module collects real-time wastewater parameters through a sensor array installed in each stage of the filtration unit. The sensor array includes a pH sensor, a turbidity sensor, and a flow meter. The data sensing module acquires sensor readings at a predetermined sampling frequency and adds device identifiers and timestamps to generate a raw data set. The edge processing module receives the raw data set, performs integrity checks and noise filtering on the data, calculates the baseline value of the filtering rate and the reference value of the amount of drug added, and generates a preprocessed data stream. Based on the preprocessed data stream, the digital twin module constructs a dynamic simulation model of the virtual filtration system, simulates the material transport and reaction kinetics during the filtration process, adjusts the input of the dynamic simulation model, and outputs an optimized set of solutions. The intelligent control module analyzes the control target and generates actuator drive signals based on the optimized scheme set. The drive signals include valve opening adjustment commands and pump power setting commands, forming a control command sequence. The platform monitoring module uses the control command sequence to monitor the operating status of the filtering unit through the communication network, execute fault diagnosis routines, and integrate data from multiple units to generate collaborative management commands.

2. The intelligent multi-stage wastewater filtration system based on the Internet of Things according to claim 1, characterized in that, The data sensing module includes: The sensor reading acquisition submodule is configured with multiple sensor interfaces to cyclically acquire analog and digital signals from each sensor, convert them into digital readings, and store them in a buffer. The data calibration submodule performs deviation correction and linear compensation processing on the digital readings, adjusts the reading error using a calibration curve, and outputs a calibrated data point sequence. The data encapsulation submodule packages the calibrated data point sequence according to the communication protocol format, adds the source address and checksum, and forms a transmission data packet; The data transmission submodule sends the data packets to the edge processing module via the wireless network, confirms the reception status, and completes the generation of the original data set; the original data set includes an array of sensor readings, a timestamp sequence, and a list of device identifiers.

3. The intelligent multi-stage wastewater filtration and treatment system based on the Internet of Things according to claim 1, characterized in that, The edge processing module includes: The data receiving submodule parses the transmission data packets in the original data set, extracts the valid data payload, and performs format standardization processing. The data cleaning submodule performs outlier detection and missing value imputation on the standardized data, calculates data consistency and integrity scores, and generates a clean dataset. The feature extraction submodule calculates time-series and statistical features from the cleaning dataset, including the moving average of the filtration rate and the trend slope of the reagent addition amount; The optimization calculation submodule calculates preliminary optimized values ​​of filtration rate and drug dosage based on the time-series and statistical characteristics using a linear regression model, generating a preprocessed data stream; the preprocessed data stream includes the calculated filtration rate value, the estimated drug dosage value, and data quality indicators.

4. The intelligent multi-stage wastewater filtration and treatment system based on the Internet of Things according to claim 1, characterized in that, The digital twin module includes: The model loading submodule reads the geometric and physical parameters of the filtering system from the database and sets the initial boundary conditions based on the preprocessed data stream. The geometric parameters include the structural dimensions and volume data of the filter unit, and the physical parameters include the porosity and density data of the filter medium. The simulation execution submodule runs the differential equation solver of the digital twin model to simulate the flow of sewage through the filter media and the adsorption process of pollutants, and predicts the effluent water quality parameters. The optimization analysis submodule compares the simulation output with the target value, and uses the gradient descent algorithm to adjust the operating parameters, including the filtration speed and backwashing frequency, to generate an optimization scheme set. The optimization scheme set includes model parameter adjustment values, simulation output data, and control target descriptions.

5. The intelligent multi-stage wastewater filtration and treatment system based on the Internet of Things according to claim 1, characterized in that, The intelligent control module includes: The control strategy submodule parses the control objective description in the set of optimization schemes and formulates priority rules and constraints. The signal generation submodule calculates the change in valve opening and the increment of pump power according to the control strategy, and generates analog control signals. The instruction encoding submodule converts analog control signals into digital instructions, adds device addresses and execution delay times, and assembles them into a control instruction sequence; the control instruction sequence includes valve control code, pump group control code, and execution timing data.

6. The intelligent multi-stage wastewater filtration system based on the Internet of Things according to claim 1, characterized in that, The platform monitoring module includes: The status monitoring submodule collects actuator feedback data from the filtering unit in real time, compares the control command sequence with the actual state, and calculates the deviation value. The fault diagnosis submodule analyzes the deviation value sequence, matches it with a predefined fault mode library, identifies potential fault types, and generates a diagnostic report. The collaborative management submodule integrates diagnostic reports and operational data from multiple units, calculates resource allocation schemes and scheduling plans, and outputs collaborative management instructions; the collaborative management instructions include unit coordination strategies, fault handling codes, and operational mode codes.

7. The IoT-based intelligent multi-stage wastewater filtration system according to claim 2, characterized in that, The sensor reading acquisition submodule supports configurable sampling frequency and measurement range. The sampling frequency is dynamically adjusted according to the sewage flow rate, and the measurement range is preset according to the sensor type.

8. The intelligent multi-stage wastewater filtration and treatment system based on the Internet of Things according to claim 1, characterized in that, The method for constructing the dynamic simulation model in the digital twin module includes: The precise geometric structure of each level of the filter unit is obtained through 3D scanning technology, and a 3D mesh model is generated. Based on the material property library and chemical reaction kinetic formula of the filter medium, physical properties are assigned to the three-dimensional mesh model and control equations are set. The model parameters are calibrated using historical operating data. The model parameters are adjusted by comparing the simulation output values ​​with the actual measured values ​​until the error is lower than the preset threshold.

9. The intelligent multi-stage wastewater filtration system based on the Internet of Things according to claim 5, characterized in that, When the signal generation submodule generates the analog control signal, it adopts the following steps: parsing the priority rules and constraints output by the control strategy submodule; calculating the change in valve opening based on the calculated filtration rate and the estimated amount of reagent added; The increment of pump power is calculated based on the pump power setting command and real-time flow data; the change and increment are converted into analog signal waveforms, and analog control signals are generated through a digital-to-analog converter.

10. The intelligent multi-stage wastewater filtration system based on the Internet of Things according to claim 6, characterized in that, When the fault diagnosis submodule performs fault diagnosis processing, it adopts the following steps: collecting the actuator feedback data sequence, calculating the deviation value sequence from the control command sequence; matching the deviation value sequence with a predefined fault mode library to identify deviation trend characteristics; When the deviation trend characteristics match a specific fault mode, a diagnostic report containing the fault type, location, and severity is generated; the diagnostic report is then sent to the collaborative management submodule to generate collaborative management instructions.

Images