Heat pump cold and hot cycle control system based on intelligent AI

Abstract

The invention discloses a heat pump cold and hot cycle control system based on an intelligent AI, and relates to the technical field of heat pump control, and the system comprises a heat pump system digital twinborn body module which obtains the historical operation data of a heat pump, analyzes the multi-dimensional characteristics of the heat pump under a specific working condition, couples a heat pump mechanism model, and builds a virtual simulation heat pump digital twinborn body; the dynamic load prediction curve module is used for acquiring real-time operation data of the heat pump, introducing an environmental factor as an external dynamic disturbance variable and generating a dynamic load prediction curve of the heat pump; the heat pump system optimal operation mode module is used for generating a user personality portrait, integrating a heat pump load prediction curve, real-time electricity price information and an equipment health state evaluation index, and substituting the user personality portrait, the heat pump load prediction curve, the real-time electricity price information and the equipment health state evaluation index into the virtual simulation heat pump digital twin body to obtain a heat pump optimal operation mode; and the global optimal control instruction set module is used for generating a heat pump global optimal control instruction set by taking the heat pump comprehensive energy efficiency ratio and the user comfort as collaborative optimization targets and utilizing a multi-target decision algorithm. The comprehensive energy efficiency of the system is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of heat pump control, in particular to a heat pump cold-heat cycle control system based on intelligent AI. BACKGROUND

[0002] The current traditional heat pump control system mostly adopts a control strategy based on fixed set values or simple PID feedback, generally lacks forward-looking perception and self-adaptive ability, cannot effectively predict building dynamic load changes and adjust in advance, leads to system response lag and large energy efficiency fluctuation, the control target is single, often focuses on maintaining temperature stability and ignores multi-target cooperation such as energy efficiency, comfort and equipment operation life, and it is difficult to achieve global optimization under complex working conditions, the existing system relies on manual experience for parameter setting and mode switching, which is low in intelligent level, cannot realize real-time dynamic optimization according to equipment performance degradation or external electricity price signals, and leads to poor economy and reliability of long-term system operation. SUMMARY

[0003] To solve the above technical problems, the application provides a heat pump cold-heat cycle control system based on intelligent AI, which solves the above problems.

[0004] To achieve the above purposes, the technical scheme adopted by the application is as follows:

[0005] A heat pump cold-heat cycle control system based on intelligent AI, comprising:

[0006] a heat pump system digital twin module, a dynamic load prediction curve module, a heat pump system optimal operation mode module and a global optimal control instruction set module;

[0007] The heat pump system optimal operation mode module is electrically connected with the heat pump system digital twin module and the dynamic load prediction curve module, and the global optimal control instruction set module is electrically connected with the heat pump system optimal operation mode module.

[0008] The heat pump system digital twin module acquires historical operation data of the heat pump system, analyzes multi-dimensional characteristics of the heat pump system under specific working conditions, couples a heat pump system mechanism model, and establishes a virtual simulation heat pump system digital twin.

[0009] The dynamic load prediction curve module acquires real-time operation data of the heat pump system based on a multi-source heterogeneous sensor, introduces environmental factors as external dynamic disturbance variables, and generates a heat pump system dynamic load prediction curve.

[0010] The heat pump system optimal operation mode module obtains target area user interaction data, generates a user individual portrait, comprehensively considers heat pump system load prediction curves, real-time electricity price information and equipment health state evaluation indexes, substitutes into a virtual simulation heat pump system digital twin for multi-objective coupled simulation deduction, and obtains a heat pump system optimal operation mode;

[0011] The global optimal control instruction set module, based on the obtained heat pump system optimal operation mode, takes heat pump system comprehensive energy efficiency ratio and user comfort degree as collaborative optimization targets, generates a heat pump system global optimal control instruction set by using a multi-objective decision algorithm.

[0012] Preferably, the heat pump system digital twin module specifically comprises the following:

[0013] A data preprocessing unit obtains heat pump system historical operation data based on sensors and control signals, unifies timestamps, and performs data preprocessing;

[0014] A relationship curve drawing unit, based on the obtained heat pump system historical operation data, performs standardization processing, divides each data point into K clusters by using a K-means clustering algorithm, calculates error sum of squares corresponding to each cluster K value, and draws a K value and error sum of squares relationship curve;

[0015] An optimal cluster number unit, based on the K value and error sum of squares relationship curve, finds a K value corresponding to an inflection point at which a chart error sum of squares descending speed significantly slows down as the optimal cluster number K of the heat pump system historical operation data;

[0016] A historical operation condition division result unit, based on the obtained optimal cluster number K of the heat pump system historical operation data, randomly selects K initial centroids, assigns each data point to a cluster with the nearest initial centroid, calculates the mean of each cluster centroid, and iteratively updates until the centroid no longer changes, to obtain a heat pump system historical operation condition division result;

[0017] The operation conditions include winter heating conditions, summer refrigeration conditions and transition season conditions;

[0018] A heat pump system energy consumption characteristic unit, based on the obtained heat pump system historical operation condition division result, calculates heat pump system average power, peak power and cumulative energy consumption under each condition, and obtains heat pump system energy consumption characteristics;

[0019] A heat pump system performance characteristic unit calculates heat pump system average COP performance coefficients under each condition, generates heat pump system average COP performance coefficient time sequence change curves in combination with time data, and obtains heat pump system performance characteristics;

[0020] The heat pump system operation mode feature unit calculates the compressor start-stop frequency and average running time of the heat pump system under each working condition to obtain heat pump system operation mode features.

[0021] The heat pump system statistical feature unit calculates the mean, variance and extreme value of the evaporation temperature and condensation temperature of the heat pump system under each working condition to obtain heat pump system statistical features.

[0022] The multi-dimensional feature unit fuses the energy consumption features, performance features, operation mode features and statistical features of the heat pump system to obtain multi-dimensional features of the heat pump system under specific working conditions.

[0023] Preferably, the heat pump system digital twin body module further comprises:

[0024] The mass flow model and power consumption model unit obtains the volumetric efficiency and isentropic efficiency of the compressor through a polynomial fitting method based on the performance data table provided by the compressor manufacturer, and obtains the compressor mass flow model and power consumption model using physical formulas.

[0025] The heat exchanger model unit calculates the heat exchanger NTU and fluid heat capacity ratio using the performance-heat transfer unit method to construct a heat exchanger model.

[0026] The throttling device model unit constructs a throttling device model based on the mass flow equation of the throttling principle.

[0027] The heat pump system refrigerant property unit calculates the enthalpy, entropy and density of the refrigerant under different pressures and temperatures based on the integrated REFPROP refrigerant property database to obtain the heat pump system refrigerant properties.

[0028] The heat pump system mechanism model unit uses the mass and energy conservation equations to couple the compressor mass flow model and power consumption model, the heat exchanger model, the throttling device model and the heat pump system refrigerant properties, uses the compressor frequency as the boundary condition, and uses the iterative algorithm to calculate the point pressure, temperature, flow and power consumption of the heat pump system under stable working conditions to establish a heat pump system mechanism model.

[0029] The digital twin body unit establishes a virtual simulation heat pump system digital twin body based on the heat pump system mechanism model and the multi-dimensional features under specific working conditions of the heat pump system.

[0030] The correction unit takes 24 hours as the collection period, substitutes it into the virtual simulation heat pump system digital twin body, calculates the total power consumption and outlet temperature of the system within the period, and compares them with the actual corresponding measured values of the heat pump system mechanism model. If the average deviation between the two within the period exceeds 10%, the parameter self-calibration process of the virtual simulation heat pump system digital twin body is automatically triggered, and the compressor volumetric efficiency fitting parameters of the virtual simulation heat pump system digital twin body are fine-tuned using the actual running data of the last 24 hours to ensure the simulation accuracy.

[0031] Preferably, the dynamic load prediction curve module specifically comprises the following:

[0032] A multi-source heterogeneous sensor unit acquires real-time operation data of the heat pump system based on multi-source heterogeneous sensors, performs data preprocessing, and automatically switches to a backup strategy if data of a certain sensor is continuously missing for more than 5 minutes. For outdoor sensors, corresponding data obtained from an authoritative weather API interface is used, and for indoor temperature sensors, the average value of other available sensors in the same area is used.

[0033] The multi-source heterogeneous sensor includes the following:

[0034] Temperature sensors are installed around the indoor return air inlet, air outlet, key rooms, and outdoor units to monitor the environment temperature and coil temperature, and the communication protocol uses Modbus RTU.

[0035] Pressure sensors are installed at the suction and exhaust ports of the compressor and before and after the throttling device to monitor the system pressure, and the output signal is a 4-20 mA analog signal.

[0036] Power sensors are installed in the heat pump main machine power supply circuit to monitor the system operation power in real time, and the communication protocol uses Modbus TCP.

[0037] Humidity sensors are installed at the indoor return air inlet to monitor the environment humidity.

[0038] Flow sensors are installed in the circulating water pipeline to monitor the water flow.

[0039] All sensor data is aggregated through an Internet of Things gateway and uploaded to the cloud platform data center using the MQTT protocol.

[0040] The hourly forecast data unit obtains hourly forecast data for the target area within the next 24 hours based on a weather forecast APP, and obtains the dry bulb temperature, wet bulb temperature, humidity, wind speed, wind direction, and cloud cover of the area.

[0041] The heat pump system operation-weather condition data set unit establishes a unified time-space reference heat pump system operation-weather condition data set based on real-time operation data of the heat pump system and hourly weather forecast data for the target area within the next 24 hours.

[0042] The historical load sequence unit sets the past 24 hours as a fixed sliding time window, extracts historical load data of the heat pump system within the sliding time window, and establishes a historical load sequence of the heat pump system.

[0043] An environmental factor unit records periodic characteristics of each time step, extracts environmental factors within a sliding time window, and calculates the average load of the heat pump system, the maximum outdoor temperature, and the temperature change rate within a past sliding time window.

[0044] Preferably, the dynamic load prediction curve module further includes:

[0045] The dynamic load prediction model unit encodes the historical load sequence and periodic characteristics of the heat pump system into vectors, respectively, introduces environmental factors as external dynamic disturbance variables, uses a long short-term memory network, designs two hidden layers with 128 neurons each, uses tanh as the activation function, uses the past 24-hour historical load sequence of the heat pump system, periodic characteristics, and environmental factors as input features, minimizes the mean square error between the predicted load and the actual load as the target, learns the long-term dependencies between different time points of the heat pump system load, and establishes a dynamic load prediction model for the heat pump system.

[0046] The dynamic load prediction curve unit establishes a dynamic load prediction curve for the heat pump system based on the dynamic load prediction model for the heat pump system and the unified spatiotemporal benchmark heat pump system operation-weather condition dataset.

[0047] Preferably, the optimal operation mode module for the heat pump system specifically includes:

[0048] The individual portrait unit obtains target area user interaction data based on a sensor background, analyzes user set temperature and humidity ranges, identifies user multi-scenario temperature and humidity comfort intervals, obtains user individual preference characteristics, and establishes a user individual portrait.

[0049] The user multi-scenario temperature and humidity comfort intervals include residential scenarios, office building scenarios, and hotel scenarios.

[0050] For residential scenarios, the temperature and humidity setting preferences of different family members at different times and in different rooms are analyzed, a user comprehensive portrait is generated in units of families through long-term learning, home, work, and sleep modes are automatically identified, and the corresponding comfort intervals and energy-saving strategies are dynamically switched to.

[0051] For office building scenarios, the group preference in public areas is analyzed, the temperature and humidity range comfortable for the majority of people is set by collecting feedback from the majority of employees and monitoring the occupancy rate of employees, and the comfortable priority mode is entered in advance before a meeting starts, combined with calendar information and dynamic load prediction of the heat pump system.

[0052] For hotel scenarios, the room status, including occupancy and vacancy, is used to obtain the historical preferences of guests through an Internet of Things cloud platform, and the environment parameters of individual preferences are automatically restored when the guests check in.

[0053] The load prediction range unit extracts a heat pump system load prediction range in a future period based on a heat pump system dynamic load prediction curve;

[0054] The electricity price information unit obtains target area real-time electricity price information based on electricity market electricity prices;

[0055] The health state index unit monitors heat pump system equipment states in real time, calculates compressor efficiency attenuation coefficients, heat exchanger fouling coefficients and filter screen resistance coefficients, and evaluates equipment health state indexes based on an equipment health management cloud platform;

[0056] The compressor efficiency attenuation coefficient is specifically: setting 30 days as a collection period, selecting a heat pump system stable operation condition, substituting into a virtual simulation heat pump system digital twin, calculating a ratio of a compressor theoretical isentropic efficiency to an actual isentropic efficiency calculated according to suction and discharge pressures and temperatures, obtaining the compressor efficiency attenuation coefficient, and if the compressor efficiency attenuation coefficient continuously is lower than 0.9, triggering a maintenance alarm automatically;

[0057] The heat exchanger fouling coefficient is specifically: taking two weeks as a calculation period, calculating a heat exchanger actual heat transfer coefficient through monitoring heat exchanger two-side inlet and outlet temperature differences and flow rates, obtaining a heat exchanger theoretical heat transfer coefficient in a clean state in combination with the virtual simulation heat pump system digital twin, calculating a ratio of the heat exchanger actual heat transfer coefficient to the theoretical heat transfer coefficient as the heat exchanger fouling coefficient, and if the heat exchanger fouling coefficient is lower than 0.85, suggesting cleaning;

[0058] The filter screen resistance coefficient specifically includes: judging through monitoring filter two-side pressure differences, calculating the filter screen resistance coefficient in real time, and if the pressure difference exceeds 2 times of an initial value monitoring value of the filter screen, prompting to replace or clean the filter screen.

[0059] Preferably, the heat pump system optimal operation mode module further includes:

[0060] The state space unit constructs a state space based on a user individual portrait, a heat pump system load prediction range in a future period, real-time electricity price information and equipment health state indexes, and unifies a time reference;

[0061] The candidate optimal operation mode strategy set unit takes the state space as input, substitutes into a virtual simulation heat pump system digital twin for coupling simulation deduction, takes user individual preference characteristics, equipment physical limits and safe operation ranges as constraint conditions, and generates a heat pump system candidate optimal operation mode strategy set;

[0062] The initial population unit initializes the heat pump system candidate optimal operation mode strategy set, and randomly generates N heat pump system candidate optimal operation mode strategy individual populations;

[0063] The non-dominated sorting unit sorts each heat pump system candidate optimization operation mode strategy individual into different front levels based on the heat pump system energy consumption, total operation cost and user comfort as multi-objective functions.

[0064] The crowding degree calculation unit calculates the crowding degree of each heat pump system candidate optimization operation mode strategy individual in the same front level.

[0065] The new offspring individual unit selects the individuals with high front level and large crowding degree as parents based on the non-dominated sorting and the crowding degree of each heat pump system candidate optimization operation mode strategy individual, and generates a new offspring individual population through crossover and mutation operations.

[0066] The optimal operation mode strategy unit combines the parent individuals and the new offspring individual population, and selects the optimal N heat pump system candidate optimization operation mode strategy individuals to form a new generation population through non-dominated sorting and crowding degree comparison, and iteratively updates to obtain the optimal operation mode strategy of the heat pump system.

[0067] Preferably, the global optimal control instruction set internal module specifically comprises:

[0068] The multi-objective decision model unit constructs a multi-objective decision model by taking the optimal operation mode of the heat pump system as input and taking the comprehensive energy efficiency ratio and user comfort of the heat pump system as collaborative optimization targets.

[0069] The weight unit calculates the target information entropy of the comprehensive energy efficiency ratio and user comfort of the heat pump system by using an entropy weight method to obtain corresponding weights.

[0070] Preferably, the global optimal control instruction set internal module further comprises:

[0071] The ideal solution unit calculates the ideal solution and negative ideal solution of each target weight of the comprehensive energy efficiency ratio and user comfort of the heat pump system based on the obtained weights.

[0072] The Euclidean distance unit calculates the Euclidean distance of the ideal solution and negative ideal solution of each target weight of the comprehensive energy efficiency ratio and user comfort of the heat pump system.

[0073] The optimal control instruction set unit calculates the Euclidean distance of each optimal operation mode of the heat pump system and the ideal solution and negative ideal solution, calculates the relative closeness of each optimal operation mode of the heat pump system and the ideal solution, performs descending sorting, and obtains the global optimal control instruction set of the heat pump system.

[0074] Compared with the prior art, the present application has the following advantages:

[0075] The application provides a heat pump cold-heat cycle control scheme based on intelligent AI, which realizes high-precision simulation and forward-looking perception of the system operation state by constructing a heat pump system digital twin and dynamic load prediction, performs multi-target coupled simulation on the basis of user individual portraits, real-time electricity prices and equipment health states, and generates an optimal operation mode covering multiple dimensions such as efficiency, comfort, economy, defrosting and maintenance; the multi-target decision algorithm is used to cooperatively optimize energy efficiency and comfort, and a globally optimal control instruction set is generated, so that the user's comfortable experience is ensured, the system's comprehensive energy efficiency and economy are improved, and the service life of the equipment is effectively prolonged. BRIEF DESCRIPTION OF DRAWINGS

[0076] Figure 1 It is a kind of heat pump cold and hot cycle control system framework based on intelligent AI. DETAILED DESCRIPTION

[0077] The following description is used to disclose the present application so that those skilled in the art can implement the present application. The preferred embodiments in the following description are only as examples, and other obvious modifications can be made by those skilled in the art.

[0078] Referring to Figure 1 The heat pump cold-heat cycle control system based on intelligent AI comprises:

[0079] The heat pump system digital twin module, the dynamic load prediction curve module, the heat pump system optimal operation mode module and the globally optimal control instruction set module are electrically connected to each other.

[0080] The heat pump system optimal operation mode module is electrically connected to the heat pump system digital twin module and the dynamic load prediction curve module, and the globally optimal control instruction set module is electrically connected to the heat pump system optimal operation mode module.

[0081] The heat pump system digital twin module acquires historical operation data of the heat pump system, analyzes multi-dimensional characteristics of the heat pump system under specific working conditions, couples a heat pump system mechanism model, and establishes a virtual simulation heat pump system digital twin. The historical operation data includes compressor frequency, valve opening degree, coil temperature and pressure.

[0082] The dynamic load prediction curve module acquires real-time operation data of the heat pump system based on multi-source heterogeneous sensors, introduces environmental factors as external dynamic disturbance variables, and generates a heat pump system dynamic load prediction curve.

[0083] The heat pump system optimal operation mode module obtains target area user interaction data, generates a user individual portrait, comprehensively considers heat pump system load prediction curves, real-time electricity price information and equipment health state evaluation indexes, substitutes into a virtual simulation heat pump system digital twin for multi-objective coupled simulation deduction, and obtains a heat pump system optimal operation mode; the operation mode includes: an efficient steady state mode, a comfort priority mode, an economic operation mode, a preventive defrosting mode and an equipment maintenance mode;

[0084] The global optimal control instruction set module, based on the obtained heat pump system optimal operation mode, takes the heat pump system comprehensive energy efficiency ratio and user comfort degree as the collaborative optimization target, generates a heat pump system global optimal control instruction set by using a multi-objective decision algorithm.

[0085] The heat pump system digital twin module specifically includes:

[0086] A data preprocessing unit obtains heat pump system historical operation data based on sensors and control signals, unifies timestamps, and performs data preprocessing;

[0087] A relationship curve unit, based on the obtained heat pump system historical operation data, performs standardization processing, divides each data point into K clusters by using a K-means clustering algorithm, calculates the error sum of squares corresponding to each cluster K value, and draws a K value and error sum of squares relationship curve;

[0088] An optimal cluster number unit, based on the K value and error sum of squares relationship curve, finds the K value corresponding to the inflection point at which the error sum of squares of the graph significantly slows down as the descending speed, as the optimal cluster number K of the heat pump system historical operation data;

[0089] A historical operation condition division result unit, based on the obtained optimal cluster number K of the heat pump system historical operation data, randomly selects K initial centroids, assigns each data point to the cluster with the nearest initial centroid, calculates the mean of each cluster centroid, and iteratively updates until the centroid no longer changes, to obtain the heat pump system historical operation condition division result;

[0090] The operation condition includes: a winter heating condition, a summer refrigeration condition and a transition season condition;

[0091] A heat pump system energy consumption characteristic unit, based on the obtained heat pump system historical operation condition division result, calculates the average power, peak power and cumulative energy consumption of the heat pump system under each condition, and obtains the heat pump system energy consumption characteristics;

[0092] A heat pump system performance characteristic unit calculates the average COP performance coefficient of the heat pump system under each condition, generates a heat pump system average COP performance coefficient time sequence change curve in combination with time data, and obtains the heat pump system performance characteristics;

[0093] The heat pump system operation mode feature unit calculates the compressor start-stop frequency and average running time of the heat pump system under each working condition to obtain heat pump system operation mode features.

[0094] The heat pump system statistical feature unit calculates the mean, variance and extreme value of the evaporation temperature and condensation temperature of the heat pump system under each working condition to obtain heat pump system statistical features.

[0095] The multi-dimensional feature unit fuses the energy consumption features, performance features, operation mode features and statistical features of the heat pump system to obtain multi-dimensional features of the heat pump system under specific working conditions.

[0096] The heat pump system digital twin body module further comprises:

[0097] The mass flow model and power consumption model unit obtains the volumetric efficiency and isentropic efficiency of the compressor through a polynomial fitting method based on the performance data table provided by the compressor manufacturer, and obtains the compressor mass flow model and power consumption model by using physical formulas.

[0098] The heat exchanger model unit calculates the heat exchanger NTU and fluid heat capacity ratio by using the performance-heat transfer unit method to construct a heat exchanger model.

[0099] The throttling device model unit constructs a throttling device model based on the mass flow equation of the throttling principle.

[0100] The heat pump system refrigerant property unit calculates the enthalpy, entropy and density of the refrigerant under different pressures and temperatures based on the integrated REFPROP refrigerant property database to obtain the heat pump system refrigerant properties.

[0101] The heat pump system mechanism model unit uses the mass and energy conservation equations to couple the compressor mass flow model and power consumption model, the heat exchanger model, the throttling device model and the heat pump system refrigerant properties, uses the compressor frequency as the boundary condition, and uses the iterative algorithm to calculate the pressure, temperature, flow and power consumption of each point under the stable working condition of the heat pump system to establish a heat pump system mechanism model.

[0102] The digital twin body unit establishes a virtual simulation heat pump system digital twin body based on the heat pump system mechanism model and the multi-dimensional features under specific working conditions of the heat pump system.

[0103] The correction unit takes 24 hours as the collection period, substitutes it into the virtual simulation heat pump system digital twin body, calculates the total power consumption and outlet temperature of the system within the period, and compares it with the actual corresponding measured values of the heat pump system mechanism model. If the average deviation between the two within the period exceeds 10%, the parameter self-calibration process of the virtual simulation heat pump system digital twin body is automatically triggered, and the compressor volumetric efficiency fitting parameters of the virtual simulation heat pump system digital twin body are fine-tuned using the actual running data of the last 24 hours to ensure the simulation accuracy.

[0104] In use, in combination with the above:

[0105] In the current field of heat pump system modeling, the main defects of the prior art are that: the traditional mechanism model excessively relies on design parameters and ignores actual operation data, resulting in insufficient simulation accuracy, poor working condition adaptability, lack of multi-dimensional feature mining of historical operation characteristics, and difficulty in realizing dynamic mapping of digital twin and physical system; this step improves the performance of the digital twin in working condition recognition accuracy, energy efficiency evaluation comprehensiveness, and state perception real-time performance by fusing data-driven and mechanism modeling, introducing a multi-working-condition automatic division method based on cluster analysis, integrating energy consumption, performance, operation mode, and statistical characteristics, and providing a high-fidelity simulation foundation for heat pump system optimization control and fault diagnosis.

[0106] The dynamic load prediction curve module specifically includes:

[0107] The multi-source heterogeneous sensor unit acquires real-time operation data of the heat pump system based on multi-source heterogeneous sensors, performs data preprocessing, and if the data of a sensor is continuously missing for more than 5 minutes, the system automatically switches to a backup strategy, uses corresponding data obtained from an authoritative weather API interface for outdoor sensors, and uses the average value of other available sensors in the same area for indoor temperature sensors;

[0108] The multi-source heterogeneous sensor includes:

[0109] The temperature sensor is installed around the indoor return air inlet, air outlet, key room, and outdoor unit, and is used to monitor the environment temperature and coil temperature, and the communication protocol uses Modbus RTU;

[0110] The pressure sensor is installed at the suction and exhaust ports of the compressor and before and after the throttling device, and is used to monitor the system pressure, with an output signal of 4-20mA analog quantity;

[0111] The power sensor is installed in the power supply circuit of the heat pump main unit, and is used to monitor the system operation power in real time, with a communication protocol of Modbus TCP;

[0112] The humidity sensor is installed at the indoor return air inlet, and is used to monitor the environment humidity;

[0113] The flow sensor is installed in the circulating water pipeline, and is used to monitor the water flow;

[0114] All sensor data is aggregated through an IoT gateway and uploaded to the cloud platform data center using the MQTT protocol;

[0115] The hourly forecast data unit acquires hourly forecast data in the target area for the next 24 hours based on a weather forecast APP, and obtains the dry bulb temperature, wet bulb temperature, humidity, wind speed, wind direction, and cloud cover of the area;

[0116] A heat pump system operation-weather condition dataset unit, based on heat pump system real-time operation data and target area future 24-hour hourly weather forecast data, unified timestamp, establishes a unified space-time benchmark heat pump system operation-weather condition dataset;

[0117] A historical load sequence unit, sets the past 24 hours as a fixed sliding time window, extracts heat pump system historical load data within the sliding time window, and establishes a heat pump system historical load sequence;

[0118] An environmental factor unit, records the periodic characteristics of each time step, extracts environmental factors within the sliding time window, and calculates the average load of the heat pump system, the maximum outdoor temperature and the temperature change rate within the past sliding time window.

[0119] The dynamic load prediction curve module further includes:

[0120] A dynamic load prediction model unit, which encodes the heat pump system historical load sequence and periodic characteristics into vectors respectively, introduces environmental factors as external dynamic disturbance variables, uses a long short-term memory network, designs two hidden layers with 128 neurons each, uses tanh as the activation function, uses the past 24 hours of heat pump system historical load sequence, periodic characteristics and environmental factors as input features, and minimizes the mean square error between predicted load and actual load as the target. Learn the long-term dependence between different time points of the heat pump system load, and establish a heat pump system dynamic load prediction model;

[0121] A dynamic load prediction curve unit, based on the heat pump system dynamic load prediction model and the unified space-time benchmark heat pump system operation-weather condition dataset, establishes a heat pump system dynamic load prediction curve.

[0122] In use, in combination with the above:

[0123] In the current heat pump system load prediction research, there are generally insufficient fusion of multi-source heterogeneous data, insufficient consideration of dynamic meteorological disturbance factors, and limited ability to capture long-term time series dependence, etc. Traditional methods often rely on single historical load data or static meteorological parameters, making it difficult to accurately reflect the load fluctuation characteristics under complex environments; This step fuses multi-source sensor real-time operation data and high-precision weather forecast information to construct a unified space-time benchmark dataset, combined with sliding time window feature extraction and long short-term memory network, which can effectively learn the long-term dependence of the load and dynamically respond to environmental factor disturbances, improve the accuracy and adaptability of the 24-hour load prediction curve, and provide reliable data support for heat pump system optimization scheduling and energy management.

[0124] The heat pump system optimal operation mode module specifically includes:

[0125] a personalized portrait unit, based on a sensor background, acquires target area user interaction data, analyzes user set temperature and humidity ranges, identifies user multi-scenario temperature and humidity comfort intervals, obtains user personalized preference characteristics, and establishes a user personalized portrait;

[0126] The user multi-scenario temperature and humidity comfort interval includes a residential scenario, an office building scenario, and a hotel scenario.

[0127] For the residential scenario, the temperature and humidity setting preferences of different family members at different times and in different rooms are analyzed, a user comprehensive portrait is generated in units of families through long-term learning, home, work, and sleep modes are automatically identified, and dynamic switching to corresponding comfort intervals and energy-saving strategies is performed.

[0128] For the office building scenario, the group preference of the public area is analyzed, the comfort temperature and humidity range of the majority of people is set by collecting feedback from the majority of employees and monitoring the in-room rate of the employees, and before the start of a meeting, the calendar information and dynamic load prediction of the heat pump system are combined to enter the comfort priority mode in advance.

[0129] For the hotel scenario, according to the guest room state, including occupancy and vacancy, the personalized setting is performed according to the guest historical preference obtained from the Internet of Things cloud platform data, and when the guest checks in, the environment parameters of the personalized preference are automatically restored.

[0130] A load prediction range unit extracts a heat pump system load prediction range in a future period based on a heat pump system dynamic load prediction curve.

[0131] A price information unit acquires real-time electricity price information in a target area based on electricity market prices.

[0132] A health status index unit real-time monitors the heat pump system device state based on a device health management cloud platform, calculates a compressor efficiency attenuation coefficient, a fouling coefficient of a heat exchanger, and a resistance coefficient of a filter screen, and evaluates the device health status index.

[0133] The compressor efficiency attenuation coefficient is specifically set to 30 days as a collection period, a stable operation condition of the heat pump system is selected, a virtual simulation heat pump system digital twin is substituted, a ratio of a compressor theoretical isentropic efficiency to an actual isentropic efficiency calculated according to suction and discharge pressures and temperatures is calculated, a compressor efficiency attenuation coefficient is obtained, and if the compressor efficiency attenuation coefficient continuously decreases below 0.9, a maintenance alarm is automatically triggered.

[0134] The fouling coefficient of the heat exchanger is specifically: taking two weeks as a calculation period, the actual heat transfer coefficient of the heat exchanger is calculated by monitoring the temperature difference and flow rate at the inlet and outlet of the heat exchanger on both sides, the theoretical heat transfer coefficient of the heat exchanger in a clean state is obtained in combination with the virtual simulation heat pump system digital twin, and the ratio of the actual heat transfer coefficient to the theoretical heat transfer coefficient is taken as the fouling coefficient of the heat exchanger; if the fouling coefficient of the heat exchanger is lower than 0.85, cleaning of the heat exchanger is recommended;

[0135] The resistance coefficient of the filter screen specifically includes: the resistance coefficient of the filter screen is calculated in real time by monitoring the pressure difference on both sides of the filter; if the pressure difference exceeds twice the initial value of the filter screen, the filter screen is prompted to be replaced or cleaned.

[0136] The heat pump system optimal operation mode module further includes:

[0137] A state space unit is configured to construct a state space based on a user individual portrait, a heat pump system load prediction range in a future time period, real-time electricity price information and a device health state index, and unify a time reference;

[0138] A candidate optimal operation mode strategy set unit is configured to take the state space as an input, couple simulation deduction with the virtual simulation heat pump system digital twin, take user individual preference characteristics, device physical limits and safe operation ranges as constraint conditions, and generate a heat pump system candidate optimal operation mode strategy set;

[0139] An initial population unit is configured to initialize the heat pump system candidate optimal operation mode strategy set, and randomly generate N heat pump system candidate optimal operation mode strategy individual populations;

[0140] A non-dominated sorting unit is configured to take the heat pump system energy consumption, total operation cost and user comfort as multi-objective functions, divide each heat pump system candidate optimal operation mode strategy individual into different front levels, and perform non-dominated sorting;

[0141] A crowding degree calculation unit is configured to calculate the crowding degree of each heat pump system candidate optimal operation mode strategy individual in the same front level;

[0142] A new offspring individual unit is configured to preferentially select individuals with high front level and large crowding degree as parents based on the different front levels after non-dominated sorting and the crowding degree of each heat pump system candidate optimal operation mode strategy individual, perform crossover and mutation operations, and generate a new offspring individual population;

[0143] An optimal operation mode strategy unit is configured to combine the parent individuals and the new offspring individual population, filter optimal N heat pump system candidate optimal operation mode strategy individuals to form a new generation population through non-dominated sorting and crowding degree comparison, and constantly iterate and update to obtain a heat pump system optimal operation mode strategy.

[0144] In use, in combination with the above:

[0145] The current heat pump system optimization research focuses on single objective static optimization, and the following defects are common: lack of user personalized comfort modeling, not combined with real-time health status of equipment for dynamic constraint, ignoring the game relationship between multi-objective, traditional optimization algorithm is difficult to achieve Pareto optimal in complex nonlinear system; this step realizes the multi-objective dynamic collaborative optimization of energy consumption, cost and comfort by constructing a multi-dimensional state space integrating user individual portrait, load forecasting, electricity price and equipment health status, and introducing digital twin simulation and non-dominated sorting genetic algorithm, which not only guarantees the personalized comfort demand of users, but also avoids operation risks through equipment state early warning, and significantly reduces the total cost by using electricity price mechanism, and improves the system adaptability and robustness.

[0146] The global optimal control instruction set internal module specifically includes:

[0147] A multi-objective decision model unit, taking the optimal operation mode of the heat pump system as input, taking the comprehensive energy efficiency ratio of the heat pump system and user comfort as collaborative optimization targets, and constructing a multi-objective decision model;

[0148] A weight unit, using an entropy weight method, calculating the information entropy of the comprehensive energy efficiency ratio of the heat pump system and user comfort, and obtaining corresponding weights.

[0149] The global optimal control instruction set internal module further includes:

[0150] An ideal solution unit, based on the obtained weights of the comprehensive energy efficiency ratio of the heat pump system and user comfort, calculating the ideal solution and negative ideal solution of each target of the comprehensive energy efficiency ratio of the heat pump system and user comfort under each weight;

[0151] An Euclidean distance unit, calculating the Euclidean distance of the ideal solution and negative ideal solution of each target of the comprehensive energy efficiency ratio of the heat pump system and user comfort under each weight;

[0152] An optimal control instruction set unit, calculating the Euclidean distance of each optimal operation mode of the heat pump system and the ideal solution and negative ideal solution, calculating the relative closeness of each optimal operation mode of the heat pump system and the ideal solution, sorting in descending order, and obtaining the global optimal control instruction set of the heat pump system.

[0153] In use, in combination with the above:

[0154] Current heat pump system optimization control research mostly uses single objective or linear weighting method, which is difficult to dynamically balance the nonlinear conflict between energy efficiency and comfort, and the weight setting relies on experience and lacks objectivity, leading to decision bias; This step introduces entropy weight method to objectively quantify the weight of double objectives, and dynamically calculates the Euclidean distance between each mode and ideal solution by combining ideal solution sorting, realizing data-driven multi-objective collaborative optimization, improving the overall energy efficiency of the system while ensuring user comfort, and solving the defects of subjective weight and target synergy in traditional methods.

[0155] In combination with the above, the specific implementation is as follows:

[0156] Taking an office building in East China for winter heating as an example, the area is 2000 square meters, and an air source heat pump water heater with the model of Gree GMV-560WM / B is used;

[0157] The system collects the historical operation data of the heat pump unit for the past three complete years, a total of about 26000 operation hours, of which the data sources include the unit's own controller, the power monitoring system and the newly added Internet of Things sensors. About 5% of the missing values and outliers are processed, including zero values or out-of-range values caused by sensor instantaneous interruption. The data from different systems are unified to 1-minute interval timestamps using time alignment algorithm, which constitutes the data basis for building digital twin;

[0158] Based on the pre-processed data, five key feature variables are extracted, including outdoor dry-bulb temperature, relative humidity, total power of heat pump system, compressor frequency, and supply and return water temperature. After standardization, K-means clustering algorithm is used for working condition division;

[0159] The error sum of squares of the clustering results is calculated when K value is from 2 to 10, and the elbow rule curve is drawn. The curve shows that when K=3, the descending rate of error sum of squares has a significant inflection point. When K value increases from 2 to 3, the error sum of squares decreases by 42%, and when K value increases from 3 to 4, the error sum of squares only decreases by 15%. The optimal clustering number K=3 is determined;

[0160] Among them, cluster 1 is the winter heating working condition, outdoor temperature ≤10℃, cluster 2 is the transition season working condition, 10°C<outdoor temperature<25℃, and cluster 3 is the summer refrigeration working condition, outdoor temperature ≥25℃;

[0161] Based on the performance data table provided by the manufacturer of this type of compressor, a cubic polynomial fitting is used to obtain the functional relationship between its volumetric efficiency and isentropic efficiency with respect to compression ratio, and the goodness of fit is greater than 0.98;

[0162] The system integrates NIST REFPROP 9.1 database and calls the thermophysical property parameters of R410A refrigerant for calculation;

[0163] According to the design drawings, the effective heat exchange areas of the condenser and the evaporator are 45 m 2 and 38 m 2 , respectively;

[0164] By coupling the mechanism model of the heat pump system with the extracted multi-dimensional features of each working condition, a high-fidelity digital twin of the heat pump system is established. It is verified that, under the steady-state working condition, the average deviation of the total power simulated by the digital twin and the measured value is less than 5%;

[0165] Using the complete data of the past year as the training set, receiving the past 24 hours of 144 data points and the historical load sequence with a time step of 10 minutes, outdoor dry bulb temperature, and hourly features representing periodicity, two stacked LSTM layers are designed, the first LSTM layer has 128 neurons, the second layer has 64 neurons, the activation function is tanh, a Dropout layer is inserted between the two LSTM layers with a dropout rate of 0.3 to prevent overfitting, and the output is the future 24-hour load prediction value;

[0166] The Adam optimizer is trained for 200 rounds, and on the test set, the mean absolute percentage error of the future 24-hour load prediction is stable at 5.5%;

[0167] On a specific winter heating day, the system runs as follows: the input office comfort temperature setting is 20-22℃, the future 24-hour load prediction curve, the peak and valley electricity price of the power grid, the valley electricity price is 0.3 yuan / kWh, 22:00-8:00 the next day, the peak electricity price is 1.0 yuan / kWh, 8:00-22:00, the compressor efficiency attenuation coefficient is 0.92, and it is in good condition. Input the above state space into the digital twin, use the fast non-dominated sorting genetic algorithm with elitist strategy for simulation and deduction, the population size is set to 100, the iteration is 500 generations, and the optimization goal is: the total energy consumption is the lowest, the operation cost is the lowest, the comfort level is the highest, and the cumulative deviation of the indoor temperature from the set value of 20℃ is the smallest;

[0168] A strategy set containing dozens of Pareto optimal solutions is obtained, the objective weights of the three targets are calculated by the entropy weight method: energy consumption 0.35, cost 0.45, and comfort level 0.20, and the relative closeness of each strategy to the ideal solution is calculated by the ideal solution ordering method. The strategy with the highest relative closeness of 0.82 is selected to generate the global optimal control instruction set. In the valley electricity price period 2:00-6:00, the compressor frequency is increased to 55Hz in advance, the heating water temperature is increased to 45℃, and the heat is stored. In the peak electricity price period 9:00-11:00, the compressor frequency is limited to 35Hz, and the room temperature is maintained by using the thermal inertia of the building to save electricity bills. The indoor temperature fluctuation is guaranteed within the range of 20.5℃±0.5℃ throughout the day;

[0169] By executing this control instruction set, compared with the old PID strategy actually run on this day, the total operation cost of the day is reduced by 22%, the average operation COP of the system is increased from 2.9 to 3.4, and the comfort requirement of the user is fully met. This embodiment fully proves that the system is efficient, economical and superior in practical application.

[0170] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above embodiments, and the above embodiments and descriptions in the specification are only the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application. The scope of protection required by the present application is defined by the appended claims and their equivalents.

Claims

1. A heat pump cooling and heating cycle control system based on intelligent AI, characterized in that, include: The heat pump system digital twin module, dynamic load prediction curve module, heat pump system optimal operation mode module, and global optimal control instruction set module; Among them, the optimal operation mode module of the heat pump system is electrically connected to the digital twin module and the dynamic load prediction curve module of the heat pump system, and the global optimal control instruction set module is electrically connected to the optimal operation mode module of the heat pump system. The heat pump system digital twin module acquires historical operating data of the heat pump system, analyzes the multi-dimensional characteristics of the heat pump system under specific operating conditions, couples the heat pump system mechanism model, and establishes a virtual simulation heat pump system digital twin. The dynamic load prediction curve module acquires real-time operating data of the heat pump system based on multi-source heterogeneous sensors, introduces environmental factors as external dynamic disturbance variables, and generates a dynamic load prediction curve for the heat pump system. The optimal operation mode module of the heat pump system acquires user interaction data in the target area, generates user profiles, integrates the load prediction curve of the heat pump system, real-time electricity price information and equipment health status assessment indicators, and substitutes them into the digital twin of the virtual simulation heat pump system to perform multi-objective coupled simulation and deduction to obtain the optimal operation mode of the heat pump system. The global optimal control instruction set module, based on the obtained optimal operating mode of the heat pump system, takes the comprehensive energy efficiency ratio of the heat pump system and user comfort as the collaborative optimization objectives, and uses a multi-objective decision algorithm to generate the global optimal control instruction set of the heat pump system.

2. The heat pump cooling and heating cycle control system based on intelligent AI according to claim 1, characterized in that, The digital twin module of the heat pump system specifically includes: The data preprocessing unit acquires historical operating data of the heat pump system based on sensor and control signals, unifies the timestamps, and performs data preprocessing. The relationship curve unit is based on the historical operating data of the heat pump system. After standardization, the K-means clustering algorithm is used to divide each data point into K clusters. The sum of squared errors corresponding to the K value of each cluster is calculated, and the relationship curve between the K value and the sum of squared errors is plotted. The optimal number of cluster units is determined by finding the K value corresponding to the inflection point where the rate of decrease of the sum of squared errors in the graph significantly slows down, based on the relationship between the K value and the sum of squared errors. This K value is used as the optimal number of clusters for the historical operating data of the heat pump system. The historical operating condition division result unit is based on the optimal number of clusters K obtained from the historical operating data of the heat pump system. K initial centroids are randomly selected, and each data point is assigned to the cluster to which the nearest initial centroid belongs. The mean of the centroid of each cluster is calculated, and the process is iteratively updated until the centroids no longer change, thus obtaining the historical operating condition division result of the heat pump system. The operating conditions include: winter heating mode, summer cooling mode, and transitional season mode; The heat pump system energy consumption characteristic unit calculates the average power, peak power and cumulative energy consumption of the heat pump system under each operating condition based on the historical operating condition classification results of the heat pump system, and obtains the energy consumption characteristics of the heat pump system. The heat pump system performance characteristic unit calculates the average COP performance coefficient of the heat pump system under each operating condition, and generates a time-series variation curve of the average COP performance coefficient of the heat pump system by combining time data, thereby obtaining the performance characteristics of the heat pump system. The heat pump system operation mode characteristic unit calculates the compressor start-stop frequency and average running time of the heat pump system under each operating condition to obtain the heat pump system operation mode characteristics. The heat pump system statistical feature unit calculates the mean, variance, and extreme values ​​of the evaporation and condensation temperatures of the heat pump system under each operating condition, thus obtaining the statistical features of the heat pump system. Multidimensional feature units integrate the energy consumption characteristics, performance characteristics, operation mode characteristics and statistical characteristics of the heat pump system to obtain multidimensional features of the heat pump system under specific operating conditions.

3. A heat pump cooling and heating cycle control system based on intelligent AI according to claim 2, characterized in that, The digital twin module of the heat pump system also includes: The mass flow rate model and power consumption model unit, based on the performance data sheet provided by the compressor manufacturer, obtains the compressor volumetric efficiency and isentropic efficiency through a polynomial fitting method, and uses physical formulas to obtain the compressor mass flow rate model and power consumption model; The heat exchanger model unit is constructed by calculating the ratio of the heat exchanger NTU to the fluid heat capacity using the efficiency-heat transfer unit number method. The throttling device model unit is constructed based on the throttling principle and mass flow equation. The refrigerant property unit for heat pump systems calculates the enthalpy, entropy, and density of the refrigerant under different pressures and temperatures based on the integrated REFPROP refrigerant property database, thus obtaining the refrigerant properties of the heat pump system. The heat pump system mechanism model unit utilizes the mass conservation and energy conservation equations, coupled with the compressor mass flow rate model and power consumption model, heat exchanger model, throttling device model and refrigerant properties of the heat pump system, and uses the compressor frequency as the boundary condition. Using an iterative algorithm, the pressure, temperature, flow rate and power consumption at each point under the steady-state operating condition of the heat pump system are calculated to establish the heat pump system mechanism model. A digital twin unit is established to create a virtual simulation digital twin of a heat pump system based on the heat pump system mechanism model and the multi-dimensional characteristics of the heat pump system under specific operating conditions. The calibration unit uses a 24-hour data acquisition cycle, inputs the data into the digital twin of the virtual simulation heat pump system, calculates the total power consumption and outlet temperature of the system within the cycle, and compares them with the actual corresponding measured values ​​of the heat pump system mechanism model. If the average deviation between the two exceeds 10% within the cycle, the self-calibration process of the parameters of the digital twin of the virtual simulation heat pump system is automatically triggered. The compressor volumetric efficiency fitting parameters of the digital twin of the virtual simulation heat pump system are fine-tuned using the actual operating data of the most recent 24 hours to ensure simulation accuracy.

4. The heat pump cooling and heating cycle control system based on intelligent AI according to claim 1, characterized in that, The dynamic load forecasting curve module specifically includes: The multi-source heterogeneous sensor unit acquires real-time operating data of the heat pump system based on multi-source heterogeneous sensors and performs data preprocessing. If data from a certain sensor is missing for more than 5 minutes, the system automatically switches to the backup strategy. For outdoor sensors, the corresponding data obtained from the authoritative weather API interface is used, and for indoor temperature sensors, the average value of other available sensors in the same area is used. The multi-source heterogeneous sensor includes: Temperature sensors are installed at indoor return air inlets, supply air inlets, key rooms, and around the outdoor unit to monitor ambient temperature and coil temperature. The communication protocol used is Modbus RTU. The pressure sensor is installed at the compressor's suction and discharge ports, before and after the throttling device, to monitor the system pressure. The output signal is a 4-20mA analog signal. A power sensor is installed in the power supply circuit of the heat pump unit to monitor the system's operating power in real time. The communication protocol used is Modbus TCP. A humidity sensor, installed at the indoor return air vent, is used to monitor ambient humidity. A flow sensor, installed in the circulating water pipeline, is used to monitor water flow. All sensor data is aggregated through an IoT gateway and uploaded to the cloud platform data center using the MQTT protocol. The hourly forecast data unit, based on the weather forecast APP, obtains hourly forecast data for the target area for the next 24 hours, and obtains the dry-bulb temperature, wet-bulb temperature, humidity, wind speed, wind direction and cloud cover of the area. The Heat Pump System Operation-Meteorological Conditions Data Set Unit is based on real-time operation data of the heat pump system and hourly weather forecast data of the target area for the next 24 hours, with a unified timestamp, to establish a unified spatiotemporal benchmark for the Heat Pump System Operation-Meteorological Conditions Data Set. The historical load sequence unit sets the past 24 hours as a fixed sliding time window, extracts the historical load data of the heat pump system within the sliding time window, and establishes the historical load sequence of the heat pump system. The environmental factor unit records the periodic characteristics of each time step, extracts environmental factors within the sliding time window, and calculates the average load, maximum outdoor temperature, and temperature change rate of the heat pump system within the past sliding time window.

5. A heat pump cooling and heating cycle control system based on intelligent AI according to claim 4, characterized in that, The dynamic load forecasting curve module also includes: The dynamic load prediction model unit encodes the historical load sequence and periodic features of the heat pump system into vectors, introduces environmental factors as external dynamic perturbation variables, and uses a long short-term memory network to design two hidden layers with 128 neurons in each layer. The activation function is tanh. The input features are the historical load sequence, periodic features and environmental factors of the heat pump system in the past 24 hours. The objective is to minimize the mean square error between the predicted load and the actual load. The model learns the long-term dependence of the heat pump system load at different time points and establishes a dynamic load prediction model for the heat pump system. The dynamic load prediction curve unit establishes dynamic load prediction curves for heat pump systems based on the dynamic load prediction model of heat pump systems and the heat pump system operation-meteorological condition dataset with a unified spatiotemporal reference.

6. A heat pump cooling and heating cycle control system based on intelligent AI according to claim 1, characterized in that, The optimal operating mode module for a heat pump system specifically includes: The personalized profiling unit, based on the sensor backend, acquires user interaction data in the target area, analyzes the temperature and humidity range set by the user, identifies the user's comfortable temperature and humidity range in multiple scenarios, obtains the user's personalized preference characteristics, and establishes a personalized user profile. The user's multi-scenario temperature and humidity comfort range includes: residential scenario, office building scenario, and hotel scenario; Specifically, for residential scenarios, the system analyzes the temperature and humidity setting preferences of different family members at different times and in different rooms, and generates a comprehensive user profile based on the family unit through long-term learning. It automatically identifies home, work, and sleep modes and dynamically switches to the corresponding comfort zone and energy-saving strategy. For office building scenarios, we analyze the group preferences in public areas, collect feedback from most employees and monitor employee attendance rates, set a temperature and humidity range that is comfortable for most people, and enter a comfort-first mode in advance by combining calendar information and dynamic load forecast of the heat pump system before the meeting starts. For hotel scenarios, based on the room status, including whether it is occupied or vacant, personalized settings are made according to the guest's historical preferences obtained from the IoT cloud platform. When the guest checks in, the personalized environmental parameters are automatically restored. The load forecast range unit extracts the load forecast range of the heat pump system for future periods based on the dynamic load forecast curve of the heat pump system. The electricity price information unit obtains real-time electricity price information for the target area based on electricity market prices. The health status indicator unit, based on the equipment health management cloud platform, monitors the equipment status of the heat pump system in real time, calculates the compressor efficiency decay coefficient, the heat exchanger fouling coefficient and the filter resistance coefficient, and evaluates the equipment health status indicators. The compressor efficiency decay coefficient is specifically calculated as follows: a 30-day data collection period is set, the stable operating conditions of the heat pump system are selected, the virtual simulation of the heat pump system digital twin is substituted, the ratio of the theoretical isentropic efficiency of the compressor to the actual isentropic efficiency calculated based on the suction and discharge pressure and temperature is calculated, and the compressor efficiency decay coefficient is obtained. If the compressor efficiency decay coefficient is continuously lower than 0.9, a maintenance alarm is automatically triggered. The fouling factor of the heat exchanger is calculated as follows: with a two-week calculation cycle, the actual heat transfer coefficient of the heat exchanger is calculated by monitoring the temperature difference and flow rate on both sides of the heat exchanger. Combined with the digital twin of the virtual simulation heat pump system, the theoretical heat transfer coefficient of the heat exchanger under clean conditions is obtained. The ratio of the actual heat transfer coefficient to the theoretical heat transfer coefficient is used as the fouling factor of the heat exchanger. If the fouling factor of the heat exchanger is lower than 0.85, cleaning is recommended. The resistance coefficient of the filter screen is specifically determined by monitoring the pressure difference across the filter. The resistance coefficient of the filter screen is calculated in real time. If the pressure difference exceeds twice the initial monitoring value of the filter screen, it will prompt the filter screen to be replaced or cleaned.

7. A heat pump cooling and heating cycle control system based on intelligent AI according to claim 6, characterized in that, The optimal operating mode module for the heat pump system also includes: The state space unit is constructed based on user profiles, the predicted load range of the heat pump system in the future time period, real-time electricity price information and equipment health status indicators, and a unified time benchmark. The candidate optimization operation mode strategy set unit takes the state space as input and substitutes it into the digital twin of the virtual simulation heat pump system for coupled simulation and deduction. It uses user's individual preference characteristics, equipment physical limits and safe operating range as constraints to generate the candidate optimization operation mode strategy set of the heat pump system. Initial population unit: Initialize the set of candidate optimization operation modes for the heat pump system, and randomly generate a population of N individuals for candidate optimization operation modes for the heat pump system. The non-dominated ranking unit, for each individual population of candidate optimization operation mode strategies for heat pump systems, uses the heat pump system energy consumption, total operating cost and user comfort as multi-objective functions, divides each individual individual of candidate optimization operation mode strategies for heat pump systems into different frontier levels, and performs non-dominated ranking. The congestion calculation unit calculates the individual congestion of each heat pump system's candidate optimized operating mode strategy for the same frontier level. New offspring individual units are generated by selecting individuals with superior frontier levels and high crowding levels as parents based on different frontier levels after non-dominated sorting and individual crowding levels of each heat pump system candidate optimization operation mode strategy, and then performing crossover and mutation operations to generate a new offspring individual population. The optimal operating mode strategy unit merges the parent individuals with the new offspring individuals population, and selects the best N heat pump system candidate optimized operating mode strategy individuals through non-dominated sorting and crowding comparison to form a new generation population. Iterative updates are continuously performed to obtain the optimal operating mode strategy of the heat pump system.

8. A heat pump cooling and heating cycle control system based on intelligent AI according to claim 7, characterized in that, The internal modules of the global optimal control instruction set specifically include: The multi-objective decision-making model unit takes the optimal operating mode of the heat pump system as input and the comprehensive energy efficiency ratio of the heat pump system and user comfort as the collaborative optimization objectives to construct a multi-objective decision-making model. The weighted unit uses the entropy weighting method to calculate the entropy of each objective information of the heat pump system's comprehensive energy efficiency ratio and user comfort, and obtains the corresponding weights.

9. A heat pump cooling and heating cycle control system based on intelligent AI according to claim 8, characterized in that, The internal modules of the global optimal control instruction set also include: The ideal solution unit calculates the ideal and negative ideal solutions for each objective of the overall energy efficiency ratio of the heat pump system and user comfort, based on the weights of each objective. Euclidean distance unit, used to calculate the Euclidean distance between the ideal and negative ideal solutions of the overall energy efficiency ratio and user comfort of the heat pump system at each weight; The optimal control instruction set unit calculates the Euclidean distance between the optimal operating mode and the ideal solution and the negative ideal solution of each heat pump system, calculates the relative closeness between the optimal operating mode and the ideal solution of each heat pump system, sorts them in descending order, and obtains the global optimal control instruction set of the heat pump system.

Images