Ground source heat pump and solar photovoltaic integrated management system based on multi-energy complementation

Abstract

The invention relates to the technical field of building energy collaborative management and renewable energy grid-connected regulation and control, in particular to a ground source heat pump and solar photovoltaic integrated management system based on multi-energy complementation, which comprises a data acquisition module for acquiring a data output real-time operation data set. The multi-time scale prediction module generates a prediction curve and calculates a time sequence matching degree index. And the energy coordinated optimization module constructs a decision model by taking maximization of renewable energy local consumption as a target, and generates a storage battery charging and discharging instruction and a heat pump adjusting instruction according to a matching degree index. The execution control module converts the instruction into an equipment-level operation signal, controls the operation of the photovoltaic inverter, the heat pump frequency converter and the storage battery management system, and collects equipment state data and feeds the data back to the data collection module to form closed-loop control. Through multi-time scale prediction and real-time optimization scheduling, the problem that photovoltaic power generation intermittency is not matched with a building thermal load time sequence is effectively solved, and the energy utilization efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of building energy collaborative management and renewable energy grid connection control technology, and in particular to an integrated management system for ground source heat pumps and solar photovoltaics based on multi-energy complementarity. Background Technology

[0002] Ground source heat pumps and solar photovoltaics are two common renewable energy technologies widely used in building energy systems. Ground source heat pump systems circulate heat transfer media through underground pipes, utilizing the stability of underground temperatures for heat transfer. Underground temperatures vary little throughout the year, being lower than air temperatures in summer and higher than air temperatures in winter. Therefore, ground source heat pumps can efficiently absorb heat from the ground for heating or release heat into the ground for cooling. This process relies on thermodynamic principles, achieving energy savings. Solar photovoltaic technology directly converts solar radiation energy into electrical energy. Photovoltaic modules are made of semiconductor materials; when sunlight shines on them, photon energy excites electrons, generating direct current (DC). An inverter converts this DC to alternating current (AC), which can be used by building electrical equipment. Solar energy is a widely distributed resource, and photovoltaic systems operate without fuel consumption, reducing carbon emissions. Combining ground source heat pumps and solar photovoltaics can form a complementary energy system.

[0003] Existing grid-connected photovoltaic (PV) and ground-source heat pump co-generation technologies suffer from the following technical challenges: The output power of solar PV is significantly volatile and intermittent due to variations in solar irradiance, while the heating and cooling loads of public buildings exhibit relatively stable temporal characteristics. In hot-summer, cold-winter regions, during the peak PV power generation period at midday in summer, although the building's cooling load is high, the ground-source heat pump units, limited by their own power regulation capabilities and the building's thermal inertia, cannot absorb all the surplus PV power in real time. Conversely, during the peak load periods in the morning and evening in winter, PV power generation is at a low point. This mismatch in source-load temporal characteristics necessitates frequent reverse transmission of PV power to the municipal grid, exacerbating grid regulation pressure and reducing the proportion of renewable energy absorbed locally by buildings. For example, during the rainy season, continuous cloudy and rainy weather causes a sharp drop in PV power generation, requiring the ground-source heat pump to heavily rely on grid power, significantly degrading the system's overall energy efficiency. The root cause lies in the lack of a multi-energy flow dynamic coordination mechanism across time scales, failing to achieve adaptive coupling between the PV power generation curve and the building's energy consumption curve. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an integrated management system for ground source heat pumps and solar photovoltaics based on multi-energy complementarity. This system solves the technical problem that in public buildings in hot-summer and cold-winter regions, solar photovoltaic power generation is intermittent due to weather and seasonal influences, which does not match the heating and cooling load time characteristics of ground source heat pumps, resulting in a low local consumption rate of renewable energy and a decline in system operating efficiency.

[0005] To solve the above-mentioned technical problems, the specific contents of the present invention are as follows: The integrated management system for ground source heat pumps and solar photovoltaics based on multi-energy complementarity provided by this invention includes: The data acquisition module continuously acquires real-time power generation of the solar photovoltaic array, operating status parameters of the ground source heat pump unit, charging and discharging capacity of the battery energy storage unit, real-time energy consumption data inside the building, and environmental meteorological data, and outputs a real-time operating dataset. The multi-timescale prediction module receives the real-time running dataset, uses machine learning algorithms to perform rolling predictions of photovoltaic power generation and building heat load for future periods, outputs photovoltaic power generation prediction curves and heat load prediction curves, and calculates the time series matching index based on the photovoltaic power generation prediction curves and the heat load prediction curves. The energy coordination and optimization module receives the photovoltaic power generation forecast curve, the heat load forecast curve, and the time-series matching degree index. It constructs a decision model with the optimization objective of maximizing the local consumption ratio of renewable energy. When the time-series matching degree index indicates that the photovoltaic power generation is higher than the heat load demand, it generates a battery charging command and a ground source heat pump pre-adjustment command. When the time-series matching degree index indicates that the photovoltaic power generation is lower than the heat load demand, it generates a battery discharging command and a ground source heat pump power adjustment command. The execution control module receives the battery charging command and the ground source heat pump pre-adjustment command or the battery discharging command and the ground source heat pump power adjustment command, and converts the command into equipment-level operation signals. The equipment-level operation signals are sent to the photovoltaic inverter, the ground source heat pump frequency converter and the battery management system respectively to adjust the photovoltaic power generation output, the heat pump operating power and the battery charging and discharging status. The execution control module also collects the output frequency of the photovoltaic inverter, the speed of the ground source heat pump compressor, and the charging and discharging current of the battery as equipment status data, and feeds the equipment status data back to the data acquisition module to update the real-time operation dataset.

[0006] Furthermore, in the multi-energy complementary ground source heat pump and solar photovoltaic integrated management system of the present invention, the data acquisition module includes: The sensor data acquisition unit collects DC-side electrical parameters through a photovoltaic power sensor, reads compressor frequency and refrigerant temperature through a heat pump unit controller, acquires state of charge data through a battery management system, collects temperature and humidity parameters through sensors inside the building, and collects solar intensity and ambient temperature through a weather station. The data preprocessing unit receives the DC-side electrical parameters, the compressor frequency and refrigerant temperature, the state of charge data, the temperature and humidity parameters, and the solar radiation intensity and ambient temperature. It performs timestamp alignment and filtering on the received data and outputs standardized data frames. The communication interface unit receives the standardized data frame and uploads the standardized data frame to the multi-timescale prediction module.

[0007] Furthermore, in the multi-energy complementary ground source heat pump and solar photovoltaic integrated management system of the present invention, the multi-time-scale prediction module includes: The short-term forecasting unit uses a long short-term memory network, takes historical photovoltaic power generation sequences and real-time meteorological data as input, and outputs the photovoltaic power generation point prediction value for the next few hours; The curve generation unit receives the predicted value of the photovoltaic power generation point and generates the photovoltaic power generation prediction curve; at the same time, it receives historical and real-time data of building heat load and generates the heat load prediction curve. The matching degree evaluation unit receives the photovoltaic power generation prediction curve and the heat load prediction curve, calculates the Pearson correlation coefficient and power deviation rate between the two curves, and outputs the time series matching degree index.

[0008] Furthermore, in the multi-energy complementary ground source heat pump and solar photovoltaic integrated management system of the present invention, the energy coordination and optimization module includes: A multi-objective decision-making unit is used to construct a linear programming model that includes constraints on battery charge / discharge rate, heat pump operating temperature range, and building comfort. The real-time scheduling unit receives the time-series matching degree index, the photovoltaic power generation prediction curve and the heat load prediction curve. Based on the time-series matching degree index, it calls the linear programming model to calculate the battery charging power command and the ground source heat pump pre-cooling / preheating capacity during the photovoltaic power surplus period, and optimizes the battery discharge sequence and heat pump set temperature during the power shortage period. The grid interaction unit receives time-of-use electricity price signals and the scheduling results of the real-time scheduling unit, and dynamically adjusts the active power reference value of the grid connection point.

[0009] Furthermore, in the integrated management system for ground source heat pump and solar photovoltaic based on multi-energy complementarity described in this invention, the execution control module includes: The instruction parsing unit receives the battery charging instruction and the ground source heat pump pre-adjustment instruction or the battery discharging instruction and the ground source heat pump power adjustment instruction, and decodes the instruction into the photovoltaic inverter active power setting value, the ground source heat pump compressor frequency setting value and the battery charging and discharging current threshold. The equipment control unit receives the active power setpoint, the compressor frequency setpoint, and the charging and discharging current threshold, and sends control words including each setpoint to the photovoltaic inverter, the ground source heat pump frequency converter, and the battery management system via the fieldbus. The status feedback unit collects the output frequency of the photovoltaic inverter, the speed of the ground source heat pump compressor, and the charging and discharging current of the battery, and performs deviation verification with the set values ​​in the control word. When the deviation exceeds the preset threshold, it sends a re-optimization request to the energy coordination and optimization module.

[0010] Furthermore, in the multi-energy complementary ground source heat pump and solar photovoltaic integrated management system of the present invention, the short-term prediction unit is configured as follows: Extract special event markers from the building operations and maintenance database; these special event markers include holiday identifiers and schedules of major events. The special event markers, the historical photovoltaic power generation sequence, and the real-time meteorological data are combined into an input feature vector; The input feature vector is input into the long short-term memory network; During the computation of the Long Short-Term Memory network, dynamic weight coefficients are configured for the special event markers; Based on the input feature vector configured with the dynamic weight coefficients, the predicted value of the photovoltaic power generation point is output.

[0011] Furthermore, in the multi-energy complementary ground source heat pump and solar photovoltaic integrated management system of the present invention, the real-time scheduling unit is configured as follows: Analyze the decay curve of building indoor temperature during heating or cooling cycles by accessing historical operating data. Based on the attenuation curve, the thermal time constant of the building envelope is calculated. Based on the thermal time constant, calculate the start time and duration of the precooling or preheating operation relative to the predicted peak heat load period; The start-up time and the duration are written into the ground source heat pump pre-adjustment command.

[0012] Furthermore, in the multi-energy complementary ground source heat pump and solar photovoltaic integrated management system of the present invention, the state feedback unit is configured as follows: Continuously monitor the real-time deviation between the actual speed of the compressor and the set frequency of the compressor; Determine whether the absolute value of the real-time deviation exceeds a preset threshold for three consecutive control cycles; When the judgment result is yes, a re-optimization request signal is generated; The re-optimization request signal is sent to the real-time scheduling unit of the energy coordination and optimization module.

[0013] Furthermore, in the multi-energy complementary ground source heat pump and solar photovoltaic integrated management system of the present invention, the multi-time-scale prediction module is configured as follows: Access the meteorological satellite ground receiving system to obtain visible light and infrared spectral satellite cloud images covering the geographical location of the building; Perform geographic coordinate registration and radiometric calibration on the satellite cloud image; For the processed satellite cloud images, a convolutional neural network is used to extract the transmittance gradient distribution and movement vector features of the cloud clusters; The extracted transmittance gradient distribution and moving vector features are fused with the ground solar radiation intensity data acquired by the data acquisition module. Based on the fused data, a revised predicted value of solar radiation intensity is generated. The photovoltaic power generation prediction curve is dynamically corrected using the corrected solar irradiance prediction value.

[0014] Furthermore, the integrated management system for ground source heat pump and solar photovoltaic based on multi-energy complementarity described in this invention also includes: The status feedback unit of the execution control module feeds back the collected photovoltaic inverter output frequency, ground source heat pump compressor speed and battery charging and discharging current as equipment status data to the data acquisition module. The data acquisition module receives the device status data and uses the device status data to update the real-time running dataset; The multi-timescale prediction module receives the updated real-time running dataset for model self-correction during subsequent rolling prediction processes.

[0015] Beneficial effects of this invention: This invention continuously acquires real-time power generation data of solar photovoltaic arrays, operating status parameters of ground source heat pump units, charging and discharging capacity of battery energy storage units, real-time energy consumption data inside buildings, and environmental meteorological data through a data acquisition module, forming a real-time operation dataset to provide a comprehensive basis for system operation status perception. A multi-timescale prediction module receives the real-time operation dataset and uses machine learning algorithms to perform rolling predictions of photovoltaic power generation and building heat load, outputting photovoltaic power generation prediction curves and heat load prediction curves. It also quantifies the source-load matching degree by calculating a time-series matching degree index. An energy coordination and optimization module, based on the time-series matching degree index, photovoltaic power generation prediction curves, and heat load prediction curves, aims to maximize the local consumption ratio of renewable energy. A decision-making model is constructed to generate battery charging commands and ground source heat pump pre-regulation commands, or battery discharging commands and ground source heat pump power adjustment commands, to achieve dynamic energy scheduling across time periods. The execution control module converts the optimized commands into equipment-level operation signals to adjust the operating status of the photovoltaic inverter, ground source heat pump frequency converter, and battery management system. At the same time, the status feedback unit collects equipment status data and feeds it back to the data acquisition module to update the real-time operation dataset, forming a closed-loop control. This multi-module collaborative working mechanism effectively smooths out the volatility of photovoltaic power generation, uses battery energy storage and building thermal inertia to buffer source-load mismatch, thereby significantly improving the local consumption ratio of renewable energy in public buildings in hot-summer and cold-winter regions, and enhancing the system's operational stability and energy efficiency. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on the accompanying drawings without creative effort.

[0017] Figure 1 This is a system architecture diagram of the integrated management system of ground source heat pump and solar photovoltaic based on multi-energy complementarity of the present invention. Detailed Implementation

[0018] To make the technical solution of the present invention clearer, the present invention will be clearly and completely described below with reference to specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. The present invention provided by various embodiments will be described in detail below with reference to the accompanying drawings. To better understand the purpose of the present invention, the present invention will be described in further detail below.

[0019] Please see Figure 1The present invention provides an integrated management system for ground source heat pumps and solar photovoltaics based on multi-energy complementarity, comprising: The data acquisition module continuously acquires real-time power generation of the solar photovoltaic array, operating status parameters of the ground source heat pump unit, charging and discharging capacity of the battery energy storage unit, real-time energy consumption data inside the building, and environmental meteorological data, and outputs a real-time operating dataset. The multi-timescale prediction module receives the real-time running dataset, uses machine learning algorithms to perform rolling predictions of photovoltaic power generation and building heat load for future periods, outputs photovoltaic power generation prediction curves and heat load prediction curves, and calculates the time series matching index based on the photovoltaic power generation prediction curves and the heat load prediction curves. The energy coordination and optimization module receives the photovoltaic power generation forecast curve, the heat load forecast curve, and the time-series matching degree index. It constructs a decision model with the optimization objective of maximizing the local consumption ratio of renewable energy. When the time-series matching degree index indicates that the photovoltaic power generation is higher than the heat load demand, it generates a battery charging command and a ground source heat pump pre-adjustment command. When the time-series matching degree index indicates that the photovoltaic power generation is lower than the heat load demand, it generates a battery discharging command and a ground source heat pump power adjustment command. The execution control module receives the battery charging command and the ground source heat pump pre-adjustment command or the battery discharging command and the ground source heat pump power adjustment command, and converts the command into equipment-level operation signals. The equipment-level operation signals are sent to the photovoltaic inverter, the ground source heat pump frequency converter and the battery management system respectively to adjust the photovoltaic power generation output, the heat pump operating power and the battery charging and discharging status. The execution control module also collects the output frequency of the photovoltaic inverter, the speed of the ground source heat pump compressor, and the charging and discharging current of the battery as equipment status data, and feeds the equipment status data back to the data acquisition module to update the real-time operation dataset.

[0020] The data acquisition module continuously collects DC-side electrical parameters of the solar photovoltaic array, compressor frequency and refrigerant temperature of the ground source heat pump unit, state of charge data of the battery energy storage unit, temperature and humidity parameters inside the building, and environmental meteorological data such as solar radiation intensity and ambient temperature through the sensor data acquisition unit. The data preprocessing unit performs timestamp alignment and filtering on the collected DC-side electrical parameters, compressor frequency and refrigerant temperature, state of charge data, temperature and humidity parameters, and solar radiation intensity and ambient temperature, and outputs standardized data frames. The communication interface unit uploads the standardized data frames to the multi-timescale prediction module to form a real-time operating dataset.

[0021] The multi-timescale prediction module receives real-time operational datasets. The short-term prediction unit employs a Long Short-Term Memory (LSTM) network, taking historical photovoltaic (PV) power generation sequences and real-time meteorological data as inputs, and outputs predicted PV power output values ​​for the next few hours. The short-term prediction unit extracts special event markers from the building operations and maintenance database, including holiday identifiers and large-scale event schedules. It combines these special event markers, historical PV power generation sequences, and real-time meteorological data into an input feature vector. During the LTM network's computation, dynamic weight coefficients are configured for these special event markers. Based on this dynamically weighted input feature vector, the predicted PV power output values ​​are output. The curve generation unit receives the predicted PV power output values ​​and generates a PV power generation prediction curve. Simultaneously, it receives historical and real-time building heat load data and generates a heat load prediction curve. The matching degree evaluation unit receives the PV power generation prediction curve and the heat load prediction curve, calculates the Pearson correlation coefficient and power deviation rate between the two curves, and outputs a time-series matching degree index. The multi-timescale prediction module also connects to the meteorological satellite ground receiving system to acquire visible light and infrared spectral satellite cloud images covering the geographical location of the building. It performs geographic coordinate registration and radiometric calibration on the satellite cloud images, and uses a convolutional neural network to extract the transmittance gradient distribution and movement vector features of the cloud clusters. The extracted transmittance gradient distribution and movement vector features are then fused with the ground solar radiation intensity data acquired by the data acquisition module. Based on the fused data, a corrected solar radiation intensity prediction value is generated, and the corrected solar radiation intensity prediction value is used to perform dynamic correction on the photovoltaic power generation prediction curve.

[0022] The energy coordination and optimization module receives photovoltaic power generation forecast curves, heat load forecast curves, and time-series matching degree indicators. The multi-objective decision-making unit constructs a linear programming model including constraints on battery charge / discharge rate, heat pump operating temperature range, and building comfort. The real-time scheduling unit, based on the time-series matching degree indicators, calls the linear programming model. When the time-series matching degree indicators suggest that photovoltaic power generation is higher than the heat load demand, it calculates the battery charging power command and the ground source heat pump pre-cooling or pre-heating capacity; when the time-series matching degree indicators suggest that photovoltaic power generation is lower than the heat load demand, it optimizes the battery discharge sequence and the heat pump set temperature. The real-time scheduling unit calls historical operating data to analyze the decay curve of building indoor temperature during the heating or cooling cycle. Based on the decay curve, it calculates the thermal time constant of the building envelope and, based on the thermal time constant, calculates the start time and duration of pre-cooling or pre-heating operations relative to the predicted peak heat load period. The start time and duration are then written into the ground source heat pump pre-regulation command. The grid interaction unit receives time-of-use electricity price signals and the scheduling results from the real-time scheduling unit, dynamically adjusting the active power reference value at the grid connection point.

[0023] The execution control module receives battery charging commands and ground source heat pump pre-adjustment commands or battery discharging commands and ground source heat pump power adjustment commands generated by the energy coordination and optimization module. The command parsing unit decodes the commands into photovoltaic inverter active power setpoints, ground source heat pump compressor frequency setpoints, and battery charging and discharging current thresholds. The equipment control unit receives the active power setpoints, compressor frequency setpoints, and charging and discharging current thresholds, and sends control words containing each setpoint to the photovoltaic inverter, ground source heat pump inverter, and battery management system via fieldbus. The status feedback unit collects the photovoltaic inverter output frequency, ground source heat pump compressor speed, and battery charging and discharging current as equipment status data, and performs deviation verification with the setpoints in the control words. The status feedback unit continuously monitors the real-time deviation between the actual compressor speed and the compressor frequency setpoint, and determines whether the absolute value of the real-time deviation exceeds the preset threshold for three consecutive control cycles. When the determination result is yes, a re-optimization request signal is generated and sent to the real-time scheduling unit of the energy coordination and optimization module. The status feedback unit also feeds back the collected photovoltaic inverter output frequency, ground source heat pump compressor speed and battery charging and discharging current as equipment status data to the data acquisition module. The data acquisition module receives the equipment status data and uses the equipment status data to update the real-time running dataset. The multi-timescale prediction module receives the updated real-time running dataset and uses it for model self-correction in the subsequent rolling prediction process.

[0024] The data acquisition module's sensor data acquisition unit collects DC-side electrical parameters, including voltage and current values, through a photovoltaic power sensor; reads compressor frequency and refrigerant temperature parameters through the heat pump unit controller; acquires state of charge data through the battery management system; collects temperature and humidity parameters through building interior sensors; and collects solar radiation intensity and ambient temperature data through a weather station. The data preprocessing unit receives data from various sources, performs timestamp alignment to unify data timing, and uses filtering algorithms to eliminate high-frequency noise, outputting standardized data frames. The communication interface unit receives the standardized data frames and uploads the data to the multi-timescale prediction module via Ethernet or wireless communication protocols, thereby ensuring the integrity and consistency of the real-time running dataset.

[0025] The short-term forecasting unit of the multi-timescale forecasting module employs a long short-term memory network model, using historical photovoltaic (PV) power generation sequences and real-time meteorological data as input features. Through forward propagation of the neural network, it calculates and outputs predicted PV power output values ​​for the next few hours. The curve generation unit receives the predicted values ​​and combines them with timestamps to generate a smooth PV power generation forecast curve. Simultaneously, it generates a heat load forecast curve based on historical building heat load data and real-time monitoring data. The matching degree evaluation unit receives the two forecast curves, calculates the Pearson correlation coefficient to assess the similarity of the curve shapes, and calculates the power deviation rate to quantify energy differences. Finally, it outputs a time-series matching degree index, providing a quantitative basis for energy coordination.

[0026] The multi-objective decision-making unit of the energy coordination and optimization module constructs a linear programming model. This model incorporates constraints on battery charge / discharge rates to prevent battery overload, heat pump operating temperature range constraints to maintain equipment safety, and building comfort constraints to ensure a safe indoor environment. The real-time scheduling unit receives time-series matching indicators, photovoltaic power generation forecast curves, and heat load forecast curves. Based on the indicator values, it calls the linear programming model to solve the problem. During periods of photovoltaic power surplus, it calculates battery charging power commands and ground-source heat pump pre-cooling or pre-heating capacity. During periods of power shortage, it optimizes the battery discharge sequence and heat pump set temperature. The grid interaction unit receives time-of-use pricing signals and the output of the real-time scheduling unit, dynamically adjusting the active power reference value at the grid connection point to achieve coordinated operation with the grid.

[0027] The instruction parsing unit of the execution control module receives instructions generated by the energy coordination and optimization module, such as battery charging instructions and ground source heat pump pre-adjustment instructions. It decodes these instructions into photovoltaic inverter active power setpoints, ground source heat pump compressor frequency setpoints, and battery charging / discharging current thresholds. The equipment control unit receives these setpoints, encapsulates them into control words conforming to the fieldbus protocol, and sends them to the photovoltaic inverter, ground source heat pump inverter, and battery management system. The status feedback unit collects actual operating parameters of the equipment, including the photovoltaic inverter output frequency, ground source heat pump compressor speed, and battery charging / discharging current. It compares these parameters with the setpoints in the control words in real time. When the deviation exceeds a preset threshold, it sends a re-optimization request to the energy coordination and optimization module, forming a control closed loop.

[0028] The short-term forecasting unit extracts special event markers from the building operations and maintenance database, such as holiday identifiers and large-scale event schedules. These markers are combined with historical photovoltaic power generation sequences and real-time meteorological data to form an input feature vector. This input feature vector is fed into a long short-term memory (LSTM) network. During network training, dynamic weight coefficients are assigned to the special event markers to adjust the intensity of the events' impact on the prediction results. Based on the weighted input feature vector, the LTM network outputs predicted photovoltaic power generation points, thereby improving the prediction model's adaptability to abnormal events.

[0029] The real-time scheduling unit retrieves historical operating data to analyze the decay curve of building indoor temperature during heating or cooling cycles. Through curve fitting, it calculates the thermal time constant of the building envelope. Based on the thermal time constant, it derives the start time and duration of pre-cooling or pre-heating operations relative to the predicted peak heat load period, thus utilizing the building's thermal inertia. The start time and duration are written into the ground source heat pump pre-regulation command, enabling proactive energy scheduling.

[0030] The status feedback unit continuously monitors the real-time deviation between the actual compressor speed and the compressor frequency setpoint, and uses a sliding window method to determine whether the absolute value of the deviation exceeds a preset threshold for three consecutive control cycles. When the condition is met, the status feedback unit generates a re-optimization request signal and sends the signal to the real-time scheduling unit of the energy coordination and optimization module, triggering the system to recalculate the optimization instructions.

[0031] The multi-timescale prediction module connects to the meteorological satellite ground receiving system to acquire visible and infrared spectral satellite cloud images covering the building's geographical location. Geographic coordinate registration and radiometric calibration are performed on the cloud images to eliminate geometric and radiometric errors. A convolutional neural network is used to extract the transmittance gradient distribution and movement vector features of cloud clusters. These features are then fused with ground solar radiation intensity data acquired by the data acquisition module to generate a corrected solar radiation intensity prediction value. This corrected solar radiation intensity prediction value is used to dynamically correct the photovoltaic power generation prediction curve, improving prediction accuracy.

[0032] The status feedback unit of the execution control module feeds back the collected photovoltaic inverter output frequency, ground source heat pump compressor speed, and battery charging and discharging current as equipment status data to the data acquisition module. The data acquisition module receives the equipment status data and uses it to update the real-time operating dataset. The multi-timescale prediction module receives the updated real-time operating dataset and performs model parameter self-calibration during subsequent rolling prediction processes, enabling continuous learning and optimization of the system.

[0033] This invention addresses the technical problem of the intermittent nature of photovoltaic (PV) power generation and the mismatch between the timing of heat load and the demand in public buildings in hot-summer and cold-winter regions by constructing a closed-loop control system. A data acquisition module continuously acquires real-time power generation data from the solar PV array, operating parameters of the ground-source heat pump unit, charging and discharging capacity of the battery storage unit, real-time energy consumption data within the building, and environmental meteorological data, forming a real-time operational dataset to provide comprehensive operational status awareness for the system. A multi-timescale prediction module receives the real-time operational dataset and uses machine learning algorithms to perform rolling predictions of PV power generation and building heat load, outputting PV power generation prediction curves and heat load prediction curves, and calculating the timing match index between them to identify source-load differences in advance. An energy coordination and optimization module, based on the timing match index, PV power generation prediction curves, and heat load prediction curves, constructs a decision model with the optimization objective of maximizing the local consumption ratio of renewable energy. When PV power generation exceeds heat load demand, it generates battery charging instructions and ground-source heat pump pre-adjustment instructions; when PV power generation is lower than heat load demand, it generates battery discharging instructions and ground-source heat pump power adjustment instructions, achieving cross-time-period energy scheduling. The execution control module converts optimization commands into device-level operation signals and sends them to the photovoltaic inverter, ground source heat pump frequency converter, and battery management system, directly adjusting the photovoltaic power output, heat pump operating power, and battery charging and discharging status. Simultaneously, the execution control module collects device status data and feeds it back to the data acquisition module, updating the real-time operating dataset. This enables the prediction module to perform model self-correction, forming an adaptive optimization loop. This dynamic coordination mechanism effectively smooths the volatility of photovoltaic power generation, utilizing battery energy storage and building thermal inertia to buffer source-load mismatch, thereby improving the local consumption ratio of renewable energy and system operating efficiency.

[0034] This invention relates to an integrated management system for ground-source heat pumps and solar photovoltaics based on multi-energy complementarity. Addressing the technical problem of the mismatch between the intermittent nature of solar photovoltaic power generation and the temporal characteristics of heating and cooling loads from ground-source heat pumps in public buildings in hot-summer and cold-winter regions, this invention improves the local consumption ratio of renewable energy by constructing a closed-loop control architecture. During system operation, the data acquisition module collects DC-side electrical parameters of the solar photovoltaic array, compressor frequency and refrigerant temperature of the ground-source heat pump unit, state-of-charge data of the battery energy storage unit, temperature and humidity parameters inside the building, and environmental meteorological data such as solar radiation intensity and ambient temperature through the sensor data acquisition unit. The data preprocessing unit performs timestamp alignment and filtering on the collected data, outputting standardized data frames. The communication interface unit uploads the standardized data frames to the multi-timescale prediction module to form a real-time operating dataset.

[0035] After receiving the real-time operational dataset, the multi-timescale prediction module employs a Long Short-Term Memory (LSTM) network, using historical photovoltaic (PV) power generation sequences and real-time meteorological data as input features to output predicted PV power output values ​​for the next few hours. The LSTM unit extracts special event markers from the building operations and maintenance database, including holiday identifiers and large-scale event schedules. These special event markers are combined with historical PV power generation sequences and real-time meteorological data to form an input feature vector. Dynamic weighting coefficients are assigned to these special event markers during the LTM network calculation process, and the predicted PV power output values ​​are output based on the weighted input feature vector. The curve generation unit receives the predicted PV power output values ​​and generates a PV power generation prediction curve. Simultaneously, it receives historical and real-time building heat load data to generate a heat load prediction curve. The matching degree evaluation unit receives the PV power generation prediction curve and the heat load prediction curve, calculates the Pearson correlation coefficient and power deviation rate, and outputs a time-series matching degree index. The multi-timescale prediction module also connects to the meteorological satellite ground receiving system to acquire visible light and infrared spectral satellite cloud images. It performs geographic coordinate registration and radiometric calibration on the satellite cloud images, and uses a convolutional neural network to extract the transmittance gradient distribution and movement vector features of cloud clusters. The extracted features are then fused with the ground solar intensity data acquired by the data acquisition module to generate a corrected solar intensity prediction value. This corrected solar intensity prediction value is then used to dynamically correct the photovoltaic power generation prediction curve.

[0036] The energy coordination and optimization module receives photovoltaic power generation forecast curves, heat load forecast curves, and time-series matching indexes. The multi-objective decision-making unit constructs a linear programming model, including constraints on battery charge / discharge rates, heat pump operating temperature ranges, and building comfort. The real-time scheduling unit invokes the linear programming model based on the time-series matching index, calculating battery charging power commands and ground-source heat pump pre-cooling or pre-heating capacity during periods of photovoltaic power surplus, and optimizing battery discharge sequences and heat pump set temperatures during periods of power shortage. The real-time scheduling unit retrieves historical operating data, analyzes the decay curve of building indoor temperature during heating or cooling cycles, calculates the thermal time constant of the building envelope based on the decay curve, and calculates the start time and duration of pre-cooling or pre-heating operations relative to the predicted peak heat load period, writing the start time and duration into the ground-source heat pump pre-regulation command. The grid interaction unit receives time-of-use electricity price signals and scheduling results from the real-time scheduling unit, dynamically adjusting the active power reference value at the grid connection point.

[0037] The execution control module receives battery charging commands and ground source heat pump pre-adjustment commands or battery discharging commands and ground source heat pump power adjustment commands generated by the energy coordination and optimization module. The command parsing unit decodes the commands into photovoltaic inverter active power setpoints, ground source heat pump compressor frequency setpoints, and battery charging and discharging current thresholds. The equipment control unit receives each setpoint and sends control words to the photovoltaic inverter, ground source heat pump inverter, and battery management system via fieldbus. The status feedback unit collects the photovoltaic inverter output frequency, ground source heat pump compressor speed, and battery charging and discharging current as equipment status data and performs deviation verification with the setpoints in the control words. The status feedback unit continuously monitors the real-time deviation between the actual compressor speed and the compressor frequency setpoint, determines whether the absolute value of the deviation exceeds the preset threshold for three consecutive control cycles, and generates a re-optimization request signal when the condition is met, sending it to the real-time scheduling unit of the energy coordination and optimization module. The status feedback unit also feeds back the equipment status data to the data acquisition module, which uses the equipment status data to update the real-time operating dataset. The multi-timescale prediction module receives the updated real-time operating dataset for model self-correction, forming an adaptive optimization loop. This invention effectively mitigates photovoltaic power generation fluctuations through multi-timescale prediction and real-time coordination optimization, and utilizes battery energy storage and building thermal inertia to buffer source-load mismatch, thereby improving the renewable energy absorption efficiency and system operation stability of public buildings in hot-summer and cold-winter regions.

[0038] Embodiment 1 of this invention: In hot-summer and cold-winter regions, during the peak midday photovoltaic power generation period in summer, the output power of the solar photovoltaic array is high, but the building's cooling load cannot be matched in real time due to thermal inertia. In this invention's system, the data acquisition module collects photovoltaic DC-side electrical parameters, ground source heat pump compressor frequency and refrigerant temperature, battery state of charge data, building interior temperature and humidity parameters, and environmental meteorological data through the sensor data acquisition unit. The data preprocessing unit performs timestamp alignment and filtering on the data to generate standardized data frames. The communication interface unit uploads the data to the multi-timescale prediction module. The short-term prediction unit of the multi-timescale prediction module uses a long short-term memory network to process historical photovoltaic power generation sequences and real-time meteorological data, outputting photovoltaic power generation point prediction values. The curve generation unit generates a photovoltaic power generation prediction curve based on the point prediction values, and simultaneously generates a heat load prediction curve by combining historical building heat load data. The matching degree evaluation unit calculates the Pearson correlation coefficient and power deviation rate of the two curves and outputs a time series matching degree index. The multi-objective decision-making unit of the energy coordination and optimization module constructs a linear programming model. The real-time scheduling unit calculates the battery charging power command and the ground source heat pump precooling capacity based on the time-series matching degree index indicating that the photovoltaic power generation is higher than the heat load demand. The grid interaction unit adjusts the grid-connected power reference value in conjunction with the time-of-use electricity price signal. The command parsing unit of the execution control module decodes the command into the active power setpoint of the photovoltaic inverter, the frequency setpoint of the ground source heat pump compressor, and the battery charging and discharging current threshold. The equipment control unit issues control words through the fieldbus. The status feedback unit collects the actual operating parameters of the equipment for deviation verification. When the deviation exceeds the limit, it sends a re-optimization request to the energy coordination and optimization module and simultaneously feeds back the equipment status data to the data acquisition module to update the real-time operating dataset, thus realizing closed-loop optimization.

[0039] Embodiment 2 of this invention: During peak load periods in the morning and evening of winter, photovoltaic power generation is low while building heating demand is significant. The data acquisition module continuously monitors environmental meteorological data and equipment operating parameters. The multi-timescale prediction module connects to the meteorological satellite ground receiving system to acquire visible and infrared spectral satellite cloud images. Geographic coordinate registration and radiometric calibration are performed on the cloud images. A convolutional neural network is used to extract the gradient distribution and movement vector features of cloud transmittance. These are then fused with ground solar radiation intensity data to generate a corrected solar radiation intensity prediction value, used to dynamically correct the photovoltaic power generation prediction curve. The short-term prediction unit extracts special event markers such as holiday identifiers from the building operation and maintenance database and configures dynamic weighting coefficients to optimize the prediction output. The real-time scheduling unit of the energy coordination and optimization module calls historical operating data to analyze the building's indoor temperature decay curve, calculates the building envelope thermal time constant, calculates the preheating operation start time and duration based on the thermal time constant, and writes it into the ground source heat pump pre-adjustment command. When the time-series matching index indicates that the photovoltaic power generation is lower than the heat load demand, the battery discharge sequence and heat pump set temperature are optimized. The status feedback unit of the execution control module continuously monitors the deviation between the actual compressor speed and the set value. When the limit is exceeded for three consecutive control cycles, re-optimization is triggered. The equipment control unit adjusts the heat pump operating power and the battery discharge status, and at the same time feeds back the equipment status data to the data acquisition module, so that the multi-timescale prediction module can realize model self-correction based on the updated data, thereby improving the system adaptability.

Claims

1. A multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system, characterized in that, Comprise: a data acquisition module that continuously acquires real-time power generation of a solar photovoltaic array, operating state parameters of a ground source heat pump unit, charge and discharge capacity of a battery energy storage unit, real-time energy consumption data inside a building, and environmental meteorological data, and outputs real-time operation data sets; a multi-time scale prediction module that receives the real-time operation data sets, uses a machine learning algorithm to perform rolling prediction on photovoltaic power generation and building heat load in a future period, outputs a photovoltaic power generation prediction curve and a heat load prediction curve, and calculates a time sequence matching degree index based on the photovoltaic power generation prediction curve and the heat load prediction curve; an energy coordination optimization module that receives the photovoltaic power generation prediction curve, the heat load prediction curve, and the time sequence matching degree index, constructs a decision-making model with the optimization goal of maximizing the proportion of renewable energy consumption on site, generates battery charging instructions and ground source heat pump pre-regulation instructions when the time sequence matching degree index indicates that photovoltaic power generation is higher than heat load demand, generates battery discharge instructions and ground source heat pump power adjustment instructions when the time sequence matching degree index indicates that photovoltaic power generation is lower than heat load demand; an execution control module that receives the battery charging instructions and ground source heat pump pre-regulation instructions or the battery discharge instructions and ground source heat pump power adjustment instructions, converts the instructions into device-level operation signals, and sends the device-level operation signals to a photovoltaic inverter, a ground source heat pump frequency converter, and a battery management system, respectively, for adjusting photovoltaic power output, heat pump operating power, and battery charge and discharge state; wherein the execution control module further collects photovoltaic inverter output frequency, ground source heat pump compressor speed, and battery charge and discharge current as device state data, and feeds back the device state data to the data acquisition module for updating the real-time operation data sets.

2. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 1, characterized in that, The data acquisition module comprises: a sensing data acquisition unit that acquires direct current side electrical parameters through a photovoltaic power sensor, reads compressor frequency and refrigerant temperature through a heat pump unit controller, acquires state of charge data through a battery management system, collects temperature and humidity parameters through a building interior sensor, and acquires solar intensity and ambient temperature through a weather station; a data preprocessing unit that receives the direct current side electrical parameters, the compressor frequency and refrigerant temperature, the state of charge data, the temperature and humidity parameters, and the solar intensity and ambient temperature, performs timestamp alignment and filtering processing on the received data, and outputs a standardized data frame; a communication interface unit that receives the standardized data frame and uploads the standardized data frame to the multi-time scale prediction module.

3. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 2, characterized in that, The multi-time scale prediction module comprises: a short-term prediction unit that uses a long short-term memory network to input historical photovoltaic power generation sequences and real-time meteorological data, and outputs photovoltaic power generation point prediction values in the future for several hours; a curve generation unit that receives the photovoltaic power generation point prediction values, generates the photovoltaic power generation prediction curve, and simultaneously receives building heat load historical data and real-time data, generates the heat load prediction curve; The matching degree evaluation unit receives the photovoltaic power generation prediction curve and the heat load prediction curve, calculates a Pearson correlation coefficient and a power deviation rate between the two curves, and outputs the time sequence matching degree index.

4. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 3, characterized in that, The energy coordination optimization module comprises: A multi-objective decision unit constructs a linear programming model comprising a battery charge / discharge rate constraint, a heat pump operating temperature range constraint, and a building comfort constraint; A real-time scheduling unit receives the time sequence matching degree index, the photovoltaic power generation prediction curve, and the heat load prediction curve, calls the linear programming model according to the time sequence matching degree index, and calculates a battery charging power instruction and a ground source heat pump pre-cooling / pre-heating capacity in a photovoltaic output surplus period, and optimizes a battery discharging sequence and a heat pump set temperature in a power gap period; A grid interaction unit receives a time-of-use electricity price signal and a scheduling result of the real-time scheduling unit, and dynamically adjusts an active power reference value of a grid connection point.

5. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 4, characterized in that, The execution control module comprises: An instruction analysis unit receives the battery charging instruction and the ground source heat pump pre-conditioning instruction or the battery discharging instruction and the ground source heat pump power adjustment instruction, decodes the instructions into a photovoltaic inverter active power set value, a ground source heat pump compressor frequency set value, and a battery charge / discharge current threshold value; A device control unit receives the active power set value, the compressor frequency set value, and the charge / discharge current threshold value, and issues a control word comprising the set values to a photovoltaic inverter, a ground source heat pump frequency converter, and a battery management system through a field bus; A state feedback unit collects a photovoltaic inverter output frequency, a ground source heat pump compressor speed, and a battery charge / discharge current, performs a deviation check with the set values in the control word, and sends a re-optimization request to the energy coordination optimization module when the deviation exceeds a preset threshold value.

6. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 5, characterized in that, The short-term prediction unit is configured to: extract special event markers from a building operation and maintenance database, the special event markers comprising holiday identifiers and large event schedules; combine the special event markers, the historical photovoltaic power generation sequence, and the real-time weather data into an input feature vector; input the input feature vector into the long short-term memory network; configure a dynamic weight coefficient for the special event markers in the calculation process of the long short-term memory network; output the photovoltaic power point prediction value based on the input feature vector with the dynamic weight coefficient configured.

7. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 6, characterized in that, The real-time scheduling unit is configured to: call historical operation data to analyze a decay curve of a building indoor temperature in a heating or cooling period; back-calculate a thermal time constant of a building envelope based on the decay curve; calculate a start time and a duration of a pre-cooling or pre-heating operation relative to a predicted heat load peak period according to the thermal time constant; write the start time and the duration into the ground source heat pump pre-conditioning instruction.

8. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 7, characterized in that, The state feedback unit is configured to: continuously monitor a real-time deviation of the compressor actual speed and the compressor frequency set value; determine whether an absolute value of the real-time deviation exceeds a preset threshold value for three consecutive control periods; generate a re-optimization request signal when the determination result is yes; The heavy optimization request signal is sent to the real-time scheduling unit of the energy coordination optimization module.

9. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 8, characterized in that, The multi-time scale prediction module is configured to: Access a meteorological satellite ground receiving system to obtain a visible light and infrared spectrum satellite cloud image covering the geographical location of the building site; Perform geographic coordinate registration and radiation calibration processing on the satellite cloud image; Using a convolutional neural network, extract the transmittance gradient distribution and movement vector features of the cloud cluster from the processed satellite cloud image; Fuse the extracted transmittance gradient distribution and movement vector features with the ground solar radiation intensity data obtained by the data acquisition module; Based on the fused data, generate a corrected solar radiation intensity prediction value; Using the corrected solar radiation intensity prediction value, perform dynamic correction on the photovoltaic power generation prediction curve.

10. The multi-energy complementary based ground source heat pump and solar photovoltaic integrated management system according to claim 8, characterized in that, Further comprising: The state feedback unit of the execution control module feeds the collected photovoltaic inverter output frequency, ground source heat pump compressor speed and battery charge and discharge current as equipment state data to the data acquisition module; The data acquisition module receives the equipment state data and updates the real-time operation data set using the equipment state data; The multi-time scale prediction module receives the updated real-time operation data set for model self-correction in subsequent rolling prediction processes.

Images