A multi-terminal environmental protection equipment collaborative control method based on digital twinning

Abstract

The application relates to the technical field of device control, in particular to a multi-terminal environmental protection device cooperative control method based on digital twinning, which comprises the following steps: collecting real-time operation parameters of multi-terminal environmental protection devices and inputting the real-time operation parameters into a digital twinning model, identifying a dynamic coupling relationship to establish a dynamic control model; connecting the model to a model predictive controller, calculating time delay nodes of load peak transmission, predicting an extrusion pressure evolution track and comparing the extrusion pressure evolution track with a limit pressure bearing threshold value to generate control boundary conditions; solving a multivariable optimal cooperative control sequence based on the time delay nodes and the control boundary conditions, and adjusting conveying rates and buffer loads of related devices in advance before a load peak arrives. The application quantifies time delay nodes in a long-distance material link, converts lagging feedback into leading feedforward adjustment, simultaneously adaptively plans a control envelope to suppress transient impact, and realizes cooperative control of multiple devices.

Description

Technical Field

[0001] This invention relates to the field of equipment control technology, specifically to a collaborative control method for multi-terminal environmental protection equipment based on digital twins. Background Technology

[0002] In modern environmental solid-liquid separation and continuous material handling engineering, the process flow typically consists of multiple continuously operating processing units connected in series. For example, a typical processing flow includes an adsorption unit at the front end, a conveying and physical sedimentation unit in the middle, and a high-pressure filtration unit at the end of the flow. With the development of industrial automation, most of these environmental protection devices are now equipped with basic sensors and actuators, enabling them to operate automatically.

[0003] Currently, for such multi-terminal environmental protection equipment, most existing control systems adopt traditional single-loop feedback control (such as PID regulation) or simple interlocking start-stop control based on local liquid level / pressure. This distributed control architecture can maintain basic operation under stable conditions, but it exposes significant limitations when facing complex actual conditions. Specifically, environmental protection materials exhibit extremely significant and nonlinear process time delays during the physical flow across equipment and over long distances. When the upstream feed end experiences sudden changes in flow rate or concentration due to process fluctuations, this disturbance will spread downstream along the spatial link. Existing feedback control mechanisms lack a global feedforward adjustment loop and can only wait for the load peak to actually impact the downstream terminal equipment, causing local sensors to detect a surge in pressure or load before the controller passively calculates and outputs adjustment commands. This severe control lag not only fails to effectively decouple the dynamic coupling variables between equipment at each level but also easily leads to pump stalling, filter plate damage, or control system collapse in the terminal equipment due to instantaneous overload. To address this, a collaborative control method for multi-terminal environmental protection equipment based on digital twins is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide a collaborative control method for multi-terminal environmental protection equipment based on digital twins, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A collaborative control method for multi-terminal environmental protection equipment based on digital twins includes: Real-time operating parameters of the control loop of multi-terminal environmental protection equipment are collected. The multi-terminal environmental protection equipment includes, in sequence, a primary adsorption device, a variable frequency conveying device, a physical sedimentation device, and a terminal filter press device in the material handling link. The real-time operating parameters are input into a digital twin model to identify the dynamic coupling relationship between each device node and establish a dynamic control model. The dynamic control model is connected to the model prediction controller. Based on the load disturbance change at the inlet of the primary adsorption device, the time delay node of the load peak being transmitted to each downstream device is calculated along the material connection sequence, and the extrusion pressure evolution trajectory of the terminal filter press is predicted. The extrusion pressure evolution trajectory is compared with the control deviation of the ultimate pressure threshold of the terminal filter press to generate the control boundary conditions of the terminal filter press. Based on the time delay node and control boundary conditions, a multivariate optimal collaborative control sequence including action timing is calculated and issued for execution; using the control boundary conditions as global constraints, the conveying rate of the variable frequency conveyor and the buffer load of the physical settlement equipment are adjusted in advance, and collaborative control is performed before the load peak arrives.

[0006] Preferably, the real-time operating parameters include the inlet flow rate and feed concentration of the primary adsorption device, the motor speed and conveying flow rate of the variable frequency conveying device, the buffer level and bottom sludge concentration of the physical sedimentation device, and the feed pressure and real-time extrusion pressure of the terminal filter press. The real-time operating parameters are input into the digital twin model to identify the dynamic coupling relationship between each device node and establish a dynamic control model.

[0007] Preferably, the digital twin model is a mechanism-data hybrid driven model that integrates the device's operating mechanism and historical operating data; Virtual equipment nodes containing operational mechanisms are pre-trained using historical operational data. The inlet flow rate, feed concentration, motor speed, conveying flow rate, buffer level, bottom sludge concentration, feed pressure, and real-time extrusion pressure are synchronously input into the virtual equipment nodes for full-link virtual simulation. The mass transfer coefficient is calculated using the inlet flow rate and feed concentration as initial disturbance boundaries. Combined with the time-series correlation analysis of the motor speed, conveying flow rate, buffer level, and bottom sludge concentration, the dynamic coupling relationship between equipment nodes is identified. The dynamic coupling relationship is transformed into a state-space matrix composed of matrix parameters as the dynamic control model, and the residual data between the simulated predicted value and the actual measured value of the feed pressure and the real-time extrusion pressure is calculated. The matrix parameters are then updated online using the residual data.

[0008] Preferably, the model predictive controller calls the dynamic control model as the predictive equation for rolling optimization; The abrupt slope and duration of the real-time operating parameters at the inlet of the primary adsorption device are extracted, and the load peak characteristics are quantified as the load disturbance. Based on the material connection sequence, the load peak characteristics are input into the prediction equation. Combining the real-time conveying flow rate of the frequency converter and the available buffer volume of the physical sedimentation device, the transmission delay and retention delay in the material transmission process are calculated step by step. The delays at each level are accumulated in time sequence to solve for the time delay node of the load peak being transmitted to each downstream device. Within the set prediction time domain of the model prediction controller, the predicted arrival load of the terminal filter press based on the time delay node is substituted into the prediction equation, mapped as a pressure state variable, and the extrusion pressure evolution trajectory under the future time series is obtained by iterative solution.

[0009] Preferably, the control deviation comparison process involves comparing the predicted pressure values ​​at each time point in the extrusion pressure evolution trajectory with the ultimate pressure threshold of the terminal filter press within a set prediction time domain. The pressure difference exceeding the ultimate pressure threshold in the differential comparison results is extracted as the control deviation, and the overpressure time node corresponding to the control deviation is extracted. Using the dynamic control model, the control deviation is back-calculated as the load reduction amount at the feed end of the terminal filter press required to offset the control deviation. Combining the overpressure time node and the load reduction amount at the feed end, the maximum allowable feed flow rate and maximum feed pressure of the terminal filter press when the load peak is reached are calculated and used as the control boundary conditions of the terminal filter press.

[0010] Preferably, the process for determining the ultimate pressure threshold of the terminal filter press is to extract the total number of historical pressing cycles of the terminal filter press and the pressure drop slope during the holding stage to calculate the real-time health coefficient, and then multiply the static mechanical rated pressure of the equipment by the real-time health coefficient to obtain the basic safe pressure threshold after derated. Extract the sludge specific resistance and material compressibility coefficient of the physical sedimentation equipment, and use the product of the sludge specific resistance and material compressibility coefficient as a characteristic variable to calculate the local stress concentration coefficient of the material applied to the filter plate at the end of dewatering; divide the basic safe pressure bearing threshold by the local stress concentration coefficient for dynamic numerical compensation to calculate the ultimate pressure bearing threshold.

[0011] Preferably, the multivariable optimal collaborative control sequence uses the motor frequency adjustment of the variable frequency conveying equipment, the bottom sludge discharge rate of the physical sedimentation equipment, and the feed pump speed of the terminal filter press as collaborative control variables; combined with the control boundary conditions, multi-objective numerical optimization is performed within the set controller prediction time domain to calculate the target action values ​​and corresponding trigger time nodes of each collaborative control variable on the future time axis, forming a time-series control matrix containing discrete control commands and execution order.

[0012] Preferably, the buffer load is the current storage material holding amount calculated by multiplying the real-time buffer level of the physical settling device by the geometric dimensions of the settling zone; the advance adjustment of the buffer load of the physical settling device is to increase the bottom sludge discharge rate of the physical settling device in advance before the time delay node when the load peak reaches the physical settling device, based on the predicted total load peak input, thereby reducing the current storage material holding amount to the preset safe storage volume limit and expanding the effective volume space of the physical settling device to receive the load peak.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention embeds a digital twin model into a model predictive controller, accurately quantifying the time delay nodes of upstream load disturbances during long-distance material transport, thereby transforming traditional lag-based feedback control into proactive feedforward regulation. This control strategy effectively overcomes the control blind spots caused by nonlinear transmission time delays in multi-device series operation, enabling the control system to generate action sequences in advance before the load peak actually impacts downstream equipment, fundamentally avoiding the problem of untimely regulation caused by the controller passively waiting for sensor alarms.

[0014] 2. This invention combines the material connection sequence of multi-terminal environmental protection equipment, uses predictive equations to iteratively solve the pressure evolution trajectory of the terminal equipment, and dynamically generates control boundary conditions that match the real-time operating conditions of the equipment. Compared with the traditional simple interlocking start-stop based on a single static mechanical rating, this method can adaptively plan the safety control envelope according to the input disturbance, effectively suppressing the transient impact of sudden load changes on the terminal filter press equipment, and preventing hard mechanical failures such as pump stalling and filter plate damage caused by instantaneous overload.

[0015] 3. This invention achieves dynamic variable decoupling and multi-objective coordination of the entire process link by using time difference to proactively adjust the conveying rate of the variable frequency conveyor and the buffer load of the physical settling equipment after calculating the optimal multivariable coordinated control sequence. This mechanism proactively reduces the holding capacity to expand the receiving space before the load peak reaches the midstream node, cleverly utilizing the volumetric characteristics of the spatial link to smooth out upstream disturbances. This not only ensures the overall stability of the material handling chain but also eliminates control conflicts that easily arise when local control loops of each level of equipment operate independently. Attached Figure Description

[0016] Figure 1 A schematic diagram of a collaborative control method for multi-terminal environmental protection equipment based on digital twins is provided in this embodiment of the invention. Figure 2 This is a schematic diagram illustrating the digital twin model construction and online update process provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the process for generating control boundary conditions for a terminal filter press provided in an embodiment of the present invention. Detailed Implementation

[0017] 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.

[0018] Please see Figure 1 This invention provides a collaborative control method for multi-terminal environmental protection equipment based on digital twins, the technical solution of which is as follows: A collaborative control method for multi-terminal environmental protection equipment based on digital twins includes: Real-time operating parameters of the control loop of multi-terminal environmental protection equipment are collected. The multi-terminal environmental protection equipment includes, in sequence, a primary adsorption device, a variable frequency conveying device, a physical sedimentation device, and a terminal filter press device in the material handling link. The real-time operating parameters are input into a digital twin model to identify the dynamic coupling relationship between each device node and establish a dynamic control model. The dynamic control model is connected to the model prediction controller. Based on the load disturbance change at the inlet of the primary adsorption device, the time delay node of the load peak being transmitted to each downstream device is calculated along the material connection sequence, and the extrusion pressure evolution trajectory of the terminal filter press is predicted. The extrusion pressure evolution trajectory is compared with the control deviation of the ultimate pressure threshold of the terminal filter press to generate the control boundary conditions of the terminal filter press. Based on the time delay node and control boundary conditions, a multivariate optimal collaborative control sequence including action timing is calculated and issued for execution; using the control boundary conditions as global constraints, the conveying rate of the variable frequency conveyor and the buffer load of the physical settlement equipment are adjusted in advance, and collaborative control is performed before the load peak arrives.

[0019] Example 1: Real-time operating parameters of the control loop of multi-terminal environmental protection equipment are collected. The multi-terminal environmental protection equipment includes, in sequence, a primary adsorption device, a variable frequency conveying device, a physical sedimentation device, and a terminal filter press device in the material handling link. The real-time operating parameters are input into a digital twin model to identify the dynamic coupling relationship between each device node and establish a dynamic control model.

[0020] See Figure 2The real-time operating parameters include the inlet flow rate and feed concentration of the primary adsorption device, the motor speed and conveying flow rate of the variable frequency conveyor, the buffer level and bottom sludge concentration of the physical sedimentation device, and the feed pressure and real-time extrusion pressure of the terminal filter press. The real-time operating parameters are input into the digital twin model to identify the dynamic coupling relationship between each device node and establish a dynamic control model.

[0021] The digital twin model is a mechanism-data hybrid driven model that integrates the equipment's operating mechanism and historical operating data; Virtual equipment nodes containing operational mechanisms are pre-trained using historical operational data. The inlet flow rate, feed concentration, motor speed, conveying flow rate, buffer level, bottom sludge concentration, feed pressure, and real-time extrusion pressure are synchronously input into the virtual equipment nodes for full-link virtual simulation. The mass transfer coefficient is calculated using the inlet flow rate and feed concentration as initial disturbance boundaries. Combined with the time-series correlation analysis of the motor speed, conveying flow rate, buffer level, and bottom sludge concentration, the dynamic coupling relationship between equipment nodes is identified. The dynamic coupling relationship is transformed into a state-space matrix composed of matrix parameters as the dynamic control model, and the residual data between the simulated predicted value and the actual measured value of the feed pressure and the real-time extrusion pressure is calculated. The matrix parameters are then updated online using the residual data.

[0022] Specifically, the inlet flow rate and feed concentration are obtained through electromagnetic flow meters and ultrasonic concentration meters installed on the feed pipe of the primary adsorption equipment; the motor speed and conveying flow rate of the variable frequency conveying equipment are obtained through the motor encoder built into the frequency converter and the flow meter on the conveying pipe; the buffer level and bottom sludge concentration are obtained through ultrasonic level gauges installed on the inner wall of the physical sedimentation equipment and sludge concentration detectors at the bottom of the equipment; and the feed pressure and real-time extrusion pressure are obtained through pressure transmitters installed at the feed inlet of the terminal filter press and pressure sensors installed on the clamping hydraulic cylinder.

[0023] The real-time operating parameters of each equipment node over the past thirty consecutive production cycles are extracted as historical operating data. Based on the law of conservation of mass, an operating mechanism equation for the equipment is constructed, whereby the rate of change of material mass within any equipment node equals the input material mass minus the output material mass. A long short-term memory neural network is constructed as a data-driven model, with the output layer of this neural network containing two types of nodes: one type outputs predicted values ​​used to fit the true values. The predicted value is the corresponding target fitting variable, and the other type of output represents the predicted material mass representing the internal state of the equipment node. The For predicting material quality based on network output Dynamic variables obtained by discrete-time differential; The historical operating data is input into the Long Short-Term Memory (LSTM) neural network for iteration. During the iteration process, the theoretical calculation result of the operating mechanism equation is added as a penalty term to the loss function of the LSM neural network. The formula for the total pre-training loss function is:

[0024] in, The total loss function during the pre-training process. The total number of historical operational data samples. These are the actual output values ​​from historical operational data. This is the predicted value output by the Long Short-Term Memory neural network; To adjust the physical mechanism penalty coefficient for the strength of the mechanism constraint, and For each device node Input flow rate and input concentration at any given time. and For each device node The output flow and output concentration at any given time; pre-training is completed by adjusting the network weights through the backpropagation algorithm, and virtual device nodes containing physical mechanism constraints are constructed.

[0025] Multiply the inlet flow rate by the feed concentration and integrate over time within a set integration time window to obtain the total mass of solid material input to the primary adsorption device; similarly, calculate the total mass of solid material output from the physical sedimentation device within the same integration time window; divide the total mass of output solid material by the total mass of input solid material to calculate the mass transfer coefficient, the formula of which is:

[0026] in The mass transfer factor is... The starting time of the integration. The set integration time window length, and These are the integral variables. The inlet flow rate and feed concentration of the primary adsorption equipment at all times. Calculated from the transport flow rate The output of the physical settling equipment at any given time is specifically converted as follows: the output is obtained by multiplying the obtained flow rate of the variable frequency conveying equipment by the preset pipeline cross-sectional area conversion coefficient and the pump volumetric efficiency coefficient. Characterized by bottom sludge concentration The solid concentration output by the physical sedimentation equipment at any given time; Using a cross-correlation function, the correlation coefficients between the upstream transport variable sequence composed of the motor speed and the transport flow rate and the downstream sedimentation variable sequence composed of the buffer level and the bottom sludge concentration are calculated under different time delay steps. All time delay steps are traversed, and the time delay step that makes the correlation coefficient reach the maximum value is extracted as the pure lag time of material transport between equipment nodes. The corresponding maximum correlation coefficient is used as the dynamic coupling strength between variables to complete the identification of the dynamic coupling relationship between equipment nodes. A discrete state-space equation is constructed, with the inlet flow rate, feed concentration, and motor speed as input vectors, and the buffer level and bottom sludge concentration as state vectors, and the feed pressure and real-time extrusion pressure as output vectors; the discrete state-space equation is:

[0027]

[0028] in, For the current discrete control cycle, for dimensional state vector, for 3D input vector, for 3D output vector simulation prediction value, for dimensional state transition matrix, for 3D input control matrix, for 3D output matrix, for Dimensional direct transfer matrix; the identified material transport pure time delay Introduced into the input vector as a time bias term The delay term causes a corresponding timing delay in the mathematical expression of the response to the upstream input; the state transition matrix is ​​pre-initialized. The identity matrix is ​​the input control matrix. The zero matrix is ​​used, and the constant matrix is ​​identified using measured steady-state data as the output matrix. With direct transfer matrix The identified dynamic coupling strength is then filled into the state transition matrix. The off-diagonal elements are used to characterize the cross-influence between liquid level and concentration state, and the mass transfer coefficient is used to represent the mass transfer coefficient. As a matrix element, it is filled into the input control matrix. The column vector positions corresponding to the input flow rate and concentration are used to generate a state space matrix composed of matrix parameters; Within each control cycle, the output vector value calculated from the state space matrix is ​​used as the simulation prediction value. This value is subtracted from the actual feed pressure and real-time extrusion pressure measurements collected by the sensor to obtain residual data. A recursive least squares algorithm is used to calculate the parameter correction gain based on the residual data of the current control cycle. The parameter correction gain is multiplied by the input vector of the current control cycle to obtain the parameter compensation amount. The parameter compensation amount is accumulated in real time into the current matrix parameters of the state transition matrix and the input control matrix for online adaptive updating of the dynamic control model. By comprehensively collecting multi-dimensional real-time operating parameters of multi-terminal environmental protection equipment across the entire material handling chain, reliable data support is provided for identifying the dynamic coupling relationship between various equipment nodes. At the same time, a mechanism-data hybrid driving model that integrates equipment operating mechanisms and historical operating data is constructed as a digital twin model. This not only ensures the scientific nature of the full-chain virtual simulation, but also enables the use of residual data between real-time measured values ​​and simulation prediction values ​​to update the state space matrix composed of matrix parameters online, allowing the dynamic control model to be corrected in real time and adapt to the dynamic changes of actual working conditions.

[0029] The dynamic control model is connected to the model prediction controller. Based on the load disturbance change at the inlet of the primary adsorption device, the time delay node of the load peak being transmitted to each downstream device is calculated along the material connection sequence, and the evolution trajectory of the extrusion pressure of the terminal filter press is predicted. The model predictive controller calls the dynamic control model as the predictive equation for rolling optimization; The abrupt change slope and duration of the real-time operating parameters at the inlet of the primary adsorption device are extracted, and the load peak characteristics are quantified as the load disturbance. Based on the material connection sequence, the load peak characteristics are input into the prediction equation. Combining the real-time conveying flow rate of the frequency converter and the available buffer volume of the physical sedimentation device, the transmission delay and retention delay in the material transmission process are calculated step by step. The delays at each level are accumulated in time sequence to solve for the time delay node of the load peak being transmitted to each downstream device. Within the set prediction time domain of the model prediction controller, the predicted arrival load of the terminal filter press based on the time delay node is substituted into the prediction equation, mapped as a pressure state variable, and the extrusion pressure evolution trajectory under the future time series is obtained by iterative solution.

[0030] Specifically, a sliding time window of a set length is set to capture continuous time series data of the inlet flow rate and feed concentration at the inlet of the primary adsorption device in real time; the real-time change rate of the real-time operating parameters is obtained by performing first-order difference calculation on adjacent data points within the sliding time window; the statistical variance of the fluctuation of the real-time operating parameters under historical no-load steady-state operation is extracted as a benchmark threshold; when the real-time change rate is found to be greater than the benchmark threshold for three consecutive sampling periods, the current time node is recorded as the mutation start point; when the real-time change rate falls back and is equal to or less than the benchmark threshold, the current time node is recorded as the mutation end point; the difference between the time node values ​​of the mutation end point and the mutation start point is used to obtain the duration, and the average real-time change rate within the duration is used as the mutation slope; the product of the inlet flow rate and the feed concentration is used as the instantaneous load, and the portion of the instantaneous load within the duration that exceeds the steady-state load benchmark of the device is calculated by definite integral calculation, and the calculated total mass value is quantified and extracted as the load peak feature as the load disturbance amount; Based on the material connection sequence of the material handling link, the load peak characteristics are input into the prediction equation; the actual total physical length and internal cross-sectional area of ​​the pipeline connected to the variable frequency conveyor are read from the field measurements; the average flow velocity of the material in the pipeline is obtained by dividing the real-time conveying flow rate of the variable frequency conveyor by the internal cross-sectional area of ​​the pipeline; the material transfer delay in the pipeline is calculated by dividing the actual total physical length by the average flow velocity; the nominal total physical volume of the physical settling device is obtained, and the real-time buffer level of the physical settling device is read and... Multiply by the measured bottom area of ​​the settling equipment to obtain the currently occupied material volume. Subtract the currently occupied material volume from the physical nominal total volume to calculate the available buffer volume of the physical settling equipment. Divide the available buffer volume by the real-time input flow rate of the physical settling equipment to calculate the retention delay of the material before it is filled in the settling equipment. Add the calculated transmission delay and the retention delay on the time axis to calculate the exact time delay node corresponding to the load peak being transmitted from the primary adsorption equipment to each downstream equipment. Within the model predictive controller, a prediction time domain is set from the current moment to the total number of fixed future steps. Within this predicted time domain, the specific time point at which the load peak arrives at the terminal filter press is deduced based on the time delay nodes, and the load disturbance at that time point is converted into the corresponding predicted load amount. The predicted load amount is substituted into the prediction equation as an input variable column vector. Using the discrete state space equation formed by the established state transition matrix and the input control matrix, with the current actual collected pressure of the terminal filter press as the initial state variable, a recursive operation is performed according to the set discrete control cycle step size. The mass load at the input end is mapped to the pressure state variable at the corresponding time through matrix multiplication. By continuously executing the recursive operation, the predicted value of the extrusion pressure corresponding to each discrete time node within the set prediction time domain is iteratively solved. All predicted values ​​are spliced ​​together in chronological order to obtain the extrusion pressure evolution trajectory under the future time series. This application combines the real-time flow rate of the frequency converter with the available buffer volume of the settling device to calculate and accumulate the material conveying delay and retention delay in segments, reflecting the actual transmission time delay of the material in the multi-level physical link; the evolution trajectory of the extrusion pressure in the future time domain provides reliable data support for the implementation of feedforward coordinated regulation, and improves the regulation lag problem caused by the lack of predictive ability in traditional feedback control.

[0031] See Figure 2 The control boundary conditions of the terminal filter press are generated by comparing the extrusion pressure evolution trajectory with the ultimate pressure threshold of the terminal filter press through control deviation comparison. The control deviation comparison process involves comparing the predicted pressure values ​​at each time point in the extrusion pressure evolution trajectory with the ultimate pressure threshold of the terminal filter press within a set prediction time domain. The pressure difference exceeding the ultimate pressure threshold in the differential comparison results is extracted as the control deviation, and the overpressure time node corresponding to the control deviation is extracted. Using the dynamic control model, the control deviation is back-calculated as the load reduction amount at the feed end of the terminal filter press required to offset the control deviation. Combining the overpressure time node and the load reduction amount at the feed end, the maximum allowable feed flow rate and maximum feed pressure of the terminal filter press when the load peak is reached are calculated and used as the control boundary conditions of the terminal filter press.

[0032] The process for determining the ultimate pressure bearing threshold of the terminal filter press equipment is to extract the total number of historical pressing cycles of the terminal filter press equipment and the pressure drop slope during the holding stage to calculate the real-time health coefficient, and then multiply the static mechanical rated pressure of the equipment by the real-time health coefficient to obtain the basic safe pressure bearing threshold after derated. Extract the sludge specific resistance and material compressibility coefficient of the physical sedimentation equipment, and use the product of the sludge specific resistance and material compressibility coefficient as a characteristic variable to calculate the local stress concentration coefficient of the material applied to the filter plate at the end of dewatering; divide the basic safe pressure bearing threshold by the local stress concentration coefficient for dynamic numerical compensation to calculate the ultimate pressure bearing threshold.

[0033] Specifically, the total number of historical pressing cycles recorded in the programmable logic controller (PLC) inside the terminal filter press is extracted through the industrial control network. The pressure difference and corresponding holding time at both ends of the pressure sensor in the feed pipeline of the terminal filter press during the holding phase are read. The pressure difference is divided by the holding time to obtain the pressure drop slope during the holding phase. A multiple linear regression algorithm is used to fit the attenuation data in the historical overhaul records of the equipment, extracting the first and second weights, and then using the formula... ;in This is a real-time health coefficient. This represents the total number of historical pressing cycles. The pressure drop slope during the pressure holding phase. As the first weight, The second weight is determined by reading the device's electronic nameplate from the factory to obtain the static mechanical rated pressure of the device. The static mechanical rated pressure is then multiplied by the real-time health coefficient to obtain the basic safe pressure threshold after devaluation. The bottom sludge concentration is obtained using an online concentration meter at the bottom of the physical sedimentation equipment, and the buffer level is obtained using an ultrasonic level gauge. The specific resistance of the sludge is calculated by multiplying the square of the bottom sludge concentration by the empirical viscosity coefficient. The hydrostatic pressure is obtained by multiplying the buffer level by the sludge density and the acceleration due to gravity. Based on the Terzaghi one-dimensional consolidation theory, the compressibility coefficient of the material is calculated by dividing the hydrostatic pressure by the sludge volumetric deformation modulus. The product of the sludge specific resistance and the compressibility coefficient is used as a characteristic variable. With 1 as a reference constant, the formula is used... Calculate the local stress concentration factor applied to the filter plate by the material at the end of the dewatering process, where... This is the local stress concentration factor. As characteristic variables, The filter plate deformation sensitivity coefficient is pre-calibrated through finite element simulation; the basic safety pressure threshold is divided by the local stress concentration coefficient for dynamic numerical compensation, and the ultimate pressure threshold of the terminal filter press is calculated.

[0034] The control deviation comparison process specifically involves comparing the predicted pressure values ​​at each time point in the extrusion pressure evolution trajectory with the ultimate pressure threshold of the terminal filter press within the set prediction time domain. When the predicted pressure value at a certain time point is greater than the ultimate pressure threshold, the predicted pressure value at the time point is subtracted from the ultimate pressure threshold, and the absolute difference of the difference comparison result is extracted as the control deviation. The predicted time coordinate corresponding to the control deviation is extracted as the overpressure time point. Using the dynamic control model, the system output matrix and system input matrix in the model state space equation are extracted. The product matrix of the system output matrix and system input matrix is ​​subjected to the Mohr-Ponros generalized inverse matrix operation to obtain the transfer inverse matrix. The control deviation is subjected to the linear algebra matrix multiplication operation with the transfer inverse matrix to reverse calculate the load reduction amount at the feed end of the terminal filter press required to offset the control deviation. The predicted initial total feed load at the overpressure time point is extracted, and the maximum allowable load is obtained by subtracting the load reduction amount at the feed end from the initial total feed load. The current feed end material concentration is extracted by the online instrument, and the maximum allowable load is divided by the current feed end material concentration to calculate the maximum allowable feed flow rate of the terminal filter press when the load peak is reached. The calibration data of the feed pump is retrieved, and the least squares method is used to perform polynomial fitting on multiple sets of calibration operating points to establish a quadratic function relationship between the pump's flow rate and pressure. The maximum feed flow rate is substituted into the quadratic function relationship to solve for the maximum allowable feed pressure of the terminal filter press when the load peak is reached. Finally, the maximum feed flow rate and the maximum feed pressure are used as the control boundary conditions of the terminal filter press.

[0035] By comprehensively evaluating the historical fatigue loss of the filter press equipment and the local stress concentration effect of the material at the end of dewatering, the ultimate pressure threshold of the equipment is dynamically adjusted and derated to reduce the risk of filter plate damage caused by equipment aging or extreme working conditions due to the use of fixed mechanical rated pressure. At the same time, based on the dynamic threshold, the prediction deviation is compared, and the matrix inverse operation of the dynamic control model is used to quantitatively back-calculate the terminal overpressure risk into the load reduction on the input side, and further transform it into the control boundary conditions of maximum feed flow and pressure, providing a quantitative global constraint basis for the advanced coordinated control of upstream equipment. Based on the time delay nodes and control boundary conditions, a multivariate optimal cooperative control sequence including action timing is calculated and then executed. The multivariable optimal collaborative control sequence uses the motor frequency adjustment of the variable frequency conveying equipment, the bottom sludge discharge rate of the physical sedimentation equipment, and the feed pump speed of the terminal filter press as collaborative control variables. Combined with the control boundary conditions, multi-objective numerical optimization is performed within the set controller prediction time domain to calculate the target action values ​​and corresponding trigger time nodes of each collaborative control variable on the future time axis, forming a time-series control matrix that includes discrete control commands and execution order.

[0036] Specifically, the multivariable optimal collaborative control sequence uses the motor frequency adjustment of the variable frequency conveyor, the bottom sludge discharge rate of the physical sedimentation equipment, and the feed pump speed of the terminal filter press as collaborative control variables; it retrieves the inherent mechanical response lag time of the variable frequency conveyor, physical sedimentation equipment, and terminal filter press recorded in the equipment ledger database; for each device in the material connection link, it subtracts the inherent mechanical response lag time of the device from the time delay node of the load peak calculated in the previous steps, calculates the advance action time coordinate required to offset the mechanical inertia, and uses it as the trigger time node corresponding to each collaborative control variable; Combining the aforementioned control boundary conditions, namely the maximum allowable feed flow rate and maximum feed pressure of the terminal filter press, these are used as the upper and lower limits of the hard constraints for optimization. Within the set controller prediction time domain, a multi-objective quadratic cost function is constructed with the objectives of minimizing system operating load deviation and minimizing equipment control energy consumption increment. The fitness value of the multi-objective numerical optimization is calculated using the following formula:

[0037] in, For the cost function value of multi-objective optimization, For the terminal to reach the predicted load, For the target reference load, This refers to the motor frequency adjustment amount of the variable frequency conveyor equipment. This refers to the bottom sludge discharge rate of the physical sedimentation equipment. The feed pump speed of the terminal filter press equipment. To identify and extract various weighting coefficients from historical engineering operation data, for example, if there is a strong concern about overpressure at the end of the pump, the coefficients would be increased. , The discrete time step index represents the future nth time step within the prediction window. Each time point; The cost function is iteratively optimized using a sequential quadratic programming algorithm to find the value that satisfies the cost function. smallest , and The numerical solution is used as the target action value of each cooperative control variable on the future time axis; the device identifier corresponding to the cooperative control variable is used as the row index of the matrix, the discrete time step containing the trigger time node is used as the column index of the matrix, and the calculated target action value is filled into the corresponding row and column intersection position to form a timing control matrix containing discrete control instructions and execution order. The timing control matrix is ​​sent to the programmable logic controllers of each terminal device via industrial Ethernet. Using this control boundary condition as a global constraint, each underlying controller adjusts the conveying rate of the variable frequency conveyor (i.e., executes the motor frequency adjustment amount) and the buffer load of the physical settling device (i.e., executes the bottom sludge discharge rate) in advance according to the trigger time node before the load peak arrives, based on the received timing control matrix. By deducting the inherent mechanical response lag time of each device, the triggering time node of the action command is calculated, which effectively avoids the timing misalignment between the control signal issuance and the actual physical action of the device. At the same time, a multi-objective numerical optimization mechanism is introduced to solve multiple variables such as frequency conversion transmission, physical settling and terminal pressure filtration within the set safety control boundary, thereby achieving a good balance between load stability control and minimizing the energy consumption of comprehensive equipment regulation. Using the aforementioned control boundary conditions as global constraints, the conveying rate of the variable frequency conveyor and the buffer load of the physical settling equipment are adjusted in advance, and coordinated control is performed before the load peak arrives. The buffer load is the current storage material holding amount calculated by multiplying the real-time buffer level of the physical settling equipment by the geometric dimensions of the settling zone; the advance adjustment of the buffer load of the physical settling equipment is to increase the bottom sludge discharge rate of the physical settling equipment before the time delay node when the load peak reaches the physical settling equipment, based on the predicted total load peak input, thereby reducing the current storage material holding amount to the preset safe storage volume limit and expanding the effective volume space of the physical settling equipment to receive the load peak. Specifically, the real-time buffer level is obtained by reading the ultrasonic level gauge installed inside the physical sedimentation equipment via an industrial IoT network; the 3D engineering drawing data of the equipment stored in the digital twin model is retrieved, and the horizontal cross-sectional area of ​​the water surface inside the physical sedimentation equipment is extracted as the geometric dimension of the sedimentation zone; the real-time buffer level is mathematically multiplied by the geometric dimension of the sedimentation zone to calculate the current storage material holding amount as the buffer load of the physical sedimentation equipment, and the calculation formula is as follows: ,in This represents the current inventory level of stored materials. For real-time buffer level, These are the geometric dimensions of the settlement zone; Extract the start and end time coordinates of the load peak, and use a numerical integration algorithm to perform definite integral calculation on the feed flow evolution curve within this time interval to obtain the predicted total load peak input. The physical overflow height alarm limit of the physical settling equipment is retrieved, and the physical overflow height alarm limit is multiplied by the geometric dimension of the settling zone to obtain the maximum physical receiving volume of the equipment. The predicted total load peak input is subtracted from the maximum physical receiving volume, and the minimum dead water level volume set based on the anti-dry running mechanism of the equipment's sludge pump is deducted to calculate the dynamic safe retention volume limit, which is used as the safe retention volume limit. Before the load peak reaches the time delay node of the physical settling device, extract the remaining time difference between the current time node and the time delay node; subtract the upper limit of the safe storage volume from the current storage material holding amount to calculate the excess volume that needs to be emptied in advance before the peak arrives; divide the excess volume by the remaining time difference to calculate the additional sludge discharge rate required to offset the excess volume. The target sludge discharge rate instruction is obtained by adding the current basic sludge discharge rate of the physical sedimentation equipment to the additional sludge discharge rate, thereby increasing the bottom sludge discharge rate of the physical sedimentation equipment in advance. By issuing the target sludge discharge rate instruction, the current storage material holding amount is precisely reduced to the upper limit of the safe holding volume, thereby expanding the effective volume space of the physical sedimentation equipment to accept the load peak. By introducing an architecture that integrates digital twins and model predictive control, the dynamic coupling relationship between multiple terminal devices and the delay nodes of material transfer are identified. The traditional lag response is transformed into an advance prediction. At the same time, the pressure trajectory of the terminal filter press equipment can be predicted in advance and a global control boundary can be set before the load peak actually reaches the end. Then, based on the optimal control sequence including the action timing, the speed of the upstream conveying equipment and the buffer load of the settling equipment are scheduled in advance to smooth out the peak and fill the valley. This global feedforward coordination mechanism effectively avoids instantaneous overload or shutdown failure of the terminal equipment due to sudden load changes.

[0038] Example 2: This embodiment applies a collaborative control method for multi-terminal environmental protection equipment based on digital twins to the continuous dewatering treatment section of a large industrial electroplating wastewater treatment plant. The material handling chain includes a primary adsorption device (heavy metal activated carbon adsorption tower), a variable frequency conveying device (variable frequency rotor pump), a physical sedimentation device (high-density sludge thickening tank), and a terminal filter press device (high-pressure diaphragm plate and frame filter press). At this time, it is facing a load peak impact caused by a sudden surge in feed concentration due to centralized sewage discharge from the workshop.

[0039] The inlet flow rate and feed concentration of the heavy metal activated carbon adsorption tower, the speed and delivery flow rate of the variable frequency rotor pump, the buffer level and bottom sludge concentration of the sludge thickening tank, and the feed pressure and real-time extrusion pressure of the filter press are collected in real time by sensors. The above parameters are input into the mechanism-data hybrid driven digital twin model for full-link virtual simulation. The residual data between the predicted and measured values ​​are used to adaptively update the state space matrix parameters representing the dynamic coupling relationship of the equipment online. When a sudden change is detected on the feed side of the adsorption tower, the slope and duration of the sudden change in the high-concentration sludge load are quantified and extracted as the load disturbance. Based on the material connection sequence, the transmission delay of the pipeline where the rotor pump is located and the retention delay brought by the available buffer volume of the thickener are added together to calculate that the time delay node for the load peak to be transmitted to the sludge thickener is 30 minutes, and the time delay node for the transmission to the end plate and frame filter press is 45 minutes.

[0040] Within the prediction time domain set by the model predictive controller, dynamic numerical compensation is performed by combining the historical fatigue health of the filter press and the local stress concentration coefficient of the sludge at the end of dewatering to obtain the current ultimate pressure threshold. When the prediction equation shows that the filter press will face overpressure risk in 45 minutes, the linear algebra operation of the inverse matrix of the dynamic control model is used to reverse calculate the load reduction required to offset the overpressure deviation, and then the maximum feed flow rate and maximum feed pressure allowed to be input to the filter press when the load peak arrives are generated, which are used as the control boundary conditions of the global constraint.

[0041] Using the motor frequency adjustment of the variable frequency rotor pump, the sludge discharge rate at the bottom of the thickener, and the feed pump speed of the filter press as collaborative control variables, under the hard constraints of the maximum feed flow rate and pressure generated above, a multi-objective numerical optimization function that minimizes system load deviation and control energy consumption is constructed. The inherent mechanical response lag time of each device is deducted in advance, and a timing control matrix containing the target action value and trigger time node on the future time axis is calculated.

[0042] The underlying controller receives and executes the control sequence, strictly using the control boundary conditions as global constraints. Before the load peak reaches the 30-minute delay node of the thickener, it increases the bottom sludge discharge rate of the thickener in advance and adjusts the rotor pump delivery rate to reduce the current storage material holding amount to the preset safe retention volume limit in advance. It uses the effective volume space of the thickener to smooth out the upstream disturbance, thereby stably limiting the load that will reach the end filter press 45 minutes later within the safe pressure envelope, fundamentally avoiding pump stalling or filter plate damage caused by sudden load overload.

[0043] 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. A collaborative control method for multi-terminal environmental protection equipment based on digital twins, characterized in that, include: Real-time operating parameters of the multi-terminal environmental protection equipment control loop are collected. These real-time operating parameters include the inlet flow rate and feed concentration of the primary adsorption equipment, the motor speed and conveying flow rate of the variable frequency conveyor, the buffer level and bottom sludge concentration of the physical sedimentation equipment, and the feed pressure and real-time extrusion pressure of the terminal filter press. The multi-terminal environmental protection equipment sequentially includes a primary adsorption equipment, a variable frequency conveyor, a physical sedimentation equipment, and a terminal filter press in the material handling chain. The real-time operating parameters are input into a digital twin model to identify the dynamic coupling relationship between each equipment node and establish a dynamic control model. The dynamic control model is connected to the model predictive controller. Based on the load disturbance change at the inlet of the primary adsorption device, the time delay node of the load peak being transmitted to each downstream device is calculated along the material connection sequence, and the evolution trajectory of the extrusion pressure of the terminal filter press is predicted. The model predictive controller calls the dynamic control model as the predictive equation for rolling optimization. The abrupt slope and duration of the real-time operating parameters at the inlet of the primary adsorption device are extracted, and the load peak characteristics are quantified as the load disturbance. Based on the material connection sequence, the load peak characteristics are input into the prediction equation. Combining the real-time conveying flow rate of the variable frequency conveyor and the available buffer volume of the physical sedimentation device, the transmission delay and retention delay during material transport are calculated level by level. The delays at each level are then accumulated sequentially to determine the time delay nodes from which the load peak is transmitted to each downstream device. Within the set prediction time domain of the model prediction controller, the predicted arrival load of the terminal filter press, derived from the time delay nodes, is substituted into the prediction equation and mapped as a pressure state variable. The extrusion pressure evolution trajectory under future time series is iteratively solved. The extrusion pressure evolution trajectory is compared with the ultimate pressure threshold of the terminal filter press to generate the control boundary conditions for the terminal filter press. Based on the time delay node and control boundary conditions, a multivariate optimal collaborative control sequence including action timing is calculated and executed. Using the control boundary conditions as global constraints, the conveying rate of the variable frequency conveyor and the buffer load of the physical settling device are adjusted in advance, and collaborative control is performed before the load peak arrives. The buffer load is the current storage material holding amount calculated by multiplying the real-time buffer level of the physical settling device by the geometric dimensions of the settling zone. The advance adjustment of the buffer load of the physical settling device is to increase the bottom sludge discharge rate of the physical settling device before the load peak arrives at the time delay node of the physical settling device, based on the predicted total load peak input, and reduce the current storage material holding amount to the preset safe retention volume limit, thereby expanding the effective volume space of the physical settling device to receive the load peak.

2. The method for collaborative control of multi-terminal environmental protection equipment based on digital twins according to claim 1, characterized in that, The digital twin model is a mechanism-data hybrid driven model that integrates the equipment's operating mechanism and historical operating data; A virtual device node containing the operating mechanism is constructed by pre-training with historical operating data, and the inlet flow rate, feed concentration, motor speed, conveying flow rate, buffer level, bottom sludge concentration, feed pressure and real-time extrusion pressure are synchronously input into the virtual device node to perform full-link virtual simulation; The mass transfer coefficient is calculated using the inlet flow rate and feed concentration as the initial disturbance boundary. Combined with the time-series correlation analysis of the motor speed, conveying flow rate, buffer level, and bottom sludge concentration, the dynamic coupling relationship between equipment nodes is identified. The dynamic coupling relationship is transformed into a state-space matrix composed of matrix parameters as the dynamic control model, and the residual data between the simulated predicted value and the actual measured value of the feed pressure and the real-time extrusion pressure is calculated. The matrix parameters are then updated online using the residual data.

3. The method for collaborative control of multi-terminal environmental protection equipment based on digital twins according to claim 1, characterized in that, The control deviation comparison process involves comparing the predicted pressure values ​​at each time point in the extrusion pressure evolution trajectory with the ultimate pressure threshold of the terminal filter press within a set prediction time domain. The pressure difference exceeding the ultimate pressure threshold in the differential comparison results is extracted as the control deviation, and the overpressure time node corresponding to the control deviation is extracted. Using the dynamic control model, the control deviation is back-calculated as the load reduction amount at the feed end of the terminal filter press required to offset the control deviation. Combining the overpressure time node and the load reduction amount at the feed end, the maximum allowable feed flow rate and maximum feed pressure of the terminal filter press when the load peak is reached are calculated and used as the control boundary conditions of the terminal filter press.

4. The method for collaborative control of multi-terminal environmental protection equipment based on digital twins according to claim 3, characterized in that, The process for determining the ultimate pressure bearing threshold of the terminal filter press equipment is to extract the total number of historical pressing cycles of the terminal filter press equipment and the pressure drop slope during the holding stage to calculate the real-time health coefficient, and then multiply the static mechanical rated pressure of the equipment by the real-time health coefficient to obtain the basic safe pressure bearing threshold after derated. Extract the sludge specific resistance and material compressibility coefficient of the physical sedimentation equipment, and use the product of the sludge specific resistance and material compressibility coefficient as a characteristic variable to calculate the local stress concentration factor of the material applied to the filter plate at the end of dewatering. The ultimate bearing capacity threshold is calculated by dividing the basic safe bearing capacity threshold by the local stress concentration factor and performing dynamic numerical compensation.

5. The method for collaborative control of multi-terminal environmental protection equipment based on digital twins according to claim 1, characterized in that, The multivariable optimal collaborative control sequence uses the motor frequency adjustment of the variable frequency conveying equipment, the bottom sludge discharge rate of the physical sedimentation equipment, and the feed pump speed of the terminal filter press as collaborative control variables. Combined with the control boundary conditions, multi-objective numerical optimization is performed within the set controller prediction time domain to calculate the target action values ​​and corresponding trigger time nodes of each collaborative control variable on the future time axis, forming a time-series control matrix that includes discrete control commands and execution order.

Images