Method for calculating and controlling energy saving amount of split air conditioner

Abstract

The invention discloses a split air conditioner energy saving amount calculation and control method, and relates to the technical field of heating ventilation air conditioner intelligent control, and the split air conditioner energy saving amount calculation and control method comprises the following steps: inputting a collected multi-source environment equipment data set into an LSTM neural network, generating a refrigerant dryness prediction value, and calculating a theoretical refrigerating capacity according to the refrigerant dryness prediction value; a reference energy efficiency ratio is calculated based on the theoretical refrigerating capacity, dynamic correction is conducted according to the refrigerant dryness predicted value, and a corrected energy efficiency ratio is generated; constructing a reinforcement learning decision model, inputting the corrected energy efficiency ratio and the multi-source environment equipment data set into the reinforcement learning decision model, and generating an optimized parameter combination; and the compressor frequency is adjusted according to the optimized parameter combination, and electronic expansion valve control is triggered based on the refrigerant dryness predicted value. According to the method, the LSTM neural network is constructed, a traditional sensor is replaced, high-precision indirect measurement is achieved, the adaptability to the dynamic working condition is improved, the energy efficiency ratio is dynamically corrected in combination with the optimal value defined by the refrigerant type, and error accumulation caused by working condition change of the fixed energy efficiency ratio is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent control of heating, ventilation and air conditioning, and in particular to a method for calculating and controlling energy of a split air conditioner. Background Art

[0002] In recent years, with the deep integration of artificial intelligence and the Internet of Things (IoT) technologies, the intelligent control of air-conditioning systems has gradually evolved from single-parameter adjustment to multi-dimensional dynamic optimization. Data-driven energy efficiency management technology has become a research hotspot, particularly in the field of split air conditioners. Scholars and companies are committed to collecting environmental and equipment operating data through sensor networks and combining them with machine learning models to achieve energy consumption prediction and control strategy optimization. Research on the refrigerant circulation characteristics of air-conditioning systems continues to deepen. As a key parameter affecting cooling efficiency, the dynamic monitoring and compensation algorithms for refrigerant dryness have become core directions for improving system energy efficiency.

[0003] Existing methods often evaluate air conditioner performance based on static or seasonal energy efficiency ratios (EERs), but they fail to fully consider the nonlinear impact of real-time changes in refrigerant dryness on cooling capacity and energy consumption. When refrigerant dryness deviates from the design value, there is a significant deviation between theoretical cooling capacity and actual output, resulting in distorted EER calculations. Furthermore, the intelligent control of split air conditioners often relies on preset rules or offline training models, making it difficult to cope with real-time changes under complex operating conditions. This results in the generated action parameters lacking a precise response to internal thermodynamic processes. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a split air conditioner energy calculation and control method to solve the problems of insufficient compensation for refrigerant dryness and insufficient response of control parameters.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides a split air conditioning energy calculation and control method, which includes inputting a collected multi-source environmental equipment data set into an LSTM neural network, generating a refrigerant dryness prediction value and calculating a theoretical cooling capacity based on the refrigerant dryness prediction value; calculating a baseline energy efficiency ratio based on the theoretical cooling capacity, and dynamically correcting it based on the refrigerant dryness prediction value to generate a corrected energy efficiency ratio; constructing a reinforcement learning decision model, inputting the corrected energy efficiency ratio and the multi-source environmental equipment data set into the reinforcement learning decision model, and generating an optimized parameter combination; The compressor frequency is adjusted according to the optimized parameter combination, and the electronic expansion valve control is triggered based on the refrigerant dryness prediction value. The temperature change rate and compressor power deviation are monitored in real time, and the transfer learning mechanism is triggered based on the monitoring results. The LSTM neural network parameters are updated and the refrigerant dryness prediction value and the corrected energy efficiency ratio are recalculated to generate the final optimized parameter combination.

[0007] As a preferred solution of the split air conditioning energy calculation and control method described in the present invention, the multi-source environmental equipment data set includes indoor and outdoor temperatures, compressor current and voltage, evaporator inlet and outlet temperature difference, condenser inlet and outlet temperature difference and user-set temperature.

[0008] As a preferred solution of the split air conditioner energy calculation and control method of the present invention, wherein: the generating of the refrigerant dryness prediction value and the calculation of the theoretical cooling capacity according to the refrigerant dryness prediction value are specifically performed as follows: Define the six-dimensional input layer, the first LSTM layer, the second LSTM layer, and the fully connected layer to build an LSTM neural network; The multi-source environmental equipment dataset is input into the LSTM neural network to generate the refrigerant dryness prediction value; The compressor power is calculated by the compressor current and voltage, and the theoretical cooling capacity is calculated by correlating it with the predicted value of the refrigerant dryness.

[0009] As a preferred solution of the split air conditioner energy calculation and control method of the present invention, wherein: the calculation of the benchmark energy efficiency ratio and the dynamic correction based on the predicted value of the refrigerant dryness to generate the corrected energy efficiency ratio are as follows: Generate a benchmark energy efficiency ratio based on theoretical cooling capacity and compressor power, identify the refrigerant type according to the split air conditioner design manual, and define the optimal refrigerant dryness value based on the refrigerant type; Calculate the difference between the predicted refrigerant dryness value and the optimal refrigerant dryness value to generate the refrigerant dryness deviation, and define the attenuation coefficient according to the refrigerant type; The benchmark energy efficiency ratio is adjusted according to the refrigerant dryness deviation and the attenuation coefficient to generate a corrected energy efficiency ratio.

[0010] As a preferred solution of the split air conditioning energy calculation and control method of the present invention, wherein: the generation of the optimized parameter combination, the specific steps are as follows: Define a state space based on the modified energy efficiency ratio and a multi-source environmental equipment dataset; Identify the controllable parameters of the split air conditioner according to the split air conditioner design manual, define the action space based on the controllable parameters of the split air conditioner, and define the output layer based on the driver interface; Construct a reinforcement learning decision model based on the state space, action space and output layer; The corrected energy efficiency ratio and multi-source environmental equipment dataset are merged into a state vector through the state space, and the original action parameters are generated through the action space, and the optimized parameter combination is generated through the output layer.

[0011] As a preferred solution of the split air conditioning energy calculation and control method of the present invention, wherein: the compressor frequency is adjusted according to the optimized parameter combination, and the electronic expansion valve control is triggered based on the refrigerant dryness prediction value, the specific steps are as follows: Extract the parameter values ​​in the optimized parameter combination, convert the compressor frequency into a PWM signal recognized by the split air conditioning control protocol, and transmit it to the inverter; The allowable deviation threshold is defined according to the experimental calibration method, the refrigerant dryness deviation is compared with the allowable deviation threshold, and the electronic expansion valve control is triggered according to the judgment result.

[0012] As a preferred solution of the split air conditioning energy calculation and control method of the present invention, wherein: the real-time monitoring of the temperature change rate and the compressor power deviation, and triggering the transfer learning mechanism according to the monitoring results, the specific steps are as follows: Collect real-time indoor temperature and compressor power, calculate temperature change rate and compressor power deviation; Define temperature change rate thresholds and power deviation thresholds based on thermal comfort standards and compressor safety margins; The triggering condition of the transfer learning mechanism is determined according to the temperature change rate threshold and the power deviation threshold, and the transfer learning mechanism is started according to the determination result.

[0013] As a preferred solution of the split air conditioning energy calculation and control method of the present invention, the specific steps of generating the final optimized parameter combination are as follows: The stochastic gradient descent method is used to update the LSTM neural network parameters, regenerate the refrigerant dryness prediction value, and update the corrected energy efficiency ratio; Incorporating the temperature change rate and the compressor power deviation into the multi-source environmental equipment dataset to generate a corrected multi-source environmental equipment dataset; The updated corrected energy efficiency ratio and corrected multi-source environmental equipment dataset are input into the reinforcement learning decision model to generate the final optimized parameter combination.

[0014] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the split air conditioning energy calculation and control method as described in the first aspect of the present invention is implemented.

[0015] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the split air conditioning energy calculation and control method as described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: by constructing an LSTM neural network, high-precision indirect measurement is achieved instead of traditional sensors, thereby improving adaptability to dynamic working conditions, and dynamically correcting the energy efficiency ratio in combination with the optimal value defined by the refrigerant type, thereby avoiding the error accumulation of the fixed energy efficiency ratio due to changes in working conditions; through the dual monitoring of the temperature change rate and the compressor power deviation, the transfer learning mechanism is triggered, and the LSTM parameters are updated online using random gradient descent, thereby improving the generalization ability of the LSTM network under environmental mutations. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 The figure is a flow chart of the energy calculation and control method for split air conditioners.

[0019] Figure 2 Flowchart for generating corrected energy efficiency ratio.

[0020] Figure 3 Flowchart for building a reinforcement learning decision model.

[0021] Figure 4 Flowchart for triggering the transfer learning mechanism. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0023] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in various places throughout this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive of other embodiments.

[0025] Reference Figures 1 to 4 , is an embodiment of the present invention, which provides a method for calculating and controlling energy of a split air conditioner, comprising the following steps: S1. Input the collected multi-source environmental equipment data set into the LSTM neural network to generate the refrigerant dryness prediction value and calculate the theoretical cooling capacity based on the refrigerant dryness prediction value.

[0026] S1.1. Collect indoor and outdoor temperatures, compressor current and voltage, evaporator inlet and outlet temperature difference, condenser inlet and outlet temperature difference, and user-set temperature, and perform preprocessing to generate a multi-source environmental equipment data set; It should be noted that the indoor unit evaporator surface temperature is collected as the indoor temperature using a PT100 temperature sensor at a sampling frequency of 1Hz, and the outdoor unit condenser ambient temperature is collected as the outdoor temperature; the original waveforms of the compressor three-phase current and DC bus voltage are captured using a current clamp and voltage probe at a sampling frequency of 10kHz, and the current total harmonic distortion rate characteristics are calculated in real time; dual PT100 probes are used to measure the evaporator inlet and outlet pipe wall temperatures respectively, and the evaporator inlet and outlet temperature difference is generated by subtracting the outlet temperature from the inlet temperature; dual PT100 probes are deployed on the condenser inlet and outlet pipe walls to generate the condenser inlet and outlet temperature difference; the user-set temperature value is read through the RS485 bus of the air-conditioning controller; the collected sensor data are subjected to median filtering with a sliding window of 30 seconds to eliminate transient interference pulses, and linear interpolation is used to fill in missing data points caused by communication packet loss. Finally, Z-score normalization is performed on all feature fields to generate a multi-source environmental equipment dataset with unified dimension.

[0027] S1.2. Define the six-dimensional input layer, the first LSTM layer, the second LSTM layer, and the fully connected layer, and build the LSTM neural network; It should be noted that a six-dimensional input layer is defined with 6 neurons to match the six-dimensional feature vectors in the multi-source environmental equipment dataset; the first LSTM layer is constructed and configured with 64 neuron units, and the hidden state output is passed to the second LSTM layer as a time series memory carrier; 64 neuron units are deployed in the second LSTM layer and receive the hidden state sequence of the first LSTM layer, and the refrigerant state features across time steps are extracted through a gating mechanism; finally, the final time step output vector of the second LSTM layer is input into the fully connected layer, which uses the Sigmoid activation function to map the 64-dimensional feature vector to a refrigerant dryness prediction value between 0 and 1, completing the topology construction of the LSTM neural network; It should also be noted that the gating mechanism is the core structure of the LSTM neural network, which collaboratively controls information flow through the forget gate, input gate, and output gate. The evaporator inlet and outlet temperature difference threshold is defined based on the measurement error tolerance of the PT100 temperature sensor and the maximum safe temperature difference allowed by the split air conditioner evaporator design. The forget gate dynamically clears outdated state memories of evaporator inlet and outlet temperature differences exceeding the evaporator inlet and outlet temperature difference threshold to avoid interference with the current prediction. The input gate filters the harmonic components in the compressor current total harmonic distortion rate characteristics and writes them into the cell state to enhance load change sensitivity. The cell state continuously memorizes the condenser inlet and outlet temperature difference response within ten seconds after the compressor frequency is adjusted. The output gate suppresses high-frequency fluctuation noise in outdoor temperature and only extracts the dominant characteristics of the condenser inlet and outlet temperature difference to ensure prediction stability. The multi-source environmental equipment dataset is divided into a training set (70%), a validation set (15%), and a test set (15%). High-precision pressure sensors and platinum resistance temperature probes are installed on the evaporator and condenser pipes to simultaneously collect the refrigerant saturation pressure and temperature. The refrigerant type of the current air-conditioning equipment is identified according to the split air-conditioning design manual, and the actual refrigerant dryness value of the current refrigerant type is determined according to the refrigerant physical property table. The training set is input into the LSTM neural network, and the mean square error is used as the loss function to measure the deviation between the predicted refrigerant dryness value and the actual refrigerant dryness value. The expression is: ; in, is the loss function value, which represents the overall error of the LSTM neural network in predicting the refrigerant dryness; Represents the total number of training set data records for a single split air conditioner laboratory prototype; Indicates the training set data index, the value range is 1- ; Indicates the Predicted value of refrigerant dryness of group data; Indicates the The actual value of the refrigerant dryness of the group data; Use the Adam optimizer, set the initial learning rate to 0.001, set the batch size to 32, and the training rounds to 100. Calculate the gradient through the time backpropagation algorithm and update the LSTM neural network weights until the maximum training rounds are reached. After training is completed, input the validation set and test set into the LSTM neural network for verification and testing, and save the final LSTM neural network. The expression is, ; in, Indicates the output refrigerant dryness prediction value; Represents the Sigmoid activation function; represents the weight matrix of the fully connected layer; represents the output gate vector; Represents the element-wise product operator; represents the cell state activation function; Indicates cell status; represents the bias term; Indicates a moment.

[0028] S1.3. Input the multi-source environmental equipment data set into the LSTM neural network to generate the refrigerant dryness prediction value, calculate the compressor power through the compressor current and voltage, and calculate the theoretical cooling capacity by associating it with the refrigerant dryness prediction value.

[0029] It should be noted that the LSTM neural network is loaded, and the multi-source environmental equipment data set is input into the LSTM neural network in time series. The time series features are extracted through the double LSTM layer and the refrigerant dryness prediction value in the range of 0 to 1 is generated; based on the three-phase current of the compressor and the DC bus voltage, the actual power of the compressor is calculated by the instantaneous power integration algorithm; the refrigerant physical property table is called by different paths, and the liquid phase enthalpy value and the gas phase enthalpy value in the refrigerant physical property table are queried according to the current condenser temperature, and the evaporator phase change latent heat value is calculated according to the liquid phase enthalpy value and the gas phase enthalpy value; the compressor exhaust enthalpy value and the compressor suction enthalpy value in the refrigerant physical property table are queried according to the exhaust pressure sensor and the suction pressure sensor, and the enthalpy difference between the compressor exhaust enthalpy value and the compressor suction enthalpy value is calculated. The actual power of the compressor and the enthalpy difference are converted into refrigerant mass flow through the power consumption-enthalpy difference equivalent conversion, and the theoretical cooling capacity is calculated according to the refrigerant mass flow, the phase change latent heat value and the refrigerant dryness prediction value. The expression is, ; in, Indicates the theoretical cooling capacity, the unit is kw; Indicates the refrigerant mass flow rate in kg / s; is the latent heat of phase change, unit is kJ / kg; Indicates the predicted value of refrigerant dryness.

[0030] S2. Calculate the benchmark energy efficiency ratio based on the theoretical cooling capacity, and make dynamic corrections based on the predicted refrigerant dryness value to generate a corrected energy efficiency ratio.

[0031] S2.1. Generate a benchmark energy efficiency ratio based on theoretical cooling capacity and compressor power, and define the optimal refrigerant dryness value based on the refrigerant type; It should be noted that the ratio of theoretical cooling capacity to compressor power is calculated to obtain the benchmark energy efficiency ratio under the current operating conditions, and the benchmark energy efficiency ratio is used as a dynamic reference value to represent the energy efficiency status of the split air conditioner; through a two-dimensional mapping table (refrigerant dryness optimal value mapping table), with the current refrigerant type and benchmark energy efficiency ratio as indexes, the corresponding refrigerant dryness optimal value in the refrigerant dryness optimal value mapping table is matched and output.

[0032] S2.2. Calculate the difference between the predicted refrigerant dryness value and the optimal refrigerant dryness value to generate a refrigerant dryness deviation, and define an attenuation coefficient based on the refrigerant type; It should be noted that the refrigerant dryness deviation is obtained by calculating the difference between the predicted refrigerant dryness value and the optimal refrigerant dryness value; the saturated gas phase density at the current exhaust pressure is obtained by querying the refrigerant physical property table through the current compressor exhaust pressure and refrigerant type. ; The theoretical value of valve control flow gain is derived based on the refrigerant flow momentum equation Formula; select five groups of operating condition combinations covering the typical operating range of split air conditioners (including rated operating condition 40℃ condensing temperature / 10℃ evaporating temperature, high load operating condition 50℃ condensing temperature / 15℃ evaporating temperature, low load operating condition 35℃ condensing temperature / 5℃ evaporating temperature, high condensing and low evaporating operating condition 50℃ condensing temperature / 5℃ evaporating temperature and low condensing and high evaporating operating condition 35℃ condensing temperature / 15℃ evaporating temperature), and keep the ambient temperature and compressor frequency constant in each group of operating condition combinations; perform an electronic expansion valve opening step response experiment in each group of operating condition combinations, increase the electronic expansion valve opening step by 10%, and record the refrigerant dryness response curve in real time, and adjust the valve control flow gain theoretical value. Until there is no overshoot and the adjustment time is ≤30 seconds, record the theoretical value of the valve control flow gain at this time As a calibration constant , all five groups of operating parameters, refrigerant types and corresponding optimal values ​​are integrated into a three-dimensional mapping table. When the split air conditioner is running, the three-dimensional mapping table is queried by the refrigerant type, the collected condensing temperature and evaporating temperature, and the Euclidean distance nearest neighbor matching algorithm is used to automatically select the calibration constant closest to the five operating conditions. , according to the calibration constant and saturated gas density Calculating the attenuation coefficient , the expression is, ; in, represents the attenuation coefficient; represents the calibration constant; represents the saturated gas phase density; Indicates the theoretical value of valve control flow gain; Indicates the cross-sectional area of ​​the electronic expansion valve port, which is set by the electronic expansion valve at the factory; Indicates the flow resistance coefficient in the electronic expansion valve, which is set by the electronic expansion valve at the factory.

[0033] S2.3. Adjust the baseline energy efficiency ratio according to the refrigerant dryness deviation and the attenuation coefficient to generate a corrected energy efficiency ratio.

[0034] It should be noted that the gain adjustment factor is obtained by performing a scalar multiplication operation on the refrigerant dryness deviation and the attenuation coefficient; the energy efficiency ratio sensitivity coefficient is calculated based on the compressor frequency using the sensitivity coefficient formula; the corrected energy efficiency ratio is calculated based on the baseline energy efficiency ratio, the gain adjustment factor and the energy efficiency ratio sensitivity coefficient, and the expression is, ; ; ; in, represents the modified energy efficiency ratio; represents the benchmark energy efficiency ratio; represents the gain adjustment factor; Indicates the refrigerant dryness deviation; represents the energy efficiency ratio sensitivity coefficient; Indicates the compressor frequency.

[0035] S3. Build a reinforcement learning decision model, input the corrected energy efficiency ratio and multi-source environmental equipment data set into the reinforcement learning decision model, and generate an optimized parameter combination.

[0036] S3.1. Define the state space based on the modified energy efficiency ratio and the multi-source environmental equipment dataset, identify the controllable parameters of the split air conditioner according to the split air conditioner design manual, define the action space based on the controllable parameters of the split air conditioner, and define the output layer based on the driver interface; It should be noted that the corrected energy efficiency ratio, real-time indoor and outdoor temperatures, current total harmonic distortion rate characteristics, evaporator inlet and outlet temperature difference, and condenser inlet and outlet temperature difference are combined in a fixed order into a 6-dimensional state vector; state space constraints are imposed on each dimension of the state vector (corrected energy efficiency ratio 1.5 to 5.0, indoor temperature 16°C to 32°C, outdoor temperature -7°C to 42°C, current total harmonic distortion rate characteristics 0% to 12%, evaporator inlet and outlet temperature difference 2°C to 15°C, condenser inlet and outlet temperature difference 3°C to 25°C) to form a closed and bounded state space S, and the linear normalization formula is used to map the corrected energy efficiency ratio and multi-source environmental equipment data set to the [-1,1] interval; according to the split air conditioner design manual, the controllable parameters of the split air conditioner are identified as the electronic expansion valve opening (0-100%) and the compressor frequency (10-120H z); based on the controllable parameters of the split air conditioner, the action space is defined as a two-dimensional continuous action vector [expansion valve opening increment, compressor frequency increment]. According to the flow control stability requirements of the electronic expansion valve and the anti-liquid hammer protection mechanism, the expansion valve opening increment range is set to -5% to +5%. According to the torque response characteristics of the compressor motor and the bearing mechanical stress limit, the compressor frequency increment range is set to -10Hz to +10Hz; based on the RS-485 drive interface protocol, the output layer is defined as two independent drive interfaces, the electronic expansion valve drive interface and the compressor drive interface. The electronic expansion valve drive interface converts the expansion valve opening increment into a 0-10V analog voltage signal through a linear mapping formula to drive the electronic expansion valve. The compressor drive interface converts the compressor frequency increment into a PWM duty cycle signal through a piecewise linear formula to drive the compressor inverter.

[0037] S3.2. Construct a reinforcement learning decision model based on the state space, action space, and output layer; It should be noted that the policy network and value network are constructed based on the state space and action space. The policy network input is a 6-dimensional normalized state vector, and it outputs the mean and standard deviation of the Gaussian distribution through two fully connected hidden layers. The value network is divided into an action value network and a state value network. The action value network input is an 8-dimensional concatenation of the state vector and the action vector, and it outputs a scalar Q value through the same hidden layer. The state value network input is a 6-dimensional state vector, and it outputs a scalar V value through the same hidden layer. Initialize the weight matrix and bias vector of the policy network and the value network. The policy network uses the policy function of the Gaussian distribution sampling action, and the value network contains the state value function and the action value function. ; In each iteration of the training cycle, the policy network randomly samples action outputs based on the current state space, collects environmental feedback data after executing physical control actions through the output layer interface, and stores the environmental feedback data in the experience replay buffer; randomly samples batch samples from the buffer to calculate the temporal difference error, updates the state value network parameters by minimizing the value loss function, and optimizes the Bellman equation to update the action value network parameters; updates the policy network parameters through policy gradients, and synchronizes the target network parameters every 1000 steps until the average reward of 10 consecutive training cycles converges to a stable variance threshold of less than 0.01 to complete the training. The expression of the reinforcement learning decision model is, ; in, represents the optimization parameter combination; represents the state-action value function; Indicates the state of the environment; represents the action vector.

[0038] S3.3. The corrected energy efficiency ratio and multi-source environmental equipment data set are merged into a state vector through the state space, and the original action parameters are generated through the action space, and the optimized parameter combination is generated through the output layer.

[0039] It should be noted that the corrected energy efficiency ratio is combined with the multi-source environmental equipment data collected in real time in a fixed order to generate a 6-dimensional original state vector; the predefined state space constraint rules are applied to perform boundary truncation processing on the original state vector; the constrained original state vector is mapped to a standardized state in the interval [-1,1] using a linear normalization method; the standardized state is input into the trained strategy network, and the normalized original action parameters [a1,a2] are output; the denormalization calculation is performed based on the expansion valve opening increment range and the compressor frequency increment range to generate an optimized parameter combination [expansion valve opening increment, compressor frequency increment]; the optimized parameter combination is converted into a physical control signal through the output layer drive interface to perform closed-loop control.

[0040] S4. Adjust the compressor frequency according to the optimized parameter combination, and trigger the electronic expansion valve control based on the predicted refrigerant dryness value.

[0041] S4.1. Extract the parameter values ​​in the optimized parameter combination, convert the compressor frequency into a PWM signal recognized by the split air conditioning control protocol, and transmit it to the inverter; It should be noted that the compressor frequency increment is extracted from the optimized parameter combination; the current operating frequency of the compressor returned by the inverter is read through the Modbus-RTU protocol; the target frequency is calculated based on the compressor frequency increment and the current operating frequency; the piecewise linear conversion rule is applied, if the target frequency is ≤80Hz, the PWM duty cycle is (target frequency / 120)×100%, and if the target frequency is >80Hz, the PWM duty cycle is 70%+0.75%×(target frequency-80); the PWM of the split air conditioner microcontroller unit is configured to generate a pulse signal with a carrier frequency of 10kHz and a duty cycle parameter of Duty_cycle; the PWM control instructions are encapsulated in the Modbus-RTU protocol format through the RS-485 communication interface, and the Modbus-RTU protocol data frame is transmitted to the compressor inverter to perform speed control.

[0042] S4.2. Define the permissible deviation threshold value based on the experimental calibration method, compare the refrigerant dryness deviation with the permissible deviation threshold value, and trigger the electronic expansion valve control based on the judgment result; It should be noted that the tested variable frequency split air conditioner is arranged in an enthalpy difference laboratory that complies with the ISO 5151 standard, and is connected to the refrigerant circulation monitoring unit, vibration sensor element and temperature and pressure acquisition unit; the liquid hammer critical point and the overheating critical point are defined by the shear strength limit of the compressor body and the Biot number deterioration criterion of the evaporator tube wall, and the allowable deviation threshold is calculated to be ±0.05 based on the minimum absolute value of the two critical points; when the refrigerant dryness deviation is satisfy When the expansion valve opening increment remains unchanged; when When , a correction increment is generated ,in Indicates the expansion valve opening increment; when When , a correction increment is generated ; According to the judgment result, the correction increment is input to the electronic expansion valve drive interface Execute electronic expansion valve control, where Indicates the opening value of the electronic expansion valve, which is obtained by decoding the valve position sensor signal.

[0043] S5. Monitor the temperature change rate and compressor power deviation in real time, and trigger the transfer learning mechanism based on the monitoring results. Update the LSTM neural network parameters and recalculate the refrigerant dryness prediction value and correct the energy efficiency ratio to generate the final optimized parameter combination.

[0044] S5.1. Collect real-time indoor temperature and compressor power, and calculate the temperature change rate and compressor power deviation; It should be noted that the original electrical signal of the indoor ambient temperature is collected by a PT1000 temperature sensor, processed into a digital temperature value by a 24-bit converter and recorded in a timing queue; the Hall current probe is used synchronously to obtain the three-phase working current waveform of the compressor, and the DC bus voltage is collected and read, and the real-time power is obtained by multiplying the current and voltage every 5 seconds; the first and last temperature values ​​50 seconds apart in the queue (the earliest and latest recording points) are extracted, and the temperature change rate is generated based on the ratio of the temperature difference to the time span of 45 seconds; the compressor rated power reference value is determined based on the operating conditions of the compressor under the ISO 5151 standard conditions of the enthalpy difference laboratory, and the compressor power deviation is obtained by calculating the real-time power and the compressor rated power reference value.

[0045] S5.2. Define the temperature change rate threshold based on the thermal comfort standard and the power deviation threshold based on the compressor safety margin; It should be noted that when calling the temperature change rate and compressor power deviation, based on the critical perception equation of human skin heat flux density, the maximum allowable change rate of 0.003℃ / s corresponding to the thermal comfort quantitative index <10% is defined as the temperature change rate threshold; based on the motor stator winding insulation thermal aging equation, ±10% of the compressor manufacturer's rated power is taken as the safety tolerance threshold boundary as the power deviation threshold.

[0046] S5.3. Determine the triggering condition of the transfer learning mechanism based on the temperature change rate threshold and the power deviation threshold, and initiate the transfer learning mechanism based on the determination result; It should be noted that the temperature change rate threshold and power deviation threshold are called, and the temperature change rate and compressor power deviation are read in real time. When it is detected that the absolute value of the temperature change rate is greater than 0.003℃ / s and the compressor power deviation exceeds ±10% for three consecutive cycles, a 32-bit trigger instruction is sent to the interrupt controller. When the 32-bit trigger instruction is received, the current compressor control thread is immediately interrupted, the compressor three-phase current waveform collected by the Hall current probe is called, the PT1000 temperature sensor is used to collect the evaporator wall temperature and exhaust pipe temperature, and the pressure sensor is used to collect the compressor exhaust pressure. The compressor's three-phase current waveform is analyzed using FFT to extract the current frequency. The evaporator wall temperature is calculated using thermal resistance compensation to generate the evaporator refrigerant temperature. The actual superheat is calculated based on the exhaust pipe temperature and the compressor exhaust pressure. The current frequency, evaporator refrigerant temperature, and actual superheat are combined to generate a three-dimensional operating condition characteristic vector. The historical optimization library is called to perform a cosine similarity comparison between the three-dimensional operating condition feature vector and the historical operating condition feature vectors stored in the historical optimization library. When a historical operating condition feature vector with a similarity greater than 85% is matched, the historical PID gain parameters and historical compressor frequency limits in the historical operating condition feature vector are extracted. The real-time compressor power deviation and evaporator refrigerant temperature are synchronously called, and new PID gain parameters are calculated based on the real-time compressor power deviation, and new compressor frequency limits are calculated based on the evaporator refrigerant temperature. The new PID gain parameters and new compressor frequency limit are transferred to the EEPROM control parameter table, and the compressor is driven synchronously. The compressor power deviation is continuously monitored within 30 seconds. If the compressor power deviation drops to within ±3%, the migration parameters are appended to the historical optimization library. Otherwise, the parameters are rolled back to the parameters before migration and the abnormal event code 0xE1 is recorded. When there is no simultaneous condition that the absolute value of the temperature change rate is greater than 0.003℃ / s and the compressor power deviation exceeds ±10% for three consecutive cycles, the real-time temperature change rate and compressor power deviation are re-read.

[0047] S5.4. Use the stochastic gradient descent method to update the LSTM neural network parameters, regenerate the refrigerant dryness prediction value, and update the corrected energy efficiency ratio; It should be noted that when the transfer learning mechanism updates the compressor frequency limit, the maximum speed range allowed by the compressor changes. The change in speed limit will directly affect the circulation speed of the refrigerant in the evaporator and condenser: when the compressor speed increases, the refrigerant flow increases, and when the speed decreases, the refrigerant flow decreases. The change in refrigerant flow causes the refrigerant evaporation rate in the evaporator and the refrigerant condensation efficiency in the condenser to change, causing the evaporator inlet and outlet temperature difference and the condenser inlet and outlet temperature difference to change; the changed evaporator inlet and outlet temperature difference, condenser inlet and outlet temperature difference, indoor and outdoor temperature, current total harmonic distortion rate characteristics and user set temperature are used together to generate a new six-dimensional input vector; when the LSTM neural network generates a refrigerant dryness prediction value based on the new six-dimensional input vector, the refrigerant dryness prediction value is compared with the refrigerant dryness optimal value, and the square of the difference is calculated as the loss amount; through The backpropagation algorithm starts from the fully connected layer of the LSTM neural network and traces back layer by layer to calculate the sensitivity of the loss to the change of the connection weights of the 128 neurons in the hidden layer, that is, the contribution of each weight parameter to the loss; according to the preset learning rate of 0.001, each weight parameter is fine-tuned in the direction of reducing the error, and the adjustment amount is the product of the sensitivity corresponding to each weight and the learning rate; in this process, 32 groups of continuous time series data are taken as calculation batches each time, and a single training is completed after 50 iterations; the updated four types of weight parameters (input gate, forget gate, output gate and memory cell weight) are permanently overwritten and stored in the flash memory to complete the LSTM neural network parameter update; the updated LSTM neural network is used to regenerate the refrigerant dryness prediction value, and the refrigerant dryness prediction value is substituted into the corrected energy efficiency ratio formula to regenerate the corrected energy efficiency ratio.

[0048] S5.5. Merge the temperature change rate and the compressor power deviation into the multi-source environmental equipment dataset, input the updated corrected energy efficiency ratio and the multi-source environmental equipment dataset into the reinforcement learning decision model, and generate the final optimized parameter combination.

[0049] It should be noted that the temperature change rate and compressor power deviation are used as new features and merged with the multi-source environmental equipment data set (indoor and outdoor temperature, current total harmonic distortion rate, evaporator / condenser inlet and outlet temperature difference and user set temperature) to generate an eight-dimensional state vector; the updated corrected energy efficiency ratio and the multi-source environmental equipment data set are obtained and input into the reinforcement learning decision model; through The greedy strategy selects the optimal action vector; the optimal action vector is decoded by the fully connected hidden layer to generate the final optimized parameter combination of the expansion valve opening increment and the compressor frequency increment; the expansion valve opening increment is written into the electronic expansion valve control register, and the compressor frequency increment is written into the inverter address.

[0050] This embodiment also provides a computer device, which is suitable for the split air conditioning energy calculation and control method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the split air conditioning energy calculation and control method proposed in the above embodiment.

[0051] The computer device may be a terminal, comprising a processor, memory, a communication interface, a display, and an input device connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores an operating system and computer programs. The internal memory provides an environment for the operating system and computer programs stored in the non-volatile storage media. The communication interface of the computer device is used to communicate with external terminals via wired or wireless communication. Wireless communication may be achieved via Wi-Fi, a carrier network, NFC (near-field communication), or other technologies. The display of the computer device may be a liquid crystal display or an electronic ink display. The input device may be a touchscreen overlay on the display, buttons, a trackball, or a touchpad on the computer device housing, or an external keyboard, touchpad, or mouse.

[0052] This embodiment also provides a storage medium having a computer program stored thereon. When the program is executed by a processor, the program implements the method for calculating and controlling the energy of a split air conditioner as proposed in the above embodiment. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0053] In summary, the present invention achieves high-precision indirect measurement by constructing an LSTM neural network to replace traditional sensors, thereby improving adaptability to dynamic working conditions, and dynamically correcting the energy efficiency ratio based on the optimal value defined by the refrigerant type, thereby avoiding the error accumulation of the fixed energy efficiency ratio due to changes in working conditions; triggering the transfer learning mechanism through dual monitoring of the temperature change rate and the compressor power deviation, and using stochastic gradient descent to update the LSTM parameters online, thereby improving the generalization ability of the LSTM network under environmental mutations.

[0054] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for calculating and controlling energy of a split air conditioner, characterized by: include, The collected multi-source environmental equipment data set is input into the LSTM neural network to generate the refrigerant dryness prediction value and calculate the theoretical cooling capacity based on the refrigerant dryness prediction value; Calculate the benchmark energy efficiency ratio based on the theoretical cooling capacity, and dynamically correct it according to the predicted value of the refrigerant dryness to generate a corrected energy efficiency ratio; Construct a reinforcement learning decision model, input the corrected energy efficiency ratio and multi-source environmental equipment data set into the reinforcement learning decision model, and generate an optimized parameter combination; Adjust the compressor frequency based on the optimized parameter combination and trigger the electronic expansion valve control based on the predicted refrigerant dryness value; The temperature change rate and compressor power deviation are monitored in real time, and the transfer learning mechanism is triggered based on the monitoring results. The LSTM neural network parameters are updated, and the refrigerant dryness prediction value and the corrected energy efficiency ratio are recalculated to generate the final optimized parameter combination.

2. The method for calculating and controlling energy of a split air conditioner according to claim 1, wherein: The multi-source environmental equipment data set includes indoor and outdoor temperatures, compressor current and voltage, evaporator inlet and outlet temperature difference, condenser inlet and outlet temperature difference, and user-set temperature.

3. The method for calculating and controlling energy of a split air conditioner according to claim 2, wherein: The specific steps of generating the predicted value of refrigerant dryness and calculating the theoretical cooling capacity according to the predicted value of refrigerant dryness are as follows: Define the six-dimensional input layer, the first LSTM layer, the second LSTM layer, and the fully connected layer to build an LSTM neural network; The multi-source environmental equipment dataset is input into the LSTM neural network to generate the refrigerant dryness prediction value; The compressor power is calculated by the compressor current and voltage, and the theoretical cooling capacity is calculated by correlating it with the predicted value of the refrigerant dryness.

4. The method for calculating and controlling energy of a split air conditioner according to claim 3, wherein: The calculation of the benchmark energy efficiency ratio and the dynamic correction based on the predicted value of the refrigerant dryness to generate the corrected energy efficiency ratio are as follows: Generate a benchmark energy efficiency ratio based on theoretical cooling capacity and compressor power, identify the refrigerant type according to the split air conditioner design manual, and define the optimal refrigerant dryness value based on the refrigerant type; Calculate the difference between the predicted refrigerant dryness value and the optimal refrigerant dryness value to generate the refrigerant dryness deviation, and define the attenuation coefficient according to the refrigerant type; The benchmark energy efficiency ratio is adjusted according to the refrigerant dryness deviation and the attenuation coefficient to generate a corrected energy efficiency ratio.

5. The method for calculating and controlling energy of a split air conditioner according to claim 4, wherein: The specific steps of generating the optimized parameter combination are as follows: Define a state space based on the modified energy efficiency ratio and a multi-source environmental equipment dataset; Identify the controllable parameters of the split air conditioner according to the split air conditioner design manual, define the action space based on the controllable parameters of the split air conditioner, and define the output layer based on the driver interface; Construct a reinforcement learning decision model based on the state space, action space and output layer; The corrected energy efficiency ratio and multi-source environmental equipment dataset are merged into a state vector through the state space, and the original action parameters are generated through the action space, and the optimized parameter combination is generated through the output layer.

6. The method for calculating and controlling energy of a split air conditioner according to claim 5, wherein: The compressor frequency is adjusted according to the optimized parameter combination, and the electronic expansion valve control is triggered based on the refrigerant dryness prediction value. The specific steps are as follows: Extract the parameter values ​​in the optimized parameter combination, convert the compressor frequency into a PWM signal recognized by the split air conditioning control protocol, and transmit it to the inverter; The allowable deviation threshold is defined according to the experimental calibration method, the refrigerant dryness deviation is compared with the allowable deviation threshold, and the electronic expansion valve control is triggered according to the judgment result.

7. The method for calculating and controlling energy of a split air conditioner according to claim 6, wherein: The real-time monitoring of the temperature change rate and the compressor power deviation, and triggering the transfer learning mechanism according to the monitoring results, are as follows: Collect real-time indoor temperature and compressor power, calculate temperature change rate and compressor power deviation; Define temperature change rate thresholds and power deviation thresholds based on thermal comfort standards and compressor safety margins; The triggering condition of the transfer learning mechanism is determined according to the temperature change rate threshold and the power deviation threshold, and the transfer learning mechanism is started according to the determination result.

8. The split air conditioner energy calculation and control method according to claim 7, characterized in that: The specific steps for generating the final optimization parameter combination are as follows: The stochastic gradient descent method is used to update the LSTM neural network parameters, regenerate the refrigerant dryness prediction value, and update the corrected energy efficiency ratio; Incorporating the temperature change rate and the compressor power deviation into the multi-source environmental equipment dataset to generate a corrected multi-source environmental equipment dataset; The updated corrected energy efficiency ratio and corrected multi-source environmental equipment dataset are input into the reinforcement learning decision model to generate the final optimized parameter combination.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the split air conditioning energy calculation and control method described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the split air conditioning energy calculation and control method according to any one of claims 1 to 8 are implemented.

Images