A heat pump multi-mode cooperative intelligent control method and system for heat storage and defrosting

Abstract

The application provides a heat pump multi-mode cooperative heat storage defrosting intelligent control method and system, which comprises the following steps: constructing a system state vector by collecting external environment temperature, heat storage pool outlet temperature, evaporator average temperature and circulation flow; adaptively selecting heat storage defrosting, mixed defrosting or maintaining heating mode by using a reinforcement learning model; dynamically setting a heat storage pool target outlet temperature and adjusting a three-way valve opening degree according to the selected mode; monitoring a thermal stability factor and an evaporator temperature response rate in real time during the defrosting process, and adjusting the temperature or switching the mode if necessary; calculating an energy efficiency index based on the evaporator temperature rise effect and the system energy consumption after the defrosting is completed, and feeding back the energy efficiency index for strategy optimization. The application realizes efficient, stable and low disturbance operation of the defrosting process, and significantly improves the energy efficiency and reliability of the heat pump system in a low-temperature and high-humidity environment.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of heat pumps, and particularly relates to a heat pump multi-mode coordinated heat storage defrosting intelligent control method and system. BACKGROUND

[0002] At present, as an energy-saving device widely used in cold and heat combined supply and domestic hot water systems, an air source heat pump generally faces the problems of serious frosting on the surface of an evaporator, frequent defrosting and low efficiency when operating in winter. Especially under the condition of low external environment temperature and high humidity, rapid frosting of the evaporator will lead to a decrease in heat exchange capacity, deterioration of energy efficiency ratio, and even cause interruption of equipment operation or damage to components. The existing system mostly adopts a defrosting strategy of timing control or based on a temperature threshold, and has problems of response lag and rough control, and cannot effectively adapt to dynamic changes of different operating loads and external environments. In addition, the mainstream heat pump system usually uses a water tank as a defrosting heat source, and during defrosting, heat is extracted from the water tank to supply the evaporator for reverse heating, which causes a sharp fluctuation of the temperature of the water tank, affecting the stable supply of domestic hot water. Due to the lack of heat storage buffering and energy fine management mechanism, the current heat pump system generally has the problems of high energy consumption, long defrosting time, slow control response and the like in the defrosting control process. Especially in the aspect of multi-mode coordinated control, the existing method cannot dynamically determine the optimal defrosting path according to the real-time state, and it is also difficult to realize stable switching between different defrosting modes, causing problems of unreasonable mode selection, chaotic heat source configuration and intensified system fluctuation and the like. SUMMARY

[0003] The purpose of the present application is to design a heat pump multi-mode coordinated heat storage defrosting intelligent control method and system, which can improve the defrosting efficiency and energy efficiency level and solve the problems of slow response of the heat source of the heat storage pool, excessive temperature disturbance of the water tank and discontinuous mode switching.

[0004] In order to achieve the above purpose, in a first aspect of the present application, a heat pump multi-mode coordinated heat storage defrosting intelligent control method is provided, which comprises the following steps:

[0005] Collecting external environment temperature, heat storage pool outlet temperature, evaporator average temperature and heat storage pool circulation flow to form a system state vector;

[0006] Based on the system state vector, determining the current optimal defrosting mode through a reinforcement learning model, wherein the defrosting mode includes a heat storage pool defrosting mode, a mixed defrosting mode or a maintenance heating mode;

[0007] According to the selected defrosting mode, dynamically setting a target outlet temperature of the heat storage pool, and adjusting the opening degree of a three-way valve to control the heat distribution path;

[0008] During the defrosting execution, the heat stability factor of the heat storage pool and the temperature response rate of the evaporator are monitored in real time. If the heat release is unstable or the defrosting response is slow, the target outlet temperature is dynamically adjusted or the mixed defrosting mode is switched to.

[0009] After the defrosting operation is completed, the energy efficiency index is calculated based on the evaporator temperature recovery effect and the total energy consumption of the system, and the defrosting result is marked and fed back to the reinforcement learning model for strategy optimization.

[0010] Further, the system state vector is collected in real time by sensors arranged at the outdoor air inlet, the heat storage coil outlet, multiple points on the evaporator surface, and the main branch pipe of the heat storage circuit, and is formed after sliding average and low-pass filtering processing.

[0011] Further, the reinforcement learning model is an improved deep Q network, which introduces a temperature difference constraint term and a heat storage pool flow reverse constraint term in the decision-making process to prevent energy sudden change and ensure the rationality of mode selection under low flow conditions.

[0012] Further, when the heat storage defrosting mode is selected, the three-way valve is switched to full open heat storage branch; when the mixed defrosting mode is selected, the three-way valve is simultaneously turned on according to the preset proportion of the heat storage pool and the water tank branch; when the heating maintenance mode is selected, the heat storage pool enters the heat preservation or energy storage state.

[0013] Further, the target outlet temperature is dynamically corrected according to the current evaporator temperature and its change rate, and the heat release intensity is smoothly adjusted through a nonlinear response function to avoid sharp fluctuations in the evaporator temperature.

[0014] Further, the heat stability factor is calculated based on the heat storage pool flow and the outlet temperature deviation, which is used to evaluate whether the current heat release process is in a stable state, and accordingly to link the heat pump frequency and the three-way valve opening degree.

[0015] Further, when the evaporator temperature response rate continues to be lower than the empirical threshold, the heat storage compensation mechanism is triggered to temporarily increase the target outlet temperature of the heat storage pool without changing the current defrosting mode.

[0016] Further, the energy efficiency index is the normalized value of the evaporator temperature recovery quality per unit energy consumption, which is used to determine whether the current defrosting is successful, and serves as the basis for subsequent strategy fine-tuning.

[0017] Further, after each defrosting operation is completed, the defrosting start and end time, the maximum temperature rise rate of the evaporator, the energy efficiency index, and the determination result of whether the defrosting is successful are written into the structured log, which is used for subsequent retraining of the reinforcement learning model and maintenance diagnosis of the system running state.

[0018] In a second aspect of the present application, a heat pump multi-mode coordinated heat storage defrosting intelligent control system is provided, the system comprising:

[0019] a data acquisition module for acquiring external environment temperature, heat storage pool outlet temperature, evaporator average temperature and heat storage pool circulation flow, forming a system state vector;

[0020] a learning decision module for determining the current optimal defrosting mode based on the system state vector through a reinforcement learning model, the defrosting mode including heat storage pool defrosting mode, mixed defrosting mode or maintaining heating mode;

[0021] a temperature control and adjustment module for dynamically setting the target outlet temperature of the heat storage pool according to the selected defrosting mode, and adjusting the three-way valve opening degree to control the heat distribution path;

[0022] a process control module for monitoring the heat stability factor of the heat storage pool and the evaporator temperature response rate in real time during the defrosting execution process, and dynamically adjusting the target outlet temperature or switching to the mixed defrosting mode if unstable heat release or slow defrosting response is detected;

[0023] an energy efficiency monitoring module for calculating the energy efficiency index based on the evaporator temperature recovery effect and the total system energy consumption after the defrosting operation is completed, and feeding back the defrosting result to the reinforcement learning model for strategy optimization.

[0024] The present application has at least the following beneficial technical effects:

[0025] To solve the above problems, the application provides a heat pump multi-mode coordinated heat storage defrosting intelligent control method and system, which constructs a dynamic closed loop system from data sensing, mode decision to heat control execution by intelligently adjusting the heat release strategy of the heat storage tank. The system first collects real-time temperature and flow data through multiple sensors deployed in the external environment, the evaporator and the heat storage circuit to form a state vector for strategy input, then constructs a state-action-reward mapping relationship in the training stage by the reinforcement learning model, and outputs the optimal defrosting mode selection and action value evaluation under the current system state in the running time. The system switches to the heat storage defrosting, mixed defrosting or heating maintenance mode in real time according to the model output, and dynamically sets the heat storage tank outlet temperature and three-way valve adjustment strategy according to the defrosting priority. In the heat control execution process, the system introduces an estimation of the heat storage and heat release stability and a dynamic detection mechanism of the evaporator response speed to adjust the target temperature output in real time, thereby ensuring the stability of heat supply in the defrosting process and the smooth recovery of the evaporator surface temperature. Finally, the system calculates the defrosting effect index under unit energy consumption after each defrosting cycle, and records whether the current defrosting behavior is successful through the result marking mechanism to provide reliable feedback for long-term strategy optimization. The application breaks through the response lag problem of the traditional rule-based control method by constructing a four-level linkage mechanism of multi-mode defrosting strategy selection, heat energy regulation response, dynamic behavior compensation and energy efficiency evaluation, significantly improves the defrosting efficiency and energy efficiency level, solves the problems of heat source response lag of the heat storage tank, excessive temperature disturbance of the water tank and discontinuous mode switching, and can be widely applied to defrosting optimization control of heat pump systems in various cold regions. BRIEF DESCRIPTION OF DRAWINGS

[0026] The application is further described by the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the application. Other drawings can also be obtained by those skilled in the art without creative labor on the basis of the following drawings.

[0027] Figure 1 A flow chart of the heat pump multi-mode coordinated heat storage defrosting intelligent control method of the application.

[0028] Figure 2 A framework diagram of the heat pump multi-mode coordinated heat storage defrosting intelligent control system of the application. DETAILED DESCRIPTION

[0029] The embodiments of the application are described in detail below, and examples of the embodiments are shown in the drawings, wherein the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the application, and cannot be understood as a limitation on the application.

[0030] In one or more embodiments, asFigure 1 As shown, an intelligent control method for heat pump multi-mode coordinated heat storage defrosting is disclosed, the method comprising the following:

[0031] S1: Collecting external environment temperature, heat storage pool outlet temperature, evaporator average temperature and heat storage pool circulation flow to form a system state vector;

[0032] Specifically, this step realizes real-time acquisition of core variables required by defrosting control logic through a simplified and high-correlation data acquisition mechanism. The data sources are limited to four physical quantities closely related to the heat pump defrosting process: external environment temperature (Tamb), heat storage pool outlet temperature (Tout), evaporator average temperature (Tave) and heat storage pool circulation flow (Q). To ensure controllability and real-time performance of the acquisition process, all data are acquired by dedicated sensors arranged at key positions and synchronously input into the central control unit for subsequent mode selection and energy management.

[0033] Ambient temperature Acquired by a temperature sensor installed near the air inlet channel on the outdoor side of the heat pump, a digital temperature probe such as DS18B20 is recommended. When arranging, heat source interference and direct sunlight should be avoided. The sampling period is controlled at one second, and is stably written into the controller buffer area for determining whether the defrosting condition has the initial triggering condition.

[0034] Heat storage pool temperature Measured by a thermistor arranged at the outlet of the heat storage coil, connected to the analog acquisition port of the microcontroller, an NTC series probe is recommended. The measured value is used to determine whether the current heat storage has heat release capability. After data acquisition, it needs to be processed by a sliding window average to eliminate fluctuation noise caused by thermal inertia.

[0035] Evaporator temperature The collection needs to cover multiple representative positions on the surface of the heat exchanger. Usually, three points are arranged, such as K-type thermocouples, and the data is input through a multi-channel signal input module. The controller performs mean value processing on the data of each measuring point to reflect whether the current evaporator is in a frosting critical state. To enhance stability, a first-order low-pass filter is recommended for the average value to weaken short-period disturbances caused by refrigerant flow.

[0036] Heat storage pool flow Acquired by an electromagnetic flowmeter installed on the heat exchange branch, the device should support standard analog signal output or Modbus RTU communication, and the output is used to infer the current heat transfer efficiency of the heat exchange branch. If the flow is below the set threshold, it will affect whether the heat storage defrosting strategy is allowed to be executed subsequently.

[0037] ​​​The four types of key data collected are structured by the controller and form a state vector , which is called by the subsequent reinforcement learning model. The relationship can be expressed as:

[0038] ;

[0039] wherein is the state vector, representing the current thermal state of the system, is the data processing function, including sampling, moving average, filtering, synchronization packaging and other processing processes. The variable meanings are as follows: is the ambient temperature collected by the sensor at the outdoor air inlet, reflecting the change of climate conditions; is the outlet temperature of the heat storage coil, which judges the current heat release capacity of the heat storage tank; is the average temperature of multiple points on the evaporator surface, which is used to judge whether there is frost layer influence; is the actual flow rate obtained through the heat storage circuit main branch flow meter, which is used to judge the current heat availability.

[0040] Finally, the system outputs the structured state data vector and a flag bit , to indicate whether the current data collection is all valid. For example, in the case of sensor drop or data sampling loss, the flag will be set to invalid, so that the subsequent module can temporarily suspend the strategy update. The core of this step is to retain only the minimum variable set directly related to the defrosting performance of the heat pump, and to complete the formation of the data vector by accurately arranging the sensor chain that can be realized by engineering.

[0041] S2: Based on the system state vector, the optimal defrosting mode is determined by the reinforcement learning model, including heat storage tank defrosting mode, mixed defrosting mode or maintaining heating mode;

[0042] Specifically, the core task of this step is to receive the system state vector output in step one, to realize intelligent judgment and selection of the heat pump system in multiple working modes by constructing a mode selection network based on reinforcement learning, so as to obtain the optimal balance between defrosting, energy distribution and system stability. This step couples the heat pump running state with the dynamic behavior of the heat storage tank, so that the system can autonomously select the best defrosting path under complex environmental conditions, and realize the maximization of energy utilization rate and dynamic optimization of thermal efficiency.

[0043] The input is the state vector formed in the previous step, including the external environment temperature , the heat storage tank temperature , the evaporator temperature and the heat storage tank flow These data are provided by sensors in real time, filtered and time-synchronized, and then input to the reinforcement learning decision module. The structure of this module consists of a perception layer, a state evaluation layer, and an action decision layer, which runs on an embedded controller and performs decision updates periodically (about every few seconds). The reinforcement learning framework uses an improved deep Q network (DQN) structure, combined with a "temperature balance constraint term" and a "heat storage utilization rate regularization term" for the heat pump system, to ensure the executability and energy efficiency orientation of the decision under actual engineering conditions.

[0044] Based on the traditional DQN, the model calculates the comprehensive score function of the current state in the state evaluation stage:

[0045] ;

[0046] wherein, is the long-term return estimate value of taking action in state ; is the immediate reward function, representing the defrosting efficiency and energy efficiency score under this mode, obtained from historical operation data and expert experience statistics; is the temperature difference constraint term of the evaporator and the heat storage pool, used to prevent energy jumps or evaporator supercooling phenomena during heat conversion; is the heat storage pool flow reverse constraint term, which ensures that the model tends to select the heat source stable mixing mode to prevent energy retention in low flow states; and are experience adjustment parameters, representing the relative weights of the two constraints, is a small constant to prevent the denominator from being zero.

[0047] On this basis, the model updates the optimal action of the current strategy through action value iteration:

[0048] ;

[0049] wherein, is the optimal mode selection, , , correspond to the heat storage defrosting mode, the mixed defrosting mode, and the maintenance heating mode, respectively. represents the expected energy change rate of the heat storage pool after executing a certain mode, calculated by the system thermal model, and is used to evaluate the support capacity of the remaining energy of the heat storage pool for the next round of defrosting demand in advance. The coefficient is used to adjust the influence of energy reserves in the decision-making process. When the energy of the heat storage pool is too low, this term will automatically reduce the probability of selecting high-energy consumption modes, thereby maintaining stable energy supply in the long-term control process.

[0050] The actual execution logic of the decision-making process is as follows: the reinforcement learning module receives the updated state from the sensing layer at each control cycle , inputs the normalized and state feature extracted state into the network model, calculates the values of all possible actions, and selects the current optimal action after adding the energy dynamic term . The controller then executes the actual mode switching according to the output action. For example, when the external temperature is low and the evaporator temperature drops too fast, the system judges that the term is large, the reward function weight is high, and the output is , i.e., entering the heat storage tank defrosting mode; if the external temperature is slightly high and the energy reserve of the heat storage tank is sufficient, the mixed defrosting mode may be selected ; when the system temperature gradient is relatively flat, the is maintained, and the regular heating continues to avoid unnecessary mode switching.

[0051] In order to adapt to the input distribution shift caused by climate change in different regions, a lightweight parameter update mechanism is introduced in the embedded control system. The control system records the average defrosting cycle data of the past 24 hours to adaptively correct the and in the network parameters, ensuring that the model maintains the energy efficiency optimization trend in the long-term operation. For example, in a high-humidity environment, the system automatically increases to strengthen the temperature difference constraint and prevent excessive defrosting caused by sudden evaporator temperature drop; in the scene with large flow fluctuation, the is increased to maintain energy transmission balance.

[0052] Finally, the reinforcement learning module outputs two results: the optimal action , i.e., the most suitable defrosting mode at the current time; and the corresponding action value , which is an evaluation index of the current energy efficiency state of the system and is passed to the next heat storage tank temperature control module. This output forms a closed loop from data to strategy to execution in the entire control process, enhancing the system's adaptive ability and environmental response speed.

[0053] This step introduces a specific constraint reinforcement mechanism for heat pump systems, which embeds temperature difference balance and flow constraint terms in the reinforcement learning network, allowing the decision-making process to reflect the real thermophysical behavior. At the same time, the energy dynamic adjustment term is added, making the control strategy predictive of the future heat storage state. Compared with traditional defrosting logic based on static rules, this model can achieve dual optimization of defrosting efficiency and energy utilization rate under multi-mode coordination conditions, and reduce the number of defrosting and system energy consumption in actual operation, thereby significantly improving the long-term operation performance of the heat pump system.

[0054] S3: According to the selected defrosting mode, dynamically set the target outlet temperature of the heat storage pool, and adjust the three-way valve opening to control the heat distribution path;

[0055] Specifically, this step is based on the optimal mode selection output by the reinforcement learning module in step two and the corresponding state value , the temperature control strategy of the heat storage pool and the heat energy release mode are dynamically refined. This step plays a key role in bridging the "intelligent strategy decision" and "physical energy scheduling" in the entire heat pump system, and is the physical fulcrum for the multi-mode cooperative defrosting control system to execute. It not only reflects in the timing and amplitude control of the heat release behavior of the heat storage pool, but also reflects in the structural optimization of energy release, so that the system has self-adaptive response capability to different defrosting modes and environmental states, avoiding the problems of defrosting failure, energy efficiency decline or evaporator supercooling caused by the lag of heat storage regulation in traditional systems.

[0056] The input variables received by this step and come from the reinforcement learning strategy selection module of the previous step. Among them is the defrosting operation mode selected by the system under the current working condition, and its value includes (heat storage defrosting), (mixed defrosting) or (keep heating); is the long-term value estimate of the strategy, which is used to assist the adjustment of the energy distribution aggressiveness in temperature control. In addition to the above two inputs, the system also synchronously calls the temperature variables and collected in real time in step one , and the heat storage pool flow to form a complete heat scheduling state field.

[0057] In terms of control structure, this step operates with the core logic of "mode-driven-temperature difference correction-energy response". First, according to the value of , determine the basic heat release behavior path: when , the system should take the heat storage pool as the main heat source, and the three-way valve is switched to the full open heat storage branch; when , the three-way valve is opened to 50%-50%, part of the heat comes from the water tank and part comes from the heat storage pool; when , the system does not perform defrosting operation, and the heat storage pool is in the heat preservation or energy storage state.

[0058] Second, to prevent the problem of "thermal jump" in the process of switching the heat source, that is, the heat storage pool releases too much heat in a short time, causing the evaporator temperature to fluctuate sharply, this step introduces a class of temperature difference adjustment terms to control the heat release rate target:

[0059] ;

[0060] where, represents the target outlet temperature of the thermal storage tank; is the current evaporator temperature (from step one); is the minimum heating temperature difference set empirically to prevent the evaporator from freezing, usually a constant between 1.5 and 2.5; is the rate of change of the evaporator temperature over two time steps; is a compression-type response function used to smooth the system response; and are adjustable coefficients that determine the sensitivity and upper limit of the temperature difference response.

[0061] This formula avoids excessive heat release by the system due to sudden temperature differences through the nonlinear dynamic response function , and automatically increases the target output temperature of the thermal storage tank when the evaporator temperature drops rapidly ( ), achieving heat feedforward regulation in the temperature control strategy. This response structure is particularly suitable for the characteristics of the evaporator and the thermal storage tank in the system, which have no direct coupling feedback, allowing the system to have a "self-stable" heat response capability.

[0062] In terms of heat distribution, to ensure the stability and efficiency of heat energy output, this step no longer directly controls the flow, but introduces a "thermal stability factor of the thermal storage tank" as an effective estimate of the current controllable heat of the thermal storage system, used to control the coupling relationship between the heat pump operating frequency and the three-way valve opening:

[0063] ;

[0064] where, is the current flow of the thermal storage tank (already collected in step one); is the deviation between the actual outlet temperature and the current target temperature; is the controller's empirical coefficient. This index reflects whether the current heat release process is in a stable state. If tends to a constant or increases, it indicates that the thermal storage tank can sustain stable heat release; otherwise, if decreases significantly, it means that the temperature target is too high or the heat is insufficient, and the system should automatically reduce the heat release rate or enter the next charging cycle early.

[0065] The system adjusts the three-way valve angle and the heat pump main cycle frequency according to the trend of : when increases, the heat pump frequency is increased to match the heat release; when decreases, the frequency is reduced to avoid energy retention. In addition, This step plays the role of weight adjustment. If the current defrosting strategy has a high estimated value (i.e., a large long-term benefit), the system allows more aggressive short-term energy expenditure; if is lower, the energy-saving control strategy is entered.

[0066] Finally, this step outputs two quantities: one is , which is the heat storage pool control target temperature of the heat pump system in the current period, used to control the three-way valve hot water path; the other is , which is the core index for the heat source stability judgment of the subsequent dynamic mode control module.

[0067] S4: During the defrosting execution process, the heat storage pool thermal stability factor and the evaporator temperature response rate are monitored in real time. If it is detected that the heat release is unstable or the defrosting response is slow, the target outlet temperature is dynamically adjusted or switched to the mixed defrosting mode;

[0068] Specifically, after completing the reinforcement learning strategy selection and heat storage pool temperature control output, this step receives the control instructions and heat storage state evaluation indexes from the previous stage, and further executes real-time adjustment and thermal efficiency optimization control of the defrosting mode. The operation goal is to dynamically adjust the execution rhythm and heat output level of the defrosting strategy according to the current running feedback of the system without retraining the strategy or modifying the decision structure. This step is the key bridge from "strategy output" to "system landing", ensuring that the control system has sufficient real-time adaptability and anti-disturbance ability to complex running environments, solving the problem that traditional defrosting strategies cannot respond to dynamic thermal disturbances.

[0069] The input data of this step mainly includes: the heat storage pool target outlet temperature output by the previous step, the heat stability factor , and the long-term value of the strategy action . In addition, the evaporator temperature from step one is continued to be cited to form a complete dynamic response analysis basis. is the expected heat release target calculated by step three according to the evaporator temperature state and defrosting rate history; reflects whether the current heat release is smooth and continuous, which is calculated by coupling the heat storage pool flow and outlet temperature deviation; and indicates the importance of the current execution strategy in long-term energy efficiency evaluation, which directly participates in the control intensity adjustment of this step.

[0070] The dynamic adjustment of the defrosting mode first relies on the continuous monitoring and judgment of . When the heat pump enters the heat storage pool defrosting mode ( ), if it is detected that is less than the threshold , it is considered that the current heat storage tank heat dissipation is unstable or insufficient, at which time the system automatically switches to the mixed defrosting mode , and the three-way valve angle is adjusted to 50% at the same time, so that part of the heat is provided by the water tank to ensure the defrosting persistence. This judgment logic is realized by interrupt judgment in the embedded controller, and the real-time data obtained by the electromagnetic flow meter (such as E+HPromag) and temperature sensor (such as NTC thermistor) at the outlet of the heat storage tank are calculated , and the comparison and logical judgment are completed by the controller in the main loop task.

[0071] In the dynamic execution process, in order to further improve the delicacy of the heat response adjustment, the system introduces an "energy efficiency fluctuation coefficient" , which is used to evaluate the deviation between the current evaporator response speed and the output capacity of the heat storage tank. The calculation of this parameter is based on the multi-point thermocouple sensors (such as K-type thermocouples) installed at the evaporator inlet and the surface, which record at a frequency of 1 second, and the temperature change rate is obtained in combination with the previous period data. In combination with , the system calculates:

[0072] ;

[0073] This index essentially reflects the speed ratio between the heating target and the actual response. When continues to exceed the empirical threshold (reversed from the average defrosting response curve of the previous day), it indicates that the evaporator response is slow or the frost layer is too thick and not effectively removed, and the system will trigger the "heat storage compensation mechanism" to temporarily increase the output temperature target of the heat storage tank without changing the mode. This logic will avoid the waste of energy consumption or the repeated problem of frosting on the surface of the heat exchanger caused by misjudgment and premature switching of the defrosting mode.

[0074] The specific execution logic of this compensation mechanism is represented by the following formula:

[0075] ;

[0076] Where, is the weighted average in the past three control periods, which is output after being processed by the filter module built-in the controller; is the adjustment amplitude limiting factor to prevent temporary temperature rise too fast causing system impact; The inverse Sigmoid structure of introduces the "execution guarantee weight" of the current strategy value, which guarantees that high-value strategies will be given priority in system compensation resources when encountering heat response blockage, while the fluctuation of low-value strategies will not cause unnecessary system adjustment.

[0077] Adjustment result The three-way valve angle setting and the heat pump speed control module are applied by the PWM control logic in the controller. In actual execution, the system maintains the heat storage pool outlet temperature near the target within a tolerance interval of ±0.3, and realizes temperature stability by controlling the heat pump frequency and flow linkage.

[0078] Finally, the system outputs two key values: one is , which is the dynamic corrected heat storage pool output target temperature in this period, and controls the three-way valve and pump speed; the other is , which is the current defrost mode state flag of the system, indicates the current mode, and if the mode is switched, it is automatically updated and written into the log system for subsequent performance backtracking and strategy fine-tuning. Through this step, the strategy decision is converted from static execution to dynamic closed-loop behavior, which not only retains the global optimization ability of the reinforcement learning strategy, but also introduces system anti-disturbance, fast response and energy efficiency optimization performance through the dynamic adjustment mechanism of the heat release path. Especially and joint judgment provides a low-overhead, high-real-time energy efficiency feedback mechanism, so that the system has intelligent defrosting control capability in long-term operation without complex secondary training. This mechanism enhances the robustness and adaptability of the heat pump system under actual working condition changes.

[0079] S5: After completing the defrosting operation, calculate the energy efficiency index based on the evaporator temperature recovery effect and the total energy consumption of the system, and feed back the defrosting result to the reinforcement learning model for strategy optimization;

[0080] Specifically, based on the heat storage pool dynamic adjustment temperature target output in step four and the current defrost mode state identifier , the specific control of the defrosting operation is completed at the execution end, and energy efficiency monitoring and feedback generation are simultaneously performed, which is the final implementation link of "from control intention to physical action" in the entire technical solution. Its task is not only to mechanically start a certain defrosting mode, but also to efficiently, stably and low-redundantly implement the control strategy based on consideration of heat storage capacity, thermal response characteristics and external environmental changes, while evaluating the energy efficiency performance of this defrosting process through multi-dimensional data acquisition and index calculation. The input variables of this step are ​is the target outlet temperature of the thermal storage tank, which is derived after the optimization of the previous four steps. It is the main setting parameter for heat control during defrosting. This temperature target value will be directly transmitted to the PID or fuzzy controller of the control system, which will control the angle of the three-way valve, the speed of the electric pump, and the water flow state in the heat exchanger. This value is refreshed every second by the controller and compared with the real-time temperature of the outlet of the thermal storage tank, forming a temperature control closed loop. The three-way valve device is recommended to use an electric regulating valve that supports proportional control (such as the BELIMOTR series), and the pump speed control unit is recommended to use a brushless DC water pump module with a PWM interface, which can achieve 0~100% continuous adjustment. Real-time outlet temperature is provided by the PT100 temperature probe installed in the outlet pipe of the thermal storage tank, with an accuracy of better than 0.2℃. The data is sampled by the ADC module and uploaded to the main control system.

[0081] is the current defrosting operation mode of the system, which is a logical variable with discrete values ( ), where indicates that the system is in the defrosting mode of the thermal storage tank, and the main heat source is the thermal storage tank; indicates that the system enters the mixed defrosting mode, and the heat comes from the combination of the thermal storage tank and the water tank; indicates that the system is in a non-defrosting state, i.e. no frost layer is detected or the rate of change of the evaporator temperature is not sufficient to trigger the defrosting judgment logic. In actual control, controls the four-way reversing valve, the electromagnetic valve on-off state, and the defrosting timing module. The logic execution is completed by the state machine in the PLC control system, and each state corresponds to different IO combination output.

[0082] During the execution of the control logic, to prevent excessive defrosting or incomplete defrosting, a set of defrosting process effectiveness monitoring mechanism based on "energy efficiency response" judgment is set in this step. This mechanism is based on the trend of the evaporator surface temperature and the time integral of the total power consumption of the system to build a kind of thermal efficiency index , which is used to reflect the thermal recovery quality of the defrosting behavior under unit energy consumption:

[0083] ;

[0084] where, is the target evaporator temperature reference value, which comes from the median value of the temperature curve in the "no frost state" in the system historical data; is the real-time evaporator surface temperature, which is collected by sensors installed at multiple representative positions of the evaporator. It is recommended to install at least 3 K-type thermocouples (top, middle, bottom) to take the average value, with a sampling period of 1s; This represents the total electrical power during system operation. The data comes from a three-phase power meter (such as a Schneider PM5560) installed on the power supply side and is uploaded via the Modbus protocol. Integration time period. A complete defrosting cycle is marked when the system starts defrosting and closed at the end.

[0085] This indicator uses "evaporator temperature recovery quality" as the input for higher-level judgment and normalizes it using energy consumption as the denominator, thus solving the problem of traditional "whether a certain temperature has been reached" defrosting judgment lacking system cost awareness. System Settings As an upper limit for experience reference, when This indicates that the energy consumption per unit during this defrosting process was too high, and it needs to be appropriately reduced in the next defrosting control cycle. The upper limit can be set, or the defrosting mode can be limited to continue in the next defrosting cycle. To prevent policy misactivation, the system remains unchanged. In this step, to simplify the overreaction problem caused by threshold fluctuations, the system... A sliding window smoothing process is employed, and the average value of three consecutive defrosting cycles is used as a reference for fine-tuning the current control strategy. Furthermore, to enhance the traceability of system operation, this step records the start time, end time, maximum evaporator temperature recovery rate, and final defrosting success determination result for each defrosting operation. (1 indicates success, 0 indicates failure) is written to the built-in memory to form a structured log of the defrosting process data. This data will be used for future reinforcement learning model retraining (step two) and system maintenance diagnosis.

[0086] In one or more embodiments, such as Figure 2 As shown, a multi-mode coordinated heat pump thermal storage defrosting intelligent control system is disclosed, the system comprising:

[0087] The data acquisition module is used to collect external ambient temperature, heat storage tank outlet temperature, evaporator average temperature, and heat storage tank circulation flow rate to form a system state vector.

[0088] The learning decision module is used to determine the current optimal defrosting mode based on the system state vector through a reinforcement learning model. The defrosting mode includes a thermal storage tank defrosting mode, a hybrid defrosting mode, or a heating maintenance mode.

[0089] The temperature control module is used to dynamically set the target outlet temperature of the heat storage tank according to the selected defrosting mode, and adjust the opening of the three-way valve to control the heat distribution path.

[0090] The process control module is used to monitor the thermal stability factor of the heat storage tank and the temperature response rate of the evaporator in real time during the defrosting process. If unstable heat release or slow defrosting response is detected, the target outlet temperature is dynamically adjusted or the system is switched to a hybrid defrosting mode.

[0091] The energy efficiency monitoring module is configured to calculate an energy efficiency index based on the evaporator temperature recovery effect and the total energy consumption of the system after the defrosting operation is completed, and to feed back the defrosting result mark to the reinforcement learning model for policy optimization.

[0092] It is worth noting that the specific working process of the heat pump multi-mode cooperative heat storage defrosting intelligent control system provided by the embodiment of the present application is the same as the process of the heat pump multi-mode cooperative heat storage defrosting intelligent control method described in the above embodiment, and will not be repeated here.

[0093] The embodiment of the present application also provides a heat pump multi-mode cooperative heat storage defrosting intelligent control device, which comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the steps in the above-mentioned heat pump multi-mode cooperative heat storage defrosting intelligent control method embodiment, such as steps S1-S5 described in the above embodiment. Figure 1 Or, the processor implements the functions of the modules in the above-mentioned system embodiments when executing the computer program.

[0094] For example, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the heat pump multi-mode cooperative heat storage defrosting intelligent control device.

[0095] The heat pump multi-mode cooperative heat storage defrosting intelligent control device can be a desktop computer, a notebook computer, a palm computer and a cloud server, etc. The heat pump multi-mode cooperative heat storage defrosting intelligent control device can include, but is not limited to, a processor, a memory. Those skilled in the art can understand that the heat pump multi-mode cooperative heat storage defrosting intelligent control device can also include input and output devices, network access devices, buses, etc.

[0096] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASAC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or can also be any conventional processor, etc. The processor is the control center of the heat pump multi-mode cooperative heat storage defrosting intelligent control device, and connects various parts of the heat pump multi-mode cooperative heat storage defrosting intelligent control device through various interfaces and lines.

[0097] The memory can be used to store the computer programs and / or modules, and the processor realizes various functions of the heat pump multi-mode cooperative heat storage defrosting intelligent control device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, etc.; and the data storage area can store data created according to the running of the air conditioner controller, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0098] If the modules of the heat pump multi-mode cooperative heat storage defrosting intelligent control device are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware. The computer program can be stored in a computer readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0099] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware. The program can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned various method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0100] The above is the preferred embodiment of the present application. It should be noted that those skilled in the art can make several improvements and refinements without departing from the principles of the present application. These improvements and refinements are also considered within the scope of protection of the present application.

Claims

1. A heat pump multi-mode cooperative heat storage defrosting intelligent control method, characterized in that, The method comprises: Collecting external environment temperature, heat storage pool outlet temperature, evaporator average temperature and heat storage pool circulation flow to form a system state vector; Based on the system state vector, the current optimal defrosting mode is determined by a reinforcement learning model, and the defrosting mode includes heat storage pool defrosting mode, mixed defrosting mode or maintaining heating mode; According to the selected defrosting mode, the target outlet temperature of the heat storage pool is dynamically set, and the three-way valve opening degree is adjusted to control the heat distribution path; when the heat storage pool defrosting mode is selected, the three-way valve switches to full opening of the heat storage branch; when the mixed defrosting mode is selected, the three-way valve simultaneously conducts the heat storage pool and the water tank branch according to the preset proportion; when the maintaining heating mode is selected, the heat storage pool enters the heat preservation or energy storage state; During the defrosting execution process, the heat stability factor of the heat storage pool and the evaporator temperature response rate are monitored in real time, and if the heat release is unstable or the defrosting response is slow, the target outlet temperature is dynamically adjusted or switched to the mixed defrosting mode; the heat stability factor is calculated based on the heat storage pool flow and the deviation between the actual outlet temperature and the current target outlet temperature, which is used to evaluate whether the current heat release process is in a stable state, and the heat pump frequency and the three-way valve opening degree are adjusted accordingly; After the defrosting operation is completed, the energy efficiency index is calculated based on the evaporator temperature recovery effect and the total energy consumption of the system, and the defrosting result is marked and fed back to the reinforcement learning model for strategy optimization; the energy efficiency index is the normalized value of the evaporator temperature recovery quality per unit energy consumption, which is used to determine whether the current defrosting is successful, and is used as the basis for subsequent strategy fine-tuning.

2. The intelligent control method for heat pump multi-mode coordinated heat storage defrosting according to claim 1, characterized in that, The system state vector is formed by real-time collection of sensors arranged at the outdoor air inlet for obtaining the external environment temperature, at the heat storage coil outlet for obtaining the heat storage pool outlet temperature, at multiple points on the evaporator surface for obtaining the evaporator average temperature, and at the main branch pipe of the heat storage circuit for obtaining the heat storage pool circulation flow, and is processed by sliding average and low-pass filtering.

3. The intelligent control method for heat pump multi-mode coordinated heat storage defrosting according to claim 1, characterized in that, The reinforcement learning model is an improved deep Q network, which introduces a temperature difference constraint term and a heat storage pool flow reverse constraint term in the decision-making process to prevent energy from suddenly changing and ensure the rationality of mode selection under low flow conditions.

4. The intelligent control method for heat pump multi-mode coordinated heat storage defrosting according to claim 1, characterized in that, The target outlet temperature is dynamically corrected according to the current evaporator temperature and its change rate, and the heat release intensity is adjusted smoothly through a nonlinear response function to avoid sharp fluctuations in the evaporator temperature.

5. The intelligent control method for heat pump multi-mode coordinated heat storage defrosting according to claim 1, characterized in that, When the evaporator temperature response rate continues to be lower than the empirical threshold, the heat storage compensation mechanism is triggered to temporarily increase the target outlet temperature of the heat storage pool without changing the current defrosting mode.

6. The intelligent control method of heat pump multi-mode coordinated heat storage defrosting according to claim 1, characterized in that, After each defrosting operation is completed, the defrosting start and end time, the maximum temperature rising rate of the evaporator, the energy efficiency index and the determination result of whether the defrosting is successful are written into a structured log, which is used for subsequent retraining of the reinforcement learning model and maintenance diagnosis of the system running state.

7. A heat pump multi-mode cooperative heat storage defrosting intelligent control system, characterized in that, The system comprises: A data acquisition module for collecting external environment temperature, heat storage pool outlet temperature, evaporator average temperature and heat storage pool circulation flow to form a system state vector; a learning decision module configured to determine a current optimal defrosting mode from the system state vector by a reinforcement learning model, the defrosting mode including a heat storage tank defrosting mode, a hybrid defrosting mode or a maintaining heating mode; a temperature control adjustment module configured to dynamically set a target outlet temperature of the heat storage tank according to the selected defrosting mode, and to adjust the opening degree of the three-way valve to control the heat distribution path; when the heat storage tank defrosting mode is selected, the three-way valve is switched to the full opening heat storage branch; when the hybrid defrosting mode is selected, the three-way valve is simultaneously turned on according to a preset proportion of the heat storage tank and the water tank branch; when the maintaining heating mode is selected, the heat storage tank enters a heat preservation or energy storage state; a process control module configured to monitor the heat stability factor of the heat storage tank and the temperature response rate of the evaporator in real time during the defrosting execution process, and to dynamically adjust the target outlet temperature or switch to the hybrid defrosting mode if the heat release is unstable or the defrosting response is slow; the heat stability factor is calculated based on the heat storage tank flow and the deviation between the actual outlet temperature and the current target outlet temperature, and is used to evaluate whether the current heat release process is in a stable state, and to adjust the heat pump frequency and the opening degree of the three-way valve accordingly; an energy efficiency monitoring module configured to calculate an energy efficiency index based on the evaporator temperature recovery effect and the total system energy consumption after the defrosting operation is completed, and to feed back the defrosting result to the reinforcement learning model for strategy optimization; the energy efficiency index is a normalized value of the evaporator temperature recovery quality per unit energy consumption, which is used to determine whether the current defrosting is successful, and serves as a basis for subsequent strategy fine-tuning.

Images