Model prediction based energy saving optimization control method for single-screw dehumidification

Abstract

A model prediction-based single-rotor dehumidification energy-saving optimization control method, the method comprising: collecting temperature and humidity data and flow data in the single-rotor airflow pipeline of an air handling system through an integrated sensor to establish original time series data and temperature smoothing curves with synchronous time stamps. Based on the isotherm parameter set and the temperature smoothing curve, the equilibrium adsorption curve is calculated, and combined with the kinetic constant, a mechanism framework model for predicting temperature disturbance on adsorption capacity is constructed. Through distributed robust mechanism, quadratic programming is performed according to the sample mean and variance of the adsorption capacity to obtain the speed distribution and temperature compensation of the single-rotor. Based on the speed distribution and temperature compensation, combined with the predicted disturbance of temperature on adsorption capacity output by the mechanism framework model, the MPC algorithm is used to solve the optimal speed and temperature control sequence of the single-rotor in real time. The optimal speed and temperature control sequence are executed to optimize the control of single-rotor adsorption dehumidification, and the model correction parameters of the execution feedback are obtained.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of air treatment, and more particularly to a single-rotor dehumidification energy-saving optimization control method based on model prediction. BACKGROUND

[0002] In a precision manufacturing electronic factory (such as a semiconductor packaging / printed circuit board), such a place needs to maintain constant temperature and humidity (such as 23±1℃, 45±2% RH) for a long time, but there are a large number of intermittent heat source disturbances, such as welding, hot air drying, UV curing, and other periodic heat source operations.

[0003] A single-rotor is a heat and moisture treatment device, which is a key structural component in an air treatment system (especially a fresh air treatment unit or a dehumidification air conditioning system). Its main function is to absorb and remove moisture from the air to achieve air dehumidification, and it is often used in the preprocessing link of an air conditioning ventilation system. The single-rotor mainly uses hygroscopic materials (such as silica gel, molecular sieve) to rotate on the rotor structure for periodic hygroscopic-regeneration operation. Understandably, the operation logic of the air treatment system is as follows: fresh air inlet; pre-coarse filter; surface cooler / heater (optional); single-rotor dehumidification section; rear cooling coil (fine adjustment); fan; air supply outlet. The single-rotor section is responsible for active dehumidification, reducing the latent heat load of the subsequent cooling coil.

[0004] Currently, the hygroscopic efficiency of the rotor hygroscopic material (such as silica gel) is highly sensitive to its own temperature. When the temperature of the hygroscopic section suddenly rises from 25℃ to 35℃ due to heat source disturbance, the unit hygroscopic capacity decreases by more than 40%. If the rotor speed and regeneration air temperature are not pre-adjusted in advance through model prediction, the system dehumidification capacity will be instantly unbalanced, the RH will rapidly rise, and production abnormalities will occur. Therefore, the instantaneous temperature rise of the rotor hygroscopic zone caused by high-frequency heat disturbance is more likely to lead to a weakening of the dynamic dehumidification capacity. SUMMARY

[0005] To solve the problems in the prior art, the purpose of the present application is to solve the above-mentioned defects, and further to provide a single-rotor dehumidification energy-saving optimization control method based on model prediction.

[0006] The application adopts the following technical solutions.

[0007] The first aspect of the present application discloses a single-rotor dehumidification energy-saving optimization control method based on model prediction, which comprises:

[0008] Collecting temperature and humidity data and flow data in the single-rotor airflow duct in the air treatment system by integrating sensors to establish original time series data and temperature smooth curves with synchronous time stamps;

[0009] An equilibrium adsorption curve is calculated based on the isotherm parameter set and the temperature smoothing curve, and a mechanism framework model for predicting temperature disturbance to adsorption amount is constructed in combination with a kinetic constant;

[0010] A quadratic programming is performed on the basis of the adsorption amount sample mean and variance by means of a distributed robust mechanism to obtain a single-rotor rotating speed distribution and temperature compensation;

[0011] Based on the rotating speed distribution and temperature compensation, in combination with the predicted disturbance of temperature to adsorption amount output by the mechanism framework model, an MPC algorithm is adopted to solve the optimal rotating speed and temperature control sequence of the single-rotor in real time;

[0012] The optimal rotating speed and temperature control sequence are executed to optimize the control of single-rotor adsorption dehumidification, and model correction parameters of execution feedback are obtained.

[0013] Further, the temperature and humidity data and flow data in the single-rotor airflow duct in the air handling system are collected by means of the integrated sensor to establish the original time series data and the temperature smoothing curve with a synchronous timestamp, which comprises:

[0014] A plurality of temperature and humidity sensors and heat flow sensors are arranged in the airflow duct at the inlet of the humidification section according to a circumferential equal interval to construct an integrated sensor for monitoring the temperature difference and flow of the airflow inlet and outlet, and each sensor in the integrated sensor is connected to an edge node;

[0015] The temperature and humidity sensors and the heat flow sensors are polled according to a set sampling frequency to obtain temperature and humidity data and gas flow data, and the temperature and humidity data and the gas flow data are time-synchronized by means of the edge node to obtain the original time series data with a synchronous timestamp;

[0016] The temperature and humidity sensors are used to collect the temperature and humidity data in the single-rotor airflow duct, the heat flow sensors are used to collect the flow data, and the edge nodes are in communication connection with the centralized controller.

[0017] Further, the temperature and humidity data and flow data in the single-rotor airflow duct in the air handling system are collected by means of the integrated sensor to establish the original time series data and the temperature smoothing curve with a synchronous timestamp, which further comprises:

[0018] A plurality of sampling points are extracted from the original time series data according to a set time window, and the temperature mean value, average heat flow and heat flow mutation value of all the extracted sampling points are calculated based on the plurality of sampling points to construct a feature vector;

[0019] Based on the feature vector, a first-order prediction formula is iteratively calculated by means of an autoregressive plus exogenous input model to obtain discrete prediction values in a future first time period, and a three-point moving average processing is performed on the discrete prediction values to obtain the temperature smoothing curve.

[0020] Further, the equilibrium adsorption curve is calculated based on the isotherm parameter set and the temperature smoothing curve, and a mechanism framework model is constructed for predicting the temperature disturbance on the adsorption amount by combining the kinetic constant, including:

[0021] Obtain the isotherm parameter set and the kinetic constant, and the isotherm parameter set includes the maximum adsorption amount, the adsorption energy constant and the empirical index;

[0022] Based on the temperature smoothing curve and the isotherm parameter set, the Dubinin-Astakhov isotherm model is called to construct a discrete curve, and the equilibrium adsorption curve is calculated according to the discrete curve.

[0023] Further, the equilibrium adsorption curve is calculated based on the isotherm parameter set and the temperature smoothing curve, and a mechanism framework model is constructed for predicting the temperature disturbance on the adsorption amount by combining the kinetic constant, including:

[0024] Based on the equilibrium adsorption curve and the kinetic constant, when the adsorption amount changes with time satisfies the first-order kinetics, the discrete time in the equilibrium adsorption curve is updated to calculate the dynamic adsorption curve according to the updated discrete time;

[0025] Obtain the reference adsorption amount, and calculate the instantaneous efficiency decay ratio according to the dynamic adsorption curve to construct an adsorption efficiency curve, the instantaneous efficiency decay ratio being the ratio between the dynamic adsorption curve and the reference adsorption amount.

[0026] Further, the single-rotor speed distribution and temperature compensation are obtained by quadratic programming according to the adsorption amount sample mean and variance based on the distributed robust mechanism, including:

[0027] Obtain the difference between the expected humidity and the actual humidity of multiple groups of inlet and outlet air ports of the single-rotor, and use the multiple difference values as historical observation gap airflow rate samples to generate Chebyshev inequality;

[0028] Write the sensitivity coefficient into the Chebyshev inequality to obtain the robustness constraint, and use the robustness constraint to calibrate the sensitivity coefficient according to the trial operation data set to output the single-rotor speed distribution and temperature compensation as the reference working point.

[0029] Further, based on the speed distribution and temperature compensation, the optimal speed and temperature control sequence of the single-rotor are solved in real time by using the MPC algorithm combined with the predicted disturbance of the temperature on the adsorption amount output by the mechanism framework model, including:

[0030] Based on the current humidity error in the single-rotor airflow pipeline, the minimum of the rotational speed distribution and the temperature compensation, combined with the predicted disturbance and the configuration parameters, a discrete model is constructed by using the MPC algorithm;

[0031] The prediction step, the energy consumption weight, the balance factor and the robust safety constraint are obtained by configuration, and based on the prediction step, the energy consumption weight, the balance factor and the robust safety constraint, the discrete model is optimized in combination with the temperature threshold and the airflow rate threshold to obtain an optimization target and constraint.

[0032] Further, based on the rotational speed distribution and the temperature compensation, combined with the predicted disturbance of the temperature to the adsorption amount output by the mechanism framework model, the optimal rotational speed and temperature control sequence of the single-rotor are solved in real time by using the MPC algorithm, which further includes:

[0033] The current temperature error, the predicted disturbance, the optimization target and constraint, and the configuration parameters of the discrete model are input into a QP solver, and the optimal rotational speed and temperature control sequence are calculated in the QP solver according to the rolling time domain.

[0034] The second aspect of the present application discloses a single-rotor dehumidification energy-saving optimization control system based on model prediction, which comprises:

[0035] A data acquisition and processing module is configured to acquire temperature and humidity data and flow data in a single-rotor airflow pipeline of an air handling system through an integrated sensor to establish original time series data and a temperature smoothing curve with a synchronous time stamp.

[0036] An adsorption disturbance quantification module is configured to calculate a balance adsorption curve based on an isotherm parameter set and the temperature smoothing curve, and to construct a mechanism framework model for predicting temperature-to-adsorption disturbance in combination with kinetic constants.

[0037] A robustness compensation module is configured to perform quadratic programming according to mean and variance of adsorption amount samples by a distributed robust mechanism to obtain rotational speed distribution and temperature compensation of the single-rotor.

[0038] An optimal control sequence solving module is configured to solve the optimal rotational speed and temperature control sequence of the single-rotor in real time by using the MPC algorithm based on the rotational speed distribution and the temperature compensation, combined with the predicted disturbance of the temperature to the adsorption amount output by the mechanism framework model.

[0039] An optimization control execution module is configured to execute the optimal rotational speed and temperature control sequence to perform optimization control on single-rotor adsorption dehumidification and to obtain model correction parameters of execution feedback.

[0040] The third aspect of the present application discloses a terminal comprising a processor and a storage medium.

[0041] The storage medium is configured to store instructions.

[0042] The processor is configured to operate according to the instructions to perform the steps of the method of the first aspect.

[0043] A fourth aspect of the present application discloses a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method of the first aspect.

[0044] The present application has the following advantages compared with the prior art:

[0045] (1) The present application captures and quantizes the impact of intermittent heat source disturbance on the temperature of the moisture absorption section in real time by integrating the temperature and humidity sensor and the flow sensor, lays a data foundation for subsequent model prediction input. At the same time, the influence of temperature rise on adsorption efficiency is quantized by using the mechanism framework model, which can intelligently predict the dynamic adsorption capacity at the current and future time, and reduce the problem that the dynamic dehumidification capacity is weakened due to the temperature transient of the moisture absorption zone of the rotary dehumidifier caused by high-frequency heat disturbance.

[0046] (2) The present application uses a distributed robust (Distributionally Robust) mechanism, estimates the worst case of the gap between the sample mean and variance, and then uses quadratic programming to allocate the speed and temperature compensation, which can take into account the safety margin and energy-saving control optimization, ensure a higher confidence coverage of the real gap under any distribution premise, and effectively reduce the risk of system out of control. In addition, in the rolling time domain, the distributed robust compensation and disturbance prediction are used to construct a discrete-time system model and an optimization objective, and the model predictive control (MPC) algorithm is used to solve the optimal rotary dehumidifier speed and regeneration air temperature control sequence in real time, which can further take into account humidity stability and energy saving. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 is a flow diagram of the single-rotor dehumidification energy-saving optimization control method based on model prediction provided by the present application;

[0048] Figure 2 is a structural diagram of the single-rotor dehumidification energy-saving optimization control system based on model prediction provided by the present application. DETAILED DESCRIPTION

[0049] The present application will be further described below in conjunction with the drawings. The following examples are only used to more clearly illustrate the technical solutions of the present application, and cannot be used to limit the protection scope of the present application.

[0050] As shown in Figure 1 in one embodiment, a single-rotor dehumidification energy-saving optimization control method based on model prediction includes the following steps:

[0051] Step S110, collecting temperature and humidity data and flow data in the single-rotor airflow duct of the air treatment system through the integrated sensor to establish original time series data with synchronous timestamps and a temperature smooth curve.

[0052] In some embodiments, the model prediction-based single-rotor dehumidification energy-saving optimization control method provided by the present application specifically comprises the following steps of step S110:

[0053] Step S111, a plurality of groups of temperature and humidity sensors and heat flow sensors are arranged in the airflow duct at the inlet of the moisture absorption section according to a circumferential equal interval to construct integrated sensors for monitoring the temperature difference and flow of the airflow at the inlet and outlet, and each sensor in the integrated sensors is connected to an edge node.

[0054] Step S112, polling the temperature and humidity sensors and the heat flow sensors according to a set sampling frequency to obtain temperature and humidity data and gas flow data, and performing time synchronization on the temperature and humidity data and the gas flow data through the edge node to obtain original time series data with synchronous timestamps;

[0055] The temperature and humidity sensors are used to collect temperature and humidity data in the single-rotor airflow duct, the heat flow sensors are used to collect flow data, and the edge node is in communication connection with the centralized controller.

[0056] In some embodiments, the model prediction-based single-rotor dehumidification energy-saving optimization control method provided by the present application specifically further comprises the following steps of step S110:

[0057] Step S113, extracting a plurality of sampling points from the original time series data according to a set time window, and calculating the temperature mean value, the average heat flow and the heat flow mutation value of all the extracted sampling points based on the plurality of sampling points to construct a feature vector.

[0058] Step S114, based on the feature vector, using an autoregressive plus exogenous input model to iteratively calculate discrete prediction values in a first future time period by calling a first-order prediction formula, and performing three-point sliding average processing on the discrete prediction values to obtain a temperature smooth curve.

[0059] In specific embodiments, the model prediction-based single-rotor dehumidification energy-saving optimization control method provided by the present application comprises steps 1-4:

[0060] Step 1, heat source disturbance detection and moisture absorption section temperature prediction.

[0061] Real-time capture and quantification of the impact of intermittent heat source disturbance on the temperature of the moisture absorption section provide input for subsequent model prediction.

[0062] Comprising the following sub-steps:

[0063] Sub-step 1.1, sensor and device structure deployment.

[0064] Specifically, based on the factory process map, the inlet space size of the moisture absorption section, and the pipeline layout map, at the air flow pipeline at the inlet of the moisture absorption section, temperature and humidity sensors (such as dry and wet ball type or capacitive type) are arranged at equal intervals (recommended every 90°) along the circumferential direction of the pipeline, a total of 4, which are used to obtain the average temperature and humidity of the cross section. A heat flow sensor (such as a Pitot tube anemometer combined with a thermocouple) is installed 10 cm downstream of the same position to monitor the temperature difference and flow rate of the air inlet and outlet to estimate the heat flux. All sensors are connected to an edge computing node (or PLC expansion I / O module) through shielded twisted pair lines, which has 4 analog input, 4 digital input and Ethernet output. This step is used to ensure that the measurement points cover the entire cross section of the inlet of the moisture absorption section, and obtain uniform temperature and humidity and heat flux information, laying a hardware foundation for subsequent prediction modeling.

[0065] Among them, the edge node communicates with the centralized controller (MPC or DCS) through Modbus / TCP protocol, and uploads the raw data in real time.

[0066] Substep 1.2, real-time data acquisition and time synchronization.

[0067] Specifically, the edge node polls each sensor at a sampling rate of 50Hz to obtain temperature signals (converted to actual temperature values by sensor calibration, unit °C), relative humidity signals (converted to humidity values), and wind speed signals and temperature difference signals combined with heat flow sensor calibration curve to calculate instantaneous heat flux. Then, the edge node performs time synchronization according to IEEE 1588 or NTP, packages the above three groups of data as timestamp records, i.e. synchronized timestamp raw time series data, and pushes them to the controller database in real time through TCP / IP. This step realizes high-frequency synchronous acquisition of multiple physical quantities, ensuring that the data used by the subsequent prediction model has sufficient temporal and spatial resolution and consistency.

[0068] Substep 1.3, feature extraction and data preprocessing.

[0069] Specifically, taking 60s as a prediction window, sample data of N (such as N=3000) sampling points in the original time series data is extracted, and the average temperature, average heat flux, and heat flux mutation value (the difference between adjacent heat flux sample data, and the larger the difference, the more likely there is a mutation) of the sample data are calculated. Then, the sampling points with temperature or flow mutation exceeding the set threshold (such as ±5℃ or ±20%) are difference filled, and finally the key feature vector in the short time window is output. This step reduces the massive raw data to a 4-dimensional feature vector, which not only retains the strength and fluctuation characteristics of the heat source disturbance, but also greatly reduces the input dimension of the prediction model, improving the online calculation efficiency.

[0070] Substep 1.4, short-term temperature trend prediction.

[0071] Specifically, based on the key feature vector obtained in sub-step 1.3, a first-order prediction formula is called by using an autoregressive plus exogenous input model (ARX):

[0072]

[0073] wherein, is the temperature prediction value in the future time, is the average temperature of the sample data, , is the average heat flow and the heat flow mutation value of the sample data, respectively; is the temperature autoregressive coefficient, the value range is [0.5, 0.9], which is obtained by fitting the historical temperature sample data; is the heat flow weight on temperature, the value range is [0.1, 0.3]; is the heat flow mutation compensation coefficient, the value range is [0.05, 0.15].

[0074] Then, starting from the latest second feature vector, the discrete prediction values in the next 30 seconds are calculated by iteration using the above formula, that is, ~ The discrete prediction values in 30 seconds are converted into a prediction value and time curve, and a three-point moving average is applied to eliminate peak errors, and the final temperature prediction curve is obtained. This step uses the key features of heat source disturbance and moisture absorption section temperature to quickly generate a short-term temperature trend, providing accurate temperature input for the subsequent adsorption efficiency model and quantifying the impact amplitude and duration of heat source disturbance in advance.

[0075] Step S120, based on the isotherm parameter set and the temperature smoothing curve, the equilibrium adsorption curve is calculated, and the mechanism framework model for predicting the disturbance of temperature on the adsorption amount is constructed combined with the kinetic constant.

[0076] In some embodiments, the model prediction-based single-rotary dehumidification energy-saving optimization control method provided by the application specifically comprises the following steps:

[0077] Step S121, obtaining the isotherm parameter set and the kinetic constant, the isotherm parameter set includes the maximum adsorption capacity, the adsorption energy constant and the empirical index.

[0078] Step S122, based on the temperature smoothing curve and the isotherm parameter set, the Dubinin-Astakhov isotherm model is called to construct a discrete curve, and the equilibrium adsorption curve is calculated according to the discrete curve.

[0079] In some embodiments, the model prediction-based single-rotary dehumidification energy-saving optimization control method provided by the application, step S120 specifically further comprises the following steps:

[0080] Step S123, based on the equilibrium adsorption curve and the kinetic constant, when the adsorption capacity changes with time satisfies the first-order kinetics, the discrete time in the equilibrium adsorption curve is updated to calculate the dynamic adsorption curve according to the updated discrete time.

[0081] Step S124, obtaining the reference adsorption capacity, and calculating the instantaneous efficiency decay ratio according to the dynamic adsorption curve to construct the adsorption efficiency curve, the instantaneous efficiency decay ratio being the ratio between the dynamic adsorption curve and the reference adsorption capacity.

[0082] In a specific embodiment, the model prediction-based single-rotor dehumidification energy-saving optimization control method provided by the application comprises the following steps:

[0083] The following sub-steps are included:

[0084] Sub-step 2.1, adsorbent parameter calibration.

[0085] Specifically, an isotherm parameter set and a kinetic constant are obtained, the isotherm parameter set including a maximum adsorption capacity (unit: kg / ㎡) of the adsorbent, the value range being [0.05, 0.15]; an adsorption energy constant (unit: J / mol), the value range being [5000, 8000]; an empirical index, dimensionless, the value range being [1.5, 2.5]. The kinetic constant (unit: s -1 ) is obtained by experiment calibration, and the typical value is [0.01, 0.05].

[0086] It should be noted that the experiment calibration process of the kinetic constant is as follows:

[0087] First, the sample is inspected as above, and it must be in the initial state after drying. The sample is first placed in a low-humidity environment (such as 10% RH) for equilibrium, and then quickly transferred to a high-humidity environment (such as 50% RH), and the moment is recorded as t=0. The sample weight change is recorded every few seconds or tens of seconds (according to the speed of your instrument), until there is no weight change after a few minutes. Then, plot the adsorption capacity-time curve with time as the horizontal axis and the adsorbed water amount at the corresponding time as the vertical axis. Assuming that the adsorption process satisfies the simplified single-exponential model, the expression is as follows:

[0088]

[0089] In the formula, is the amount of water adsorbed at t, is the equilibrium adsorption capacity, is the kinetic constant, indicating the speed of the adsorbent to adsorb water.

[0090] Sub-step 2.2, calculate the instantaneous equilibrium adsorption amount curve.

[0091] Specifically, the simplified expression of the Dubinin-Astakhov isotherm model is called:

[0092]

[0093] In the formula, is the equilibrium adsorption amount at time t (unit: kg / m 2 ), is the maximum adsorption amount, is the adsorption energy constant; is the gas constant, a fixed value of 8.314 J / (mol·K); is the predicted temperature value, is an empirical index, dimensionless.

[0094] Then, for the future prediction interval (30 seconds), the above formula is called once every second, and the corresponding discrete curve is finally obtained.

[0095] Sub-step 2.3, coupling kinetics to calculate dynamic adsorption amount.

[0096] Specifically, assuming that the adsorption process satisfies the first-order kinetics, the discrete time update formula is:

[0097]

[0098] In the formula, is the actual adsorption amount at time t, is the kinetic constant, is the equilibrium adsorption amount, is the time step (e.g., 1 s).

[0099] Then, initialize , and then iteratively calculate for the next 30 seconds, and finally build the dynamic adsorption amount curve.

[0100] Sub-step 2.4, calculate the instantaneous adsorption efficiency decay ratio.

[0101] Specifically, set the reference adsorption amount , where is the equilibrium capacity at 25°C (298K), is obtained by averaging the values of multiple equilibrium adsorption amounts obtained at 298K. Then, calculate the instantaneous efficiency decay ratio:

[0102]

[0103] Finally, for The corresponding curves are smoothed (e.g., by a three-second moving average) to obtain the adsorption efficiency curve.

[0104] Step S130: Based on the mean and variance of the adsorption amount sample, a secondary planning is performed using a distributed robust mechanism to obtain the rotational speed distribution and temperature compensation of a single rotor.

[0105] In some embodiments, the model-predictive-based single-rotor dehumidification energy-saving optimization control method provided by the present invention includes the following steps in step S130:

[0106] Step S131: Obtain the difference between the expected humidity and the actual humidity of multiple sets of air inlets and outlets of a single rotor, and use multiple sets of differences as historical observation gap airflow rate samples to generate the Chebyshev inequality.

[0107] Step S132: Write the sensitivity coefficient into the Chebyshev inequality to obtain the robustness constraint, and use the robustness constraint to calibrate the sensitivity coefficient according to the trial run dataset, with the speed distribution and temperature compensation of the output single rotor as the reference operating point.

[0108] In a specific embodiment, the single-rotor dehumidification energy-saving optimization control method based on model prediction provided by the present invention includes step 3, which is based on distributed robust constraints and quadratic programming feedforward compensation. Utilizing the distributed robust mechanism, the worst-case scenario of the gap is estimated through the sample mean and variance (based on the Chebyshev inequality), and then quadratic programming is used to allocate rotational speed and temperature compensation, balancing safety margin and energy saving.

[0109] Among them, the worst-case compensation boundary obtained by the Chebyshev inequality ensures that the real gap is covered with 95% confidence under any distribution, effectively reducing the risk of system runaway; the quadratic programming aims to minimize heating and speed increments while meeting the minimum compensation requirements, taking into account energy saving and stable equipment operation; the feedforward compensation can be directly issued without iterative solution, significantly shortening the disturbance response time.

[0110] Sub-step 3.1: Determine the sample mean and variance of the historical observation gap rate samples in real time.

[0111] Specifically, the historical observation gap rate sample is represented as . The number of sampling points is consistent with that in step 1.3; Indicates the first The group gap rate is the difference between the target absolute humidity of the inlet and outlet air vents and the actual absolute humidity of the inlet and outlet air vents.

[0112] Sub-step 3.2, Chebyshev distributed robust security conditions.

[0113] Specifically, based on the historical observed gap rate samples, a Chebyshev inequality is generated, as shown in the following formula:

[0114]

[0115] In the formula, represents a probability, and are the sample mean and variance of the historical observed gap rate samples, respectively. For describing the risk tolerance, in this case, can take values .

[0116] Substitute the sensitivity coefficient into the Chebyshev inequality to obtain a simplified robustness certainty constraint, as shown in the following formula:

[0117]

[0118] In the formula, are the sensitivity coefficients, is the parameter to be controlled, and represents the value that the rotor motor speed needs to increase or decrease relative to the reference speed, and the amplitude that the regenerative section inlet air temperature needs to increase or decrease relative to the reference temperature under the current operating condition.

[0119] Sub-step 3.3, sensitivity coefficient calibration.

[0120] Specifically, first, the commissioning data set, i.e., the temperature data and humidity data collected in step 1, is obtained, combined with the outputted to-be-controlled parameters in sub-step 3.2, and then local one-dimensional linear regression is used again to process, so as to output the real-time rotor motor speed and the reference temperature under the calibrated sensitivity coefficient of the current operating condition.

[0121] Step S140, based on the speed distribution and temperature compensation, combined with the predicted disturbance of the temperature to the adsorption amount outputted by the mechanism framework model, the MPC algorithm is used to solve the optimal speed and temperature control sequence of the single-rotor in real time.

[0122] In some embodiments, the model prediction-based single-rotor dehumidification energy-saving optimization control method provided by the present application specifically comprises the following steps:

[0123] Step S141, based on the current humidity error in the single-rotor air flow pipeline and the minimum value of the speed distribution and temperature compensation, combined with the predicted disturbance and configuration parameters, a discrete model is constructed by using the MPC algorithm.

[0124] Step S142: Obtain the configuration to get the prediction step size, energy consumption weight, balance factor and robust safety constraints, and optimize the discrete model based on the prediction step size, energy consumption weight, balance factor and robust safety constraints, combined with temperature threshold and airflow rate threshold, to obtain the optimization objective and constraints.

[0125] Step S143: Input the current temperature error, predicted disturbance, optimization objective and constraints, and configuration parameters of the discrete model into the QP solver. Calculate the optimal rotational speed and temperature control sequence in the rolling time domain within the QP solver.

[0126] In a specific embodiment, the single-rotor dehumidification energy-saving optimization control method based on model prediction provided by the present invention includes step 4, joint optimization scheduling based on model predictive control. In the rolling time domain, a discrete-time system model and optimization objective are constructed using distributed robust compensation (i.e., the output of step 3) and disturbance prediction. The optimal rotor speed and regeneration air temperature control sequence are solved in real time using the model predictive control (MPC) algorithm, taking into account both humidity stability and energy saving.

[0127] Feedforward + Feedback Dual-Drive: The robust compensation and initial value of the quadratic programming in step 3 are introduced into MPC, combined with rolling optimization, to achieve proactive pre-adjustment and real-time correction of future disturbances.

[0128] Multi-objective balance: The MPC objective function simultaneously minimizes the sum of squared humidity deviation and energy consumption weighting, ensuring that comfort and energy saving are achieved in tandem.

[0129] Adaptation and learning: By correcting model parameters and weights through post-execution feedback, the control strategy can be continuously improved in long-term operation to adapt to material aging or changes in operating conditions.

[0130] Includes the following sub-steps:

[0131] Sub-step 4.1, State-space model extension and introduction of feedforward initial values.

[0132] Specifically, based on the current humidity deviation, the minimum robust compensation amount, the disturbance prediction results, and the system configuration parameters, the system configuration parameters are loaded into the controller. The robust compensation amount output from sub-step 3.4 is used as an initial guess and input into the MPC algorithm in a warmstart manner to construct the final discrete model. This discrete model is used to predict the humidity deviation in the future. Feedforward ensures that the system has completed minimum robust compensation before MPC optimization; the initial value accelerates MPC convergence and reduces the online computational burden.

[0133] The expression for the discrete model is:

[0134]

[0135] In the formula, the humidity deviation for future time, 、 、 are system configuration parameters, 、 is the real-time rotational speed of the runner motor and the reference temperature under the current working condition after the sensitivity coefficient is calibrated.

[0136] Sub-step 4.2, integration of MPC optimization target and robust constraint.

[0137] Specifically, the prediction step, energy consumption weight and balance factor are first determined, while the upper and lower limits of the rotational speed and temperature control (i.e. the rotational speed adjustment and temperature control adjustment range are within the respective set upper and lower limits) and the robust safety constraint are controlled.

[0138] Among them, the energy consumption weight is divided into rotational speed energy consumption weight and temperature energy consumption weight, which are used to measure the unit energy consumption increment brought by the increase of 1% of the rotational speed adjustment and the unit energy consumption increment brought by the increase of 1℃ of the regenerative air temperature, respectively. They are the coefficients of the energy consumption term in the quadratic optimization target, used to balance the trade-off between dehumidification effect and energy consumption.

[0139] The expression of the robust safety constraint is:

[0140]

[0141] In the formula, 、 are the minimum values of the rotational speed and temperature in the robust compensation, represents the time step, is the prediction step, and when , ; .

[0142] Sub-step 4.3, online MPC rolling solution.

[0143] Specifically, based on the current humidity deviation, disturbance prediction sequence and robust safety constraint, combined with system configuration parameters, all of them are input into the QP solver (such as OSQP), in which the control of the first time is only executed in a rolling horizon way, and the remaining sequence is used for the next warm start, thereby improving the solving efficiency. This step realizes the predictive compensation of future disturbance prediction and energy consumption optimization by real-time calculation of the globally optimal multi-step control action.

[0144] Step S150, execute the optimal rotational speed and temperature control sequence to optimize the control of single-runner adsorption dehumidification, and obtain the model correction parameters of execution feedback.

[0145] The model prediction-based single-rotor dehumidification energy-saving optimization control system provided by the present application is described below, and the model prediction-based single-rotor dehumidification energy-saving optimization control system described below can be correspondingly referred to the model prediction-based single-rotor dehumidification energy-saving optimization control method described above.

[0146] As shown in Figure 2 In one embodiment, a model prediction-based single-rotor dehumidification energy-saving optimization control system includes a data acquisition and processing module, an adsorption disturbance quantification module, a robustness compensation module, an optimal control sequence solving module, and an optimization control execution module.

[0147] The data acquisition and processing module is used to collect temperature and humidity data and flow data in the single-rotor airflow duct of the air handling system through integrated sensors to establish original time series data and temperature smoothing curves with synchronous time stamps.

[0148] The adsorption disturbance quantification module is used to calculate the equilibrium adsorption curve based on the isotherm parameter set and the temperature smoothing curve, and to construct a mechanism framework model for predicting temperature disturbance of adsorption capacity in combination with kinetic constants.

[0149] The robustness compensation module is used to perform quadratic programming according to the mean and variance of the adsorption capacity sample through a distributed robust mechanism to obtain the speed distribution and temperature compensation of the single-rotor.

[0150] The optimal control sequence solving module is used to solve the optimal speed and temperature control sequence of the single-rotor in real time by using the MPC algorithm based on the speed distribution and temperature compensation, in combination with the predicted disturbance of temperature to adsorption capacity output by the mechanism framework model.

[0151] The optimization control execution module is used to execute the optimal speed and temperature control sequence to optimize the control of single-rotor adsorption dehumidification and obtain model correction parameters of execution feedback.

[0152] The applicant of the present application has made a detailed description and explanation of the embodiments of the present application in combination with the drawings of the specification, but those skilled in the art should understand that the above embodiments are only preferred embodiments of the present application, and the detailed description is only to help the reader better understand the spirit of the present application, and is not a limitation on the protection scope of the present application. On the contrary, any improvement or modification based on the spirit of the present application should fall within the protection scope of the present application.

Claims

1. A model prediction based single-screw dehumidification energy-saving optimization control method, characterized in that, The method comprises: Collecting temperature and humidity data and flow data in a single-rotor airflow duct in an air handling system through integrated sensors to establish original time series data with synchronous timestamps and a temperature smoothing curve; Calculating an equilibrium adsorption curve based on an isotherm parameter set and the temperature smoothing curve, and combining kinetic constants to construct a mechanism framework model for predicting temperature-induced perturbations in adsorption capacity; Using a distributed robust mechanism to perform quadratic programming based on sample mean and variance of adsorption capacity to obtain speed distribution and temperature compensation of the single-rotor; Based on the speed distribution and temperature compensation, combining the predicted temperature-induced perturbations in adsorption capacity output by the mechanism framework model, using an MPC algorithm to solve the optimal speed and temperature control sequence of the single-rotor in real time; Executing the optimal speed and temperature control sequence to optimize control of single-rotor adsorption dehumidification and obtaining model correction parameters for execution feedback; The method comprises: Obtaining an isotherm parameter set and kinetic constants, the isotherm parameter set including maximum adsorption capacity, adsorption energy constant, and empirical exponent; Based on the temperature smoothing curve and isotherm parameter set, calling the Dubinin-Astakhov isotherm model to construct a discrete curve, and calculating the equilibrium adsorption curve based on the discrete curve; The method further comprises: Based on the equilibrium adsorption curve and kinetic constants, updating the discrete time in the equilibrium adsorption curve when the change in adsorption capacity over time satisfies first-order kinetics, to calculate a dynamic adsorption curve based on the updated discrete time; Obtaining a reference adsorption capacity, and calculating an instantaneous efficiency decay ratio based on the dynamic adsorption curve to construct an adsorption efficiency curve, the instantaneous efficiency decay ratio being the ratio between the dynamic adsorption curve and the reference adsorption capacity; The method further comprises: Obtaining the difference between expected humidity and actual humidity of multiple groups of inlet and outlet air ports of the single-rotor, and using the multiple difference values as historical observed gap airflow rate samples to generate a Chebyshev inequality; Writing a sensitivity coefficient into the Chebyshev inequality to obtain a robustness constraint, and using the robustness constraint to calibrate the sensitivity coefficient based on a trial run data set to output the speed distribution and temperature compensation of the single-rotor as a reference operating point.

2. The model prediction based single rotary dehumidification energy saving optimization control method according to claim 1, characterized in that, The method further comprises: A plurality of temperature and humidity sensors and heat flow sensors are arranged at equal intervals in the circumferential direction in the air flow pipeline at the inlet of the dehumidification section to construct integrated sensors for monitoring the temperature difference and flow rate of the air flow at the inlet and outlet, and each sensor in the integrated sensors is connected to an edge node; The temperature and humidity sensors and the heat flow sensors are polled at a set sampling frequency to obtain temperature and humidity data and gas flow data, and the temperature and humidity data and the gas flow data are time-synchronized by the edge node to obtain original time series data with synchronized time stamps; The temperature and humidity sensors are used to collect temperature and humidity data in the single-rotor air flow pipeline, the heat flow sensors are used to collect flow data, and the edge node is in communication connection with a centralized controller.

3. The model prediction based single rotary dehumidification energy saving optimization control method according to claim 2, characterized in that, The temperature and humidity data and the flow data in the single-rotor air flow pipeline in the air handling system are collected by the integrated sensors to establish original time series data with synchronized time stamps and a temperature smooth curve, and the method further comprises: A plurality of sampling points are extracted from the original time series data according to a set time window, and the temperature mean value, average heat flow and heat flow mutation value of all extracted sampling points are calculated based on the plurality of sampling points to construct a feature vector; Based on the feature vector, a first-order prediction formula is iteratively calculated by using an autoregressive plus exogenous input model to obtain discrete prediction values in a future first time period, and a three-point moving average processing is performed on the discrete prediction values to obtain the temperature smooth curve.

4. The model prediction based single rotary dehumidification energy saving optimization control method according to claim 1, characterized in that, Based on the rotational speed distribution and the temperature compensation, the optimal rotational speed and temperature control sequence of the single-rotor are solved in real time by using the MPC algorithm in combination with the predicted disturbance of the temperature to the adsorption amount output by the mechanism framework model, which comprises: Based on the current humidity error in the single-rotor air flow pipeline and the minimum value of the rotational speed distribution and the temperature compensation, the discrete model is constructed by using the MPC algorithm in combination with the predicted disturbance and the configuration parameters; The prediction step, energy consumption weight, balance factor and robust safety constraint are obtained by configuration, and based on the prediction step, energy consumption weight, balance factor and robust safety constraint, the discrete model is optimized in combination with the temperature threshold and the air flow rate threshold to obtain an optimization target and constraint.

5. The model prediction based single rotary dehumidification energy saving optimization control method according to claim 4, characterized in that, Based on the rotational speed distribution and the temperature compensation, the optimal rotational speed and temperature control sequence of the single-rotor are solved in real time by using the MPC algorithm in combination with the predicted disturbance of the temperature to the adsorption amount output by the mechanism framework model, which further comprises: The current temperature error, predicted disturbance, optimization target and constraint, and configuration parameters of the discrete model are input into a QP solver, and the optimal rotational speed and temperature control sequence are calculated in the QP solver according to the rolling time domain.

6. A terminal comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is used to operate according to the instructions to perform the steps of the method according to any one of claims 1-5.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the method according to any one of claims 1-5.

Images