A heat pump system control method based on improved adaptive predictive control

Abstract

The application discloses a heat pump system control method based on improved adaptive prediction control and belongs to the technical field of heat pump system control, which comprises the following steps: S1, a heat pump Hammerstein-Wiener model is constructed based on the thermodynamic mechanism of heat pump core components, a discrete incremental modeling is performed on a control object, and then a prediction model is obtained; S2, model parameters are updated online according to a prediction residual error, a total disturbance estimator is constructed to correct the prediction model; S3, a control amount constraint and an output constraint are contracted, a multi-objective performance index function is established, and a weight matrix is adaptively adjusted according to a current prediction error; and S4, a rolling optimization problem is constructed, the multi-objective performance index function is optimized and solved, a control input is adjusted in real time, and system output data are predicted, so that the heat pump system control method can keep a high tracking precision of hot water outlet temperature, a small control fluctuation and a good energy-saving effect under the conditions of strong disturbance and parameter perturbation.

Description

Technical Field

[0001] This invention belongs to the field of heat pump system control technology, specifically relating to a heat pump system control method based on improved adaptive predictive control. Background Technology

[0002] Driven by the "dual carbon" goal, air source heat pumps have become the core energy-saving equipment in the building HVAC field. Building energy consumption accounts for more than 40% of the total social energy consumption in my country, of which HVAC systems account for 50%. Air source heat pumps, with their "low-energy-driven reverse Carnot cycle" characteristics, have become a key technology for reducing building energy consumption.

[0003] Air source heat pump systems are typical multivariable, strongly coupled, and highly nonlinear thermal systems. Their core components include a compressor, evaporator, condenser, and electronic expansion valve. During operation, the refrigerant flow rate of the compressor is affected by the coupling effect of volumetric efficiency and refrigerant density. The expansion valve exhibits significant laminar / turbulent flow switching characteristics. Furthermore, the heat exchange processes in the evaporator and condenser are continuously disturbed by ambient temperature, heat exchange boundaries, and load changes. Therefore, traditional linear control methods struggle to simultaneously achieve rapid response, steady-state accuracy, and energy consumption optimization.

[0004] While existing predictive control methods can handle input-output constraints to some extent, they typically rely on fixed predictive models. When ambient temperature drops sharply, load changes, or heat transfer characteristics degrade, model mismatch accumulates rapidly, leading to increased prediction bias, which in turn causes increased overshoot, prolonged settling time, and even constraint violations. On the other hand, control schemes that solely rely on offline identification models struggle to maintain long-term stable control performance when heat pump operating conditions continuously change. Therefore, providing an improved adaptive predictive control method that combines mechanistic modeling, online adaptive correction, disturbance compensation, and constraint contraction to enhance the temperature control accuracy, robustness, and energy-saving performance of air-source heat pump systems under varying operating conditions has become a pressing technical challenge in this field. Summary of the Invention

[0005] The purpose of this invention is to provide a control method for a heat pump system based on improved adaptive predictive control. This method uses the predicted residual of hot water outlet temperature to correct the model parameters online, introduces a total disturbance estimator to compensate for unmodeled errors, and constructs a predictive control optimization problem with constrained contraction and adaptive weight adjustment. This effectively improves the temperature control accuracy and robustness, and overcomes the technical problems in existing methods where the prediction accuracy of a fixed model decreases under varying operating conditions, and unmodeled disturbances and environmental temperature fluctuations have a significant impact on the control effect.

[0006] This invention is achieved through the following measures: a control method for a heat pump system based on improved adaptive predictive control, specifically including the following steps: Step S1: Construct a Hammerstein-Wiener model of the heat pump based on the thermodynamic mechanism of the core components of the heat pump, perform discrete incremental modeling of the controlled object, and then obtain the local linear incremental prediction model of the heat pump system in the discrete time domain. Step S2: Update the model parameters online based on the predicted residuals and construct a total perturbation estimator to correct the prediction model; Step S3: Shrink the control quantity constraints and output quantity constraints, establish a multi-objective performance index function, and adaptively adjust the weight matrix in the multi-objective performance index function according to the current prediction error; Step S4: Construct a rolling optimization problem, optimize and solve the multi-objective performance index function, and adjust the control input in real time to predict the system output data.

[0007] As a further improvement of the present invention, the specific implementation method of step S1 is as follows: Step S11: Construct nonlinear models of the core components of the air source heat pump, including the compressor nonlinear model, the expansion valve nonlinear model, the ambient temperature correction model, and the hot water outlet temperature model. Compressor nonlinear model: Based on thermodynamic mechanisms and the ideal gas law, a saturation function is introduced to characterize the physical saturation characteristics of the rotational speed. The model form is as follows: (1) in, For volumetric efficiency, For compressor speed input, Theoretical displacement For refrigerant density, , Where is pressure, m is molar mass, R is ideal gas constant, and T is refrigerant temperature; The compressor model with saturation characteristics is represented as: (2) in, and It is a constant.

[0008] Nonlinear model of expansion valve: based on the pressure difference across the valve , The condenser outlet pressure, The evaporator inlet pressure is used; turbulent and laminar flow states are modeled separately, and a pressure drop threshold is introduced. The model is integrated as follows: (3) in Input for expansion valve opening. The expansion valve flow coefficient; Ambient temperature correction model: An ambient temperature correction term is introduced to compensate for the decrease in the evaporator heat transfer coefficient. The equivalent input is: (4) in, This is the ambient temperature correction factor. The ambient temperature; Hot water outlet temperature model: Combining the heat transfer and exponential saturation characteristics of the condenser, the model is as follows: (5) in, This refers to the inlet temperature of the cold water. For the maximum temperature rise, These are nonlinear coefficients. This refers to the condenser temperature. Step S12: Integrate the compressor nonlinear model, expansion valve nonlinear model, ambient temperature correction model, and hot water outlet temperature model to construct the heat pump Hammerstein-Wiener model, represented as: (6) (7) (8) In the formula, t It is a time variable. and These are the system's input and output signals, respectively. It is the output of the nonlinear part inside the system. It is the linear dynamic part within the system; It is a unit shift operator: = ; , yes The constant polynomial in the equation is defined as follows: (9) (10) in It is a polynomial The factors, It is a polynomial The factors and order n are known; output and the output of the nonlinear part The relationship between them can be obtained from formula (7): (11) Based on equations (1), (3), and (4), equation (6) is updated to: (12) in, For volumetric efficiency Theoretical displacement Compared with reference density The product of the three. , It is a constant.

[0009] According to equation (5), equation (8) is updated to: (13) in, , It is a constant.

[0010] Step S13: Perform discrete incremental modeling on the controlled object, assuming the sampling period is... In the Select the current operating point near each control cycle:

[0011] Define the incremental form of each variable: (14) (15) (16) (17) Perform a first-order Taylor expansion of equation (12) at the current working point: (18) The compressor-side gain is: (19) The gain on the expansion valve side is related to the flow range it is in: when Then, from equation (3), we get: (20) when Then, from equation (3), we get: (twenty one) For equation (13) in Expand the first-order Taylor model: (twenty two) in: (twenty three) From equation (11), we can obtain the incremental form: (twenty four) Substituting equation (18) into equation (24), and then performing output mapping using equation (22), we can obtain the local linear incremental prediction model of the heat pump system in the discrete time domain: (25) in, This is the comprehensive equivalent disturbance term. .

[0012] As a further improvement of the present invention, the specific implementation method of step S2 is as follows: Step S21: Update the model parameters online based on the predicted residual of the hot water outlet temperature; To overcome the model mismatch problem of heat pump systems under varying operating conditions, this invention employs a recursive least squares algorithm with a variable forgetting factor to update the parameter vector online. Define the one-step prediction error: (26) The parameter update law is: (27) The gain matrix is: (28) The covariance matrix is ​​updated as follows: (29) Among them, the forgetting factor Designed as a forgetting factor that varies with the magnitude of the residual: (30) In the formula, , This is the error adjustment coefficient.

[0013] As can be seen from equation (30), when the prediction error is large, Reducing the weight of older data decreases the overall weight of the model, thus speeding up model updates; when the prediction error is small... As the model parameters increase, they tend to stabilize, thus balancing rapid adaptability with steady-state noise suppression.

[0014] Step S22: Construct the total disturbance estimator and further refine the model; Considering factors such as drastic changes in ambient temperature, heat exchanger frosting, sensor bias, and unmodeled dynamics, this invention uniformly characterizes these effects as a total disturbance. Construct a disturbance estimator: (31) in, The smoothing factor is estimated for the perturbation.

[0015] Adding the disturbance term to the prediction model of equation (25) yields a one-step prediction equation with compensation: (32) The total disturbance estimate is continuously corrected using real-time residuals to reduce the impact of unmodeled factors on the control effect; to perform multi-step rolling prediction, the state space of the prediction model is constructed, and an augmented state vector is defined: (33) Equation (25) can then be expressed in state-space form: (34) (35) in: (36) , , , All parameters are updated online. Real-time reconstruction, assuming the prediction time domain is... Control time domain is By recursively deriving equations (35) and (36), we can obtain the first... The matrix form of the step prediction output: (37) in:

[0016] (38) The free response matrix, To control the incremental mapping matrix, This is the perturbation mapping matrix.

[0017] As a further improvement of the present invention, the specific implementation method of step S3 is as follows: Step S31: Since there are model uncertainties in the actual operation of the heat pump system, the present invention does not directly use the original constraint boundary, but dynamically shrinks the constraint based on the disturbance estimate and residual statistics; Define the control constraint shrinkage amount: (39) Define the output constraint shrinkage amount: (40) in, , The shrinkage coefficient, This represents the standard deviation of the prediction error within the most recent sliding window. The control constraints are: (41) The output constraints are: (42) This approach can automatically reserve a safety margin for control when model uncertainty increases.

[0018] Step S32: Construct a multi-objective performance index function. To balance tracking accuracy, actuator smoothness, and energy efficiency, the following performance index function is constructed: (43) in, This serves as a reference trajectory for the hot water outlet temperature. For the output error weights, To control the incremental weight matrix, As an energy consumption penalty item, For soft-constrained slack variables, The relaxation penalty coefficient; In this invention, the weight matrix is ​​adaptively adjusted based on the current prediction error: (44) (45) In the formula, and These are the baseline weights, This is the adjustment coefficient.

[0019] As a further improvement of the present invention, the implementation method of step S4 is as follows: Step S41: Construct a rolling optimization problem. Substitute equation (37) into equation (43) to obtain the standard quadratic form: (46) in, (47) (48) Under the condition of satisfying the contraction constraints and soft constraints shown in equations (41) and (42), solve the following optimization problem: (49) Step S42: By continuously performing rolling optimization, updating parameters, estimating disturbances, and repeating the above process, the optimal control input u(k) is obtained, and the system output water temperature data is predicted, thereby efficiently completing the control process of the heat pump system.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) This invention establishes a physically interpretable Hammerstein-Wiener prediction model based on the thermodynamic mechanism of the core components of the heat pump, which can more accurately characterize the strong nonlinear dynamic characteristics of the air source heat pump system. Compared with the traditional control methods based solely on linear models or purely empirical models, this invention incorporates nonlinear mechanisms such as compressor saturation characteristics, expansion valve segmented flow characteristics, ambient temperature correction, and hot water outlet temperature into the prediction model, which can more realistically reflect the input-output relationship of the heat pump system under different operating conditions, thereby providing a more reliable model basis for subsequent rolling optimization. This not only improves the prediction accuracy but also enhances the engineering interpretability and feasibility of the control strategy.

[0021] (2) This invention improves the model adaptability of the heat pump system under varying operating conditions through a residual-driven online parameter update mechanism, effectively suppressing the control performance degradation caused by model mismatch. Traditional model predictive control mostly relies on a fixed model. When the environment changes or equipment parameters drift, the model prediction accuracy will drop rapidly, leading to an increase in system output tracking error and aggravated control fluctuations. This invention corrects the model parameters online based on the prediction residual of the hot water outlet temperature and adaptively adjusts the update speed through a variable forgetting factor mechanism. This enables the model to quickly track changes in operating conditions when the error is large, and to maintain parameter stability when the error is small, avoiding frequent oscillations. Therefore, it can maintain high prediction accuracy and control stability under complex operating environments.

[0022] (3) This invention introduces a total disturbance estimation and compensation mechanism, which can uniformly compensate for external disturbances, unmodeled dynamics, and parameter perturbations, significantly reducing the prediction error and tracking error of hot water outlet temperature and enhancing the system's anti-interference capability. Compared with traditional fixed model predictive control, this invention can significantly reduce the cumulative deviation caused by model incompleteness, improve the reliability of future output prediction, and thus improve the controller's anti-disturbance performance and steady-state control accuracy of the heat pump system. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of the improved adaptive predictive control in this invention, wherein... Configure the system output. For reference trajectory, For the actual output, For model output, To predict the output, For control input; Figure 2 This is a schematic diagram of the Hammerstein-Wiener model of the heat pump system in this invention; Figure 3This is a schematic diagram of a compressor model with saturation characteristics in this invention; Figure 4 The improved adaptive predictive control in this invention, compared with traditional predictive control, predicts the output of the heat pump control system over time t. The curves show a comparison of the changes; where MPC represents the prediction curve before improvement, RDAMPC represents the prediction curve after improvement, and True represents the actual output curve of the system. Figure 5 This is a comparison of the curves showing the change of the output prediction error of the heat pump control system over time t between the improved adaptive predictive control and the traditional predictive control in this invention; where MPC represents the prediction error curve before the improvement, and RDAMPC represents the prediction error curve after the improvement. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. Of course, the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0025] Example 1 See Figures 1 to 5 This embodiment provides a control method for a heat pump system based on improved adaptive predictive control, the core implementation steps of which are as follows: Step S1: Construct a Hammerstein-Wiener model of the heat pump based on the thermodynamic mechanism of the core components of the heat pump, perform discrete incremental modeling of the controlled object, and then obtain the local linear incremental prediction model of the heat pump system in the discrete time domain. Step S2: Update the model parameters online based on the predicted residuals and construct a total perturbation estimator to correct the prediction model; Step S3: Shrink the control quantity constraints and output quantity constraints, establish a multi-objective performance index function, and adaptively adjust the weight matrix in the multi-objective performance index function according to the current prediction error; Step S4: Construct a rolling optimization problem, optimize and solve the multi-objective performance index function, and adjust the control input in real time to predict the system output data.

[0026] In this embodiment, the specific implementation method of step S1 is as follows: Step S11: Construct nonlinear models of the core components of the air source heat pump, including the compressor nonlinear model, the expansion valve nonlinear model, the ambient temperature correction model, and the hot water outlet temperature model. Compressor nonlinear model: Based on thermodynamic mechanisms and the ideal gas law, a saturation function is introduced to characterize the physical saturation characteristics of the rotational speed. The model form is as follows: (1) in, For volumetric efficiency, For compressor speed input, Theoretical displacement For refrigerant density, , Where is pressure, m is molar mass, R is ideal gas constant, and T is refrigerant temperature; The compressor model with saturation characteristics is represented as: (2) in, and It is a constant.

[0027] Nonlinear model of expansion valve: based on the pressure difference across the valve , The condenser outlet pressure, The evaporator inlet pressure is used; turbulent and laminar flow states are modeled separately, and a pressure drop threshold is introduced. The model is integrated as follows: (3) in Input for expansion valve opening. The expansion valve flow coefficient; Ambient temperature correction model: An ambient temperature correction term is introduced to compensate for the decrease in the evaporator heat transfer coefficient. The equivalent input is: (4) in, This is the ambient temperature correction factor. The ambient temperature; Hot water outlet temperature model: Combining the heat transfer and exponential saturation characteristics of the condenser, the model is as follows: (5) in, This refers to the inlet temperature of the cold water. For the maximum temperature rise, These are nonlinear coefficients. This refers to the condenser temperature. Step S12: Integrate the compressor nonlinear model, expansion valve nonlinear model, ambient temperature correction model, and hot water outlet temperature model to construct the heat pump Hammerstein-Wiener model, represented as: (6) (7) (8) In the formula, t It is a time variable. and These are the system's input and output signals, respectively. It is the output of the nonlinear part inside the system. It is the linear dynamic part within the system; It is the unit shift operator: = ; , yes The constant polynomial in the equation is defined as follows: (9) (10) in It is a polynomial The factors, It is a polynomial The factors and order n are known; output and the output of the nonlinear part The relationship between them can be obtained from formula (7): (11) Based on equations (1), (3), and (4), equation (6) is updated to: (12) in, For volumetric efficiency Theoretical displacement Compared with reference density The product of the three. , It is a constant.

[0028] Based on equations (5) and (8), the following is updated: (13) in, , It is a constant.

[0029] Step S13: Perform discrete incremental modeling on the controlled object, assuming the sampling period is... In the Select the current operating point near each control cycle:

[0030] Define the incremental form of each variable: (14) (15) (16) (17) Perform a first-order Taylor expansion of equation (12) at the current working point: (18) The compressor-side gain is: (19) The gain on the expansion valve side is related to the flow range it is in: when Then, from equation (3), we get: (20) when Then, from equation (3), we get: (twenty one) For equation (13) in Expand the first-order Taylor model: (twenty two) in: (twenty three) From equation (11), we can obtain the incremental form: (twenty four) Substituting equation (18) into equation (24), and then performing output mapping using equation (22), we can obtain the local linear incremental prediction model of the heat pump system in the discrete time domain: (25) in, This is the comprehensive equivalent disturbance term. .

[0031] In this embodiment, the specific implementation method of step S2 is as follows: Step S21: Update the model parameters online based on the predicted residual of the hot water outlet temperature; To overcome the model mismatch problem of heat pump systems under varying operating conditions, this invention employs a recursive least squares algorithm with a variable forgetting factor to update the parameter vector online. Define the one-step prediction error: (26) The parameter update law is: (27) The gain matrix is: (28) The covariance matrix is ​​updated as follows: (29) Among them, the forgetting factor Designed as a forgetting factor that varies with the magnitude of the residual: (30) In the formula, , This is the error adjustment coefficient.

[0032] As can be seen from equation (30), when the prediction error is large, Reducing the weight of older data decreases the overall weight of the model, thus speeding up model updates; when the prediction error is small... As the model parameters increase, they tend to stabilize, thus balancing rapid adaptability with steady-state noise suppression.

[0033] Step S22: Construct the total disturbance estimator and further refine the model; Considering factors such as drastic changes in ambient temperature, heat exchanger frosting, sensor bias, and unmodeled dynamics, this invention uniformly characterizes these effects as a total disturbance. Construct a disturbance estimator: (31) in, The smoothing factor is estimated for the perturbation.

[0034] Adding the disturbance term to the prediction model of equation (25) yields a one-step prediction equation with compensation: (32) The total disturbance estimate is continuously corrected using real-time residuals to reduce the impact of unmodeled factors on the control effect; to perform multi-step rolling prediction, the state space of the prediction model is constructed, and an augmented state vector is defined: (33) Equation (25) can then be expressed in state-space form: (34) (35) in: (36) , , , All parameters are updated online. Real-time reconstruction, assuming the prediction time domain is... Control time domain is By recursively deriving equations (35) and (36), we can obtain the first... The matrix form of the step prediction output: (37) in:

[0035] (38) The free response matrix, To control the incremental mapping matrix, This is the perturbation mapping matrix.

[0036] In this embodiment, the specific implementation method of step S3 is as follows: Step S31: Since there are model uncertainties in the actual operation of the heat pump system, the present invention does not directly use the original constraint boundary, but dynamically shrinks the constraint based on the disturbance estimate and residual statistics; Define the control constraint shrinkage amount: (39) Define the output constraint shrinkage amount: (40) in, , The shrinkage coefficient, This represents the standard deviation of the prediction error within the most recent sliding window. The control constraints are: (41) The output constraints are: (42) This approach can automatically reserve a control safety margin when model uncertainty increases; Step S32: Construct a multi-objective performance index function. To balance tracking accuracy, actuator smoothness, and energy efficiency, the following performance index function is constructed: (43) in, This serves as a reference trajectory for the hot water outlet temperature. For the output error weights, To control the incremental weight matrix, As an energy consumption penalty item, For soft-constrained slack variables, The relaxation penalty coefficient; In this invention, the weight matrix is ​​adaptively adjusted based on the current prediction error: (44) (45) In the formula, and These are the baseline weights, This is the adjustment coefficient.

[0037] In this embodiment, step S4 is implemented as follows: Step S41: Construct a rolling optimization problem. Substitute equation (37) into equation (43) to obtain the standard quadratic form: (46) in, (47) (48) Under the condition of satisfying the contraction constraints and soft constraints shown in equations (41) and (42), solve the following optimization problem: (49) Step S42: By continuously performing rolling optimization, updating parameters, estimating disturbances, and repeating the above process, the optimal control input u(k) is obtained, and the system output water temperature data is predicted, thereby efficiently completing the control process of the heat pump system.

[0038] Example 2 Based on Example 1, this example employs an improved adaptive predictive control method, such as... Figure 1 As shown. The control method uses the Hammerstein-Wiener model of the heat pump established based on the thermodynamic mechanism of the core components as the initial prediction model, and introduces online parameter updates, total disturbance estimation compensation, and constraint contraction mechanisms during the rolling optimization process.

[0039] Using the Hammerstein-Wiener model of the heat pump system mentioned above, the following model can be established for this embodiment: Output stage: (50) Intermediate linear link: (51) (52) Input process: (53) (54) Comparing the above model and step S1, the initial values ​​of the parameter vector can be constructed based on the above identification results as follows:

[0040] Based on step S21, the parameters are updated using recursive least squares with a variable forgetting factor: (55) (56) The forgetting factor is taken as: (57) A total disturbance estimator is further introduced based on step S22: (58) The predictive controller in this embodiment employs a rolling optimization method, taking the prediction time domain as... Control time domain The sampling period is taken as .

[0041] The constraint contraction mechanism is introduced according to step S31. Let the compressor speed constraint, electronic expansion valve opening constraint, and hot water outlet temperature constraint be as follows: (59) (60) (61) Furthermore, this embodiment introduces a constraint contraction mechanism, wherein the input constraint contraction coefficient is taken as... The output constraint shrinkage coefficient is taken as .

[0042] The multi-objective performance index function is constructed according to step S32. The construction method is the same as that in Example 1, and will not be repeated here.

[0043] The rolling optimization problem is constructed according to step S41. Its prediction output matrix expression is the same as that in Example 1, and will not be repeated here.

[0044] The rolling optimization problem can be written as: (62) According to step S42, by continuously performing rolling optimization, updating parameters, estimating disturbances, and repeating the above process, the optimal control input u(k) is obtained, and the system output water temperature data is predicted, thereby efficiently completing the control process of the heat pump system.

[0045] The improved adaptive predictive control of this embodiment is compared with the traditional predictive control to predict the change of the output of the heat pump control system over time t, for example. Figure 4 As shown, the variation of the output prediction error of the heat pump control system with time t is compared between the improved adaptive predictive control and the traditional predictive control. Figure 5As shown in the figure. The results show that under conditions of sudden drop in ambient temperature and load fluctuation, the method of this embodiment can significantly reduce the prediction error of hot water outlet temperature, effectively suppress overshoot in the initial stage of control, reduce the drastic fluctuations in compressor speed and electronic expansion valve opening, and improve the system temperature control accuracy and operational stability, thus demonstrating that the method of this invention has good applicability to heat pump systems.

[0046] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvements made using the inventive concept and technical solution of the present invention, or the direct application of the inventive concept and technical solution of the present invention to other occasions without modification, are all within the protection scope of the present invention.

Claims

1. A control method for a heat pump system based on improved adaptive predictive control, characterized in that, Includes the following steps: Step S1: Construct a Hammerstein-Wiener model of the heat pump based on the thermodynamic mechanism of the core components of the heat pump, perform discrete incremental modeling of the controlled object, and then obtain the local linear incremental prediction model of the heat pump system in the discrete time domain. Step S2: Update the model parameters online based on the predicted residuals and construct a total perturbation estimator to correct the prediction model; Step S3: Shrink the control quantity constraints and output quantity constraints, establish a multi-objective performance index function, and adaptively adjust the weight matrix in the multi-objective performance index function according to the current prediction error; Step S4: Construct a rolling optimization problem, optimize and solve the multi-objective performance index function, and adjust the control input in real time to predict the system output data.

2. The heat pump system control method based on improved adaptive predictive control according to claim 1, characterized in that, The specific implementation method of step S1 is as follows: Step S11: Construct nonlinear models of the core components of the air source heat pump, including the compressor nonlinear model, the expansion valve nonlinear model, the ambient temperature correction model, and the hot water outlet temperature model. Step S12: Integrate the compressor nonlinear model, expansion valve nonlinear model, ambient temperature correction model, and hot water outlet temperature model to construct the heat pump Hammerstein-Wiener model; Step S13: Perform discrete incremental modeling on the controlled object to obtain the local linear incremental prediction model of the heat pump system in the discrete time domain.

3. The heat pump system control method based on improved adaptive predictive control according to claim 2, characterized in that, The compressor nonlinear model, expansion valve nonlinear model, ambient temperature correction model, and hot water outlet temperature model in step S11 are as follows: Compressor nonlinear model: Based on thermodynamic mechanisms and the ideal gas law, a saturation function is introduced to characterize the physical saturation characteristics of the rotational speed. The model form is as follows: (1) in, For volumetric efficiency, For compressor speed input, Theoretical displacement For refrigerant density, , Where is pressure, m is molar mass, R is ideal gas constant, and T is refrigerant temperature; The compressor model with saturation characteristics is represented as: (2) in, and It is a constant; Nonlinear model of expansion valve: based on the pressure difference across the valve , The condenser outlet pressure, The evaporator inlet pressure is used; turbulent and laminar flow states are modeled separately, and a pressure drop threshold is introduced. The model is integrated as follows: (3) in, Input for expansion valve opening. The expansion valve flow coefficient; Ambient temperature correction model: An ambient temperature correction term is introduced to compensate for the decrease in the evaporator heat transfer coefficient. The equivalent input is: (4) in, This is the ambient temperature correction factor. The ambient temperature; Hot water outlet temperature model: Combining the heat transfer and exponential saturation characteristics of the condenser, the model is as follows: (5) in, This refers to the inlet temperature of the cold water. For the maximum temperature rise, These are nonlinear coefficients. This refers to the condenser temperature.

4. The heat pump system control method based on improved adaptive predictive control according to claim 3, characterized in that, The heat pump Hammerstein-Wiener model in step S12 is represented as follows: (6) (7) (8) In the formula, t It is a time variable. and These are the system's input and output signals, respectively. It is the output of the nonlinear part inside the system. It is the linear dynamic part within the system; It is a unit shift operator: = ; , yes The constant polynomial in the equation is defined as follows: (9) (10) in, It is a polynomial The factors, It is a polynomial The factors and order n are known; output and the output of the nonlinear part The relationship between them can be obtained from formula (7): (11) Based on equations (1), (3), and (4), equation (6) is updated to: (12) in, For volumetric efficiency Theoretical displacement Compared with reference density The product of the three , It is a constant; According to equation (5), equation (8) is updated to: (13) in, , It is a constant.

5. The heat pump system control method based on improved adaptive predictive control according to claim 4, characterized in that, The method for discrete incremental modeling in step S13 is as follows: Let the sampling period be In the Select the current operating point near each control cycle: Define the incremental form of each variable: (14) (15) (16) (17) Perform a first-order Taylor expansion of equation (12) at the current working point: (18) The compressor-side gain is: (19) The gain on the expansion valve side is related to the flow range it is in: when Then, from equation (3), we get: (20) when Then, from equation (3), we get: (21) For equation (13) in Expand the first-order Taylor model: (22) in: (23) From equation (11), we can obtain the incremental form: (24) Substituting equation (18) into equation (24), and then performing output mapping using equation (22), we can obtain the local linear incremental prediction model of the heat pump system in the discrete time domain: (25) in, This is the comprehensive equivalent disturbance term. 。 6. The heat pump system control method based on improved adaptive predictive control according to claim 5, characterized in that, The specific implementation method of step S2 is as follows: Step S21: Update the model parameters online based on the predicted residual of the hot water outlet temperature; The parameter vector is updated online using a recursive least squares algorithm with a variable forgetting factor. Define the one-step prediction error: (26) The parameter update law is: (27) The gain matrix is: (28) The covariance matrix is ​​updated as follows: (29) Among them, the forgetting factor Designed as a forgetting factor that varies with the magnitude of the residual: (30) In the formula, , This is the error adjustment coefficient; As can be seen from equation (30), when the prediction error is large, Reducing the weight of older data decreases the overall weight of the model, thus speeding up model updates; when the prediction error is small... As the model parameters increase, they tend to stabilize, thus balancing rapid adaptability with steady-state noise suppression. Step S22: Construct the total disturbance estimator and further refine the model; The effects of drastic changes in ambient temperature, heat exchanger frosting, sensor bias, and unmodeled dynamics are uniformly represented as a total disturbance. Construct a disturbance estimator: (31) in, To estimate the smoothing factor for the perturbation; Adding the disturbance term to the prediction model of equation (25) yields a one-step prediction equation with compensation: (32) The total disturbance estimate is continuously corrected using real-time residuals to reduce the impact of unmodeled factors on the control effect; to perform multi-step rolling prediction, the state space of the prediction model is constructed, and an augmented state vector is defined: (33) Equation (25) can then be expressed in state-space form: (34) (35) in: (36) , , , All parameters are updated online. Real-time reconstruction, assuming the prediction time domain is... Control time domain is By recursively deriving equations (35) and (36), we can obtain the first... The matrix form of the step prediction output: (37) in: (38) The free response matrix, To control the incremental mapping matrix, This is the perturbation mapping matrix.

7. The heat pump system control method based on improved adaptive predictive control according to claim 6, characterized in that, The specific implementation method of step S3 is as follows: Step S31: Dynamically shrink the constraints based on the disturbance estimate and residual statistics; Define the control constraint shrinkage amount: (39) Define the output constraint shrinkage amount: (40) in, , The shrinkage coefficient, This represents the standard deviation of the prediction error within the most recent sliding window. The control constraints are: (41) The output constraints are: (42) Step S32: Construct a multi-objective performance index function, as follows: (43) in, This serves as a reference trajectory for the hot water outlet temperature. For the output error weights, To control the incremental weight matrix, As an energy consumption penalty item, For soft-constrained slack variables, This is the relaxation penalty coefficient.

8. The heat pump system control method based on improved adaptive predictive control according to claim 7, characterized in that, The weight matrix in step S32 is adaptively adjusted based on the current prediction error: (44) (45) In the formula, and These are the baseline weights, This is the adjustment coefficient.

9. The heat pump system control method based on improved adaptive predictive control according to claim 1, characterized in that, The specific implementation method of step S4 is as follows: Step S41: Construct a rolling optimization problem. Substitute equation (37) into equation (43) to obtain the standard quadratic form: (46) in, (47) (48) Under the condition of satisfying the contraction constraints and soft constraints shown in equations (41) and (42), solve the following optimization problem: (49) Step S42: By continuously performing rolling optimization, updating parameters, estimating disturbances, and repeating the above process, the optimal control input u(k) is obtained, and the system output water temperature data is predicted, thereby completing the control process of the heat pump system.

Images