Model prediction-based compressor variable frequency energy-saving control method, device and system

Abstract

The present disclosure relates to a model prediction-based compressor variable frequency energy-saving control method, device and system, which comprises: obtaining an online estimated current equivalent heat load based on current measurement data of a controlled object collected in a current sampling period; obtaining a future heat load sequence of a future preset number of steps through a prediction model based on the current equivalent heat load and historical data in a preset number of historical sampling periods; obtaining a feedforward component based on the future heat load sequence, obtaining a feedback component, superimposing the feedforward component and the feedback component to obtain a target speed instruction; and controlling a frequency converter of the controlled object according to the target speed instruction to adjust the speed of a compressor of the controlled object. The present disclosure predicts the heat load in the future preset sampling period, and corrects the generated deviation by combining the feedback component. By combining feedforward control and feedback control, the heat load disturbance is compensated in advance, and the speed of the compressor is adjusted in advance.

Description

Technical Field

[0001] This disclosure relates to the field of variable frequency control technology, specifically to a model-predictive variable frequency energy-saving control method, device, and system for compressors. Background Technology

[0002] In refrigeration systems such as constant temperature and humidity chambers, cold chain logistics boxes, and cold storage facilities, the compressor is the main energy-consuming component. The compressor's control strategy directly determines the system's energy efficiency and temperature control accuracy.

[0003] Currently, the control of variable frequency compressors in related technologies typically employs a proportional-integral-derivative (PID) control strategy. This strategy is a pure feedback control, adjusting the compressor speed based on the deviation between the currently detected temperature and the target temperature. However, due to the generally large thermal inertia of refrigeration systems, the temperature response often lags behind changes in heat load. When the system encounters a sudden change in heat load (such as opening or closing a door, placing an item inside), pure feedback control must wait for the temperature deviation to occur and accumulate to a certain level before triggering the adjustment action. This results in a significant lag in the control response, leading to large temperature fluctuations and a long recovery time, affecting temperature control accuracy and temperature recovery speed.

[0004] Furthermore, this feedback control method, which relies solely on historical and current deviations, cannot adjust the compressor speed in advance, causing the system to constantly operate in a cycle of "response lag - overcompensation - oscillation convergence," thus affecting the overall energy efficiency of the system. Summary of the Invention

[0005] The main purpose of this disclosure is to provide a model-predictive compressor variable frequency energy-saving control method, device and system to solve the problem of poor temperature control accuracy and energy saving effect caused by control response lag in related technologies.

[0006] According to some embodiments of this disclosure, a model-predictive compressor variable frequency energy-saving control method is provided. The method includes: obtaining an online estimated current equivalent heat load based on current measurement data of the controlled object collected in the current sampling period, wherein the current equivalent heat load is the total heat load of the internal space and contents of the controlled object; obtaining a future heat load sequence of a preset number of steps based on the current equivalent heat load and historical data in a preset number of historical sampling periods through a predictive model; obtaining a feedforward component based on the future heat load sequence, obtaining a feedback component based on the current measurement data and the historical data, superimposing the feedforward component and the feedback component to obtain a target speed command; and controlling the inverter of the controlled object according to the target speed command to adjust the speed of the compressor of the controlled object.

[0007] In some embodiments, obtaining the online estimated current equivalent heat load based on the current measurement data of the controlled object collected within the current sampling period includes: collecting the current temperature within the controlled object, the current input power of the compressor of the controlled object, and the power of the auxiliary heater through the controller within the current sampling period; and obtaining the current equivalent heat load based on the heat dissipation balance equation, according to the current temperature, the current input power, and the power of the auxiliary heater, wherein the current equivalent heat load is expressed as: _load(k)=C_th×[T(k)-T(k-1)] / Ts+γ(k)×P_comp(k)-P_heat(k), where _load(k) represents the current equivalent heat load, T(k) represents the current temperature, T(k-1) represents the temperature of the controlled object in the previous sampling period adjacent to the current sampling period, C_th is the heat capacity of the controlled object, Ts is the sampling period, γ(k) represents the equivalent conversion factor from electrical power to cooling capacity, P_comp(k) represents the current input power, and P_heat(k) represents the power of the auxiliary heater.

[0008] In some embodiments, obtaining a future heat load sequence for a preset number of steps based on the current equivalent heat load and historical data within a preset number of historical sampling periods through a prediction model includes: using the current equivalent heat load, historical data within a preset number of historical sampling periods, and the time code of the current moment as input to the prediction model, wherein the historical data is a multi-dimensional feature vector, including historical temperature sequences within the controlled object, historical estimated heat load sequences, and historical compressor power sequences; and using a two-layer stacked long short-term memory network structure to enable the prediction model to output a future heat load sequence for a preset number of steps.

[0009] In some embodiments, obtaining the feedforward component based on the future heat load sequence includes: obtaining a first predicted heat load adjacent to the current sampling period in the future heat load sequence; obtaining an initial speed component based on the first predicted heat load according to the inverse mapping relationship of the variable frequency compressor cooling capacity-speed characteristics; and performing boundary protection on the initial speed component to obtain the feedforward component, wherein the feedforward component represents the compressor speed required to balance the first predicted heat load without considering the current temperature deviation, wherein the feedforward component is expressed as: n_ff(k) = n_rated × [ _load(k+1) / Q_rated]^(1 / α), where n_ff(k) represents the feedforward component. _load(k+1) represents the first predicted heat load, n_rated represents the rated speed of the compressor, Q_rated represents the rated cooling capacity of the compressor at the rated speed, and α is the cooling capacity-speed power law exponent.

[0010] In some embodiments, obtaining the feedback component based on the current measurement data and the historical data includes: obtaining a current temperature deviation based on the current temperature in the current measurement data, wherein the current temperature deviation is the difference between a preset reference temperature and the current temperature; obtaining a historical temperature deviation sequence based on the historical temperature sequence in the historical data, wherein the historical temperature deviation sequence includes multiple historical temperature deviations, and the historical temperature deviation is the difference between a preset reference temperature and a corresponding historical temperature; and obtaining the feedback based on the current temperature deviation and the historical temperature deviation sequence through a proportional-integral-differential operation. The feedback component is expressed as: n_PID(k) = K_p × e_T(k) + K_i × Σ_{j=0}^{k}e_T(j) × Ts + K_d × [e_T(k) - e_T(k-1)] / Ts, where n_PID(k) represents the feedback component, e_T(k) represents the current temperature deviation, Σ_{j=0}^{k}e_T(j) represents the sum of all historical temperature deviations, e_T(k-1) represents the temperature deviation in the previous sampling period adjacent to the current sampling period, K_p is the proportional coefficient, K_i is the integral gain, and K_d is the differential gain.

[0011] In some embodiments, the step of superimposing the feedforward component and the feedback component to obtain the target speed command includes: superimposing the feedforward component and the feedback component to obtain a composite speed command; limiting the rate of change of the composite speed command; and performing boundary protection on the limited composite speed command to obtain the target speed command.

[0012] In some embodiments, the method further includes: performing rolling time-domain optimization based on preset constraints within the time-domain range of the future preset number of steps to obtain a future rotational speed sequence; the step of performing rolling time-domain optimization based on preset constraints within the time-domain range of the future preset number of steps to obtain a future rotational speed sequence includes: Based on the current temperature of the current sampling period, the future heat load sequence, and the mapping relationship between rotational speed and cooling capacity, the future temperature sequence for the next preset number of steps is obtained using the heat dissipation balance equation. Based on the total power consumption of the compressor and the speed change penalty within the time domain, an objective function is constructed, and rolling time-domain optimization is performed on the objective function. In response to each temperature deviation in the future temperature sequence satisfying the preset constraint conditions, each rotational speed within the next preset number of steps satisfying the preset constraint conditions, and the rolling time-domain optimization being successful, the future rotational speed sequence is obtained. The first element of the future rotational speed sequence is used as the current rotational speed command, and the target rotational speed command is switched to the current rotational speed command. The rotational speed of the compressor is adjusted using the current rotational speed command. In response to each temperature deviation in the future temperature sequence satisfying the preset constraint conditions, each rotational speed within the next preset number of steps satisfying the preset constraint conditions, and the rolling time-domain optimization timeout or failure, the rotational speed of the compressor is adjusted using the target rotational speed command.

[0013] According to some embodiments of this disclosure, this disclosure provides a compressor variable frequency energy-saving control device based on model prediction. The device includes: an acquisition unit, configured to acquire an online estimated current equivalent heat load based on current measurement data of the controlled object collected in the current sampling period, wherein the current equivalent heat load is the total heat load of the internal space and contents of the controlled object; a prediction unit, configured to acquire a future heat load sequence of a preset number of steps based on the current equivalent heat load and historical data within a preset number of historical sampling periods through a prediction model; a superposition unit, configured to acquire a feedforward component based on the future heat load sequence, acquire a feedback component based on the current measurement data and the historical data, and superimpose the feedforward component and the feedback component to obtain a target speed command; and a control unit, configured to control the inverter of the controlled object according to the target speed command to adjust the speed of the compressor of the controlled object.

[0014] According to some embodiments of this disclosure, an electronic device is provided, including: a processor; and a memory for storing computer programs or instructions executable by the processor; wherein the processor is configured to execute the computer programs or instructions to implement the compressor frequency conversion energy-saving control method described in any of the above embodiments.

[0015] According to some embodiments of this disclosure, a refrigeration system is provided, which includes a compressor, a frequency converter, and the electronic equipment described in the above embodiments.

[0016] By employing the above technical solution, this disclosure can predict the heat load within a preset sampling period and generate a feedforward component based on the prediction result. This feedforward component can compensate for impending changes in heat load in advance. Simultaneously, it combines a feedback component generated based on current measurement data to correct existing deviations. By combining feedforward control and feedback control, it achieves proactive compensation for heat load disturbances and advances the adjustment of the compressor speed. Attached Figure Description

[0017] The preferred embodiments of this disclosure are described below with reference to the accompanying drawings, in which: Figure 1 This is a flowchart illustrating a model-predictive compressor variable frequency energy-saving control method according to some embodiments of the present disclosure; Figure 2 This is a flowchart illustrating a method for obtaining feedforward components according to some embodiments of the present disclosure; Figure 3 This is a flowchart illustrating a method for obtaining feedback components according to some embodiments of the present disclosure; Figure 4 This is a block diagram of a model-predictive compressor variable frequency energy-saving control device according to some embodiments of the present disclosure; Figure 5 This is a block diagram illustrating an apparatus for variable frequency energy-saving control of a compressor, according to some embodiments of the present disclosure. Detailed Implementation

[0018] Some embodiments of this disclosure will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. Various changes, modifications, and equivalents of the methods, apparatus, and / or systems described herein will become readily apparent upon understanding this disclosure. For example, the order of operations described herein is merely illustrative and is not limited to those orders set forth herein, but can be changed as will become readily apparent upon understanding this disclosure, except for operations that must be performed in a particular order. Furthermore, for clarity and brevity, descriptions of features known in the art may be omitted.

[0019] The embodiments described in the following examples of this disclosure do not represent all embodiments consistent with this disclosure. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this disclosure as detailed in the appended claims. Those skilled in the art should understand that these embodiments are merely for explaining the technical principles of this disclosure and are not intended to limit the scope of protection of this disclosure.

[0020] This disclosure provides a compressor variable frequency energy-saving control method in some embodiments, which is used to pre-regulate the compressor speed to prevent response lag from affecting the temperature control accuracy of the refrigeration system.

[0021] Before providing a further detailed description of the embodiments of this disclosure, the nouns and terms involved in the embodiments of this disclosure will be explained, and the nouns and terms involved in the embodiments of this disclosure are subject to the following interpretations.

[0022] (1) Current equivalent heat load: refers to the total heat load of the internal space and contents of the controlled object (such as a refrigeration system like a constant temperature and humidity chamber), which is the comprehensive thermal disturbance that needs to be overcome to maintain its temperature stability. In this disclosure, it is estimated online through the heat dissipation balance equation based on energy conservation. This load includes not only steady-state heat leakage through the chamber walls, but also dynamic instantaneous thermal disturbances caused by factors such as opening doors, placing high-temperature items, and heating of internal components.

[0023] (2) Predictive model: refers to a mathematical or computational model that can predict future trends based on historical data. In this disclosure, a time series predictive model is preferred, such as a Long Short-Term Memory (LSTM) network, for predicting future changes in heat load. This model can make multi-step predictions of heat load over a period of time by learning complex patterns and dynamic relationships in historical data, providing a key basis for the "proactive" adjustment of the control system.

[0024] (3) Feedforward component: This refers to the control quantity calculated in advance based on the prediction of future disturbances (i.e., the predicted future heat load). Its function is to perform advance compensation, adjusting the control output before the temperature deviation actually occurs or significantly increases, so as to actively offset or reduce the impact of the expected disturbance. This is fundamentally different from traditional feedback control, which only reacts after the deviation occurs.

[0025] (4) Feedback component: This refers to the control quantity calculated based on the deviation between the current system state and the target state (i.e., temperature deviation). Its function is to eliminate the deviation that has already occurred and maintain the long-term stability and accuracy of the system. In practical applications, this component is usually implemented through a classic proportional-integral-derivative (PID) controller or its variants, and is a passive correction of the current error of the system.

[0026] (5) Receding Horizon Optimization (RHO): Also known as Model Predictive Control (MPC), it is a process control method. In each control cycle, based on the system's dynamic model and future disturbance predictions, it solves a constrained optimal control problem within a finite timeframe (i.e., the "prediction time domain") to find a control sequence that minimizes future cost functions (such as energy consumption). However, it only executes the first control action of this optimal sequence, and then in the next control cycle, it re-predicts and optimizes based on the latest system state. This "rolling" execution method gives it both optimality and robustness to change.

[0027] Figure 1 This is a flowchart illustrating a model-predictive compressor variable frequency energy-saving control method 100 according to some embodiments of this disclosure, such as... Figure 1 As shown, the method 100 includes steps S102, S104, S106 and S108.

[0028] In S102, based on the current measurement data of the controlled object collected in the current sampling period, the current equivalent heat load is estimated online, where the current equivalent heat load is the total heat load of the internal space of the controlled object and its contents.

[0029] In this embodiment of the disclosure, the purpose of this step is to quantify the total thermal disturbance acting on the controlled object in real time and accurately. Traditional control methods typically do not directly measure or estimate the thermal load, but rather infer it indirectly through temperature deviations, which leads to control lag. This disclosure provides a direct, delay-free input for subsequent prediction and control by actively estimating the current equivalent thermal load online, which is the foundation for achieving proactive control. By establishing a thermal balance model of the controlled object and combining it with real-time acquired operational data, this key physical quantity can be dynamically calculated.

[0030] In S104, based on the current equivalent heat load and historical data within a preset number of historical sampling periods, the future heat load sequence for the preset number of steps is obtained through a prediction model.

[0031] In this embodiment, based on the newly acquired current equivalent heat load and a series of historical data, a pre-trained prediction model is used to obtain the future heat load sequence for a preset number of steps, and the heat load for future sampling periods can be predicted in a rolling manner. This step embodies the "foresight" of this disclosure. Historical data may include past temperatures, heat loads, compressor power, etc., and these time series data contain the dynamic characteristics of the system and the patterns of external disturbances. The prediction model, such as a deep learning network, can learn from these complex time series and infer the trend of heat load changes over a future period. Through this step, the controller no longer merely passively responds to deviations, but has a quantitative prediction of future thermal disturbances, thereby solving the technical deficiency of traditional control methods that cannot predict load changes.

[0032] In S106, a feedforward component is obtained based on the future thermal load sequence, and a feedback component is obtained based on the current measurement data and historical data. The feedforward component and the feedback component are superimposed to obtain the target speed command.

[0033] In this embodiment, after obtaining the predicted future heat load sequence, the instruction generation stage begins. This stage decomposes the control task into two parts: feedforward control and feedback control. First, based on the predicted future heat load sequence, a feedforward component is calculated. The feedforward component represents the amount of heat load that, without considering any current temperature deviation, is sufficient to balance the predicted heat load at future times. This is an active, open-loop compensation strategy. Its principle is to output a control action in advance that is equivalent in magnitude and opposite in direction to the future disturbance, thereby "canceling" the disturbance before it actually affects the controlled object, shortening the response time to the disturbance.

[0034] Furthermore, to compensate for the inherent errors of the predictive model and to cope with unforeseen sudden disturbances, this disclosure also calculates feedback components based on current measurement data and historical data. The feedback components focus on the "present" and the "past," generating corrective control commands by measuring the deviation between the actual temperature of the controlled object and the set reference temperature, and combining this with the accumulation and rate of change of historical deviations. This closed-loop feedback mechanism ensures that even with deviations in the feedforward control, the controlled object can ultimately be accurately stabilized at the setpoint, ensuring the robustness and steady-state accuracy of the control.

[0035] In S108, the inverter of the controlled object is controlled according to the target speed command to adjust the speed of the compressor of the controlled object.

[0036] In this embodiment, the calculated feedforward and feedback components are superimposed to form a composite target speed command. This command includes both prediction and compensation for future disturbances and correction for current errors. Based on the target speed command that combines foresight and real-time performance, the inverter of the controlled object is controlled, thereby precisely adjusting the compressor speed. In this way, the cooling capacity provided by the compressor can be matched with the dynamically changing heat load in advance, ensuring strict temperature control accuracy while avoiding energy waste caused by lag response and overcompensation, thus achieving the dual goals of energy saving and precise control.

[0037] This disclosure achieves predictive and precise regulation of compressor speed by introducing prediction of future heat load and combining feedforward control and feedback control, thus solving the problems of response lag, poor temperature control accuracy and high energy consumption of traditional PID control or on-off control when dealing with dynamic heat load.

[0038] Furthermore, in a preferred embodiment, within each sampling period, the controller acquires the current temperature T(k) within the controlled object, the current input power P_comp(k) of the compressor, and (if present) the power P_heat(k) of the auxiliary heater. These are all easily measurable physical quantities. Based on these current measurement data, the current equivalent heat load is calculated using the heat dissipation balance equation. This equation is based on the law of conservation of energy, that is, the heat entering the system minus the heat dissipated by the system equals the change in energy within the system.

[0039] Specifically, the current equivalent heat load can be expressed as: _load(k)=C_th×[T(k)-T(k-1)] / Ts+γ(k)×P_comp(k)-P_heat(k) in, _load(k) represents the current equivalent heat load; T(k) represents the current temperature, and T(k-1) represents the temperature of the controlled object in the previous sampling period adjacent to the current sampling period. The difference between these two reflects the rate of temperature change, which, multiplied by the heat capacity C_th, represents the change in the system's internal energy; Ts is the sampling period; γ(k) represents the equivalent conversion factor from electrical power to cooling capacity, which converts the compressor's input electrical power P_comp(k) into the actual generated cooling capacity; P_heat(k) represents the auxiliary heater power, serving as the direct heat input. Through this formula, the controller can quantitatively "observe" the net heat load acting on the system in real time, i.e., the current equivalent heat load, providing accurate input for subsequent predictions. C_th is the heat capacity of the controlled object, which can be obtained through factory calibration or online identification. For example, online identification can be performed using the Recursive Least Square (RLS) method to adapt to changes in thermal characteristics caused by equipment aging or load variations, thereby improving the long-term accuracy of observations.

[0040] In this embodiment, C_th is identified online using the recursive least squares method. The identification parameter vector is defined as θ = [C_th, Q_loss_0]^T, where Q_loss_0 is the steady-state heat leakage baseline. A regression equation is constructed: ΔT(k) = φ(k)^T × θ, where ΔT(k) represents the temperature change in the current sampling period, i.e., the temperature difference between the current and previous sampling periods, i.e., ΔT(k) = T(k) - T(k-1), and φ(k) represents the regression vector constructed from measurable variables such as ambient temperature difference, input heat, and cooling capacity.

[0041] Where the regression vector φ(k) = [ΔT_env(k) / Ts, 1 / Ts]^T, ΔT_env is the environmental temperature difference. The RLS update law is: K(k)=P(k-1)φ(k) / [λ+φ(k)^TP(k-1)φ(k)] P(k)=[IK(k)φ(k)^T]P(k-1) / λ Where K(k) represents the RLS gain vector of the k-th sampling period, used to determine the correction magnitude of the current measurement residual to the parameter update. P(k-1) represents the parameter estimation covariance matrix of the previous sampling period, reflecting the uncertainty of the parameter estimation. λ is the forgetting factor; optionally, λ∈[0.95,0.99] is the forgetting factor to ensure that the parameters are updated as the operating conditions change. Identification Results _th is used to update the hot load observer. Let P(k) represent the parameter estimation vector updated after the k-th sampling period. Let P(k) represent the parameter estimation covariance matrix of the current sampling period, and let I represent the identity matrix, which has the same dimension as P(k-1).

[0042] because =[ _th(k), Therefore, after each RLS update, the parameter estimation vector is [loss, θ(k)]^T. The first item is the equivalent heat capacity of the box obtained online. _th.

[0043] In another preferred embodiment, the prediction model can employ a two-layer stacked Long Short-Term Memory (LSTM) network structure. LSTM is a recurrent neural network particularly adept at processing and predicting time-series data. Its internal gating mechanism can effectively learn long-term dependencies in the data, which is crucial for predicting heat loads with complex periodicity and nonlinear dynamics. Specifically, during prediction, the current equivalent heat load, historical data within a preset number of historical sampling periods, and the time code of the current moment are used as inputs to the prediction model. Here, historical data is a multi-dimensional feature vector, such as a multi-dimensional feature vector of the past N sampling periods (including the current sampling period). For example, with a historical window length of N=20, it corresponds to the historical temperature sequence, historical estimated heat load sequence, and historical compressor power sequence of the past 20 sampling periods. These features describe the historical state and dynamic behavior of the system from different dimensions. The time code can also be used as input to the prediction model. For example, using sine and cosine functions to encode the time of day helps the model learn time-related periodic load patterns (such as diurnal temperature range). The input X(k) of the prediction model can be expressed as: X(k) = [T(k-N+1:k), _load(k-N+1:k), P_comp(k-N+1:k), TOD(k)]^T. Where T(k-N+1:k) is an N×1 dimension, representing the historical temperature sequence within the controlled object, reflecting the thermal inertia state, capturing the temperature change trend, and is the core feature for predicting the temperature regression path; `_load(k-N+1:k)` is an N×1 dimension representing the historical estimated heat load sequence, directly reflecting historical thermal disturbance patterns and providing the strongest correlation features for predicting future heat load. `P_comp(k-N+1:k)` is an N×1 dimension representing the historical compressor power sequence, used to help correct cooling capacity estimation errors and improve observation accuracy. `TOD(k)` is a 2×1 dimension representing the time code of the current moment, which can be represented as [sin(2πt / T_day), cos(2πt / T_day)], used to capture diurnal periodic load patterns and avoid periodic aliasing. Here, `k` represents the sampling period at the `k`th step, i.e., the current sampling period.

[0044] These input features are fed into the input layer of a stacked LSTM network, with a history window length of N=20 and an input dimension of 62 (3 N-dimensional sequences and 2-dimensional temporal encoding). The first LSTM layer extracts preliminary short-term dynamic features from the original sequence of the input layer, and its output sequence is then fed into the second LSTM layer to learn and extract higher-order temporal features, outputting the final hidden state. Finally, through a fully connected layer (i.e., the output layer), the network outputs a sequence, namely the future hot load sequence for a preset number of steps H. _load(k+1:k+H), such as H=5.

[0045] In this way, the predictive model not only predicts the load at the next moment, but also provides a profile of load changes in the short term, enabling more complex optimization control (such as RHO).

[0046] Furthermore, the core idea of ​​feedforward components is to use predictive information for reverse engineering to calculate the control quantity that can just balance the future load. Figure 2 This is a flowchart illustrating a method 200 for obtaining feedforward components according to some embodiments of the present disclosure, such as... Figure 2 As shown, the method 200 includes steps S202, S204 and S206.

[0047] In S202, the first predicted heat load adjacent to the current sampling period in the future heat load sequence is obtained.

[0048] In this embodiment of the disclosure, the controller first obtains the first predicted heat load adjacent to the current sampling period from the predicted future heat load sequence, denoted as... _load(k+1). This value represents the thermal load that will be encountered in the next sampling period.

[0049] In S204, the initial speed component is obtained based on the first predicted heat load according to the inverse mapping relationship of the variable frequency compressor's cooling capacity-speed characteristics.

[0050] Based on the pre-calibrated inverse mapping relationship of the variable frequency compressor's cooling capacity-speed characteristics, and based on this first predicted heat load... The initial rotational speed component is calculated using _load(k+1). This characteristic can usually be described by a power-law function, and its inverse mapping is: n_ff(k)=n_rated×[ _load(k+1) / Q_rated]^(1 / α) Where n_ff(k) is the calculated feedforward component; _load(k+1) is the first predicted heat load; n_rated is the rated speed of the compressor; Q_rated is the rated cooling capacity of the compressor at its rated speed; α is the power-law exponent of cooling capacity-speed, which can be taken as α∈[1.1, 1.4]). The physical meaning of this formula is: without considering the current temperature deviation, to generate To balance the predicted heat load with such a large cooling capacity as _load(k+1), the compressor needs to operate at a speed of n_ff(k).

[0051] In S206, the initial speed component is boundary protected and a feedforward component is obtained. The feedforward component represents the compressor speed required to balance the first predicted thermal load without considering the current temperature deviation.

[0052] In this embodiment, to ensure safety, the initial speed component also needs boundary protection, that is, it needs to be limited between the minimum and maximum allowable speeds of the compressor. This can be represented by the formula n_ff(k) = clamp(n_ff(k), n_min, n_max). Here, n_min and n_max are the minimum and maximum allowable speeds of the compressor, respectively, which can be determined by the inverter protection parameters. Typical values ​​are n_min = 0.3 × n_rated and n_max = n_rated. The resulting final feedforward component characterizes the ideal compressor speed required to balance the thermal load at the next moment, without considering the current temperature deviation.

[0053] In one alternative implementation, the method for obtaining the feedback components is described in detail. Figure 3 This is a flowchart illustrating a method 300 for obtaining feedback components according to some embodiments of the present disclosure, such as... Figure 3 As shown, the method 300 includes steps S302, S304 and S306.

[0054] In S302, the current temperature deviation is obtained based on the current temperature in the current measurement data, where the current temperature deviation is the difference between the preset reference temperature and the current temperature.

[0055] In S304, based on the historical temperature sequence in the historical data, a historical temperature deviation sequence is obtained. The historical temperature deviation sequence includes multiple historical temperature deviations, and the historical temperature deviation is the difference between the preset reference temperature and the corresponding historical temperature.

[0056] In this embodiment, the feedback component serves to eliminate existing temperature deviations. First, the controller compares the current temperature T(k) from the current measurement data with a preset reference temperature T_ref to obtain the current temperature deviation e_T(k). Simultaneously, the controller maintains a historical temperature sequence; by comparing the historical temperatures with the preset reference temperature, a historical temperature deviation sequence can be obtained.

[0057] In S306, the feedback component is obtained by proportional-integral-differential operation based on the current temperature deviation and the historical temperature deviation sequence.

[0058] This feedback component can be represented as: n_PID(k)=K_p×e_T(k)+K_i×Σ_{j=0}^{k}e_T(j)×Ts+K_d×[e_T(k)-e_T(k-1)] / Ts Wherein, n_PID(k) represents the feedback component; e_T(k) is the current temperature deviation, and its product with the proportional coefficient K_p constitutes the proportional term, used for rapid response to the current deviation; Σ_{j=0}^{k}e_T(j) represents the accumulation of all historical temperature deviations from the beginning to the present, where {j=0}^{k} represents j=0, 1, 2, ..., k, and k represents the current count, and its product with the integral gain K_i constitutes the integral term, used to eliminate the steady-state error of the system; e_T(k)-e_T(k-1) represents the difference between the current temperature deviation and the temperature deviation of the upsampling period, reflecting the trend of deviation change, and its product with the derivative gain K_d constitutes the derivative term, used to predict deviation changes and provide damping to prevent overshoot. Through the combination of these three terms, the PID controller can effectively correct the current temperature deviation. To prevent integral saturation, an anti-windup mechanism is introduced: when n_PID(k) exceeds the range [n_min, n_max], the accumulation of the integral term is paused until n_PID(k) returns to the feasible region. The PID parameters (K_p, Ki, K_d) can be tuned using the Ziegler-Nichols step response method or designed based on the model identification results of the thermodynamic equations.

[0059] Furthermore, the feedforward and feedback components are superimposed to obtain the target speed command, thereby improving the stability and reliability of the system. First, the calculated feedforward component n_ff(k) and the feedback component n_PID(k) are arithmetically superimposed to obtain the original composite speed command n_raw(k).

[0060] However, this raw command may fluctuate drastically due to measurement noise or sudden load changes. To protect the compressor's mechanical structure and prevent wear and damage caused by frequent and drastic speed changes, a speed change rate limit is needed for this composite speed command. Specifically, the currently calculated raw command is compared with the command output at the previous moment. If the absolute value of the difference exceeds the product of a preset maximum allowable change rate and the sampling period (e.g., 200 rpm / s), the command value is only allowed to increase or decrease by a maximum amount from the previous moment. Otherwise, the difference between the two (the difference between the currently calculated raw command value and the command output at the previous moment) is increased or decreased from the command value at the previous moment. This step effectively smooths the speed command change curve. The speed change rate limit can be expressed as: n_cmd(k) = n_cmd(k-1) + clamp[n_raw(k) - n_cmd(k-1), -ΔR×Ts, +ΔR×Ts], where ΔR is the maximum allowable rate of change of speed (typical value ΔR = 200 rpm / s), and Ts is the sampling period (e.g., 1 second). The essence of rate of change limiting is to strike a balance between control response speed and compressor mechanical protection: too small a ΔR will weaken the feedforward's anticipatory response, while too large a ΔR will cause speed chattering. ΔR = 200 rpm / s is a value chosen by comprehensively balancing typical compressor protection specifications (maximum rate of change ≤ 300 rpm / s) and control bandwidth requirements.

[0061] Finally, the composite speed command, after being limited by the rate of change, undergoes boundary protection again to ensure that its value always falls within the safe operating range (i.e., minimum and maximum speed) allowed by the compressor and inverter. The command obtained after this series of processes is the final, safe, and smooth target speed command output to the inverter. This approach leverages the advantages of rapid response in composite control while ensuring the long-term reliable operation of the compressor.

[0062] This disclosure obtains the future heat load sequence through a predictive model and generates a feedforward component based on the prediction result. This feedforward component can compensate for impending changes in heat load in advance. Simultaneously, a feedback component generated based on current measurement data is used to correct existing deviations. By combining feedforward and feedback control, proactive compensation for heat load disturbances is achieved, effectively overcoming the response lag problem of pure feedback control. This significantly shortens the temperature recovery time and improves temperature control accuracy. Furthermore, by dynamically and precisely matching the compressor speed with the actual load demand, unnecessary energy waste and speed fluctuations are avoided. This effectively reduces system energy consumption while ensuring temperature control accuracy and helps extend equipment lifespan.

[0063] In a more preferred embodiment, the method of this disclosure may further include a rolling time-domain optimization (RHO) step to achieve globally optimal energy-saving performance while meeting temperature control accuracy requirements. This step does not simply superimpose feedforward and feedback components, but seeks the optimal speed control sequence within a future time window (i.e., a preset number of steps, such as 5 steps).

[0064] Specifically, in each sampling period, the known quantities are first determined: the current measured temperature T (k), the compressor speed n (k-1) of the previous sampling period, and the future heat load sequence predicted by the LSTM network. _load(k+1), _load(k+2),..., The mapping relationship between speed and cooling capacity is: Q_cool(n(k+j)) = Q_rated × [n(k+j) / n_rated]^α. The mapping relationship between speed and compressor input power is: P_comp(n(k+j)) = P_rated × [n(k+j) / n_rated]^β, where Q_cool(n(k+j)) represents the cooling capacity in the (k+j)th sampling period, Q_rated represents the rated cooling capacity at the rated speed n_rated, and n(k+j) is the candidate speed in the (k+j)th sampling period (or the j-th future sampling period), j = 0, 1, ..., H-1. P_comp(n(k+j)) represents the compressor input power in the (k+j)th sampling period, P_rated represents the rated power, and the exponents α and β are determined by the compressor model and refrigerant characteristics, and can be obtained by fitting the factory performance curve. The compressor speed sequence of the current sampling period and the next H-1 sampling periods is used as the optimization variable.

[0065] Based on the current measured temperature, the future heat load sequence, and the heat dissipation balance equation, the future temperature change trajectory under different future speed strategies is predicted. The future temperature sequence can be expressed as: (k+j+1)= (k+j)+[Ts / _th]×[ _load(k+j+1)-Q_cool(n(k+j))], where: (k) = T(k), with the current measured temperature as the initial value. _th represents the online heat capacity identification result. If heater or heat leakage correction is considered, it can also be written as: (k+j+1)= (k+j)+[Ts / _th]×[ _load(k+j+1)+P_heat(k+j)-Q_cool(n(k+j))-Q_loss(k+j)], where P_heat(k+j) represents the auxiliary heater power and Q_loss(k+j) represents the heat loss.

[0066] An objective function is constructed to minimize the total power consumption of the compressor in the prediction time domain, and may include a penalty term for speed changes to ensure smooth operation and suppress drastic speed changes. Simultaneously, a series of strict constraints are set, including: (1) Temperature accuracy constraint: requiring that the predicted temperature for each future step in the optimization time domain must remain within the allowable deviation range, i.e., | (k+j+1)-T_ref|≤ε_T, where ε_T is the temperature control accuracy constraint, which can be 0.3℃; (2) Rotational speed physical constraint, requiring that the rotational speed of each future step in the optimization time domain must meet the physical constraint, i.e., n_min≤n(k+j)≤n_max, and the rotational speed is between the upper and lower limits; (3) Rotational speed change rate constraint, requiring that the rotational speed change rate of each future step in the optimization time domain does not exceed the set threshold, i.e., |n(k+j)-n(k+j-1)|≤ΔR×Ts.

[0067] The controller solves this constrained optimization problem. During the solution process, the previous cycle speed n(K-1) or feedforward speed n_ff(k+j) can be used as initial values. First-order or second-order approximations are made to Q_cool(n) (cooling capacity) and P_comp(n) (compressor input power) near the current operating point, transforming the original nonlinear optimization problem into a small-scale quadratic programming (QP) problem. The solution yields: If the optimization is successful, meaning a sequence of future speeds that satisfies all constraints and minimizes energy consumption has been found, then the controller will use the first element of that sequence. As the final speed command at the current moment, it replaces the original target speed command generated by the superposition of feedforward and feedback, i.e., n_cmd(k) = The system uses the optimal RHO solution as the final command, with feedforward and PID feedback serving only as reference values ​​(used for RHO initialization and guessing values ​​to accelerate convergence). To avoid shocks during switching, a soft switch with linear interpolation can be applied between the two, with a transition time of 2-5 sampling periods to prevent sudden speed jumps. If the optimization process fails due to timeout or lack of a feasible solution, the system automatically degrades, using the target speed command generated by the superposition of feedforward and feedback as a backup, i.e., n_cmd(k) = n_ff(k) + n_PID(k), thus ensuring the continuity and safety of control.

[0068] In this embodiment of the disclosure, the controller completes the RHO solution once within each sampling period Ts, and only the first step of the optimal solution sequence is processed. The current control command output, or the target speed command generated by superimposing the feedforward and feedback, is used as the final control command. This conforms to the closed-loop process of the rolling time-domain control principle. The next sampling period is used to resolve the problem and update the time continuously. Through this rolling optimization method, this disclosure can make decisions that take into account both current needs and future energy efficiency at every moment, thereby improving energy-saving performance.

[0069] Based on the same concept, this disclosure also provides a compressor frequency converter control device 400. It is understood that the compressor frequency converter control device 400 provided in this disclosure includes hardware structures and / or software modules corresponding to each function in order to achieve the above-mentioned functions. Combining the units and algorithm steps of the various examples disclosed in this disclosure, this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the technical solutions of this disclosure.

[0070] Figure 4 This is a block diagram of a model-predictive compressor variable frequency energy-saving control device 400 shown according to some embodiments of the present disclosure, with reference to... Figure 4 The device 400 includes an acquisition unit 402, a prediction unit 404, an overlay unit 406, and a control unit 408.

[0071] The acquisition unit 402 is configured to acquire the online estimated current equivalent heat load based on the current measurement data of the controlled object collected in the current sampling period, wherein the current equivalent heat load is the total heat load of the internal space and contents of the controlled object; the prediction unit 404 is configured to acquire the future heat load sequence of a preset number of steps in the future through a prediction model based on the current equivalent heat load and historical data in a preset number of historical sampling periods; the superposition unit 406 is configured to acquire the feedforward component based on the future heat load sequence, acquire the feedback component based on the current measurement data and historical data, and superimpose the feedforward component and the feedback component to obtain the target speed command; and the control unit 408 is configured to control the inverter of the controlled object according to the target speed command to adjust the speed of the compressor of the controlled object.

[0072] Regarding the apparatus in the above embodiments, the specific manner in which each unit performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0073] Figure 5 This is a block diagram illustrating an apparatus 500 for inverter energy-saving control of a compressor according to some embodiments of the present disclosure. For example, apparatus 500 may be provided as a server, computer, or embedded system. See also... Figure 5 The device 500 includes a processing component 522, which further includes one or more processors, and memory resources represented by a memory 532 for storing instructions, such as application programs, that can be executed by the processing component 522. The application programs stored in the memory 532 may include one or more modules, each corresponding to a set of instructions. Furthermore, the processing component 522 is configured to execute instructions to perform the aforementioned compressor variable frequency energy-saving control method.

[0074] Device 500 may also include a power supply component 526 configured to perform power management of device 500, a wired or wireless network interface 550 configured to connect device 500 to a network, and an input / output (I / O) interface 558. Device 500 can operate on an operating system stored in memory 532, such as Windows Server™, MacOSX™, Unix™, Linux™, FreeBSD™, or similar.

[0075] This disclosure also provides a refrigeration system, which includes a compressor, a frequency converter, and the electronic equipment described above. The electronic equipment can implement a model-predictive compressor frequency conversion energy-saving control method to control the compressor speed through the frequency converter and achieve energy-saving temperature control regulation of the refrigeration system.

[0076] This disclosure also provides a computer program product, including a computer program that, when executed by a processor, implements the compressor variable frequency energy-saving control method involved in any of the above embodiments. This computer program product can exist in various forms; for example, it can be a firmware update package stored on a USB flash drive, CD-ROM, or network server.

[0077] It should be noted that in the description of this disclosure, the terms "upper," "lower," "left," "right," "inner," and "outer," which indicate directional or positional relationships, are based on the directional or positional relationships shown in the accompanying drawings. These are used merely for ease of description and do not indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this disclosure. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0078] Furthermore, it should be noted that, in the description of this disclosure, unless otherwise expressly specified and limited, the terms "installation," "setup," and "connection" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection, an indirect connection through an intermediate medium, or a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this disclosure according to the specific circumstances.

[0079] In this description, "multiple" means at least two, referring to two or more, such as two, three, etc., unless otherwise explicitly specified. Other quantifiers are similar. The singular forms "a," "the," and "the" are also intended to include the plural forms unless the context clearly indicates otherwise. Furthermore, unless otherwise specified or clearly indicated from the context, the articles "a" and "an" as used in this disclosure and the appended claims are generally understood to mean "one or more."

[0080] It should be understood that, unless otherwise specifically indicated, features of various embodiments of this disclosure described herein can be combined with each other. As used herein, the term "and / or" includes any one of the related listed items and any combination of two or more; "and / or" describes the association relationship between related objects, indicating that three relationships may exist, for example, A and / or B can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. Similarly, "at least one of..." includes any one of the related listed items and any combination of two or more.

[0081] Furthermore, the term "exemplary" is used herein to indicate that it serves as an example, instance, or illustration. Any aspect or design described herein as "exemplary" is not necessarily to be construed as advantageous compared to other aspects or designs. Rather, the use of the term "exemplary" is intended to present concepts in a concrete manner. As used herein, the term "or" is intended to indicate an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified or clear from the context, "X applies A or B" is intended to indicate any of the natural inclusive permutations. That is, if X applies A; X applies B; or X applies both A and B, then applying A or B satisfies the condition under any of the foregoing instances.

[0082] Similarly, although this disclosure has been shown and described with respect to one or more implementations, equivalent variations and modifications will occur to those skilled in the art upon reading and understanding the specification and drawings. This disclosure includes all such modifications and variations and is limited only by the scope of the claims. In particular, with respect to the various functions performed by the components described above (e.g., elements, resources, etc.), unless otherwise indicated, the terminology used to describe such components is intended to correspond to any component (functionally equivalent) that performs the specific function of the described component, even if structurally not equivalent to the disclosed structure. Furthermore, although specific features of this disclosure may have been disclosed with respect to only one of several implementations, such features may be combined with one or more other features of other implementations, as may be desired and advantageous to any given or particular application. Moreover, with regard to the terms “comprising,” “owning,” “having,” “having,” or variations thereof as used in this disclosure, such terms are intended to be inclusive in a manner similar to the term “including.”

[0083] The technical solutions of this disclosure have been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of this disclosure is obviously not limited to these specific embodiments. Without departing from the principles of this disclosure, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of this disclosure.

Claims

1. A compressor variable frequency energy-saving control method based on model prediction, characterized in that, The method includes: Based on the current measurement data of the controlled object collected in the current sampling period, the current equivalent heat load is estimated online, wherein the current equivalent heat load is the total heat load of the internal space and contents of the controlled object. Based on the current equivalent heat load and historical data within a preset number of historical sampling periods, a future heat load sequence with a preset number of steps is obtained through a prediction model. A feedforward component is obtained based on the future heat load sequence; a feedback component is obtained based on the current measurement data and the historical data; the feedforward component and the feedback component are superimposed to obtain the target speed command; and The inverter of the controlled object is controlled according to the target speed command to adjust the speed of the compressor of the controlled object.

2. The method according to claim 1, characterized in that, The step of obtaining the online estimated current equivalent heat load based on the current measurement data of the controlled object collected within the current sampling period includes: During the current sampling period, the controller collects the current temperature inside the controlled object, the current input power of the compressor of the controlled object, and the power of the auxiliary heater. Based on the heat dissipation balance equation, the current equivalent heat load is obtained according to the current temperature, the current input power, and the auxiliary heater power. The current equivalent heat load is expressed as: _load(k)=C_th×[T(k)-T(k-1)] / Ts+γ(k)×P_comp(k)-P_heat(k), where _load(k) represents the current equivalent heat load, T(k) represents the current temperature, T(k-1) represents the temperature of the controlled object in the previous sampling period adjacent to the current sampling period, C_th is the heat capacity of the controlled object, Ts is the sampling period, γ(k) represents the equivalent conversion factor from electrical power to cooling capacity, P_comp(k) represents the current input power, and P_heat(k) represents the power of the auxiliary heater.

3. The method according to claim 1, characterized in that, The step of obtaining the future heat load sequence for a preset number of steps based on historical data within a preset sampling period and the current equivalent heat load, through a prediction model, includes: The current equivalent heat load, a preset number of historical data within a historical sampling period, and the time code of the current moment are used as inputs to the prediction model. The historical data is a multi-dimensional feature vector, including the historical temperature sequence, the historical estimated heat load sequence, and the historical compressor power sequence within the controlled object. By using a double-layer stacked long short-term memory network structure, the prediction model outputs a future hot load sequence with a preset number of steps.

4. The method according to claim 1, characterized in that, The step of obtaining the feedforward component based on the future heat load sequence includes: Obtain the first predicted heat load adjacent to the current sampling period in the future heat load sequence; Based on the inverse mapping relationship of the variable frequency compressor's cooling capacity-speed characteristics, the initial speed component is obtained based on the first predicted heat load; and Boundary protection is applied to the initial speed component to obtain the feedforward component, which represents the compressor speed required to balance the first predicted heat load without considering the current temperature deviation. The feedforward component is expressed as: n_ff(k)=n_rated×[ _load(k+1) / Q_rated]^(1 / α), where n_ff(k) represents the feedforward component. _load(k+1) represents the first predicted heat load, n_rated represents the rated speed of the compressor, Q_rated represents the rated cooling capacity of the compressor at the rated speed, and α is the cooling capacity-speed power law exponent.

5. The method according to claim 1, characterized in that, The step of obtaining feedback components based on the current measurement data and the historical data includes: Based on the current temperature in the current measurement data, the current temperature deviation is obtained, wherein the current temperature deviation is the difference between the preset reference temperature and the current temperature; Based on the historical temperature sequence in the historical data, a historical temperature deviation sequence is obtained, which includes multiple historical temperature deviations, each of which is the difference between a preset reference temperature and a corresponding historical temperature; and Based on the current temperature deviation and the historical temperature deviation sequence, the feedback component is obtained through proportional-integral-differential operations, and the feedback component is represented as: n_PID(k) = K_p × e_T(k) + K_i × Σ_{j=0}^{k}e_T(j) × Ts + K_d × [e_T(k) - e_T(k-1)] / Ts, where n_PID(k) represents the feedback component, e_T(k) represents the current temperature deviation, Σ_{j=0}^{k}e_T(j) represents the sum of all historical temperature deviations, e_T(k-1) represents the temperature deviation in the previous sampling period adjacent to the current sampling period, K_p is the proportional coefficient, K_i is the integral gain, and K_d is the derivative gain.

6. The method according to claim 1, characterized in that, The step of superimposing the feedforward component and the feedback component to obtain the target speed command includes: The feedforward component and the feedback component are superimposed to obtain a composite speed command; Limit the rate of change of the composite speed command; and The composite speed command after being limited is subjected to boundary protection to obtain the target speed command.

7. The method according to claim 1, characterized in that, The method further includes: Within the time domain of the predetermined number of future steps, rolling time domain optimization is performed based on predetermined constraints to obtain the future rotational speed sequence; The step of performing rolling time-domain optimization based on preset constraints within the time domain of the future preset number of steps to obtain the future rotational speed sequence includes: Based on the current temperature of the current sampling period, the future heat load sequence, and the mapping relationship between rotation speed and cooling capacity, the future temperature sequence of the preset number of steps is obtained using the heat dissipation balance equation. Based on the total power consumption and speed change penalty of the compressor within the time domain, an objective function is constructed, and the objective function is optimized in the rolling time domain. In response to the fact that each temperature deviation in the future temperature sequence satisfies the preset constraint, each rotation speed within the future preset number of steps satisfies the preset constraint, and the rolling time domain optimization is successful, the future rotation speed sequence is obtained, the first element in the future rotation speed sequence is taken as the current rotation speed command, the target rotation speed command is switched to the current rotation speed command, and the rotation speed of the compressor is adjusted using the current rotation speed command; In response to the fact that each temperature deviation in the future temperature sequence meets the preset constraint, each rotational speed within the future preset number of steps meets the preset constraint, and the rolling time-domain optimization times out or fails, the rotational speed of the compressor is adjusted using the target rotational speed command.

8. A compressor variable frequency energy-saving control device based on model prediction, characterized in that, The device includes: The acquisition unit is used to acquire the online estimated current equivalent heat load based on the current measurement data of the controlled object collected in the current sampling period, wherein the current equivalent heat load is the total heat load of the internal space and contents of the controlled object. The prediction unit is used to obtain the future heat load sequence of a preset number of steps based on the current equivalent heat load and historical data within a preset number of historical sampling periods through a prediction model. The superposition unit is configured to obtain a feedforward component based on the future heat load sequence, obtain a feedback component based on the current measurement data and the historical data, and superimpose the feedforward component and the feedback component to obtain a target rotational speed command; and The control unit is used to control the inverter of the controlled object according to the target speed command, so as to adjust the speed of the compressor of the controlled object.

9. An electronic device, characterized in that, include: processor; Memory for storing computer programs or instructions executable by the processor; The processor is configured to execute the computer program or instructions to implement the compressor frequency conversion energy-saving control method according to any one of claims 1-7.

10. A refrigeration system, characterized in that, It includes a compressor, a frequency converter, and the electronic equipment described in claim 9.

Images