A composite energy system based on photovoltaic panel backboard heat recovery

Abstract

The present application relates to the technical field of photovoltaic panel heat recovery, and particularly relates to a composite energy system based on photovoltaic panel backboard heat recovery, comprising: a micro-channel structure arranged on the back of a photovoltaic panel; a data monitoring unit for collecting multi-dimensional monitoring data and forming a set of operation parameters; a photovoltaic panel cooling unit for analyzing heat load distribution based on the set of operation parameters by using an adaptive fuzzy logic algorithm, generating flow regulation instructions and controlling a proportional flow valve array, so that the cooling working medium forms directional gradient flow in the micro-channel; a heat energy recovery unit for recovering heat energy carried by the cooling working medium through a multi-stage heat exchanger, recycling the cooling working medium after adjusting the temperature, and generating heat energy utilization data; and a performance optimization unit for structurally processing the heat energy utilization data, the set of operation parameters and photovoltaic power generation data, generating a performance optimization scheme by using a machine learning algorithm, and dynamically adjusting system operation parameters. The present application improves the energy utilization rate of photovoltaic panel backboard heat recovery.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic panel heat recovery technology, and more specifically to a composite energy system based on photovoltaic panel backsheet heat recovery. Background Technology

[0002] In traditional photovoltaic energy utilization systems, there is an irreconcilable contradiction between the power generation efficiency of photovoltaic panels and thermal management: to improve photoelectric conversion efficiency, the operating temperature needs to be lowered, and the resulting waste heat is often regarded as a negative byproduct that must be eliminated. Existing technologies generally adopt passive heat dissipation or single-function active cooling systems. These methods either cannot effectively recover and utilize waste heat resources or require additional energy input to maintain cooling operation, resulting in low overall energy utilization efficiency.

[0003] For example, existing patent CN119906361A discloses a photovoltaic-thermal integrated system for large buildings and its operation method. This system uses cooling water pipes installed on the back of the photovoltaic panels to lower their operating temperature through water cooling, thereby improving power generation efficiency, while simultaneously recovering waste heat for building heating. However, this technology has significant drawbacks: excessive cooling causes the photovoltaic panel operating temperature to fall below the optimal range, actually reducing power generation efficiency; the system lacks intelligent control capabilities and cannot dynamically adjust its operating strategy based on real-time operating conditions, resulting in low energy utilization.

[0004] Therefore, it is necessary to design a composite energy system based on photovoltaic backsheet heat recovery to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes a composite energy system based on heat recovery from the backsheet of a photovoltaic panel, aiming to solve the problem of low energy utilization rate of heat recovery from the backsheet of a photovoltaic panel.

[0006] This invention proposes a composite energy system based on photovoltaic panel backsheet heat recovery, comprising:

[0007] Microchannel structure, set on the surface of photovoltaic panel backsheet;

[0008] The data monitoring unit is used to collect multi-dimensional monitoring data in real time, and to preprocess and extract features from the multi-dimensional monitoring data to generate a set of operating parameters, which includes temperature gradient features, thermal characteristic parameters and light irradiance change rate features.

[0009] The photovoltaic panel cooling unit is used to analyze the heat load distribution characteristics of the photovoltaic panel based on the set of operating parameters and an adaptive fuzzy logic algorithm, and generate multi-region differentiated cooling medium flow rate adjustment instructions; and controls the proportional flow valve array according to the cooling medium flow rate adjustment instructions to make the cooling medium form a directional gradient flow within the microchannel structure.

[0010] The heat recovery unit is used to transfer the heat energy of the cooling working fluid flowing out of the microchannel structure through a multi-stage heat exchanger, adjust the temperature of the cooling working fluid after the heat energy transfer and reintroduce it into the microchannel structure to form a closed loop; and generate heat energy utilization data.

[0011] The performance optimization unit is used to perform structured processing on the thermal energy utilization data, combine the operating parameter set and photovoltaic power generation data to generate a structured thermal energy dataset; based on the structured thermal energy dataset, it optimizes the cooling fluid flow strategy and thermal energy recovery method through machine learning algorithms to generate a system performance optimization scheme, and dynamically adjusts the operating parameters according to the system performance optimization scheme.

[0012] Furthermore, when the data monitoring unit collects multi-dimensional monitoring data in real time, it includes:

[0013] The photovoltaic panel backsheet surface temperature distribution data is collected using distributed temperature sensors; the cooling medium inlet temperature data is collected using inlet temperature sensors; the cooling medium outlet temperature data is collected using outlet temperature sensors; and the ambient light intensity data is collected using light sensors. The collected backsheet surface temperature distribution data, cooling medium inlet temperature data, cooling medium outlet temperature data, and ambient light intensity data are organized into multi-dimensional monitoring data, and the timestamp information corresponding to the multi-dimensional monitoring data is acquired simultaneously.

[0014] Furthermore, when the data monitoring unit preprocesses and extracts features from the multi-dimensional monitoring data to generate the operating parameter set, it includes:

[0015] The backplate surface temperature distribution data is subjected to noise reduction processing; temperature gradient features are extracted from the backplate surface temperature distribution data;

[0016] The inlet and outlet temperature data of the cooling medium are synchronously corrected; the heat exchange rate of the cooling medium is calculated as a thermal characteristic parameter.

[0017] Extract the light change rate feature from the ambient light intensity data;

[0018] The thermal characteristic parameters, temperature gradient features, and illumination change rate features are integrated to generate a set of operating parameters.

[0019] Furthermore, when the photovoltaic panel cooling unit generates multi-region differentiated cooling fluid flow adjustment commands based on the set of operating parameters and by using an adaptive fuzzy logic algorithm to analyze the heat load distribution characteristics of the photovoltaic panel, it includes:

[0020] The temperature gradient feature is used as the first input variable; the illumination change rate feature is used as the second input variable; and the heat exchange data obtained from the thermal characteristic parameters is used as the third input variable.

[0021] Each input variable is fuzzified and converted into a fuzzy set; the fuzzy set is then used to perform inference operations through a preset fuzzy rule base to generate inference results; the inference results are then defuzzified to generate heat load distribution characteristics.

[0022] Calculate the required cooling medium flow rate ratio for different areas of the photovoltaic panel based on the heat load distribution characteristics; generate cooling medium flow rate adjustment instructions corresponding to different areas of the photovoltaic panel.

[0023] Furthermore, the photovoltaic panel cooling unit controls the proportional flow valve array according to the cooling medium flow rate adjustment command, so that the cooling medium forms a directional gradient flow within the microchannel structure, including:

[0024] The system analyzes the regional flow parameters in the cooling medium flow regulation command; converts the regional flow parameters into opening control signals for proportional flow valves; sends opening control signals to the corresponding valve bodies in the proportional flow valve array; adjusts the flow rate of the cooling medium in different regions of the photovoltaic panel; and achieves directional gradient flow of the cooling medium within the microchannel structure.

[0025] Furthermore, the heat recovery unit transfers heat energy from the cooling medium carrying heat energy flowing out of the microchannel structure through a multi-stage heat exchanger. When the temperature of the transferred cooling medium is adjusted and it is reintroduced into the microchannel structure, the process includes:

[0026] The system receives the cooling medium carrying thermal energy flowing out of the microchannel structure; introduces the cooling medium into the first stage of a multi-stage heat exchanger; performs heat energy transfer in the first-stage heat exchanger; introduces the cooling medium processed in the first stage into the second stage of the multi-stage heat exchanger; performs heat energy transfer in the second-stage heat exchanger; adjusts the temperature of the cooling medium that has completed the multi-stage heat exchange; and reintroduces the temperature-adjusted cooling medium into the inlet of the microchannel structure.

[0027] Furthermore, when the heat energy recovery unit generates heat energy utilization data, it includes:

[0028] Measure the temperature of the first cooling medium flowing out of the microchannel structure; measure the temperature of the second cooling medium flowing out of the multi-stage heat exchanger; monitor the mass flow rate of the cooling medium;

[0029] The amount of heat energy transferred is calculated based on the temperature of the first cooling medium, the temperature of the second cooling medium, and the mass flow rate.

[0030] Record the start and end times of the heat transfer process; calculate the duration parameter of the heat transfer.

[0031] The amount of heat energy transferred per unit time;

[0032] The amount of heat energy transferred, the duration parameter, and the amount of heat energy transferred per unit time are organized into heat energy utilization data.

[0033] Furthermore, when the performance optimization unit performs structured processing on the thermal energy utilization data and generates a structured thermal energy dataset by combining the operating parameter set and photovoltaic power generation data, it includes:

[0034] Obtain the power generation efficiency data of the photovoltaic power generation system as photovoltaic power generation data;

[0035] The thermal energy utilization data, operating parameter set, and photovoltaic power generation data are timestamped to establish a time index;

[0036] The thermal energy utilization data and the operating parameter set are associated and matched according to the time index; the associated and matched thermal energy utilization data and photovoltaic power generation data are matched according to the time index; and the matched thermal energy utilization data, operating parameter set and photovoltaic power generation data are cleaned to remove outliers and missing values.

[0037] Construct a data table structure, which includes timestamps, thermal energy utilization data, operating parameter sets, and photovoltaic power generation data; verify the logical consistency of each field in the data table structure; and generate a structured thermal energy dataset.

[0038] Furthermore, when the performance optimization unit optimizes the cooling fluid flow strategy and heat recovery method using machine learning algorithms to generate a system performance optimization scheme, it includes:

[0039] Based on a structured thermal energy dataset, historical thermal energy utilization data, historical operating parameter sets, and historical photovoltaic power generation data are acquired. These data are then used as input data for a machine learning algorithm. The machine learning algorithm analyzes the correlation between these data to determine the optimal parameters for the cooling medium flow strategy and the heat recovery method. Finally, the optimal parameters for the cooling medium flow strategy and the heat recovery method are integrated into a system performance optimization scheme.

[0040] Furthermore, when the performance optimization unit dynamically adjusts the operating parameters according to the system performance optimization scheme, it includes:

[0041] Analyze the cooling medium flow strategy optimization parameters in the system performance optimization scheme; adjust the flow distribution of the proportional flow valve array according to the cooling medium flow strategy optimization parameters;

[0042] Analyze the optimization parameters of the heat recovery method in the system performance optimization scheme; adjust the operating parameters of the multi-stage heat exchanger according to the optimization parameters of the heat recovery method.

[0043] Compared with existing technologies, the advantages of this invention are as follows: By setting microchannels on the backplate and implementing multi-regional differentiated cooling medium flow rate regulation, the temperature distribution and thermal uniformity of the photovoltaic panel can be improved, reducing local high-temperature zones, thereby improving the photovoltaic module conversion efficiency and reducing temperature-related power decay. Using a multi-stage heat exchanger to transfer heat energy and regulate the temperature of the outflowing cooling medium in stages can convert the heat energy generated by the photovoltaic panel into usable heat energy (such as low-temperature hot water or process heat sources), achieving electrothermal coupling utilization and improving overall energy utilization and system economy. Introducing multi-dimensional monitoring data such as distributed temperature, inlet / outlet temperature, and irradiance variation rate, and generating regionalized flow commands based on an adaptive fuzzy logic algorithm, enables the cooling strategy to adapt to irradiance fluctuations and heat load changes in real time, ensuring cooling effect and system stability. Precise control of the opening degree of each region is achieved through a proportional flow valve array, allowing the cooling medium to form the expected directional gradient flow in the microchannels, which is beneficial for enhancing local heat exchange performance and reducing cycle energy consumption. The cooling medium, after undergoing multi-stage heat exchange and temperature regulation, is returned to the microchannel, forming a closed loop. Combined with continuous monitoring and control, this reduces the thermal degradation of the cooling medium and the frequency of system maintenance, extending equipment lifespan. By structuring thermal energy utilization data, operating parameter sets, and photovoltaic power generation data, and using machine learning algorithms for correlation analysis and modeling, it is possible to uncover system operating patterns, optimize cooling and recovery parameters, and continuously improve energy efficiency and economics over time. By improving power generation efficiency, recovering and utilizing thermal energy, and optimizing operating strategies, energy consumption and operating costs per unit of power generation / heat supply can be reduced, decreasing dependence on fuel or external heat sources, thereby bringing environmental emission reduction and economic benefits. Attached Figure Description

[0044] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0045] Figure 1 This is a structural block diagram of a composite energy system based on photovoltaic backsheet heat recovery, provided in an embodiment of the present invention. Detailed Implementation

[0046] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0047] Existing photovoltaic-thermal energy utilization solutions generally suffer from simple structures and inefficient control, particularly lacking systematic intelligent optimization in the coordinated management of photovoltaic panel cooling and waste heat recovery. Traditional cooling designs using backsheet water pipes or heat sinks can lower photovoltaic panel temperatures, but the cooling intensity is often uncontrollable, easily leading to overcooling or undercooling, causing the photovoltaic panel operating temperature to deviate from the optimal photoelectric conversion range and thus weakening power generation gain. These systems generally rely on fixed flow rates or simple temperature control strategies, failing to accurately respond to fluctuations in ambient light, local temperature gradient differences, or actual load demands, resulting in low energy distribution and heat recovery efficiency. More importantly, existing systems lack intelligent assessment and dynamic scheduling capabilities for waste heat value. Waste heat utilization often remains at a single-stage heat exchange and fixed purpose, unable to adjust the recovery path and utilization method according to real-time demand, resulting in a significant waste of usable heat energy and a significantly insufficient overall energy utilization rate.

[0048] Taking a commercial building's rooftop photovoltaic system as an example, a serpentine water-cooled pipe is installed on the back of the photovoltaic panels, using continuous cooling water flow to lower the operating temperature of the panels. In actual operation, during the intense afternoon sunlight in summer, the cooling water system maintains a constant flow rate, causing the photovoltaic panel temperature to drop to near-ambient temperature, below its optimal power generation temperature range, resulting in a decrease in photoelectric conversion efficiency instead of an increase. Furthermore, maintaining the same cooling intensity in the evening when sunlight decreases leads to a low temperature of the recovered hot water, failing to meet the minimum requirements for domestic hot water use, necessitating additional electric heating compensation. The lack of intelligent analysis capabilities for real-time temperature gradients, rates of change in sunlight, and heat load demands results in the building's inability to efficiently utilize waste heat even during peak energy consumption periods. After long-term operation, the overall energy utilization rate is only slightly higher than that of a traditional standalone photovoltaic system, failing to achieve the ideal goal of solar-thermal synergy.

[0049] If the aforementioned problems remain unresolved for an extended period, the overall energy efficiency of photovoltaic systems will continue to be limited. On one hand, the inability to precisely control the operating temperature of photovoltaic panels will cause frequent fluctuations in photoelectric conversion efficiency between high and low temperatures, further amplifying the unpredictability of system operation due to the instability of power generation. On the other hand, waste heat recovery remains in an inefficient and passive utilization mode, with a large amount of heat energy that could be converted into domestic hot water, district heating, or low-grade industrial heat sources being released into the environment, resulting in irreversible energy loss.

[0050] For this, please refer to Figure 1 As shown, this application proposes a composite energy system based on photovoltaic panel backsheet heat recovery, comprising:

[0051] Microchannel structure, set on the surface of photovoltaic panel backsheet;

[0052] The data monitoring unit is used to collect multi-dimensional monitoring data in real time, and to preprocess and extract features from the multi-dimensional monitoring data to generate a set of operating parameters, including temperature gradient features, thermal characteristic parameters and light irradiance change rate features.

[0053] The photovoltaic panel cooling unit is used to analyze the heat load distribution characteristics of the photovoltaic panel based on the set of operating parameters and an adaptive fuzzy logic algorithm, and generate multi-region differentiated cooling medium flow rate adjustment instructions; according to the cooling medium flow rate adjustment instructions, it controls the proportional flow valve array to form a directional gradient flow of the cooling medium in the microchannel structure.

[0054] The heat recovery unit is used to transfer the heat energy of the cooling working medium flowing out of the microchannel structure through a multi-stage heat exchanger, regulate the temperature of the cooling working medium after the heat energy transfer and reintroduce it into the microchannel structure to form a closed loop; and generate heat energy utilization data.

[0055] The performance optimization unit is used to perform structured processing on thermal energy utilization data, and combine the operating parameter set and photovoltaic power generation data to generate a structured thermal energy dataset. Based on the structured thermal energy dataset, the unit optimizes the cooling fluid flow strategy and thermal energy recovery method through machine learning algorithms to generate a system performance optimization scheme, and dynamically adjusts the operating parameters according to the system performance optimization scheme.

[0056] Specifically, the microchannel structure is directly integrated onto the surface of the photovoltaic panel backsheet using precision micromachining technology, forming a microchannel network with an optimized flow channel layout. The microchannel network consists of three parts: an inlet distribution area, a main flow channel area, and an outlet collection area. The inlet distribution area is fan-shaped and contains multiple tapering channels to ensure uniform distribution of the cooling medium. The main flow channel area adopts a fractal tree structure design, with main and branch channels forming a multi-level network. The branch angles are optimized, mimicking efficient mass transport networks in nature, enabling uniform distribution and low-resistance flow of the cooling medium. The outlet collection area uses a spiral expanding structure with multiple converging channels to ensure stable collection of the cooling medium. The microchannel surface undergoes nanoscale processing to form a coating of a specific thickness, with surface roughness controlled to an extremely low level, improving heat transfer efficiency. The microchannel substrate uses a high thermal conductivity material and is tightly connected to the photovoltaic panel backsheet through a special process, with interfacial thermal resistance controlled within a very small range to ensure efficient heat transfer. The data monitoring unit consists of a distributed temperature sensor array, flow sensors, pressure sensors, and a light sensor, all connected to the central processing unit via an industrial-grade bus. The distributed temperature sensor array includes multiple high-precision temperature sensors evenly distributed in a grid pattern on the photovoltaic panel backsheet surface, covering most of the area. The inlet and outlet temperature sensors employ a dual-redundancy design, with multiple sensors installed at each location, and software verification ensures measurement reliability. The flow sensor is a thermal mass flow meter, offering high accuracy and good repeatability. The pressure sensor is a piezoresistive sensor, providing high precision. The light sensor is a silicon photodiode array, offering a wide spectral response range and high accuracy. All sensors are equipped with temperature compensation circuitry and electromagnetic interference shielding to ensure measurement accuracy in harsh environments. The signal conditioning circuitry of the data monitoring unit uses a high-resolution analog-to-digital converter and features a built-in digital filter to eliminate high-frequency noise. It communicates with the microcontroller via a high-speed interface, ensuring real-time data transmission and accuracy. The core of the photovoltaic panel cooling unit is an adaptive fuzzy logic controller, composed of a high-performance microcontroller and a coprocessor. The adaptive fuzzy logic algorithm comprises five key steps: First, the temperature gradient feature is used as the first input variable, obtained by calculating and normalizing the absolute value of the temperature difference between adjacent temperature sensors; the illumination change rate feature is used as the second input variable, obtained by calculating and normalizing the illumination intensity change rate over multiple consecutive sampling periods; and the heat exchange rate is obtained from thermal characteristic parameters as the third input variable, calculated by multiplying the inlet and outlet temperature difference by the flow rate. Second, each input variable is fuzzified using a Gaussian membership function, dividing each variable into multiple fuzzy sets, with the membership function parameters dynamically adjusted based on historical data. Third, inference operations are performed using a pre-set fuzzy rule base, which contains multiple rules, with rule weights dynamically adjusted based on the system's operating status.Fourth, the centroid method is used for defuzzification to generate heat load distribution characteristics. These characteristics are continuous values ​​representing the regional heat load intensity. Finally, the required cooling medium flow rate ratio for different areas of the photovoltaic panel is calculated based on the heat load distribution characteristics. The calculation formula considers factors such as the baseline flow rate, adjustment coefficient, and heat load distribution characteristics. The proportional flow valve array consists of multiple miniature proportional solenoid valves arranged in a matrix, corresponding one-to-one with the temperature sensor positions. Each proportional solenoid valve adopts advanced drive technology, featuring fast response characteristics, a wide flow control range, high control accuracy, and an appropriate working pressure range to ensure stable system operation. The valve core of the solenoid valve is made of wear-resistant material, the valve seat is made of corrosion-resistant material, and the sealing ring is made of high-performance elastic material to ensure long-term stable operation. The valve body is designed with a laminar flow channel to reduce turbulence losses, and the flow channel surface is treated with ultra-precision polishing to ensure smooth fluid flow. The proportional flow valve array is controlled by a pulse width modulation signal, with the duty cycle corresponding to the flow range. The control signal is generated by the photovoltaic panel cooling unit and transmitted through an industrial bus to ensure reliable communication. It is also equipped with a flow feedback loop, with a miniature flow meter installed at each solenoid valve outlet to monitor the actual flow rate in real time, compare it with the target flow rate, and dynamically adjust the control signal through algorithms to ensure flow control accuracy. The heat recovery unit adopts a multi-stage heat exchange structure, with each stage using a plate heat exchanger design. The heat exchange plates are made of corrosion-resistant stainless steel with moderate thickness, and the corrugation depth and angle are optimized. The heat exchange area of ​​each plate is precisely calculated. Each stage of the heat exchanger is designed according to the heat energy temperature range and connected to the corresponding building heat energy utilization system. The heat exchangers at each stage are connected by three-way regulating valves. The valve body is made of corrosion-resistant material, and the valve core is made of wear-resistant material, ensuring high adjustment accuracy and short response time. The temperature regulation device adopts a hybrid three-way valve combined with a heater design for fine adjustment of the cooling medium temperature, with high control accuracy. The heat recovery unit is also equipped with multiple high-precision temperature sensors, flow sensors, and pressure sensors with a high data sampling frequency for real-time monitoring of heat transfer efficiency. The performance optimization unit is implemented based on a deep neural network algorithm. The hardware platform uses an embedded AI computer equipped with a multi-core processor and a dedicated graphics processor, with sufficient memory and storage capacity to meet data processing requirements. The neural network employs a multi-layered structure. The input layer contains multiple nodes corresponding to historical operational data, the hidden layers contain a large number of nodes using appropriate activation functions, and the output layer contains multiple nodes corresponding to optimized parameters. Training data comes from the system's historical operational records, with complete datasets collected periodically, including timestamps, temperature distribution, flow parameters, heat transfer, and power generation efficiency. The training process utilizes optimized algorithms with appropriate learning rates, reasonable batch sizes, and sufficient training cycles. The neural network model is automatically updated periodically, retaining historical data for training, and employing a sliding window approach to ensure the model adapts to environmental changes. The performance optimization unit is also equipped with a large-capacity solid-state storage device for storing historical operational data, achieving a high data compression ratio and capable of storing several years of operational records.

[0057] This invention addresses the problems of excessive cooling leading to decreased power generation efficiency, spatiotemporal mismatch of heat energy, complex system structure, and lack of intelligent control capabilities in existing photovoltaic heat recovery technologies. Through precise design of the microfluidic structure and a regionalized cooling strategy, the operating temperature of the photovoltaic panel can be maintained within the optimal power generation efficiency range (25-45℃), avoiding energy loss caused by excessive cooling. Multi-stage heat exchangers and intelligent thermal management solve the problem of time mismatch between heat generation and demand. The simplified structure eliminates unnecessary auxiliary equipment, reducing failure rate and maintenance costs. A machine learning-based performance optimization unit enables the system to adapt, dynamically adjusting operating strategies based on real-time conditions and historical data, thereby improving overall energy utilization efficiency.

[0058] The working process and principle of this application are as follows: After the system starts, the data monitoring unit collects multi-dimensional monitoring data in real time, including the surface temperature distribution of the photovoltaic panel backsheet, the inlet temperature of the cooling medium, the outlet temperature of the cooling medium, and the ambient light intensity. These data undergo preprocessing and feature extraction to form a set of operating parameters containing temperature gradient features, thermal characteristic parameters, and light intensity variation rate features. The photovoltaic panel cooling unit receives the operating parameter set and analyzes the heat load distribution characteristics of the photovoltaic panel using an adaptive fuzzy logic algorithm: using the temperature gradient feature as the first input variable, the light intensity variation rate feature as the second input variable, and the heat exchange rate as the third input variable, after fuzzification, rule reasoning, and defuzzification, an accurate heat load distribution characteristic is generated. Based on the heat load distribution characteristic, the required cooling medium flow rate ratio for different areas of the photovoltaic panel is calculated, generating multi-region differentiated cooling medium flow rate adjustment commands. The proportional flow valve array receives the flow rate adjustment commands and precisely controls the cooling medium flow rate in each region. In high heat load areas (usually areas with high light intensity), the cooling medium flow rate is increased to reduce the photovoltaic panel temperature; in low heat load areas, the cooling medium flow rate is reduced to avoid overcooling. Through this differentiated flow control, the cooling medium forms a directional gradient flow within the microchannel structure, flowing naturally from high heat load areas to low heat load areas, fully utilizing the thermal gradient and reducing pump power consumption. The heat recovery unit receives the cooling medium carrying thermal energy flowing out of the microchannel structure. It first enters the first-stage high-temperature heat exchanger, transferring the thermal energy to the high-temperature heating system; then it enters the second-stage medium-temperature heat exchanger, transferring the remaining thermal energy to the medium-temperature heating system; finally, it enters the third-stage low-temperature heat exchanger to recover the remaining low-temperature thermal energy. After three stages of heat exchange, the cooling medium is adjusted to a suitable temperature by a temperature regulating device and reintroduced into the microchannel structure inlet, forming a closed loop. The heat recovery unit simultaneously generates heat utilization data, including the amount of heat transferred, duration parameters, and the amount of heat transferred per unit time. The performance optimization unit performs structured processing on the heat utilization data, combining it with the operating parameter set and photovoltaic power generation data to generate a structured heat energy dataset. Based on this dataset, machine learning algorithms are used to analyze the correlations between historical data, determine the optimal parameters for the cooling fluid flow strategy and the heat recovery method, and generate a system performance optimization scheme. According to this scheme, the flow distribution of the proportional flow valve array and the operating parameters of the multi-stage heat exchanger are dynamically adjusted to achieve continuous optimization of system performance.

[0059] As a preferred embodiment, the specific implementation of this application is as follows: In a 200-square-meter photovoltaic panel array, a microchannel structure is integrated on the back surface of each photovoltaic panel. The microchannel width is 0.8 mm, the depth is 0.6 mm, and the channel spacing is 3 mm. The data monitoring unit is equipped with 36 distributed temperature sensors (one per 6 square meters), 2 flow sensors, 2 pressure sensors, and 1 high-precision light sensor. After the system starts, the data monitoring unit collects data every 5 seconds and generates a set of operating parameters. When the temperature in the central area of ​​the photovoltaic panel is detected to be higher than 45℃ (the upper limit of optimal power generation efficiency), while the temperature in the edge area is lower than 30℃ (the lower limit of optimal power generation efficiency), the photovoltaic panel cooling unit uses an adaptive fuzzy logic algorithm to determine that the central area is a high heat load area and the edge area is a low heat load area. A cooling medium flow rate adjustment command is generated: the flow rate in the central area is increased by 20%, the flow rate in the edge area is decreased by 15%, and the flow rate in the intermediate transition area is adjusted linearly. After receiving the command, the proportional flow valve array precisely controls the flow rate in each area, so that the cooling medium flows directionally from the edge area to the central area. The heat recovery unit receives a cooling medium with an outlet temperature of 58°C. First, in the first-stage heat exchanger, the temperature is lowered to 48°C, recovering high-temperature heat energy for building domestic hot water. Then, in the second-stage heat exchanger, the temperature is lowered to 38°C, recovering medium-temperature heat energy for building heating. Finally, in the third-stage heat exchanger, the temperature is lowered to 32°C, recovering low-temperature heat energy for preheating the fresh air system. After temperature regulation, the cooling medium re-enters the microchannel structure at 30°C. The performance optimization unit records system operating data. After a week of operation and learning, it was found that during periods of weak morning sunlight, the cooling medium flow rate can be appropriately reduced to maintain the photovoltaic panels within the optimal temperature range; during periods of strong midday sunlight, the flow rate in the central area needs to be increased, while avoiding over-cooling. Based on these patterns, an optimization scheme is generated, automatically adjusting the flow distribution strategy.

[0060] Through the above scheme, this application reduces pump power consumption and lowers system energy consumption by adopting a multi-regional differentiated cooling strategy and directional gradient flow; it also achieves cascade utilization of heat energy by adopting a three-stage heat exchange structure, thus solving the problem of spatiotemporal mismatch of heat energy; and it enables the system to adapt to environmental changes and continuously improve energy utilization efficiency based on a machine learning-based performance optimization mechanism.

[0061] This application further proposes that when the data monitoring unit collects multi-dimensional monitoring data in real time, it includes:

[0062] Data on the surface temperature distribution of the photovoltaic panel backsheet is collected using distributed temperature sensors; data on the inlet temperature of the cooling medium is collected using inlet temperature sensors; data on the outlet temperature of the cooling medium is collected using outlet temperature sensors; and data on ambient light intensity is collected using light sensors. The collected data on the surface temperature distribution of the backsheet, the inlet temperature of the cooling medium, the outlet temperature of the cooling medium, and the ambient light intensity are organized into multi-dimensional monitoring data, and timestamp information corresponding to the multi-dimensional monitoring data is acquired simultaneously.

[0063] Specifically, the distributed temperature sensor array of the data monitoring unit employs sensor nodes based on a thermocouple-resistance device (RTD) composite principle. Each node contains a pair of K-type thermocouples and a PT1000 RTD. The thermocouples provide rapid response to temperature changes, while the RTDs offer high-precision steady-state measurements. Signals from both are acquired alternately via an analog multiplexer. The sensor nodes are arranged in a hexagonal grid on the backsheet surface of the photovoltaic panel. The node spacing is dynamically adjusted based on the photovoltaic panel size and expected thermal distribution characteristics, with a denser layout in the central region (node ​​spacing approximately 1 / 8 of the panel's side length) and a sparser layout at the edges (node ​​spacing approximately 1 / 4 of the panel's side length). Each sensor node is connected to the signal conditioning circuit via shielded twisted-pair cables. The shielding layer employs a double-layer design: an inner aluminum foil shield and an outer braided copper mesh shield. Both shielding layers are grounded at a single point at the signal conditioning circuit to suppress common-mode interference. The signal conditioning circuit employs a multi-stage filtering architecture: the first stage is a passive RC low-pass filter, with the cutoff frequency dynamically adjusted according to the sensor's response characteristics; the second stage is an active Butterworth filter, with its order and cutoff frequency remotely configured via a digital potentiometer; the third stage is a digital filter implemented in an embedded processor, using an adaptive filtering algorithm to dynamically adjust filtering parameters based on the real-time signal-to-noise ratio. After conditioning, the temperature signal is digitized by a 24-bit Σ-Δ analog-to-digital converter. This converter uses chopper stabilization technology to eliminate 1 / f noise and integrates a digital decimation filter to achieve high-resolution measurement. The inlet and outlet temperature sensors employ a dual-redundancy design, with two independent temperature sensors installed at each location, and the signals are processed through independent signal conditioning channels. A cross-validation mechanism is implemented; when the difference between the readings of the two sensors exceeds a preset threshold, a fault diagnosis program is initiated, determining a reliable reading through historical data trend analysis and physical model prediction. The illumination sensor uses a multispectral silicon photodiode array containing four photodiodes with different spectral response characteristics, corresponding to the visible, near-infrared, and two ultraviolet bands, respectively. A weighted fusion algorithm converts the multispectral response into equivalent solar radiation intensity, eliminating the influence of atmospheric condition variations on the measurement. The data synchronization mechanism employs the IEEE 1588 Precise Time Protocol (PTP), with one master clock node and other sensor nodes acting as slave clocks. The master clock acquires standard time via a GPS receiver, and the slave clocks synchronize with the master clock through the PTP protocol, achieving sub-microsecond synchronization accuracy. Each data packet contains a precise timestamp, data quality markers, and verification information. The data organization uses a hierarchical structure: the bottom layer contains raw sensor data, the middle layer contains time-synchronized data blocks, and the top layer contains multi-dimensional monitoring datasets.

[0064] Through the above technical solutions, this application achieves comprehensive and accurate monitoring of the working status of photovoltaic panels; through high-precision time synchronization technology, it ensures the time consistency of multi-source heterogeneous data, providing a key guarantee for system dynamic analysis and optimization.

[0065] This application further proposes that when the data monitoring unit preprocesses and extracts features from multi-dimensional monitoring data to generate a set of operating parameters, it includes:

[0066] Noise removal is performed on the backplate surface temperature distribution data; temperature gradient features are extracted from the backplate surface temperature distribution data; the inlet and outlet temperature data of the cooling medium are synchronously corrected; the heat exchange rate of the cooling medium is calculated as a thermal characteristic parameter; the light intensity variation rate feature is extracted from the ambient light intensity data; and the thermal characteristic parameter, temperature gradient feature, and light intensity variation rate feature are integrated to generate an operating parameter set.

[0067] Specifically, the data monitoring unit first performs multi-scale noise cancellation on the backplate surface temperature distribution data: wavelet packet decomposition is used to decompose the temperature signal into multiple frequency sub-bands, and thresholds are adaptively selected based on the energy distribution characteristics of each sub-band. A temperature signal feature database is constructed, containing various typical noise patterns and actual temperature change patterns. The optimal wavelet basis function and decomposition level are determined through pattern matching. The noise cancellation process employs a hybrid strategy of soft and hard thresholding. Soft thresholding is used for low-frequency sub-bands to preserve signal details, while hard thresholding is used for high-frequency sub-bands to completely eliminate noise. After wavelet reconstruction, anisotropic diffusion filtering based on partial differential equations is applied to further smooth the temperature distribution while preserving the abrupt changes in temperature gradient characteristics. Temperature gradient feature extraction uses a method combining an improved Sobel operator and morphological processing. The temperature distribution data is treated as a grayscale image, and the gradient components in the horizontal, vertical, and diagonal directions are calculated using a multi-directional Sobel operator. The temperature gradient magnitude and direction of each pixel are obtained through vector synthesis. To eliminate directional bias caused by discrete calculations, sub-pixel interpolation of the gradient direction is implemented, expanding the eight discrete directions into continuous directions. Morphological closing operations were applied to eliminate isolated gradient outliers, and a region growing algorithm was used to identify regions with similar gradient characteristics, forming temperature gradient region divisions. The temperature gradient features of each region include average gradient magnitude, gradient direction consistency index, and region boundary curvature. These features were dimensionality-reduced using principal component analysis to retain the most representative feature combinations. Synchronization correction of the cooling medium temperature data employed a combination of cross-correlation analysis and phase correction. The cross-correlation function of the inlet and outlet temperature signals was calculated to determine the optimal time offset between the two signals. Hilbert transform was applied to extract the instantaneous phase of the signal, and a phase correction algorithm was used to eliminate the phase difference caused by asynchronous sampling. To compensate for the dynamic response hysteresis of the temperature sensor, a transfer function model of the sensor was constructed, and the sensor hysteresis effect was eliminated using a frequency domain deconvolution method. The calculation of heat exchange considered the nonlinear thermal properties of the cooling medium, dividing the temperature range into multiple intervals. Within each interval, linear interpolation was used to approximate the changes in specific heat capacity and density. The heat exchange was calculated using numerical integration, employing an adaptive switching between the trapezoidal rule and Simpson's rule to ensure calculation accuracy. Adaptive Kalman filtering was used to extract the illumination change rate feature. A state-space model was constructed, treating illumination intensity as a state variable and measured values ​​as observation variables. The noise covariance matrix of the Kalman filter was dynamically adjusted based on historical data to reflect the statistical characteristics of illumination changes. Short-term (1-5 minute window) and long-term (30-60 minute window) rates of change were calculated, and wavelet analysis was used to identify the change characteristics at different time scales. To distinguish between real illumination changes and measurement noise, a confidence assessment of the rate of change was implemented. A probability distribution model of the rate of change was established based on historical data, and the statistical significance of the current rate of change was calculated.The generation of the operating parameter set involves multi-level feature fusion: normalization of each feature parameter to eliminate dimensional differences; mutual information analysis to evaluate the correlation between features and remove redundant features; and kernel principal component analysis to map features to a high-dimensional feature space and extract nonlinear combined features. Confidence weights are also calculated for each feature parameter and dynamically adjusted based on data quality, sensor status, and environmental conditions. The structural design of the operating parameter set considers subsequent processing needs, employing a hierarchical organization: the base layer contains original features, the intermediate layer contains combined features, and the high-level layer contains decision features.

[0068] Through the above technical solution, this application achieves efficient conversion from raw monitoring data to valuable feature parameters, which not only improves data quality but also extracts key features reflecting the system status. Through a scientific feature fusion method, a set of operating parameters that can accurately describe the thermal state of photovoltaic panels is constructed, providing precise input basis for intelligent control.

[0069] This application further proposes that when a photovoltaic panel cooling unit analyzes the heat load distribution characteristics of the photovoltaic panel based on a set of operating parameters and uses an adaptive fuzzy logic algorithm to generate multi-region differentiated cooling fluid flow adjustment commands, it includes:

[0070] Temperature gradient features are used as the first input variable; the rate of change of light intensity is used as the second input variable; heat exchange data obtained from thermal characteristic parameters are used as the third input variable; each input variable is fuzzified and converted into a fuzzy set; inference operations are performed on the fuzzy set using a preset fuzzy rule base to generate inference results; the inference results are defuzzified to generate heat load distribution features; the required cooling medium flow rate ratio for different areas of the photovoltaic panel is calculated based on the heat load distribution features; and cooling medium flow rate adjustment instructions corresponding to different areas of the photovoltaic panel are generated.

[0071] Specifically, the photovoltaic panel cooling unit performs spatial normalization on the input variables: the temperature gradient feature is normalized by dividing by the diagonal length of the photovoltaic panel to eliminate the influence of photovoltaic panel size differences and convert it into a relative temperature gradient value. A heat conduction model is applied to establish a correlation between the relative temperature gradient and heat flux density, considering the thermal conductivity of the photovoltaic panel material to achieve a preliminary estimate of the heat load from the temperature gradient. The rate of change of illumination feature is correlated with the temperature response using a dynamic time warping algorithm, constructing a photovoltaic material thermal response model. This model considers the material's heat capacity, thermal conductivity, and surface radiation characteristics to predict the impact of illumination changes on temperature. The heat exchange data undergoes temperature compensation correction, and a thermodynamic model is applied to analyze the impact of the cooling medium temperature on heat exchange efficiency, determining the compensation coefficient through iterative calculation. In the fuzzification stage, an adaptive Gaussian membership function generation technique is used: each input variable is divided into five fuzzy sets (negative large, negative small, zero, positive small, positive large), and the mean and standard deviation of the membership function are dynamically adjusted based on historical data. Online clustering analysis was implemented, clustering historical input data into five clusters. The center and dispersion of each cluster were used to determine the parameters of the Gaussian membership function. To handle the non-stationary nature of the input variables, a sliding window technique was applied, using only data from the most recent period for clustering analysis to ensure that the membership function could adapt to environmental changes. A soft boundary design was adopted for the boundaries of the fuzzy sets, with partial overlap of the membership functions of adjacent fuzzy sets to ensure a smooth transition of input variables in the boundary region. The fuzzy rule base adopted a hierarchical structure: the basic layer contained 27 core rules (3 inputs × 3 fuzzy sets) describing the basic laws of the thermal behavior of photovoltaic panels; the optimization layer contained extended rules based on historical operating experience, extracted from historical data through association rule mining; and the adaptive layer contained dynamically generated rules, adjusted according to real-time performance. Each rule was equipped with a dynamic weight coefficient, reflecting the reliability and applicability of the rule, which was updated online through a reinforcement learning algorithm. Rule reasoning employed a weighted reasoning method, considering not only the matching degree between the rule's antecedents and inputs but also the confidence level of the rule itself, generating more reliable reasoning results. A rule conflict detection and resolution mechanism was implemented; when multiple rules produced conflicting conclusions, evidence theory was applied for fusion. The defuzzification process employs an improved centroid method: the centroid position of each output fuzzy set is calculated; then, confidence weights are applied to weight the centroid positions; finally, a nonlinear transformation maps the continuous outputs to actual physical quantities. To avoid the influence of extreme values ​​on the results, output smoothing is implemented by weighted averaging of the current output and historical outputs. The heat load distribution characteristics undergo spatial interpolation, using a radial basis function interpolation algorithm to extend the discrete regional heat load estimate into a continuous heat load distribution map. The interpolation process considers the physical structure and thermal conductivity characteristics of the photovoltaic panels to ensure that the interpolation results conform to thermodynamic laws. The calculation of the cooling fluid flow rate ratio considers the temperature-efficiency characteristic curve of the photovoltaic material: a temperature-efficiency mapping model is constructed to determine the target operating temperature range for each region.Based on the deviation between the current temperature and the target temperature, and combined with the regional heat load intensity, the required cooling intensity is calculated. The flow rate ratio calculation also considers the hydrodynamic characteristics of the microchannel network, applying fluid network analysis algorithms to ensure the feasibility of flow allocation. A flow rate change limit is implemented, using a first-order inertial element to restrict the rate of flow change, avoiding system instability caused by sudden flow changes.

[0072] Through the above technical solutions, this application achieves accurate analysis and prediction of the heat load distribution of photovoltaic panels, and can dynamically adjust the cooling strategy according to real-time operating conditions; through the adaptive fuzzy logic algorithm, the uncertainties in the system are handled, the robustness and adaptability of the control are improved, and the photovoltaic panels are ensured to operate in the optimal temperature range.

[0073] This application further proposes a photovoltaic panel cooling unit that controls a proportional flow valve array according to a cooling medium flow rate adjustment command, so that the cooling medium forms a directional gradient flow within the microchannel structure, including:

[0074] The system analyzes the regional flow parameters in the cooling medium flow regulation command; converts the regional flow parameters into opening control signals for proportional flow valves; sends opening control signals to the corresponding valve bodies in the proportional flow valve array; adjusts the flow rate of the cooling medium in different regions of the photovoltaic panel; and achieves directional gradient flow of the cooling medium within the microchannel structure.

[0075] Specifically, the photovoltaic panel cooling unit first performs spatial mapping processing on the flow regulation commands: a microchannel network topology model is constructed, establishing a precise correspondence between the area identifiers in the commands and the physical locations of the proportional flow valve array. The mapping process considers the fluid dynamic characteristics of the microchannel structure, and pre-compensation is performed using computational fluid dynamics simulation results to eliminate the influence of channel geometry on flow distribution. Nonlinear correction is implemented for the area flow rate to valve opening: the static characteristic curve of each proportional flow valve is obtained through online identification, considering valve core wear and changes in fluid characteristics, and the correction parameters are updated periodically. The correction process uses a neural network model, with the target flow rate and current operating conditions as inputs, and the valve opening as the output; the network weights are updated online through supervised learning. Differential communication technology is used for control signal transmission: an RS-485 bus is used, and Manchester encoding is implemented to improve anti-interference capabilities. Each data packet contains the target opening, checksum, and timestamp; the receiving end performs data packet integrity verification and duplicate detection. Communication priority management is implemented, with critical control signals transmitted first to ensure real-time requirements. To prevent system loss of control due to communication failures, a heartbeat monitoring mechanism is implemented, automatically switching to safe mode when communication is interrupted. Flow control employs a closed-loop feedback strategy: a miniature turbine flow meter is installed at the outlet of each proportional flow valve to monitor the actual flow rate in real time. An adaptive PID algorithm is applied to dynamically adjust the opening control signal based on flow error. PID parameters are optimized online using an extreme value search algorithm, considering nonlinear and time-varying characteristics. To eliminate the propagation of flow disturbances, feedforward compensation control is implemented; when a flow change in a certain area is detected, the flow setpoint of adjacent areas is adjusted in advance to maintain overall system stability. Model predictive control technology is also applied, based on the fluid dynamics model of the microchannel network, to predict the propagation path and impact range of flow changes and implement compensation control in advance. The formation of directional gradient flow is based on fluid dynamics principles: the Bernoulli equation and the Darcy-Weisbach equation are applied to calculate the pressure distribution under different flow settings in different areas. By precisely controlling the flow ratio of each area, a pressure gradient is created from the high-flow area to the low-flow area, guiding the natural flow of the cooling medium. A gradient optimization algorithm is implemented, considering the heat load distribution of the photovoltaic panel and the structural characteristics of the microchannel, to determine the optimal pressure gradient distribution. To reduce pump power consumption, flow distribution is optimized to ensure that the flow direction of the cooling medium is consistent with the heat flow direction, fully utilizing the natural convection effect. Dynamic constraints are implemented to limit the rate of flow change: a rate limiter is applied to restrict the speed of flow change, avoiding fluid shock and pressure fluctuations caused by sudden flow abrupt changes. The rate limit value is dynamically adjusted according to the current system state, using a smaller limit value during steady-state operation and a larger limit value during transient processes. Smooth transitions in flow change are also implemented using an S-shaped curve to ensure the smoothness of the transition process. To prevent vibration of the microchannel structure, frequency analysis is performed to avoid the flow change frequency coinciding with the system's natural frequency.Fault detection and diagnosis: Real-time monitoring of the operating status of each proportional flow valve, and detection of faults such as valve core jamming and leakage through flow-opening relationship analysis. When a fault is detected, the opening of other valves is automatically adjusted to compensate for the impact of the faulty valve and maintain the overall system performance.

[0076] Through the above technical solution, this application achieves precise control of the flow of the cooling medium, and can dynamically adjust the flow rate of each region according to the heat load distribution to form a directional gradient flow that is conducive to heat exchange; through closed-loop control and coordination strategies, the stability and reliability of the system are ensured, cooling efficiency is improved and energy consumption is reduced.

[0077] This application further proposes a heat recovery unit that transfers heat energy from the cooling working fluid flowing out of the microchannel structure via a multi-stage heat exchanger. When the transferred cooling working fluid is temperature-regulated and reintroduced into the microchannel structure, the following steps are included:

[0078] The system receives the cooling medium carrying thermal energy flowing out of the microchannel structure; introduces the cooling medium into the first stage of a multi-stage heat exchanger; performs heat energy transfer in the first-stage heat exchanger; introduces the cooling medium processed in the first stage into the second stage of the multi-stage heat exchanger; performs heat energy transfer in the second-stage heat exchanger; adjusts the temperature of the cooling medium that has completed the multi-stage heat exchange; and reintroduces the temperature-adjusted cooling medium into the inlet of the microchannel structure.

[0079] Specifically, the multi-stage heat exchanger of the heat recovery unit adopts a modular plate heat exchanger design: each stage of the heat exchanger consists of multiple stacked heat exchange plates, forming alternating hot and cold fluid channels between the plates. The plate surface features a special corrugated design, with the corrugation shape optimized through computational fluid dynamics to enhance turbulence and improve the heat transfer coefficient. The corrugation depth and angle are dynamically adjusted according to the heat exchanger stage; deeper corrugations are used in the high-temperature stage to enhance turbulence, while shallower corrugations are used in the low-temperature stage to reduce pressure drop. The plate material is selected based on the operating temperature range: 316L stainless steel is used in the high-temperature stage, 304 stainless steel in the medium-temperature stage, and titanium alloy in the low-temperature stage, ensuring optimal material performance within their respective temperature ranges. A counter-current heat transfer strategy is implemented during the heat transfer process: thermodynamic analysis is applied to determine the optimal fluid flow direction for each stage of the heat exchanger, ensuring maximum temperature driving force. The high-temperature stage heat exchanger adopts a fully counter-current design, while the medium-temperature and low-temperature stages adopt a partially counter-current design based on actual application requirements. Temperature cross-monitoring is implemented; when temperature cross-monitoring is detected, the fluid flow distribution is automatically adjusted to prevent a decrease in heat exchange efficiency. Heat transfer efficiency is calculated in real time using the heat balance equation. When the efficiency falls below a threshold, a cleaning or maintenance procedure is triggered. Intelligent switching valve assemblies are installed between multi-stage heat exchangers: these valve assemblies are pneumatically controlled, offering fast response and excellent sealing performance. An inter-stage coordinated control algorithm is implemented to dynamically adjust the fluid distribution ratio based on the real-time operating status of each heat exchanger. When the efficiency of a heat exchanger decreases, the processing load of other stages is automatically increased to maintain overall heat recovery efficiency. The valve assemblies are also equipped with pressure balancing devices to ensure stable system pressure during switching and prevent pressure fluctuations from affecting the microchannel structure. A hybrid control strategy is employed in the temperature regulation stage: a combination of a three-way regulating valve and an auxiliary heating / cooling device is used to precisely control the return temperature of the cooling medium. The three-way regulating valve uses proportional-integral-derivative control, dynamically adjusting its opening based on the deviation between the target and actual temperatures. The auxiliary heating device uses electric heating elements, with power precisely adjusted through phase control; the auxiliary cooling device uses a semiconductor refrigeration module, with the temperature difference between the hot and cold ends precisely controlled through pulse width modulation. Dynamic adjustment of the temperature setpoint is implemented to determine the optimal return temperature based on the current operating status of the photovoltaic panels and environmental conditions, ensuring that the photovoltaic panels operate within their optimal temperature range. Thermodynamic optimization is implemented in the heat recovery process: entropy analysis is applied to assess irreversible losses in each stage of the heat exchange process, optimize temperature matching, and reduce entropy production. High-temperature stage heat exchangers are prioritized for high-grade heat energy demands (such as domestic hot water), medium-temperature stage for medium-grade heat energy demands (such as building heating), and low-temperature stage for low-grade heat energy demands (such as preheating fresh air). A heat energy grade matching algorithm is implemented to dynamically adjust the heat energy distribution strategy based on heat demand characteristics and heat energy grade. The re-introduction process of the cooling medium considers system pressure balance: buffer tanks and pressure regulating devices are used to ensure a smooth transition. The buffer tank is equipped with level and pressure sensors to monitor its internal status in real time.The pressure regulating device employs proportional-integral-derivative control, dynamically adjusting the opening based on the deviation between the demand pressure and the actual pressure. Flow balance control is implemented to ensure that the re-introduced cooling fluid flow matches the requirements of the microchannel structure, avoiding the impact of flow fluctuations on system stability.

[0080] Through the above technical solution, this application realizes the cascade utilization of waste heat from photovoltaic panels, solving the problem of spatiotemporal mismatch of heat energy; through multi-stage heat exchange and intelligent control, it maximizes the efficiency of heat energy recovery, ensures that the system can operate efficiently under various working conditions, and improves the overall energy utilization efficiency.

[0081] This application further proposes that when the heat recovery unit generates heat utilization data, it includes:

[0082] The process involves measuring the temperature of the first cooling medium flowing out of the microfluidic structure; measuring the temperature of the second cooling medium flowing out of the multi-stage heat exchanger; monitoring the mass flow rate of the cooling medium; calculating the amount of heat energy transferred based on the temperatures of the first and second cooling media and the mass flow rate; recording the start and end times of the heat energy transfer process; calculating the duration parameter of the heat energy transfer; statistically analyzing the amount of heat energy transferred per unit time; and organizing the amount of heat energy transferred, the duration parameter, and the amount of heat energy transferred per unit time into heat energy utilization data.

[0083] Specifically, the heat recovery unit employs dual-sensor redundancy measurement technology: both the first and second cooling medium temperatures are measured using a dual-sensor configuration, with two independent temperature sensors installed at each location. The signals are processed through independent signal conditioning channels. A cross-validation mechanism is implemented; when the difference between the two sensor readings exceeds a preset threshold, a fault diagnosis program is initiated, determining a reliable reading through historical data trend analysis and physical model prediction. The temperature sensors undergo a rigorous calibration procedure, considering environmental temperature drift and long-term stability factors, and implementing automatic compensation correction. The calibration process uses a multi-point calibration method, performing calibration at multiple reference temperature points, and determining the calibration curve through polynomial fitting. Mass flow monitoring uses a Coriolis mass flow meter: the flow meter contains a vibrating tube; when fluid flows through the vibrating tube, a Coriolis force is generated, causing a phase difference in the vibrating tube. The phase difference is measured by a high-precision phase detection circuit and converted into mass flow rate. The flow meter is equipped with temperature and pressure sensors to compensate for the influence of fluid density changes on the measurement. A zero-point calibration procedure is implemented, periodically measuring zero-point drift under no-flow conditions for automatic compensation. To improve the accuracy of low-flow-rate measurements, a small-signal cutoff technique is applied to eliminate measurement noise near the zero point. The calculation of heat transfer utilizes an integral method: temperature and flow measurement data are time-aligned, and the trapezoidal integral rule is applied to calculate the heat transfer per unit time. Considering the temperature-dependent variation of the specific heat capacity of the cooling medium, the temperature range is divided into multiple intervals, and linear interpolation is used to approximate the specific heat capacity within each interval to improve calculation accuracy. Energy balance verification is implemented, comparing the heat transfer amount with the theoretical maximum value; when the difference exceeds a threshold, a data review process is triggered. To eliminate the influence of measurement noise, a moving average filter is applied, but short-term fluctuation characteristics are preserved to ensure data integrity and accuracy. The duration parameter is calculated using dynamic threshold detection: start and end conditions for the heat transfer process are defined, and dynamic thresholds are set based on the temperature change rate and heat transfer rate. When the heat transfer rate exceeds the initial threshold, the process is marked as starting; when the heat transfer rate falls below the end threshold and persists for a certain period, the process is marked as ending. Trend analysis is implemented to distinguish between the actual heat transfer process and short-term fluctuations caused by measurement noise. To handle overlapping processes, a state machine model is applied to accurately identify the boundaries of each heat transfer process. The statistical analysis of heat transfer per unit time employs a sliding window algorithm: the window size is dynamically adjusted based on the characteristics of the heat transfer process, using a small window for short-term processes and a large window for long-term processes. Outlier detection is implemented to identify and remove data points that significantly deviate from the normal range. To reflect the changing trends of system performance, the heat transfer per unit time is calculated at multiple time scales, including instantaneous values, short-term averages, and long-term averages. The heat utilization data is organized using a structured data format: data records include data quality markers, anomaly detection indicators, and confidence assessments, facilitating subsequent analysis and utilization.Data compression technology is implemented, employing a combination of differential coding and Huffman coding to reduce storage space. Data encryption uses the AES-256 algorithm to ensure data security. A data version control system is also established to record every step of data processing, ensuring the traceability and repeatability of the data processing process.

[0084] Through the above technical solutions, this application achieves precise quantification and comprehensive recording of the thermal energy utilization process; through scientific data organization and verification methods, it ensures the accuracy and availability of thermal energy utilization data, supporting the continuous improvement and optimization of the system.

[0085] This application further proposes a performance optimization unit that performs structured processing on thermal energy utilization data, combining operating parameter sets and photovoltaic power generation data to generate a structured thermal energy dataset, including:

[0086] The process involves: acquiring power generation efficiency data from photovoltaic (PV) power generation systems as PV power generation data; timestamping thermal energy utilization data, operating parameter sets, and PV power generation data to create a time index; associating and matching thermal energy utilization data with the operating parameter sets according to the time index; matching the associated and matched thermal energy utilization data with PV power generation data according to the time index; cleaning the matched thermal energy utilization data, operating parameter sets, and PV power generation data to remove outliers and missing values; constructing a data table structure including timestamps, thermal energy utilization data, operating parameter sets, and PV power generation data; verifying the logical consistency of each field in the data table structure; and generating a structured thermal energy dataset.

[0087] Specifically, the performance optimization unit first implements data standardization: photovoltaic power generation efficiency data is unified into equivalent efficiency under standard test conditions through conversion formulas, taking into account the effects of irradiance, temperature, and spectral distribution. The correction method in the International Electrotechnical Commission standard IEC 61853 is applied to convert the measured power generation into equivalent power under standard test conditions. The data standardization process considers the configuration differences of photovoltaic arrays and measurement methods, implementing necessary corrections and normalizations to ensure the comparability of data from different sources. The time index is established using high-precision time synchronization technology: an enhanced version of the IEEE 1588 precision time protocol is implemented, considering the nonlinear characteristics of network latency and clock drift. The master clock obtains standard time through a GPS receiver, and the slave clock eliminates the impact of network latency through bidirectional timestamp exchange and filtering algorithms. A clock drift compensation algorithm is applied to predict and compensate for clock drift based on historical clock deviation data. The timestamp is in 64-bit format, achieving nanosecond-level time resolution to ensure time accuracy over long periods of operation. A timestamp verification mechanism is also implemented to detect and correct potential timestamp errors. The data association and matching process employs dynamic time warping: data sequences with different sampling frequencies are aligned using a dynamic time warping algorithm, taking into account the nonlinear time deformation of the data sequences. The algorithm performs multi-scale analysis, first determining the approximate alignment point at a coarse scale, and then precisely locating it at a fine scale. To handle local deformation of the data sequences, elastic matching technology is applied, allowing local time scaling to ensure accurate alignment of key feature points. Association matching also considers the time delay of physical processes, such as the delay effect of heat transfer, and determines the optimal time offset through cross-correlation analysis. The data cleaning stage implements multi-level anomaly detection: primary detection is based on statistical methods, calculating the Z-score of the data and identifying data that significantly deviates from the normal range; intermediate detection is based on physical models, constructing energy balance equations to check whether the data conforms to basic laws such as energy conservation; advanced detection is based on historical data patterns, applying the isolated forest algorithm to identify potential anomaly patterns. Missing value handling uses a spatiotemporal interpolation algorithm, considering not only the continuity in the time dimension but also the correlation in the spatial dimension. Kriging interpolation is applied, considering the spatial autocorrelation of the data, to generate reasonable substitute values. To improve interpolation accuracy, multi-source data fusion is implemented, utilizing information from relevant parameters to enhance the accuracy of missing value estimation. The data table structure adopts a hierarchical model: the base layer stores raw data, including timestamps, original measurements, and data quality markers; the intermediate layer stores feature extraction results, including temperature gradient features, thermal characteristic parameters, and light intensity variation rate features; and the high-level layer stores analysis conclusions, including heat load distribution characteristics, system efficiency indicators, and optimization suggestions. Each layer is tightly linked through a time index, supporting rapid data retrieval and analysis. A data sharding strategy is implemented, dividing data into multiple segments based on time ranges to improve query efficiency under large data volumes.The logical consistency verification implementation rule engine predefines multiple data logic rules, such as "thermal energy utilization should not exceed the theoretical maximum value" and "power generation efficiency should be negatively correlated with temperature." The rule engine uses the Rete algorithm for efficient matching, detecting and marking data that violates the rules in real time. It also performs cross-parameter consistency checks, such as checking whether the relationship between heat exchange and temperature difference, and flow rate, conforms to thermodynamic equations. When a logical inconsistency is detected, a diagnostic program is automatically initiated to determine possible causes and propose corrective measures.

[0088] Through the above technical solutions, this application achieves efficient integration and structured processing of multi-source heterogeneous data, and constructs high-quality, high-value data assets; through strict data quality control and logical verification, it ensures the accuracy and reliability of the structured thermal energy dataset.

[0089] This application further proposes that when the performance optimization unit optimizes the cooling fluid flow strategy and heat recovery method using machine learning algorithms to generate a system performance optimization scheme, it includes:

[0090] Based on a structured thermal energy dataset, historical thermal energy utilization data, historical operating parameter sets, and historical photovoltaic power generation data are acquired. These data are then used as input data for a machine learning algorithm. The machine learning algorithm analyzes the correlation between these data to determine the optimal parameters for the cooling medium flow strategy and the heat recovery method. Finally, the optimal parameters for the cooling medium flow strategy and the heat recovery method are integrated into a system performance optimization scheme.

[0091] Specifically, the performance optimization unit first performs feature engineering: wavelet transform is applied to extract time-frequency domain features of the signal, including short-term fluctuation features, periodic features, and trend features. Feature selection employs a recursive feature elimination method, gradually removing redundant features based on the correlation between features and the target variable. Feature cross-referencing is also implemented to generate combined features reflecting the interaction between parameters, such as the product of temperature gradient and the rate of change of illumination. Feature scaling uses a robust scaler, employing median and interquartile range for scaling to reduce the impact of outliers. The machine learning algorithm adopts an ensemble learning framework: the base models include deep neural networks, gradient boosting trees, and support vector machines, each optimized for different data characteristics and prediction objectives. The deep neural network uses a combination of convolutional layers and long short-term memory layers; convolutional layers extract spatial features, and long short-term memory layers capture temporal dependencies. The gradient boosting tree is implemented using XGBoost, optimized separately for classification and regression tasks. The support vector machine uses radial basis function kernels to provide robust predictions for small sample sizes. The ensemble strategy employs a stacked generalization method, using the output of the base model as the input of the meta-model, which is implemented using logistic regression or a neural network. A multi-objective optimization strategy is implemented using Pareto front analysis: three optimization objectives are defined: maximizing power generation efficiency, maximizing heat recovery efficiency, and minimizing system energy consumption. Non-dominated sorting genetic algorithm II (NSGA-II) is applied for multi-objective optimization to generate a Pareto optimal solution set. Each solution in the set represents an optimization scheme that balances different objectives. Preference modeling is implemented to select the best compromise from the Pareto front based on user preferences or economic analysis. The optimization process considers system constraints, such as flow rate limits, temperature limits, and pressure limits, to ensure the feasibility of the generated optimization scheme. The process for determining the optimization parameters of the cooling fluid flow strategy considers regional heat load distribution: a space-time prediction model is applied to predict the heat load distribution of each region over a future period. Based on the prediction results, model predictive control technology is applied to determine the optimal flow allocation strategy. The optimization process considers the hydrodynamic characteristics of the microchannel network to ensure the feasibility of flow allocation. A rolling optimization strategy is implemented to periodically update the optimization scheme to adapt to changes in system state. To handle uncertainty, a robust optimization method is applied, considering the possible range of prediction errors to generate an optimization scheme robust to uncertainty. The optimization parameters for heat recovery methods are determined by considering the matching degree between heat energy grade and demand: a heat energy demand model is constructed to predict heat energy demand under different application scenarios. Based on demand forecasting, thermoeconomic analysis is applied to evaluate the economics of different heat recovery schemes. The optimization process considers the principle of tiered utilization of heat energy grade, prioritizing the use of high-grade heat energy for high-value applications. Heat energy allocation optimization is implemented to determine the optimal operating parameters of each stage of heat exchanger, achieving efficient utilization of heat energy. To handle dynamic demand, a real-time pricing mechanism is applied to adjust the heat energy allocation strategy according to real-time changes in heat energy demand. An online learning mechanism implements incremental learning: an online gradient descent algorithm is used to gradually update model parameters as new data arrives.To avoid catastrophic forgetting, a resilient weighting consolidation technique is implemented to protect critical parameters from excessive modification by new data. Concept drift detection is also applied, triggering model retraining when a significant change in data distribution is detected. Model validation employs a rolling prediction method, using historical data to predict future performance and assess the model's predictive accuracy. A model fusion strategy is implemented, weighted averaging the predictions from multiple models to improve prediction stability and accuracy. Optimization scheme evaluation utilizes counterfactual analysis: applying causal inference techniques to evaluate the performance differences before and after implementing the optimization scheme. By constructing counterfactual scenarios, the expected performance of the system without implementing the optimization scheme is estimated, thereby accurately evaluating the optimization effect.

[0092] Through the above technical solutions, this application achieves intelligent optimization of system operation strategies, learns the optimal operation mode from historical data, and dynamically adjusts system parameters; through multi-objective optimization and constraint processing, the feasibility and practicality of the optimization scheme are ensured.

[0093] This application further proposes that when the performance optimization unit dynamically adjusts its operating parameters according to the system performance optimization scheme, it includes:

[0094] Analyze the optimization parameters of the cooling medium flow strategy in the system performance optimization scheme; adjust the flow distribution of the proportional flow valve array according to the optimization parameters of the cooling medium flow strategy; analyze the optimization parameters of the heat recovery method in the system performance optimization scheme; adjust the operating parameters of the multi-stage heat exchanger according to the optimization parameters of the heat recovery method.

[0095] Specifically, the performance optimization unit first performs feasibility verification: applying constraint satisfaction problem-solving techniques to check whether the optimization parameters are within the system's physical limitations. This includes checking flow rate range, temperature range, pressure range, and rate of change limits. For parameters exceeding these limits, projection correction is performed, projecting the parameters onto the boundary of the feasible region. Safety constraint checks are also performed to ensure that the optimization parameters will not lead the system into a dangerous state, such as excessively high temperature or pressure. The feasibility verification process considers the system's dynamic characteristics, predicting intermediate states during parameter changes to ensure the safety of the entire process. The analysis of the cooling medium flow strategy optimization parameters employs a hierarchical decoding method: decomposing the global optimization objective into regional flow allocation instructions, considering the topology and hydrodynamic characteristics of the microchannel network. The decoding process uses a network flow algorithm to convert the optimization parameters into flow setpoints for each microchannel. Flow coordination calculations are performed, considering the mutual influence of flow changes in adjacent regions to ensure the physical feasibility of flow allocation. To handle the nonlinear characteristics of the microchannel network, an iterative calculation method is applied to gradually approximate the optimal flow allocation. The decoding results include the target opening, rate of change, and path of change for each proportional flow valve. The flow distribution adjustment of the proportional flow valve array employs a smooth transition strategy: an S-curve is used for flow changes to ensure a smooth transition process. The parameters of the S-curve are dynamically adjusted based on the system's dynamic characteristics to ensure a smooth transition in the shortest possible time while avoiding excessive dynamic stress. A feedforward-feedback composite control is implemented, with the feedforward part predicting the control signal based on the system model and the feedback part correcting for deviations between the actual and target flow rates. To handle system nonlinearity, gain scheduling technology is applied, dynamically adjusting controller parameters based on the operating point. Flow distribution adjustment also considers system inertia, applying predictive control algorithms to compensate for system delays in advance. The analysis of heat recovery optimization parameters considers the matching degree between heat energy grade and demand: the optimized parameters are converted into specific operating parameters for each stage of the heat exchanger, including flow distribution, temperature setting, and operating mode. Thermal network analysis is applied during the analysis process to consider the thermal coupling effect between each stage of the heat exchanger. Heat balance calculations are performed to ensure reasonable heat load distribution among each stage of the heat exchanger. To handle dynamic changes in heat energy demand, a rolling planning method is applied to determine short-term and long-term operating parameter adjustment strategies. The analysis results include the target temperature, flow rate, and operating status of each stage of the heat exchanger. The adjustment of operating parameters for multi-stage heat exchangers employs a collaborative control strategy: implementing inter-stage coordinated control to ensure coordinated operation among heat exchangers at each stage. The control strategy adopts a hierarchical architecture: the upper layer defines the global optimization objective, the middle layer coordinates the operation of each heat exchanger, and the lower layer implements specific control actions. Model predictive control technology is applied, based on a dynamic model of the heat exchange process, to predict the effect of parameter adjustments and optimize the control sequence. To handle the nonlinearity of the heat exchange process, adaptive control technology is applied, dynamically adjusting control parameters according to the system response.

[0096] Through the above technical solutions, this application achieves the precise execution of the system performance optimization scheme, which can transform the optimization results into actual operating parameters and continuously improve system performance; at the same time, through security monitoring and effect evaluation, the safety and effectiveness of the parameter adjustment process are ensured, forming a complete optimization closed loop.

[0097] In summary, by setting microchannels on the backplane and implementing multi-regional differentiated cooling medium flow rate regulation, the temperature distribution and thermal uniformity of the photovoltaic panel can be improved, reducing local high-temperature zones, thereby increasing the photovoltaic module conversion efficiency and reducing temperature-related power decay. Using a multi-stage heat exchanger to transfer heat energy and regulate the temperature of the outflowing cooling medium in stages can convert the heat energy generated by the photovoltaic panel into usable heat energy (such as low-temperature hot water or process heat sources), achieving electrothermal coupling utilization and improving overall energy efficiency and system economy. Introducing multi-dimensional monitoring data such as distributed temperature, inlet / outlet temperature, and irradiance variation rate, and generating regionalized flow commands based on an adaptive fuzzy logic algorithm, enables the cooling strategy to adapt to real-time fluctuations in irradiance and changes in heat load, ensuring cooling effect and system stability. Precise control of the opening degree of each region is achieved through a proportional flow valve array, allowing the cooling medium to form the expected directional gradient flow in the microchannels, which is beneficial for enhancing local heat exchange performance and reducing cycle energy consumption. The cooling medium, after undergoing multi-stage heat exchange and temperature regulation, is returned to the microchannel, forming a closed loop. Combined with continuous monitoring and control, this reduces the thermal degradation of the cooling medium and the frequency of system maintenance, extending equipment lifespan. By structuring thermal energy utilization data, operating parameter sets, and photovoltaic power generation data, and using machine learning algorithms for correlation analysis and modeling, it is possible to uncover system operating patterns, optimize cooling and recovery parameters, and continuously improve energy efficiency and economics over time. By improving power generation efficiency, recovering and utilizing thermal energy, and optimizing operating strategies, energy consumption and operating costs per unit of power generation / heat supply can be reduced, decreasing dependence on fuel or external heat sources, thereby bringing environmental emission reduction and economic benefits.

[0098] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A composite energy system based on photovoltaic panel backsheet heat recovery, characterized in that, include: Microchannel structure, set on the surface of photovoltaic panel backsheet; The data monitoring unit is used to collect multi-dimensional monitoring data in real time, and to preprocess and extract features from the multi-dimensional monitoring data to generate a set of operating parameters, which includes temperature gradient features, thermal characteristic parameters and light irradiance change rate features. The photovoltaic panel cooling unit is used to analyze the heat load distribution characteristics of the photovoltaic panel based on the set of operating parameters and an adaptive fuzzy logic algorithm, and generate multi-region differentiated cooling medium flow rate adjustment instructions; and controls the proportional flow valve array according to the cooling medium flow rate adjustment instructions to make the cooling medium form a directional gradient flow within the microchannel structure. The heat recovery unit is used to transfer the heat energy of the cooling working medium flowing out of the microchannel structure through a multi-stage heat exchanger, adjust the temperature of the cooling working medium after the heat energy transfer, and reintroduce it into the microchannel structure to form a closed loop. And generate thermal energy utilization data; The performance optimization unit is used to perform structured processing on the thermal energy utilization data, and combine the operating parameter set and photovoltaic power generation data to generate a structured thermal energy dataset. Based on the structured thermal energy dataset, the cooling fluid flow strategy and heat recovery method are optimized through machine learning algorithms to generate a system performance optimization scheme, and the operating parameters are dynamically adjusted according to the system performance optimization scheme.

2. The composite energy system based on photovoltaic panel backsheet heat recovery according to claim 1, characterized in that, When the data monitoring unit collects multi-dimensional monitoring data in real time, it includes: The photovoltaic panel backsheet surface temperature distribution data is collected using distributed temperature sensors; the cooling medium inlet temperature data is collected using inlet temperature sensors; the cooling medium outlet temperature data is collected using outlet temperature sensors; and the ambient light intensity data is collected using light sensors. The collected backsheet surface temperature distribution data, cooling medium inlet temperature data, cooling medium outlet temperature data, and ambient light intensity data are organized into multi-dimensional monitoring data, and the timestamp information corresponding to the multi-dimensional monitoring data is acquired simultaneously.

3. The composite energy system based on photovoltaic panel backsheet heat recovery according to claim 2, characterized in that, When the data monitoring unit preprocesses and extracts features from the multi-dimensional monitoring data to generate a set of operating parameters, it includes: The backplate surface temperature distribution data is subjected to noise reduction processing; temperature gradient features are extracted from the backplate surface temperature distribution data; The inlet and outlet temperature data of the cooling medium are synchronously corrected; the heat exchange rate of the cooling medium is calculated as a thermal characteristic parameter. Extract the light change rate feature from the ambient light intensity data; The thermal characteristic parameters, temperature gradient features, and illumination change rate features are integrated to generate a set of operating parameters.

4. The composite energy system based on photovoltaic backsheet heat recovery according to claim 3, characterized in that, When the photovoltaic panel cooling unit analyzes the heat load distribution characteristics of the photovoltaic panel based on the set of operating parameters and uses an adaptive fuzzy logic algorithm to generate multi-region differentiated cooling fluid flow adjustment commands, it includes: The temperature gradient feature is used as the first input variable; the illumination change rate feature is used as the second input variable; and the heat exchange data obtained from the thermal characteristic parameters is used as the third input variable. Each input variable is fuzzified and converted into a fuzzy set; the fuzzy set is then used to perform inference operations through a preset fuzzy rule base to generate inference results; the inference results are then defuzzified to generate heat load distribution characteristics. Calculate the required cooling medium flow rate ratio for different areas of the photovoltaic panel based on the heat load distribution characteristics; generate cooling medium flow rate adjustment instructions corresponding to different areas of the photovoltaic panel.

5. The composite energy system based on photovoltaic panel backsheet heat recovery according to claim 4, characterized in that, The photovoltaic panel cooling unit controls the proportional flow valve array according to the cooling medium flow rate adjustment command, so that the cooling medium forms a directional gradient flow within the microchannel structure, including: The system analyzes the regional flow parameters in the cooling medium flow regulation command; converts the regional flow parameters into opening control signals for proportional flow valves; sends opening control signals to the corresponding valve bodies in the proportional flow valve array; adjusts the flow rate of the cooling medium in different regions of the photovoltaic panel; and achieves directional gradient flow of the cooling medium within the microchannel structure.

6. The composite energy system based on photovoltaic panel backsheet heat recovery according to claim 5, characterized in that, The heat recovery unit transfers heat energy from the cooling medium flowing out of the microchannel structure via a multi-stage heat exchanger. When the transferred cooling medium is temperature-regulated and reintroduced into the microchannel structure, the process includes: The system receives the cooling medium carrying thermal energy flowing out of the microchannel structure; introduces the cooling medium into the first stage of a multi-stage heat exchanger; performs heat energy transfer in the first-stage heat exchanger; introduces the cooling medium processed in the first stage into the second stage of the multi-stage heat exchanger; performs heat energy transfer in the second-stage heat exchanger; adjusts the temperature of the cooling medium that has completed the multi-stage heat exchange; and reintroduces the temperature-adjusted cooling medium into the inlet of the microchannel structure.

7. The composite energy system based on photovoltaic panel backsheet heat recovery according to claim 6, characterized in that, When the heat recovery unit generates heat utilization data, it includes: Measure the temperature of the first cooling medium flowing out of the microchannel structure; measure the temperature of the second cooling medium flowing out of the multi-stage heat exchanger; monitor the mass flow rate of the cooling medium; The amount of heat energy transferred is calculated based on the temperature of the first cooling medium, the temperature of the second cooling medium, and the mass flow rate. Record the start and end times of the heat transfer process; calculate the duration parameter of the heat transfer. The amount of heat energy transferred per unit time; The amount of heat energy transferred, the duration parameter, and the amount of heat energy transferred per unit time are organized into heat energy utilization data.

8. The composite energy system based on photovoltaic panel backsheet heat recovery according to claim 7, characterized in that, When the performance optimization unit performs structured processing on the thermal energy utilization data, combining the operating parameter set and photovoltaic power generation data to generate a structured thermal energy dataset, it includes: Obtain the power generation efficiency data of the photovoltaic power generation system as photovoltaic power generation data; The thermal energy utilization data, operating parameter set, and photovoltaic power generation data are timestamped to establish a time index; The thermal energy utilization data and the operating parameter set are associated and matched according to the time index; the associated and matched thermal energy utilization data and photovoltaic power generation data are matched according to the time index; and the matched thermal energy utilization data, operating parameter set and photovoltaic power generation data are cleaned to remove outliers and missing values. Construct a data table structure, which includes timestamps, thermal energy utilization data, operating parameter sets, and photovoltaic power generation data; verify the logical consistency of each field in the data table structure; and generate a structured thermal energy dataset.

9. The composite energy system based on photovoltaic backsheet heat recovery according to claim 8, characterized in that, The performance optimization unit optimizes the cooling fluid flow strategy and heat recovery method using machine learning algorithms to generate a system performance optimization scheme, including: Based on a structured thermal energy dataset, historical thermal energy utilization data, historical operating parameter sets, and historical photovoltaic power generation data are acquired. These data are then used as input data for a machine learning algorithm. The machine learning algorithm analyzes the correlation between these data to determine the optimal parameters for the cooling medium flow strategy and the heat recovery method. Finally, the optimal parameters for the cooling medium flow strategy and the heat recovery method are integrated into a system performance optimization scheme.

10. The composite energy system based on photovoltaic backsheet heat recovery according to claim 9, characterized in that, When the performance optimization unit dynamically adjusts the operating parameters according to the system performance optimization scheme, it includes: Analyze the cooling medium flow strategy optimization parameters in the system performance optimization scheme; adjust the flow distribution of the proportional flow valve array according to the cooling medium flow strategy optimization parameters; Analyze the optimization parameters of the heat recovery method in the system performance optimization scheme; adjust the operating parameters of the multi-stage heat exchanger according to the optimization parameters of the heat recovery method.

Images