A photovoltaic direct-drive heat pump control method and system

Abstract

This invention discloses a photovoltaic direct-drive heat pump control method and system. The method constructs a refrigerant dynamic state observer to estimate the refrigerant distribution state within the evaporator and condenser online, pre-defines the refrigerant safe operating envelope, and performs forward simulation prediction of the refrigerant state evolution trajectory under different candidate speed change rates based on the current refrigerant state vector in each control decision cycle. A bisection search is used to determine the maximum safe speed increase rate and the maximum safe speed decrease rate, forming a safe feasible region. The speed change rate required for photovoltaic power tracking is then clipped to this feasible region, absorbing the power difference. This invention transforms the speed change rate constraint from a fixed constant into a dynamic function of the refrigerant state, resolving the contradiction between safety and power tracking speed in existing fixed-limit schemes.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic direct-drive variable frequency heat pump control technology, and in particular to a photovoltaic direct-drive heat pump control method and system. Background Technology

[0002] A photovoltaic direct-drive heat pump system is an energy-saving heating / cooling system that directly drives a compressor via a frequency converter using the output power of a photovoltaic array, eliminating the need for a grid-connected inverter. In this system, a maximum power point tracking (MPPT) algorithm tracks the current maximum power point of the photovoltaic array to obtain available power, converts this power into a target compressor speed, and then the frequency converter outputs the corresponding frequency to drive the compressor. This direct-drive architecture has significant advantages in reducing energy conversion losses, but it also presents a significant challenge to compressor speed control—photovoltaic power can change abruptly by 50% to 80% within 2 to 5 seconds due to factors such as cloud cover, and the compressor speed must quickly follow these changes to fully utilize the photovoltaic energy.

[0003] To address the impact of speed variations on the refrigeration cycle, existing technologies generally employ a "fixed speed change rate limit" for constraint. This involves setting a constant maximum speed change rate (e.g., not exceeding 10 Hz / min or 5 Hz / s) on the inverter or controller side, causing the compressor speed to rise or fall at a fixed slope. Simultaneously, the evaporator outlet superheat is maintained by an independent PID control loop adjusting the opening of the electronic expansion valve. However, this fixed speed change rate limit scheme has the following drawbacks:

[0004] (1) The fixed limit scheme implies a physically invalid assumption: that the refrigerant system's safe tolerance to changes in speed is a constant. In reality, the refrigerant circuit's tolerance to compressor speed changes varies greatly under different dynamic distribution states. When the evaporator is in a "liquid shortage" state, the system's tolerance to speed increases is extremely weak, while its tolerance to speed decreases is relatively strong; when the evaporator is in a "liquid accumulation" state, the situation is exactly the opposite. This state-dependent directional asymmetry cannot be expressed by a fixed constant.

[0005] (2) There is a contradiction between the fixed limit and the power following speed. To ensure the system remains safe under the most unfavorable refrigerant conditions, the fixed limit must be a sufficiently small conservative value. However, when the refrigerant condition is good, this conservative limit unnecessarily restricts the compressor speed change rate, causing the system to be unable to keep up with the changes in photovoltaic power in a timely manner, resulting in a large amount of curtailment or the need for large-capacity energy storage buffer. If the fixed limit is relaxed to improve the power following speed, the system will frequently touch the safety boundary when the refrigerant condition is unfavorable, leading to liquid slugging, exhaust overheating, and protective shutdown.

[0006] (3) There is a time scale mismatch of one to two orders of magnitude between photovoltaic power abrupt changes and refrigerant redistribution. Photovoltaic power can change dramatically within 2 to 5 seconds, while the mass redistribution process of refrigerant from one steady state to another usually takes 30 to 90 seconds. When faced with such frequent and large power changes, fixed limit schemes are either too conservative and lose their ability to follow the changes, or too aggressive and cause frequent shutdowns.

[0007] (4) In the prior art, the electronic expansion valve is driven by an independent superheat PID control loop. This loop does not receive information about changes in compressor speed commands, nor does it know the global state of refrigerant distribution. Under conditions of rapid fluctuations in photovoltaic power, this pure feedback loop faces the dual difficulties of response lag and interference from non-minimum phase characteristics, and cannot effectively coordinate with frequent changes in speed.

[0008] In summary, existing technologies suffer from the problem that fixed speed change rate limits cannot express the state correlation and directional asymmetry of the refrigerant system's safety tolerance, leading to a trade-off between safety and photovoltaic power following speed. Summary of the Invention

[0009] This invention provides a photovoltaic direct-drive heat pump control method and system that can change the compressor speed change rate constraint from a fixed constant to a dynamic function of the refrigerant's real-time state. When the refrigerant is in good condition, the constraint is automatically relaxed to accelerate power following, and when the refrigerant is in unfavorable condition, the constraint is automatically tightened to ensure safety. This solves the contradiction between safety and power following speed in the existing fixed speed change rate limit scheme.

[0010] In a first aspect, this invention provides a photovoltaic direct-drive heat pump control method, applied to a photovoltaic direct-drive heat pump system comprising a photovoltaic array, a frequency converter, a variable frequency compressor, an electronic expansion valve, an evaporator, a condenser, and a buffer energy storage module. The method includes: constructing an online refrigerant dynamic state observer based on a simplified dynamic model of the refrigerant system and refrigerant state data collected by sensors; estimating the refrigerant regional distribution state within the evaporator and condenser in real time; and outputting a refrigerant state vector. A refrigerant safe operating envelope, defined by the intersection of multiple refrigerant thermodynamic state constraints, is predefined. In each control decision cycle, based on the current refrigerant state vector output by the observer, the simplified dynamic model is used to perform forward simulation prediction of the refrigerant state evolution trajectory within a prediction time window under different candidate speed change rates. The maximum speed increase rate and the maximum speed decrease rate that prevent the trajectory from touching the safe operating envelope within the prediction time window are determined, forming a safe feasible region for the speed change rate at the current moment. The target speed change rate required for photovoltaic power to follow is clipped to the safe feasible region, generating a constrained compressor speed command. The difference between the actual power consumption corresponding to the constrained compressor speed command and the available photovoltaic power is absorbed or supplemented by the buffer energy storage module.

[0011] As described above, this invention, by observing the refrigerant distribution in the region online and calculating the safety envelope distance using forward simulation, transforms the constraint on the compressor speed change rate from a static constant into a dynamic function of the refrigerant state for the first time, achieving simultaneous protection of power following speed and refrigeration system safety. The core innovation of this technical solution lies in recognizing that the refrigerant system's tolerance to speed change rate is essentially a state-dependent, directionally asymmetric dynamic quantity, and transforming this physical understanding into a dynamic constraint mechanism that can be calculated and executed in real time in each control cycle, fundamentally eliminating the contradiction between safety and speed following in existing fixed-limit schemes.

[0012] Preferably, the online refrigerant dynamic state observer is constructed based on a simplified moving boundary model. The simplified moving boundary model divides the interior of the evaporator into two regions: a two-phase region and a superheated region, and the interior of the condenser into three regions: a superheated region, a two-phase region, and a subcooled region. The proportion of each region to the total length of the heat exchanger is used as a dynamic state variable. The observer employs an extended Kalman filter framework, using the simplified moving boundary model as the state prediction equation. The observation inputs are the temperature and pressure sensor measurements at the compressor's suction and discharge ports, as well as the temperature sensor measurements at the evaporator and condenser outlets. The extended Kalman filter framework includes an observation equation, which establishes a mapping function between sensor measurements and the state vector through refrigerant thermodynamic properties. The evaporation and condensation pressures are directly obtained from the corresponding suction and discharge pressures. The evaporator outlet temperature is obtained by summing the saturation temperature and superheat corresponding to the evaporation pressure. The discharge temperature is obtained from the suction state and pressure ratio through isentropic compression. The condenser outlet temperature is obtained by the difference between the saturation temperature corresponding to the condensation pressure and the supercooling corresponding to the subcooling zone length ratio. The online estimated state vector includes the evaporator two-phase region length ratio, the condenser superheated region length ratio, the condenser two-phase region length ratio, the condenser subcooled region length ratio, the evaporation pressure, the condensation pressure, and the evaporator outlet superheat.

[0013] As described above, this invention parameterizes the refrigerant distribution information inside the heat exchanger as a region length ratio using a simplified moving boundary model, and combines this with an extended Kalman filter framework to achieve online state estimation. This transforms the refrigerant region distribution state, which cannot be directly measured, into a real-time state quantity usable for control decisions. The observer model is only of order 7, with a single-step computation time of no more than 0.5ms on an embedded DSP with a main frequency of 200MHz, meeting the real-time requirement of a 10ms sampling period. It can acquire global state information of the refrigerant distribution without adding additional high-cost sensors.

[0014] Preferably, the dynamic equation for the length ratio of the two-phase region of the evaporator in the simplified moving boundary model is determined by the mass conservation equation, and its driving quantities include the difference between the compressor mass flow rate and the electronic expansion valve mass flow rate, as well as the density change caused by the change in evaporation pressure. The compressor mass flow rate is determined by the product of volumetric efficiency, suction density, displacement, and rotational speed, where the volumetric efficiency is a nonlinear function of rotational speed and pressure ratio. The electronic expansion valve mass flow rate is determined by the flow coefficient, the flow area corresponding to the valve opening, and the difference between the condensing pressure and the evaporating pressure.

[0015] As described above, this invention accurately establishes the dynamic equations of the region boundary based on the mass conservation equation, and fully considers the nonlinear characteristics of volumetric efficiency with changes in speed and pressure ratio, as well as the pressure difference dependence of the throttling process, in the mass flow sub-models of the compressor and electronic expansion valve. This enables forward simulation prediction to accurately capture the dynamic process of refrigerant redistribution caused by changes in speed, providing a reliable model basis for the accurate calculation of the feasible region.

[0016] Preferably, the refrigerant safe operation envelope is formed by the intersection of the following constraints: the evaporator outlet superheat is not lower than the lower limit of superheat to prevent wet vapor from entering the compressor and causing liquid slugging; the evaporator outlet superheat is not higher than the upper limit of superheat to prevent the exhaust temperature from exceeding the limit; the compressor exhaust temperature is not higher than the upper limit of exhaust temperature, corresponding to the temperature tolerance limit of the compressor material and lubricating oil; the ratio of the two-phase region length of the evaporator is not lower than the lower limit of the two-phase region length ratio to ensure that the evaporator has sufficient two-phase heat exchange area to maintain the refrigeration capacity; the ratio of the subcooled region length of the condenser is not lower than the lower limit of the subcooled region length ratio to ensure that there is liquid refrigerant supplied before the electronic expansion valve; the ratio of condensing pressure to evaporating pressure is not higher than the upper limit of the pressure ratio, corresponding to the mechanical strength and efficiency constraints of the compressor.

[0017] As described above, this invention formalizes the safe operating conditions of a refrigerant system into a safety envelope in the state space formed by the intersection of six constraints. This comprehensively covers multiple dimensions of safety requirements, including liquid slugging protection, exhaust over-temperature protection, heat exchange efficiency assurance, refrigerant supply assurance, and compressor mechanical safety, providing complete safety criteria for forward simulation prediction. These constraint values ​​can be calibrated and adjusted according to specific machine models and refrigerant types, making the solution widely applicable.

[0018] Preferably, the process of determining the maximum speed increase rate and the maximum speed decrease rate respectively includes:

[0019] The current refrigerant state vector output by the observer is taken as the initial state of the forward simulation, and the current compressor speed and the current electronic expansion valve opening are taken as the initial conditions.

[0020] For a given candidate speed change rate, it is assumed that the compressor operates at a constant change rate within the prediction time window. It is also assumed that the electronic expansion valve maintains its current opening degree in the previous control update cycle of the prediction window and then calculates the opening degree response according to the current feedforward and feedback control logic. The simplified dynamic model is used as the simulation engine to solve the state equation by numerical integration to obtain the refrigerant state evolution trajectory within the prediction time window.

[0021] Check whether the refrigerant state evolution trajectory always satisfies all the constraints defined by the safe operation envelope within the prediction time window. If all are satisfied, the candidate change rate is determined to be safe and feasible. If any constraint is violated at any time, it is determined to be infeasible.

[0022] In the direction of increasing rotational speed, the maximum positive candidate rate of change that satisfies the safety envelope constraint is determined by a bisection search and is taken as the maximum rate of increase in rotational speed. In the direction of decreasing rotational speed, the maximum negative candidate rate of change that satisfies the safety envelope constraint is determined by a bisection search and is taken as the maximum rate of decrease in rotational speed.

[0023] As described above, this invention creatively transforms the physical constraints of the refrigerant into explicit dynamic constraints on the rate of change of rotational speed by combining forward simulation prediction with bisection search to calculate the feasible region boundary in real time. Using a 7th-order simplified model, a 10-second prediction window, and a simulation step size of 50ms, the bisection search achieves an accuracy of 0.5Hz / s in approximately 8 to 10 iterations in both positive and negative directions. The overall decision time is controlled within 5ms, far less than a decision cycle of 100ms, fully meeting the requirements of embedded real-time control.

[0024] Preferably, the safe and feasible region has directional asymmetry: when the ratio of the two-phase region length of the evaporator estimated by the observer is lower than the steady-state optimal value at the current speed, the absolute value of the maximum speed increase rate is less than the absolute value of the maximum speed decrease rate; when the ratio of the two-phase region length of the evaporator is higher than the steady-state optimal value, the absolute value of the maximum speed decrease rate is less than the absolute value of the maximum speed increase rate; when the refrigerant distribution is close to the steady-state optimal distribution, the safe and feasible region has a large allowable range in both the upward and downward directions.

[0025] As described above, this invention discloses for the first time and utilizes the physical characteristic of "directional asymmetry" within the safe feasible region of the rate of change of rotational speed. Existing fixed-limit schemes impose the same restriction on both the increase and decrease of rotational speed, failing to recognize that the refrigerant system exhibits significantly different sensitivities in the increasing and decreasing directions under different charge levels. This invention precisely captures this asymmetry by independently calculating the upper bound of the rate of increase and the lower bound of the rate of decrease, enabling the control strategy to fully utilize the available safety margin in each direction.

[0026] Preferably, the control method further includes feedforward-feedback coordinated control of the electronic expansion valve: the feedforward component, based on the ratio and trend of the two-phase region length of the evaporator output by the observer and the constrained compressor speed command, calculates the target valve opening required by matching the mass flow rate of the electronic expansion valve to the mass flow rate demand of the compressor at the constrained speed through the refrigerant mass flow balance equation; the feedback component performs PID correction based on the evaporator outlet superheat deviation; and the final electronic expansion valve opening command is the sum of the feedforward component and the feedback component.

[0027] As described above, this invention accelerates the refrigerant redistribution process through a feedforward-feedback coordinated control strategy, enabling the feasible region to recover to a more favorable state as quickly as possible. Since the feedforward component already undertakes the main flow matching function, the PID controller only needs to handle small deviation corrections, and its gain can be reduced to suppress oscillations caused by non-minimum phase characteristics.

[0028] Preferably, the control method further includes online adaptive updating of the compressor volumetric efficiency parameters: detecting a step change in photovoltaic power as an excitation signal for system identification, recording the time series of rotational speed and suction / discharge pressure and temperature during the transition process, periodically updating the coefficients of the volumetric efficiency parameterization model with rotational speed and pressure ratio as independent variables using the recursive least squares method, and injecting the updated parameters into the refrigerant dynamic state observer and the model used in the forward simulation prediction.

[0029] As described above, the present invention transforms the "interference" of photovoltaic power fluctuation, which originally needed to be overcome, into the "excitation signal" required for system identification, and can achieve online parameter calibration without the need to apply additional test disturbances.

[0030] Preferably, the capacity of the buffer energy storage module is determined based on the product of the maximum transition time of the refrigerant from one steady state to another and the maximum power difference during that transition period, wherein the capacity is the energy reserve for 30 to 60 seconds corresponding to the rated power of the system; the buffer energy storage module is a supercapacitor module.

[0031] As described above, this invention redefines the function of buffer energy storage based on feasible region constraints, reducing the energy storage capacity requirement from the kilowatt-hour level of traditional solutions to the tens of watt-hour level. Because the feasible region constraint ensures that the refrigerant is always transitioning to a new steady state rather than deviating uncontrollably, the buffer energy storage only needs to support this limited transition process. Its capacity is determined by the refrigerant transition time rather than by the total energy fluctuation of the photovoltaic system, a difference that can reach two orders of magnitude. Taking a 5kW system as an example, the buffer energy storage only requires approximately 40Wh of supercapacitor, reducing costs by more than 80% and volume and weight by more than 75% compared to traditional 2-3kWh lithium battery solutions.

[0032] Secondly, this invention provides a photovoltaic direct-drive heat pump system based on the real-time thermodynamic state of the refrigerant to dynamically generate a safe and feasible domain for the compressor speed change rate. The system includes a photovoltaic array, a frequency converter, a variable frequency compressor, an electronic expansion valve, an evaporator, a condenser, and temperature and pressure sensors installed at the compressor suction and discharge ports, and temperature sensors installed at the evaporator outlet and condenser outlet. It also includes: a refrigerant dynamic state observer module, used to estimate the length ratio of the two-phase region within the evaporator and the length ratio of each region within the condenser in real time, based on a simplified moving boundary model and an extended Kalman filter algorithm, using the measurement data from the temperature and pressure sensors as input, and outputting a refrigerant state vector; and a speed change rate safe and feasible domain calculation module, used to calculate the safe and feasible domain for the compressor speed change rate in each control decision cycle based on the real-time thermodynamic state of the refrigerant. The refrigerant state vector output by the observer module is used to perform forward simulation prediction of the refrigerant state evolution trajectory under different candidate speed change rates using the simplified moving boundary model. A bisection search is used to determine the maximum speed increase rate and the maximum speed decrease rate that ensure the trajectory does not touch the predefined safe operating envelope of the refrigerant within the prediction time window, forming a directionally asymmetric speed change rate safe feasible region. A speed constraint trimming module is used to trim the target speed change rate generated by photovoltaic power following to the range of the safe feasible region, generating a constrained compressor speed command. A buffer energy storage module is used to absorb or supplement the difference between the actual power consumption corresponding to the constrained compressor speed command and the available photovoltaic power, with a capacity of 30 to 60 seconds of energy reserve corresponding to the system's rated power.

[0033] The above describes the beneficial effects of the present invention:

[0034] The core contribution of this invention lies in the first identification of two key physical characteristics of the refrigerant system's safe tolerance to the rate of change of compressor speed: "state correlation" and "directional asymmetry." This physical understanding is then transformed into a dynamic constraint mechanism that can be calculated and executed in real time in each control cycle through online observers and forward simulation prediction.

[0035] Compared with existing fixed limit schemes, the present invention achieves the following improvements:

[0036] First, by changing the constraint on the rate of change of rotational speed from an indifferent static constant to a state-dependent, direction-asymmetric dynamic function, the contradiction between safety and power following speed is fundamentally resolved.

[0037] Second, it is the first time that the directional asymmetry of the feasible region has been disclosed and utilized, enabling the control strategy to make full use of the available safety margin in every direction.

[0038] Third, the function of buffer energy storage has been redefined based on feasible domain constraints, reducing the energy storage capacity requirement from the kilowatt-hour level to the tens of watt-hour level.

[0039] Fourth, a feedforward-feedback coordinated control strategy is adopted for the electronic expansion valve to reduce the superheat fluctuation amplitude and improve the transition state COP.

[0040] Fifth, the natural fluctuations in photovoltaic power are used as an identification excitation to achieve online adaptive updates of compressor parameters, ensuring long-term operational accuracy. Attached Figure Description

[0041] Figure 1 The overall control signal flow block diagram of the photovoltaic direct-drive heat pump control method provided by the present invention.

[0042] Figure 2 This is a schematic diagram of the refrigerant safe operation envelope and feasible region provided by the present invention.

[0043] Figure 3 A timeline diagram comparing the fixed limit scheme for photovoltaic power drop scenarios provided by this invention with the scheme of this invention.

[0044] Figure 4 A timeline diagram comparing the fixed limit scheme provided by this invention with the scheme of this invention under the scenario of sudden increase in photovoltaic power.

[0045] Figure 5 The graph shows the relationship between the feasible domain width and the refrigerant state deviation provided by the present invention.

[0046] Figure 6 This is a flowchart of the binary search method for finding the boundary of the feasible region provided by the present invention.

[0047] Figure 7 This is a schematic diagram of the photovoltaic direct-drive heat pump control system provided by the present invention. Detailed Implementation

[0048] The following description and accompanying drawings fully illustrate specific embodiments of this application to enable those skilled in the art to practice them. The embodiments represent only possible variations. Individual components and functions are optional unless explicitly required, and the order of operations may vary. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. The scope of embodiments of this application includes the entire scope of the technical solutions of this application, as well as all available equivalents of the technical solutions. In this document, each embodiment may be referred to individually or collectively by the term "invention," which is merely for convenience and, if more than one invention is disclosed, is not intended to automatically limit the scope of the application to any single invention or inventive concept. Relational terms such as "first" and "second" are used herein only to distinguish one entity or operation from another entity or operation, without requiring or implying any actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed. The various embodiments in this document are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the structures, products, etc., disclosed in the embodiments, since they correspond to the disclosed parts, the descriptions are relatively simple; relevant details can be found in the method section.

[0049] Terminology Explanation

[0050] Before describing the technical solution in detail, the key terms used in this application will be explained first:

[0051] Photovoltaic direct-drive heat pump system: This refers to a heat pump system in which the output power of a photovoltaic array directly drives a variable frequency compressor via a DC / AC inverter, eliminating the need for a grid-connected inverter. In this system, the photovoltaic array and the inverter are connected via a DC bus. The MPPT algorithm tracks the current maximum power point of the photovoltaic array to obtain the available power, converts this power into the target compressor speed, and then the inverter outputs the corresponding frequency to drive the compressor.

[0052] Simplified moving boundary model: This refers to a simplified thermodynamic model that divides the interior of a heat exchanger (evaporator or condenser) into several regions based on the refrigerant phase (e.g., two-phase region, superheated region, subcooled region), using the boundary positions of each region (as a proportion of the total heat exchanger length) as dynamic state variables. The moving speed of the region boundaries is determined by the mass conservation equation and driven by the compressor mass flow rate and the electronic expansion valve mass flow rate. The total length of the heat exchanger in the model is used as a known geometric parameter in the dynamic equation calculations. Compared to the complete distributed parameter model, the moving boundary model reduces the partial differential equations to ordinary differential equations, significantly reducing the computational load and making it suitable for embedded real-time control.

[0053] Extended Kalman Filter (EKF): A recursive filtering algorithm used for state estimation of nonlinear systems. Because refrigerant system models contain nonlinear elements (such as nonlinearity in volumetric efficiency and square root operations in throttling models), standard Kalman filtering cannot be directly applied. EKF approximates the nonlinear system as a linear system by locally linearizing the nonlinear model at each time step (by calculating the Jacobian matrix), and then applies standard Kalman filtering to update it, thereby achieving online estimation of internal states that cannot be directly measured (such as the proportion of region length). Implementing EKF requires setting the process noise covariance matrix (reflecting model uncertainty), the observation noise covariance matrix (reflecting sensor measurement accuracy), and the initial state estimate and its covariance matrix.

[0054] The refrigerant safe operation envelope refers to the state space region enclosed by the intersection of multiple refrigerant thermodynamic state constraints (superheat upper and lower limits, exhaust temperature upper limit, region length ratio lower limit, pressure ratio upper limit, etc.). When the state vector of the refrigerant system is within this envelope, the system is considered to be in a safe operating state; when the state vector touches or crosses the envelope boundary, the system faces safety risks such as liquid slugging, exhaust overheating, and insufficient heat exchange capacity.

[0055] The feasible region for the rate of change of compressor speed refers to the allowable range of compressor speed change that ensures the refrigerant state does not touch any boundary of the safe operating envelope within the future prediction time window, under the current refrigerant state. This feasible region is a closed interval with the maximum safe rate of decrease as the lower bound and the maximum safe rate of increase as the upper bound. It has three key physical characteristics: directional asymmetry (the absolute values ​​of the upper bound of the rate of increase and the lower bound of the rate of decrease are usually not equal), state dependence (the width varies with the refrigerant state), and time-varying nature (updated in each decision cycle).

[0056] Forward simulation prediction refers to the process of using the current refrigerant state vector output by the observer as the initial state, assuming the compressor operates at a constant rate of change of speed, and using a simplified dynamic model to solve for the refrigerant state evolution trajectory over a future period. This process solves the state equations using numerical integration methods (such as the fourth-order Runge-Kutta method or explicit Euler method) to obtain the state vector value for each simulation time step within the prediction window. Forward simulation only involves numerical integration of ordinary differential equations and does not include matrix operations such as Kalman gain; therefore, its single-step computational cost is much less than that of an EKF observer.

[0057] Directional asymmetry refers to the physical characteristic where the constraint limits of the safe and feasible region are not equal in the directions of increasing and decreasing rotational speed. In the evaporator "liquid shortage" state (smaller two-phase region length ratio, higher superheat), the system is less tolerant of increasing rotational speed but more tolerant of decreasing speed; therefore, the absolute value of the upper bound of the rate of increase is much smaller than the absolute value of the lower bound of the rate of decrease. Conversely, in the evaporator "liquid accumulation" state (larger two-phase region length ratio, lower superheat), the situation is exactly the opposite. This asymmetry precisely reflects the different sensitivities of the refrigerant thermodynamic system to increasing and decreasing rotational speed.

[0058] Buffer energy storage module: This refers to a small-capacity energy storage device connected to the DC bus of a photovoltaic direct-drive heat pump system, used to absorb or supplement the difference between the actual compressor power consumption and the available photovoltaic power caused by feasible domain constraints. A supercapacitor module is preferred, and its capacity is determined by the product of the maximum refrigerant transition time and the maximum power difference, with a preferred value being 30 to 60 seconds of energy reserve corresponding to the system's rated power.

[0059] Feedforward-feedback coordinated control: In the control strategy of electronic expansion valve, the feedforward component actively calculates the target valve opening based on the refrigerant state estimated by the observer and the constrained speed command through the mass flow balance equation, while the feedback component is corrected based on the superheat deviation using a low-gain PID. The two are superimposed to form the final control command.

[0060] A volumetric efficiency parameterized model is a mathematical model that uses a polynomial function with speed and pressure ratio as independent variables to describe the variation of compressor volumetric efficiency with operating conditions. In this application, a second-order polynomial form containing five parameters to be identified is used. The coefficients of this model are periodically updated using actual operating data through a recursive least squares method to compensate for parameter drift caused by factors such as compressor aging.

[0061] Detailed technical solution

[0062] This invention provides a control method for a photovoltaic direct-drive heat pump. For example... Figure 1 As shown, Figure 1The overall control signal flow block diagram of the control method provided by this invention is shown below. The method provided by this invention is executed by a controller in a photovoltaic direct-drive heat pump system. This controller can be implemented in software and / or hardware, preferably as an embedded DSP controller. The method includes the following steps:

[0063] Step 101: Collect refrigerant state data and construct an online refrigerant dynamic state observer to estimate the refrigerant regional distribution state in the evaporator and condenser in real time.

[0064] In this step, the controller collects refrigerant status data using temperature and pressure sensors installed at the compressor suction and discharge ports, and temperature sensors installed at the evaporator and condenser outlets. The sensor signals are then input to the controller after analog-to-digital conversion.

[0065] Specifically, the measurements collected by the sensor include: compressor suction temperature (T) suc Inspiratory pressure (P) suc ), exhaust temperature (T) dis ), exhaust pressure (P) dis ), Evaporator outlet temperature (T) e,out ) and condenser outlet temperature (T c,out ).

[0066] The regional distribution of refrigerant within the evaporator and condenser cannot be directly measured by sensors and needs to be indirectly estimated using a state observer.

[0067] The observer is constructed based on a simplified moving boundary model, which divides the evaporator interior into two regions: a two-phase region and a superheated region, and the condenser interior into three regions: a superheated region, a two-phase region, and a subcooled region. The proportion of each region to the total length of the heat exchanger is used as a dynamic state variable. Specifically, the proportion of the evaporator's two-phase region length is denoted as l. e,tp Its physical meaning is the proportion of the two-phase region in which liquid and gaseous refrigerant coexist in the evaporator to the total length of the evaporator. This value directly reflects the refrigerant charge status in the evaporator.

[0068] The length ratio of the condenser superheated zone is denoted as l. c,sh The length ratio of the two-phase region of the condenser is denoted as l. c,tp The proportion of the subcooled zone length in the condenser is denoted as l. c,sc The total length of the evaporator is denoted as L. e The total length of the condenser is denoted as L. c Both are known constants determined based on the geometry of the heat exchanger.

[0069] The core of the simplified moving boundary model is that the moving speed of each region boundary is determined by the mass conservation equation. Taking the length ratio of the two-phase region of the evaporator as an example, its dynamic equation is:

[0070] ;

[0071] Where ρ e,tp ρ is the average density of the two-phase region of the evaporator. e,sh A is the average density of the superheated region. e L is the cross-sectional area of ​​the evaporator flow path. e The total length of the evaporator. For the mass flow rate of the electronic expansion valve, P is the compressor mass flow rate. e This is the evaporation pressure. The left side of the equation... The change in mass per unit time caused by the movement of the boundary of the two-phase region is expressed as the density difference between the two-phase region and the superheated region multiplied by the cross-sectional area, the total length of the evaporator, and the proportion of the boundary. Its dimension is kg / s. The first term on the right-hand side of the equation ( - The net mass inflow rate of the evaporator is the mass flow rate into the evaporator through the electronic expansion valve minus the mass flow rate sucked away by the compressor; both are in kg / s. (Second item) This reflects the effect of changes in refrigerant density in the two-phase region caused by changes in evaporation pressure on the law of mass conservation. (Item 3) This reflects the effect of refrigerant density changes in the superheated zone caused by variations in evaporation pressure. All terms in the above equation have dimensions in kg / s. The core physical meaning of this equation is that the moving speed of the evaporator two-phase zone boundary is jointly determined by the mass flow difference between the electronic expansion valve and the compressor, as well as the density change caused by the evaporation pressure variation. When the compressor speed changes due to variations in photovoltaic power, the compressor mass flow changes, directly driving the dynamic evolution of the two-phase zone length, and this evolution process has a significant time lag.

[0072] The compressor mass flow rate is determined by the rotational speed, pressure ratio, and suction conditions, and its model is as follows:

[0073] ;

[0074] Where η v Volumetric efficiency is the ratio of rotational speed N to pressure ratio r. p =P c / P e (P) c ρ is a nonlinear function of the condensation pressure. suc V is the inhalation density; d The displacement is used. Volumetric efficiency is not a constant but a nonlinear function that varies with speed and pressure ratio, meaning that the actual mass flow rate of the compressor is not a simple linear relationship with speed. In this application, volumetric efficiency is modeled using a second-order polynomial parameterization with speed and pressure ratio as independent variables:

[0075] It contains 5 parameters to be identified (a0) to (a4).

[0076] The mass flow rate of the electronic expansion valve is determined by the valve opening and the pressure difference between the high and low pressure zones. Its model is as follows:

[0077] ; among which (C d ) is the flow coefficient, (A) EEV (θ) represents the flow area corresponding to the valve opening (θ), (ρ) l ) represents the density of the liquid refrigerant at the condenser outlet.

[0078] The observer employs an extended Kalman filter framework, using the simplified moving boundary model described above as the state prediction equation, and sensor measurements as input to the observation equation. The online estimated state vector is:

[0079] ;

[0080] Where (SH) represents the evaporator outlet superheat. This state vector contains 7 components, and the model order is 7.

[0081] The observation equations of the extended Kalman filter establish the mapping relationship between sensor measurements and the state vector. The observation vector is obtained by transforming the sensor measurements through thermodynamic relationships and is defined as:

[0082] ;

[0083] Observation function The state vector is mapped to the observation vector, and the correspondence of its components is as follows. Inhalation pressure (P) suc ) and evaporation pressure (P) e The relationships between them are direct, that is... The difference lies solely in the slight pressure drop in the intake manifold. Exhaust pressure (P) dis ) and condensing pressure (P) c There is also a direct correspondence between them, that is... The evaporator outlet temperature is given by the relationship between saturation temperature and superheat, i.e. , of which (T sat (P e The saturation temperature (T) corresponds to the evaporation pressure and can be found in the refrigerant property table. Discharge temperature (T) dis The isentropic compression process, combined with isentropic efficiency correction, can be calculated from the intake state and pressure ratio. This function includes thermodynamic relationships such as the specific enthalpy of the refrigerant in the suction state, the specific enthalpy after isentropic compression, and the isentropic efficiency correction. The condenser outlet temperature is given by subtracting the subcooling from the saturation temperature corresponding to the condensing pressure, i.e. The ratio of subcooling (SC) to the length of the subcooled zone of the condenser (l) c,sc =x4) There exists a functional relationship, which is determined by the energy conservation equation of the supercooled region. In summary, the specific form of the observation function is:

[0084] .

[0085] Of the five observations mentioned above, the evaporation pressure (P) e ), condensing pressure (P) c The three state components (l, s ... e,tp ), the ratio of the length of the superheated zone of the condenser (l) c,sh ), the ratio of the length of the two-phase region of the condenser (l) c,tp ) and the ratio of the length of the subcooled zone of the condenser (l c,sc The lengths of the four regions cannot be directly measured and must be estimated using indirect observation information provided by exhaust temperature and condenser outlet temperature, combined with the dynamic constraint relationships of the moving boundary model.

[0086] The extended Kalman filter performs two steps in each sampling period: prediction and update. The prediction step uses a simplified dynamic model to estimate the state at the next time step and its covariance matrix based on the current input (speed and valve opening). The update step calculates the observation function. The state estimate is corrected by using the Jacobian matrix at the current state estimate, combined with the deviation between the sensor measurements and the model predictions (i.e., the innovation vector), and then by using the Kalman gain matrix.

[0087] In the EKF implementation, the diagonal elements of the process noise covariance matrix (Q) reflect the model uncertainty of each state variable—the process noise of the region length proportion state is set relatively large (preferred value is (10)). -4 ) to (10 -3 (on the order of magnitude), because the simplification of the moving boundary model introduces a large modeling error; the process noise of the pressure state is set to be small (preferred value is (10)). -2 ) to (10 -1 )Pa 2 The magnitude is low because the pressure dynamics are well captured by the model. The diagonal elements of the observation noise covariance matrix (R) are determined according to the sensor accuracy specifications; the preferred observation noise value for the temperature sensor (PT1000) is 0.25℃. 2 (Corresponding to ±0.5℃ accuracy), the observation noise of the pressure sensor is calculated based on the full-scale accuracy (preferably ±0.5%FS). Initial state estimation ( After the system is powered on, the initial estimation error covariance matrix (P0) is obtained by combining the sensor steady-state readings with the refrigerant property relationship and steady-state model calculation. The initial estimation error covariance matrix (P0) is set as a diagonal matrix, and its diagonal elements reflect the uncertainty of the initial steady-state estimation.

[0088] This step corresponds to the technical feature in the first aspect of the technical solution: "Based on a simplified dynamic model of the refrigerant system and refrigerant state data collected by sensors, an online refrigerant dynamic state observer is constructed to estimate the refrigerant regional distribution state in the evaporator and condenser in real time and output a refrigerant state vector." Through this step, the refrigerant regional distribution information, which cannot be directly measured, is transformed into a real-time state vector that can be used for control decisions. Furthermore, the EKF observer's single-step computation time (including matrix operations such as Jacobian matrix calculation and Kalman gain update) on a 200MHz DSP does not exceed 0.5ms, meeting the embedded real-time control requirements with a 10ms sampling period.

[0089] Step 102: Predefine the refrigerant safe operating envelope.

[0090] In this step, the controller pre-stores the definition parameters of the refrigerant safe operating envelope. The safe operating envelope is the state space region enclosed by the intersection of the following constraints: the evaporator outlet superheat is not lower than the lower superheat limit (SH). min (Preferred value is 2℃), its physical meaning is to prevent wet steam from entering the compressor and causing liquid slugging; the superheat at the evaporator outlet should not exceed the upper limit of superheat (SH). max (Preferred value is 12℃), its physical meaning is to prevent the exhaust temperature from exceeding the limit due to excessive heat;

[0091] The compressor discharge temperature shall not exceed the upper limit of discharge temperature (T). dis,max (Preferred value is 115℃), corresponding to the temperature tolerance limit of the compressor material and lubricating oil; the length ratio of the two-phase region of the evaporator is not less than the lower limit of the length ratio of the two-phase region (l) e,tp,min (Preferred value 0.3), its physical meaning is to ensure that the evaporator has sufficient two-phase heat exchange area to maintain the refrigeration capacity; the proportion of the subcooled zone length of the condenser is not less than the lower limit of the subcooled zone length proportion (l) c,sc,min (Preferred value 0.05), its physical meaning is to ensure that there is liquid refrigerant supplied before the electronic expansion valve; the ratio of condensing pressure to evaporating pressure is not higher than the upper limit of the pressure ratio (r). p,max (Preferred value 4.5), corresponding to the mechanical strength and efficiency constraints of the compressor.

[0092] like Figure 2 As shown, Figure 2This is a schematic diagram of the refrigerant safe operation envelope and feasible region provided by the present invention. The horizontal axis is the length ratio of the two-phase region of the evaporator, the vertical axis is the superheat of the evaporator outlet, and the safety boundary is enclosed by the aforementioned constraint lines. Three preferred operating points are marked within the safety region, corresponding to the three states of "optimal refrigerant distribution", "evaporator liquid shortage", and "evaporator liquid accumulation".

[0093] The aforementioned constraint values ​​can be calibrated and adjusted according to specific machine models and refrigerant types. Essentially, they formalize the safe operating conditions of the refrigerant system as boundaries in the state space. This step corresponds to the technical feature in the first aspect of the technical solution: "pre-defining the refrigerant safe operating envelope formed by the intersection of multiple refrigerant thermodynamic state constraints."

[0094] Step 103: Perform forward simulation prediction based on the current refrigerant state vector in each control decision cycle to determine the maximum speed increase rate and the maximum speed decrease rate, and form a safe and feasible region for the speed change rate.

[0095] In this step, the controller makes decisions in each control decision cycle (T). c (Preferred value: 100ms), based on the current refrigerant state vector output by the observer in step 101 ( ), and perform the following forward simulation prediction and binary search process.

[0096] First, determine the initial conditions for prediction: take the current refrigerant state vector output by the observer ( As the initial state for the forward simulation, the current compressor speed (N) is taken. k As the initial rotational speed, the current electronic expansion valve opening (θ) is taken. k This is used as the initial valve opening.

[0097] Then, for the given candidate rate of change of rotational speed ( Perform forward simulation: Assume the compressor operates within the prediction time window (T). p (Preferred value: 10 seconds) The rotational speed is maintained at this constant rate of change, meaning the rotational speed at any time (t) within the prediction window is... Meanwhile, it is assumed that the electronic expansion valve maintains its current opening (θ) for the first 200ms of the prediction window (i.e., one electronic expansion valve control update cycle). k The PID control parameters remain unchanged. Then, within the remaining time of the prediction window, the electronic expansion valve opening response is calculated based on the current PID control parameters and feedforward logic; that is, the prediction includes the expected regulating behavior of the electronic expansion valve. A simplified moving boundary model is used as the simulation engine, and the state equations are solved using either the fourth-order Runge-Kutta method or the explicit Euler method (preferably with a step size of 50ms) to obtain the refrigerant state trajectory within the prediction window. .

[0098] Next, the safety envelope constraints are checked: it is checked whether the predicted state trajectory always satisfies all the safety envelope constraints defined in step 102 throughout the entire prediction window. If all are satisfied, the candidate rate of change is determined to be "safe and feasible"; if any constraint is violated at any time, it is determined to be "infeasible".

[0099] Finally, the feasible region boundary is determined using a binary search method. In the direction of increasing speed (positive direction), the initial upper bound is set to the maximum rate of change allowed by the inverter hardware. The initial lower bound is 0. Forward simulation and safety checks are repeatedly performed by repeatedly taking the midpoint. The search interval is narrowed based on the results until the accuracy meets the requirements (preferably 0.5 Hz / s). The maximum safe rate of ascent is then determined. In the direction of decreasing speed (negative direction), the maximum safe rate of decrease is determined using the same dichotomy method. (The maximum value in the absolute sense). For example... Figure 6 As shown, Figure 6 This is a flowchart of the binary search method for finding the boundary of the feasible region provided by the present invention.

[0100] Through the above process, the feasible region at the current moment is obtained:

[0101] ;

[0102] This feasible region has three key physical properties.

[0103] The first is directional asymmetry: when the evaporator charge is insufficient, i.e., l e,tp When the evaporator is under the condition of low speed and high SH, the absolute value of the maximum speed increase rate is much smaller than the absolute value of the maximum speed decrease rate, that is, the allowable increase rate is much smaller than the allowable decrease rate; when the evaporator is overcharged, the situation is exactly the opposite.

[0104] The second is state dependence: the feasible region is widest when the refrigerant distribution is close to the steady-state optimal distribution at the current speed, and the feasible region becomes narrower the further it deviates from the optimal distribution.

[0105] Third is time-varying nature: after a step change in photovoltaic power, the feasible region usually narrows first and then gradually recovers.

[0106] like Figure 5 As shown, Figure 5 The graph showing the relationship between the width of the feasible region and the refrigerant state deviation provided by the present invention illustrates the narrowing of the feasible region in the corresponding direction as the deviation increases.

[0107] Regarding computational efficiency, using a 7th-order simplified model, a 10-second prediction window, and a 50ms simulation step, a single forward simulation requires only 200 steps. The bisection search requires approximately 8 to 10 iterations in each of the positive and negative directions, totaling approximately 16 to 20 forward simulations, or 3200 to 4000 model computation steps. It should be noted that each step of the forward simulation only involves the numerical integration of the simplified moving boundary model's ordinary differential equations, and does not include matrix operations such as the Jacobian matrix calculation and Kalman gain matrix update required by the EKF observer. Therefore, its single-step computation is much less than that of the EKF observer in step 101 (the latter is approximately 0.5ms). On a 200MHz DSP, the model's single-step computation time for the forward simulation is approximately 0.5μs, totaling approximately 1.6 to 2.0ms for 3200 to 4000 steps. Including data preparation and communication overhead, the overall decision time is controlled within 5ms, far less than the 100ms decision cycle.

[0108] This step corresponds to the technical features of forward simulation prediction and feasible region calculation in the first aspect of the technical solution, and also reflects the technical features of forward simulation specific steps and directional asymmetry in the preferred solution.

[0109] Step 104: Trim the target speed change rate to a safe and feasible range to generate a constrained compressor speed command.

[0110] In this step, the controller first receives the current available power output from the MPPT algorithm on the photovoltaic side. Based on the compressor power-speed characteristics under the current operating conditions, it is converted into the target speed. Then calculate the required rate of change of rotational speed. Next, the required rate of change is compared with the current feasible region obtained in step 103, and a pruning constraint is applied:

[0111] ;

[0112] The final actual speed command is The meaning of the trimming operation is: when the rate of change of speed required by the change in photovoltaic power is within the feasible range, the compressor can follow the change in photovoltaic power at full speed without any following delay; when the required rate of change exceeds the feasible range, the actual rate of change is limited to the safety boundary, and the compressor approaches the target speed at the fastest rate that the refrigerant system can currently withstand.

[0113] This step corresponds to the technical feature in the first aspect of the technical solution: "to trim the photovoltaic power to the required target speed change rate to the safe and feasible range, and generate a constrained compressor speed command".

[0114] Step 105: Absorption of power difference in buffer storage and coordinated control of feedforward and feedback of electronic expansion valve.

[0115] In this step, when the feasible region constraint causes the actual speed to deviate from the target speed, the actual power consumed by the compressor (P) is... comp (N cmd )) and available photovoltaic power ( A difference arises between the two. This difference is absorbed or supplemented by a small-capacity buffer energy storage module:

[0116] ;

[0117] When (P) buf When P > 0), the buffer energy storage charge absorbs excess power; when P buf When the power density is <0), the buffer energy storage discharges to supplement insufficient power. The buffer energy storage module is preferably a supercapacitor module, and its capacity is determined by the product of the maximum transition time of the refrigerant from one steady state to another and the maximum power difference during that transition period. The preferred value is 30 to 60 seconds of energy reserve corresponding to the system's rated power. The feasibility of a small-capacity buffer energy storage is fundamentally due to the feasible region constraint mechanism, which actively manages the time window requirements of the refrigerant transition process. The feasible region ensures that the rotational speed only varies within the range that the refrigerant can withstand, allowing the refrigerant to continuously transition to the new steady state. The buffer energy storage only needs to maintain the power difference during the transition. Once the refrigerant has completed redistribution and the feasible region returns to a more relaxed state, the rotational speed can fully follow the photovoltaic power, and the buffer energy storage release capacity returns to standby. Therefore, the capacity of the buffer energy storage is determined by the refrigerant transition time, not by the total energy fluctuation of the photovoltaic system.

[0118] Simultaneously, to accelerate the refrigerant redistribution process and restore the feasible region to a more favorable state as quickly as possible, this step employs a feedforward-feedback coordinated control strategy for the electronic expansion valve. The feedforward component, based on the observer's estimation of the current evaporator two-phase region length ratio and its changing trend, along with the target rotational speed trajectory, is used to back-calculate the valve opening required to match the evaporator inlet flow rate with the current compressor extraction rate using the refrigerant mass flow balance equation.

[0119] ;

[0120] Its physical meaning is: given the target speed command and the current condensing and evaporating pressures, the mass flow rate requirement of the compressor at that speed is calculated through the compressor mass flow rate model. Then, the flow area function in the electronic expansion valve throttling model is solved to obtain the valve opening required to make the mass flow rate of the electronic expansion valve equal to the mass flow rate of the compressor.

[0121] The feedback component is used for PID correction based on the evaporator outlet superheat deviation. However, since the feedforward component already handles the main flow matching function, the PID only needs to correct small deviations, and its gain can be reduced to suppress oscillations caused by non-minimum phase characteristics. The final electronic expansion valve opening command is the sum of the feedforward and feedback components.

[0122] In addition, this step also includes online adaptive updates of the compressor volumetric efficiency parameters.

[0123] The accuracy of feasible region calculation depends on the accuracy of the compressor mass flow model in the observer, while volumetric efficiency drifts slowly with aging and operating conditions. This invention utilizes the natural fluctuations in photovoltaic power as the excitation signal for system identification: whenever a step change in photovoltaic power is detected (exceeding 10% of the rated power), the rotational speed and intake / exhaust pressure / temperature time series during the transition are recorded, and the five coefficients of the volumetric efficiency parameterized model are updated using the recursive least squares method. The parameter vector is updated every 24 hours and injected into the observer and forward simulation model to ensure the long-term accuracy of the feasible region calculation.

[0124] This step corresponds to the technical feature of buffer energy storage power difference absorption in the first aspect of the technical solution, and also reflects the technical features of feedforward-feedback coordinated control of electronic expansion valve, online adaptive updating of compressor volumetric efficiency, and buffer energy storage capacity design in the preferred solution.

[0125] Example 1

[0126] The technical solution of the present invention will be described in detail below through a specific embodiment of a 5kW residential photovoltaic direct-drive air source heat pump system.

[0127] The system hardware configuration is as follows: Photovoltaic array rated power 5.5kWp, DC bus voltage range 280 to 420V. Scroll inverter compressor with a displacement of 25cc / rev, speed range 1800 to 7200rpm (corresponding to 30 to 120Hz), refrigerant R410A, charge 2.8kg. Electronic expansion valve driven by a stepper motor, 480-step full stroke. Evaporator is finned tube type, heat exchange area 8.5m². 2 Total length (L) e =1.2m, the condenser is a finned tube type with a heat exchange area of ​​12m². 2 Total length (L) c =1.6m. The buffer energy storage module is a supercapacitor module with a rated voltage of 400V, a capacitance of 1.8F, and a stored energy of... .

[0128] The sensor configuration is as follows: one PT1000 temperature sensor and one piezoresistive pressure sensor at the compressor suction port; one PT1000 temperature sensor and one piezoresistive pressure sensor at the compressor discharge port; one PT1000 temperature sensor at the evaporator outlet; and one PT1000 temperature sensor at the condenser outlet. The total additional cost of the sensors is approximately 120 yuan.

[0129] The controller uses a TMS320F28379D dual-core DSP. Core A runs the MPPT algorithm, DC bus voltage control, and buffer energy storage charging and discharging management, with an interrupt cycle of 1ms. Core B runs the refrigerant state observer (sampling cycle 10ms), feasible region calculation and speed constraint trimming (decision cycle 100ms), and electronic expansion valve feedforward-feedback control (execution cycle 200ms).

[0130] The buffer energy storage capacity design is based on the following: the maximum refrigerant transition time is approximately 60 seconds, during which the maximum power difference is approximately 2.5kW (50% of the system's rated power). The theoretical maximum buffer energy requirement is... Considering that the power difference decreases gradually as the refrigerant is redistributed during the actual transition process (i.e., the power difference is largest in the early stage of the transition and gradually decreases to zero in the later stage), the actual cumulative buffer energy is lower than the theoretical maximum value obtained by multiplying the peak power difference by the total transition time. Therefore, a supercapacitor module with a rated voltage of 400V and a capacitance of 1.8F (energy storage of 40Wh) is selected to meet the buffer requirements of the system under various optimal operating conditions.

[0131] Observer parameter calibration: Before leaving the factory, the inlet and outlet state parameters of the evaporator and condenser were measured at five steady-state operating points (30Hz, 50Hz, 70Hz, 90Hz, and 110Hz) in the enthalpy difference laboratory to calibrate the heat transfer coefficient and refrigerant density parameters in the moving boundary model. The volumetric efficiency parameterization model obtained its initial coefficients by fitting the suction and discharge data at the above operating points. The initial values ​​of the EKF process noise covariance matrix (Q) and observation noise covariance matrix (R) were set based on the model identification residuals and sensor accuracy specifications, and were fine-tuned and optimized during the commissioning phase.

[0132] Scenario 1: A sudden drop in photovoltaic power. For example... Figure 3 As shown, Figure 3 This invention provides a time-series diagram comparing a sudden drop in photovoltaic power. Simulated cloud cover causes the photovoltaic power to drop from 4.5kW to 1.2kW within 3 seconds. Under a fixed-limit scheme (limit set at 5Hz / s), the compressor speed decreases at a fixed rate, but the high-charge refrigerant accumulated in the evaporator cannot be discharged in time. The superheat drops from 5°C to -2°C within 8 seconds, entering a wet steam state, triggering liquid slugging protection and shutdown. Recovery requires 3 to 5 minutes.

[0133] The present invention's solution is as follows: When a sudden drop occurs, the observer detects that the ratio of the two-phase region length of the evaporator is 0.65 (too high). The calculated feasible region results are a maximum drop rate of -2.1 Hz / s (the drop rate is tightened) and a maximum rise rate of 8.3 Hz / s (the rise rate is loose), which intuitively reflects the directional asymmetry of the feasible region. The controller limits the actual drop rate to -2.1 Hz / s, and the buffer energy storage discharges to compensate for the power difference for about 8 seconds (discharge energy of about 7Wh). The electronic expansion valve feedforward control synchronously reduces the opening to accelerate the evaporator discharge, and the superheat is always maintained above 3°C. After about 15 seconds, the refrigerant redistributes completely, the feasible region returns to a loose state, the rotation speed completely follows the photovoltaic power, the buffer energy storage returns to full capacity standby state, and there is no downtime throughout the entire process.

[0134] Scenario 2: A sudden increase in photovoltaic power. For example... Figure 4 As shown, Figure 4 This invention provides a time-series diagram comparing scenarios of sudden increases in photovoltaic power. The photovoltaic power increases from 1.5kW to 5kW within 3 seconds. Under the fixed limit scheme, the compressor increases at a fixed slope, the evaporator charge cannot be replenished in time, the superheat rises from 5°C to 18°C ​​within 6 seconds, and the exhaust temperature reaches 120°C, triggering over-temperature protection shutdown.

[0135] In this invention, the observer detects a two-phase region length ratio of 0.42 in the evaporator (relatively low). The feasible region calculation results show a maximum rise rate of 1.8 Hz / s (the rise rate is tightened) and a maximum fall rate of -7.5 Hz / s (the fall rate is loose), also reflecting directional asymmetry. The controller limits the rise rate to 1.8 Hz / s, and excess power is absorbed by the buffer energy storage for approximately 12 seconds (charging energy of approximately 15 Wh). The electronic expansion valve feedforward control increases the opening to accelerate evaporator charging. The superheat peak is controlled below 9°C, and the exhaust temperature does not exceed 105°C. After approximately 20 seconds, the refrigerant redistribution is completed, and the rotational speed completely follows the photovoltaic power.

[0136] Scenario 3: Continuous Fluctuations. Simulating intermittent cloud cover, the photovoltaic power randomly fluctuates between 1kW and 5kW, with a fluctuation coefficient of approximately 0.4. In a 30-minute continuous operation test, the fixed-limit scheme experienced 3 liquid hammer protection shutdowns and 2 exhaust over-temperature shutdowns, with a total downtime of approximately 12 minutes, resulting in only 60% of the effective operating time.

[0137] The solution of this invention operates without downtime in the same scenario, with 100% effective operating time. In terms of cumulative heat generation over 30 minutes, the solution of this invention is about 42% higher than the fixed limit solution, of which about 20% comes from eliminating downtime losses and about 22% comes from improving the transition state COP.

[0138] Example 2

[0139] The technical advantages of the present invention are further illustrated below through comparative examples of the synergistic relationship between buffer energy storage capacity and feasible region.

[0140] Based on the 5kW system in Example 1, the differences in energy storage requirements between the two schemes are compared and analyzed.

[0141] Option 1 is a high-capacity energy storage solution without feasible region constraints: instead of using the feasible region constraint mechanism of this invention, a large-capacity battery energy storage is configured to completely smooth photovoltaic fluctuations. To reduce the power fluctuation amplitude to an acceptable rate of change for the compressor, under the condition of a fluctuation coefficient of 0.4, the required energy storage capacity is approximately 2 to 3 kWh (lithium battery), the cost is approximately 3,000 to 4,000 yuan, the volume is approximately 10L, and the weight is approximately 15kg.

[0142] Option 2 is a feasible region constraint plus small capacity buffer energy storage scheme, which is the scheme of this invention: After using the feasible region constraint mechanism, the buffer energy storage only needs to provide buffering during the short-term power difference caused by the feasible region constraint. The capacity is 40Wh (supercapacitor, 1.8F / 400V), the cost is about 800 yuan, the volume is about 2.5L, and the weight is about 3.5kg.

[0143] Both schemes perform similarly in terms of refrigeration system safety and heat generation, but the scheme of this invention reduces energy storage costs by about 80%, volume by about 75%, and weight by about 77%. This comparison fully illustrates the redefinition of buffer energy storage function by the feasible domain constraint mechanism—the energy storage capacity is determined by the refrigerant transition time rather than by the total energy fluctuation of photovoltaics, and the difference between the two can be two orders of magnitude.

[0144] like Figure 7 As shown, the present invention also provides a photovoltaic direct-drive heat pump system based on the real-time thermodynamic state of the refrigerant to dynamically generate a safe and feasible domain for the compressor speed change rate. This system includes a photovoltaic array, a frequency converter, a variable frequency compressor, an electronic expansion valve, an evaporator, a condenser, and temperature and pressure sensors installed at the compressor suction and discharge ports, and temperature sensors installed at the evaporator outlet and condenser outlet. It also includes:

[0145] The refrigerant dynamic state observer module is used to estimate the length ratio of the two-phase region in the evaporator and the length ratio of each region in the condenser in real time based on the simplified moving boundary model and the extended Kalman filter algorithm, taking the measurement data of the temperature sensor and pressure sensor as input, and outputting the refrigerant state vector.

[0146] The speed change rate safe feasible region calculation module is used to perform forward simulation prediction of the refrigerant state evolution trajectory under different candidate speed change rates based on the refrigerant state vector output by the observer module in each control decision cycle, using the simplified moving boundary model. The maximum speed increase rate and the maximum speed decrease rate are determined by bisection search to ensure that the trajectory does not touch the predefined refrigerant safe operation envelope within the prediction time window, thus forming a directionally asymmetric speed change rate safe feasible region.

[0147] The speed constraint trimming module is used to trim the target speed change rate generated by photovoltaic power tracking to the safe and feasible range, and generate a constrained compressor speed command.

[0148] A buffer energy storage module is used to absorb or supplement the difference between the actual power consumption corresponding to the constrained compressor speed command and the available photovoltaic power. Its capacity is 30 to 60 seconds of energy reserve corresponding to the system's rated power, and it is preferably a supercapacitor module.

[0149] The specific functions and implementation methods of the above modules correspond to the corresponding steps in the aforementioned method embodiments, and will not be repeated here.

[0150] This invention also provides a photovoltaic direct-drive heat pump control device, including a processor and a memory. The memory stores a computer program and transmits the computer program to the processor. The processor executes the photovoltaic direct-drive heat pump control method as described in the foregoing method embodiments according to the instructions in the computer program. In specific implementations, the processor is preferably a TMS320F28379D dual-core DSP or an embedded processor with equivalent performance. Core A runs the MPPT algorithm, DC bus voltage control, and buffer energy storage charging and discharging management, while core B runs the refrigerant state observer, feasible region calculation and speed constraint trimming, and electronic expansion valve feedforward-feedback control.

[0151] The present invention also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a computer processor, are used to perform the photovoltaic direct-drive heat pump control method as described in the foregoing method embodiments. The computer-readable storage medium may be an on-chip flash memory of an embedded DSP, an external EEPROM, or other non-volatile memory.

Claims

1. A photovoltaic direct-drive heat pump control method, applied to a photovoltaic direct-drive heat pump system comprising a photovoltaic array, a frequency converter, a variable frequency compressor, an electronic expansion valve, an evaporator, a condenser, and a buffer energy storage module, characterized in that, include: Based on a simplified dynamic model of the refrigerant system and refrigerant state data collected by sensors, an online refrigerant dynamic state observer is constructed to estimate the refrigerant distribution state in the evaporator and condenser in real time and output the refrigerant state vector. A predefined safe operating envelope for refrigerants is formed by the intersection of multiple thermodynamic state constraints for refrigerants; In each control decision cycle, based on the current refrigerant state vector output by the observer, the simplified dynamic model is used to perform forward simulation prediction of the evolution trajectory of the refrigerant state under different candidate speed change rates within the prediction time window, and the maximum speed increase rate and the maximum speed decrease rate that make the trajectory not touch the safe operation envelope within the prediction time window are determined respectively, forming the safe and feasible region of the speed change rate at the current moment. The photovoltaic power is tailored to the required target speed change rate within the safe and feasible range to generate a constrained compressor speed command; The difference between the actual power consumption corresponding to the constrained compressor speed command and the available photovoltaic power is absorbed or supplemented by the buffer energy storage module.

2. The control method according to claim 1, characterized in that, The online refrigerant dynamic state observer is constructed based on a simplified moving boundary model. The simplified moving boundary model divides the interior of the evaporator into two regions: a two-phase region and a superheated region, and the interior of the condenser into three regions: a superheated region, a two-phase region, and a subcooled region. The proportion of each region to the total length of the heat exchanger is used as a dynamic state variable. The observer employs an extended Kalman filter framework, uses the simplified moving boundary model as the state prediction equation, and takes the temperature and pressure sensor measurements at the compressor intake and exhaust ports, as well as the temperature sensor measurements at the evaporator outlet and condenser outlet, as the observation inputs. The extended Kalman filter framework includes an observation equation, which establishes a mapping function between sensor measurements and state vectors through the thermodynamic properties of the refrigerant. The evaporation pressure and condensation pressure are directly obtained from the corresponding suction pressure and discharge pressure. The evaporator outlet temperature is obtained from the sum of the saturation temperature and superheat corresponding to the evaporation pressure. The discharge temperature is obtained from the suction state and pressure ratio through isentropic compression. The condenser outlet temperature is obtained from the difference between the saturation temperature corresponding to the condensation pressure and the supercooling corresponding to the subcooling zone length ratio. The online estimated state vector includes the proportion of the two-phase region length of the evaporator, the proportion of the superheated region length of the condenser, the proportion of the two-phase region length of the condenser, the proportion of the subcooled region length of the condenser, the evaporation pressure, the condensation pressure, and the evaporator outlet superheat.

3. The control method according to claim 2, characterized in that, In the simplified moving boundary model, the dynamic equation for the length ratio of the two-phase region of the evaporator is determined by the mass conservation equation. In the dynamic equation, the total length of the evaporator, as a known geometric parameter, together with the flow cross-sectional area, constitutes a volume factor for calculation. Its driving quantities include the difference between the compressor mass flow rate and the electronic expansion valve mass flow rate, as well as the refrigerant density changes in the two-phase region and the superheated region caused by the change in evaporation pressure. The compressor mass flow rate is determined by the product of volumetric efficiency, suction density, displacement, and speed. The volumetric efficiency is a nonlinear function of speed and pressure ratio. The electronic expansion valve mass flow rate is determined by the flow coefficient, the flow area corresponding to the valve opening, and the difference between the condensing pressure and the evaporating pressure.

4. The control method according to claim 1, characterized in that, The safe operating envelope of the refrigerant is defined by the intersection of the following constraints: The superheat at the evaporator outlet shall not be lower than the lower limit of superheat. The superheat at the evaporator outlet shall not exceed the upper limit of superheat. The compressor discharge temperature shall not exceed the upper limit of the discharge temperature. The length ratio of the two-phase region of the evaporator shall not be less than the lower limit of the length ratio of the two-phase region. The proportion of the subcooled zone length in the condenser shall not be less than the lower limit of the subcooled zone length proportion. The ratio of condensing pressure to evaporating pressure shall not exceed the upper limit of the pressure ratio.

5. The control method according to claim 1, characterized in that, The process of determining the maximum speed increase rate and the maximum speed decrease rate includes: The current refrigerant state vector output by the observer is taken as the initial state of the forward simulation, and the current compressor speed and the current electronic expansion valve opening are taken as the initial conditions. For a given candidate speed change rate, it is assumed that the compressor operates at a constant change rate within the prediction time window. It is also assumed that the electronic expansion valve maintains its current opening degree in the previous control update cycle of the prediction window and then calculates the opening degree response according to the current feedforward and feedback control logic. The simplified dynamic model is used as the simulation engine to solve the state equation by numerical integration to obtain the refrigerant state evolution trajectory within the prediction time window. Check whether the refrigerant state evolution trajectory always satisfies all the constraints defined by the safe operation envelope within the prediction time window. If all are satisfied, the candidate change rate is determined to be safe and feasible. If any constraint is violated at any time, it is determined to be infeasible. In the direction of increasing rotational speed, the maximum positive candidate rate of change that satisfies the safety envelope constraint is determined by a bisection search and is taken as the maximum rate of increase in rotational speed. In the direction of decreasing rotational speed, the maximum negative candidate rate of change that satisfies the safety envelope constraint is determined by a bisection search and is taken as the maximum rate of decrease in rotational speed.

6. The control method according to claim 1, characterized in that, The safe and feasible region exhibits directional asymmetry: when the ratio of the two-phase region length of the evaporator estimated by the observer is lower than the steady-state optimum at the current speed, the absolute value of the maximum speed increase rate is less than the absolute value of the maximum speed decrease rate; when the ratio of the two-phase region length of the evaporator is higher than the steady-state optimum, the absolute value of the maximum speed decrease rate is less than the absolute value of the maximum speed increase rate; when the refrigerant distribution is close to the steady-state optimum distribution, the safe and feasible region has a large allowable range in both the upward and downward directions.

7. The control method according to claim 1, characterized in that, It also includes feedforward-feedback coordinated control of the electronic expansion valve: the feedforward component is based on the ratio of the two-phase region length of the evaporator and its changing trend output by the observer and the constrained compressor speed command, and calculates the target valve opening required by the electronic expansion valve to match the mass flow rate of the compressor under the constrained speed by back-calculating the refrigerant mass flow balance equation. The feedback component is corrected by PID based on the evaporator outlet superheat deviation; the final electronic expansion valve opening command is the sum of the feedforward component and the feedback component.

8. The control method according to claim 1, characterized in that, It also includes online adaptive updating of compressor volumetric efficiency parameters: detecting step changes in photovoltaic power as excitation signals for system identification, recording the time series of speed and suction / discharge pressure and temperature during the transition process, periodically updating the coefficients of the volumetric efficiency parameterization model with speed and pressure ratio as independent variables using recursive least squares method, and injecting the updated parameters into the refrigerant dynamic state observer and the model used by the forward simulation prediction.

9. The control method according to claim 1, characterized in that, The capacity of the buffer energy storage module is determined based on the product of the maximum transition time of the refrigerant from one steady state to another and the maximum power difference during that transition period. The capacity is the energy reserve for 30 to 60 seconds corresponding to the rated power of the system. The buffer energy storage module is a supercapacitor module.

10. A photovoltaic direct-drive heat pump system based on dynamically generating a safe and feasible domain for the compressor speed change rate according to the real-time thermodynamic state of the refrigerant, comprising a photovoltaic array, a frequency converter, a variable frequency compressor, an electronic expansion valve, an evaporator, a condenser, and temperature and pressure sensors installed at the compressor suction port and exhaust port, and temperature sensors installed at the evaporator outlet and condenser outlet, characterized in that, Also includes: The refrigerant dynamic state observer module is used to establish the thermodynamic mapping relationship between sensor measurements and state vectors through observation equations based on the simplified moving boundary model and extended Kalman filter algorithm, taking the measurement data of the temperature sensor and pressure sensor as input, and estimating the length ratio of the two-phase region in the evaporator and the length ratio of each region in the condenser in real time, and outputting the refrigerant state vector. The speed change rate safe feasible region calculation module is used to perform forward simulation prediction of the refrigerant state evolution trajectory under different candidate speed change rates based on the refrigerant state vector output by the observer module in each control decision cycle, using the simplified moving boundary model. The maximum speed increase rate and the maximum speed decrease rate are determined by bisection search to ensure that the trajectory does not touch the predefined refrigerant safe operation envelope within the prediction time window, thus forming a directionally asymmetric speed change rate safe feasible region. The speed constraint trimming module is used to trim the target speed change rate generated by photovoltaic power tracking to the safe and feasible range, and generate a constrained compressor speed command. The buffer energy storage module is used to absorb or supplement the difference between the actual power consumption corresponding to the constrained compressor speed command and the available photovoltaic power, and its capacity is 30 to 60 seconds of energy reserve corresponding to the system's rated power.

Images