Radiation cooling surface dewing monitoring method and system based on multi-source sensing fusion

Abstract

The application relates to the field of heating, ventilation, air conditioning and indoor environment control, and discloses a radiation cooling surface dewing monitoring method and system based on multi-source sensing fusion, which aims at the problem of lagging monitoring of the traditional static dew point temperature margin method, acquires the radiation surface transient temperature, the environment and the boundary layer water vapor partial pressure, calculates the surface cooling rate and the water vapor migration driving force change rate and maps them to a two-dimensional phase plane to obtain the polar coordinate vector angle; meanwhile, an active thermal disturbance is applied to the system, the dynamic thermal impedance weight coefficient is calculated based on the surface temperature transient response; the dynamic thermal impedance weight coefficient and the polar coordinate vector angle are coupled to calculate the phase change precursor index, and a dewing prevention control signal is output when the index is greater than a preset threshold. The application evaluates the dewing risk from two dimensions of temperature and humidity state evolution dynamics and initial phase change heat resistance, can realize early warning and closed-loop intervention before the formation of macroscopic condensation water droplets, and improves the timeliness of the dewing prevention response of the system.

Description

Technical Field

[0001] This invention relates to the field of heating, ventilation, air conditioning and indoor environment control technology, specifically to a method and system for monitoring condensation on radiant cooling surfaces based on multi-source sensor fusion. Background Technology

[0002] Radiant cooling systems absorb indoor sensible heat load through thermal radiation and natural convection by lowering the temperature of the indoor building envelope or terminal equipment surfaces. Due to their high comfort level and energy-saving potential, they are increasingly widely used in modern building air conditioning systems. However, during operation, radiant cooling systems are prone to condensation when the supply water temperature is too low or a sudden increase in indoor moisture load causes the indoor air dew point temperature to exceed the radiant surface temperature. Condensation not only damages interior decoration materials and fosters mold growth, but in severe cases, it can also damage indoor facilities or cause electrical safety accidents. Therefore, condensation prevention control is a core technical aspect of radiant cooling system operation.

[0003] Current radiant cooling anti-condensation control strategies mainly rely on monitoring methods based on static dew point temperature margin. These methods typically involve placing temperature and humidity sensors indoors to calculate the ambient dew point temperature, while simultaneously attaching temperature sensors to the radiant surface to collect surface temperature data. Condensation is prevented by maintaining the surface temperature above the dew point temperature by a certain margin (i.e., static dew point temperature margin, usually set to 1°C to 2°C).

[0004] This conventional method offers some protection under steady-state conditions, but in practical engineering applications, it reveals significant monitoring lag and control inaccuracies. Significant heat transfer hysteresis exists in radiating terminals with high heat capacity (such as embedded concrete pipes or heavy capillary networks), where surface temperature changes typically lag behind the action of the cold water valve. When a building experiences sudden wet load disturbances such as open windows or a large influx of people, the indoor water vapor partial pressure rises sharply, causing the dew point temperature to rapidly approach or even exceed the surface temperature. In such cases, the static dew point margin method often only triggers the sensor-level temperature alarm limit after macroscopic condensation droplets have already formed or even converged and slid down.

[0005] Furthermore, due to the lack of dynamic assessment of the evolution trend of humid air state and the heat transfer characteristics of surface micro-phase change, existing methods cannot effectively distinguish between the margin reduction caused by normal fluctuations in ambient temperature and the actual water vapor phase change condensation process. To avoid condensation, the system usually needs to set a conservatively high supply water temperature, which directly limits the cooling capacity. When the system forcibly increases the cooling capacity, the anti-condensation control logic is easily affected by environmental noise and may malfunction, leading to frequent start-stop of the chilled water valve, which seriously affects the system's operating life and the stability of indoor thermal comfort. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and system for monitoring condensation on the surface of radiant cooling systems based on multi-source sensor fusion. This solves the problems of existing large heat capacity radiant cooling systems, which often trigger warnings only after macroscopic water droplets have formed due to heat transfer lag when dealing with sudden wet loads. Furthermore, existing methods rely on steady-state temperature comparisons and cannot distinguish between normal fluctuations in ambient temperature and water vapor phase change processes, leading to frequent malfunctions in anti-condensation control.

[0007] To address the above problems, the present invention provides the following technical solution: The first aspect of this invention provides a method for monitoring condensation on a radiative cooling surface based on multi-source sensor fusion, comprising the following steps: The operating status parameters of the radiant cooling system are obtained, including transient temperature of the radiant surface, ambient temperature and humidity, and boundary layer temperature and humidity. Based on the ambient temperature and humidity and the boundary layer temperature and humidity, calculate the ambient water vapor partial pressure and the boundary layer water vapor partial pressure respectively. The surface cooling rate is calculated based on the transient temperature of the radiating surface, and the rate of change of water vapor migration-driven characteristic quantities is calculated based on the difference between the ambient water vapor partial pressure and the boundary layer water vapor partial pressure. The surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity are normalized, and the normalized result is mapped to a two-dimensional phase plane to calculate the polar coordinate vector angle. An active thermal disturbance is applied to the radiant cooling system, and the dynamic thermal impedance weighting coefficient is calculated based on the transient response of the radiant surface to the active thermal disturbance. Based on the product of the dynamic thermal impedance weighting coefficient and the polar coordinate vector angle, the coupling phase change precursor index is calculated, and when the coupling phase change precursor index is greater than the preset safety judgment threshold, the anti-condensation control signal is output.

[0008] The above-mentioned technical solution of the present invention can produce the following technical effects: it combines the cooling process of the radiating surface with the boundary layer water vapor migration process, and transforms the dynamic changes of the above physical processes into polar coordinate vector angles through two-dimensional phase plane mapping, so as to quantitatively characterize the trend of the system state deviating from thermodynamic equilibrium; at the same time, by applying active thermal perturbation to the system and analyzing the transient temperature response, the current dynamic thermal impedance weighting coefficient of the system is calculated and obtained; by using the product of the weighting coefficient and the polar coordinate vector angle to construct the coupled phase change precursor index, the risk of water vapor condensation can be quantitatively assessed, and the single-point threshold judgment is transformed into dynamic early warning calculation based on the coupling of multiple physical quantities.

[0009] Furthermore, in the steps of calculating the surface cooling rate based on the transient temperature of the radiating surface and calculating the rate of change of the water vapor migration-driven characteristic quantity based on the difference between the ambient water vapor partial pressure and the boundary layer water vapor partial pressure: historical sampling sequences of the transient temperature of the radiating surface and historical sampling sequences of the water vapor migration-driven characteristic quantity are extracted within a sliding time window of a set length; the actual timestamp difference corresponding to the valid data at the beginning and end of the sliding time window is obtained respectively; using the actual timestamp difference as the time denominator, combined with the difference between the beginning and end data of the historical sampling sequence, the surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity are calculated respectively. The above steps, by introducing the actual timestamp difference for differential calculation, can reduce the calculation deviation caused by fluctuations in the system sampling period and improve the time accuracy of the rate calculation.

[0010] Furthermore, in the step of normalizing the surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity: the maximum value of the surface cooling rate and the maximum value of the rate of change of the water vapor migration-driven characteristic quantity recorded in the system's historical operating data are extracted and used as the first historical extreme value scaling factor and the second historical extreme value scaling factor, respectively; the first historical extreme value scaling factor and the second historical extreme value scaling factor are used as the normalization denominator to perform a division operation on the currently calculated surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity to complete the normalization process; wherein, according to a set long-term time period, the first historical extreme value scaling factor and the second historical extreme value scaling factor are multiplied by a forgetting coefficient less than one, or the extreme values ​​are reset based on the current seasonal climate characteristics to adjust the benchmark, so as to execute the extreme value forgetting and recovery logic. The above steps, by setting the extreme value forgetting or reset logic, can correct the impact of boundary condition changes caused by long-term system operation on the normalization benchmark.

[0011] Preferably, the step of applying active thermal disturbance to the radiant cooling system includes: obtaining the thermal time constant of the radiant cooling system; when the thermal time constant is less than a preset calibration threshold, confirming that the system has entered a low thermal inertia mode, and generating a continuous sinusoidal micro-disturbance signal as the active thermal disturbance; when the thermal time constant is greater than or equal to the preset calibration threshold, confirming that the system has entered a high thermal inertia mode, and generating a sparse step pulse signal as the active thermal disturbance; wherein the pulse occurrence period of the sparse step pulse signal is greater than the width of a single pulse. The above steps can select the corresponding thermal disturbance signal form according to the thermal inertia characteristics of the building structure where the radiant cooling system is located, reducing the attenuation effect of high heat capacity structures on the disturbance signal.

[0012] Furthermore, in the low thermal inertia mode, the step of calculating the dynamic thermal impedance weighting coefficient based on the transient response of the active thermal disturbance to the transient temperature of the radiating surface includes: extracting the input amplitude corresponding to the sinusoidal micro-disturbance signal; when the input amplitude is lower than a preset minimum effective amplitude, using the previous effective dynamic thermal impedance weighting coefficient, or switching to a degraded calculation mode; wherein, the preset minimum effective amplitude is determined based on the equivalent noise temperature difference of the infrared temperature sensor or the sampling resolution of the system's analog-to-digital converter; when the input amplitude meets the preset minimum effective amplitude, extracting the amplitude attenuation ratio, and constructing the dynamic thermal impedance weighting coefficient based on the amplitude attenuation ratio. The above steps utilize hardware equivalent parameters to set the minimum effective amplitude, avoiding the use of noise as an effective signal in the calculation under weak disturbance conditions.

[0013] Furthermore, in the high thermal inertia mode, the step of calculating the dynamic thermal impedance weighting coefficient based on the transient response of the radiant surface to the active thermal disturbance includes: extracting the transient temperature recovery segment data caused by the sparse step pulse signal and fitting it to obtain the response recovery segment decay time constant at the current sampling time; obtaining the reference response recovery segment decay time constant obtained by the calibration of the radiant cooling system under initial dry conditions; calculating the dynamic thermal impedance weighting coefficient using the natural constant as the base and the target product as the exponent; the target product is obtained by multiplying the system correction coefficient by the difference between the response recovery segment decay time constant and the reference response recovery segment decay time constant, and then dividing by the reference response recovery segment decay time constant. The above steps calculate the weighting coefficient by analyzing the transient recovery characteristics of temperature under sparse step pulse disturbance to quantify the change in heat transfer resistance under a high heat capacity structure.

[0014] Preferably, the steps following the output of the anti-condensation control signal further include: limiting the maximum allowable opening of the water valve of the radiant cooling system according to the anti-condensation control signal, pausing the application of the active thermal disturbance, and locking the dynamic thermal impedance weighting coefficient to the current calculated value at the time of triggering the warning; during the duration of the anti-condensation control signal, updating the coupled phase change precursor index based on the real-time calculated polar coordinate vector angle; if the coupled phase change precursor index falls below the safety release threshold, then the restriction on the maximum allowable opening of the water valve is lifted; if sufficient transient response data is not collected within a preset recovery time to update the dynamic thermal impedance weighting coefficient, then the degraded calculation mode is maintained and the priority of the anti-condensation control signal is preserved; wherein, the duration limit of the preset recovery time is determined based on the length of the sliding time window for extracting feature data. The above steps limit the opening of the cold source and lock the correlation coefficient after outputting the control signal, and at the same time set the degraded processing logic in the case of missing transient response data, ensuring the continuity of system operation during intervention control.

[0015] Preferably, in the step of obtaining the operating status parameters of the radiant cooling system: the transient temperature of the radiant surface is acquired non-contactly by an infrared temperature sensor; ambient temperature and humidity data are acquired by an ambient temperature and humidity sensor set in the indoor reference area, and the ambient water vapor partial pressure is calculated; boundary layer temperature and humidity data are acquired by a boundary layer temperature and humidity sensor set close to the radiant surface, and the boundary layer water vapor partial pressure is calculated.

[0016] A second aspect of the present invention provides a radiative cooling surface condensation monitoring system based on multi-source sensor fusion, comprising: The data acquisition module is used to acquire the operating status parameters of the radiant cooling system, including the transient temperature of the radiant surface, ambient temperature and humidity, and boundary layer temperature and humidity. The rate calculation module is used to calculate the ambient water vapor partial pressure and the boundary layer water vapor partial pressure based on the ambient temperature and humidity and the boundary layer temperature and humidity, respectively; and to calculate the surface cooling rate based on the transient temperature of the radiating surface, and to calculate the rate of change of water vapor migration-driven characteristic quantities based on the difference between the ambient water vapor partial pressure and the boundary layer water vapor partial pressure. The phase plane mapping module is used to normalize the surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity, and to map the normalized result to a two-dimensional phase plane to calculate the polar coordinate vector angle. The impedance assessment module is used to apply an active thermal disturbance to the radiant cooling system and calculate the dynamic thermal impedance weighting coefficient based on the transient response of the radiant surface to the active thermal disturbance. The early warning control module is used to calculate the coupling phase change precursor index based on the product of the dynamic thermal impedance weighting coefficient and the polar coordinate vector angle, and output an anti-condensation control signal when the coupling phase change precursor index is greater than a preset safety judgment threshold.

[0017] Furthermore, the system also includes a sensing and detection end that is communicatively connected to the data acquisition module. The sensing and detection end includes: an infrared temperature sensor for non-contact acquisition of the transient temperature of the radiating surface; an ambient temperature and humidity sensor, set in an indoor reference area, for acquiring ambient temperature and humidity data to calculate the ambient water vapor partial pressure; and a boundary layer temperature and humidity sensor, set close to the radiating surface of the radiant cooling system, for acquiring boundary layer temperature and humidity data to calculate the boundary layer water vapor partial pressure.

[0018] This invention provides a method and system for monitoring condensation on surfaces in radiative cooling systems based on multi-source sensor fusion. It offers the following advantages: 1. This invention maps the surface cooling rate to the rate of change of water vapor migration driving force in a two-dimensional phase plane to obtain a polar coordinate vector angle, and couples it with the dynamic thermal impedance weighting coefficient extracted under active thermal perturbation to calculate the pre-phase change index. This mechanism simultaneously assesses the risk of condensation from two dimensions: the dynamics of temperature and humidity changes and the heat transfer resistance caused by the latent heat of the initial phase change. It can identify changes in surface state before the formation of macroscopic condensate droplets, solving the monitoring lag problem of the traditional static dew point temperature margin method and improving the proactive response capability of the system's anti-condensation control.

[0019] 2. In calculating the surface cooling rate and the rate of change of water vapor migration driving force, this invention extracts the difference between the actual timestamps at the beginning and end of the feature data sequence as the time denominator for the differential operation. This calculation logic replaces the conventional method that uses a fixed sampling period based on the control system theory, eliminating timing jitter interference caused by packet loss in communication networks or congestion of the underlying data bus in practical engineering applications, and ensuring the numerical stability of the state evolution rate calculation results.

[0020] 3. This invention sets a minimum effective disturbance amplitude based on the inherent hardware noise threshold of the sensor to filter out false heat transfer identification results generated under weak excitation. Furthermore, it introduces extreme value forgetting and recovery logic, causing the scaling factor in the algorithm's normalization process to dynamically decay with the operating cycle or be reset seasonally. This mechanism prevents the algorithm's calculation denominator from becoming excessively large due to single, occasional extreme operating condition records, thus avoiding degradation of the system's state mapping sensitivity during cross-seasonal operation. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the architecture of the radiation cooling surface condensation monitoring system of the present invention; Figure 2 This is a flowchart of the radiant cooling surface condensation monitoring method of the present invention; Figure 3 This is a timing diagram of the thermal time constant calibration response of the present invention; Figure 4 This is a schematic diagram of the active thermal disturbance command waveform generation of the present invention; Figure 5 This is a schematic diagram of the timing of multi-source data synchronous acquisition and preprocessing according to the present invention; Figure 6 This is a schematic diagram illustrating the principle of dynamic thermal impedance feature extraction in this invention. Figure 7 This is a schematic diagram illustrating the calculation principle of microenvironment water vapor dynamic parameters in this invention. Figure 8 This is a schematic diagram illustrating the feature transformation principle of phase plane vector mapping in this invention. Figure 9 This is a schematic diagram of the coupled phase transition precursor index calculation and state monitoring logic of the present invention; Figure 10 This is the logic diagram for condensation risk assessment and closed-loop intervention control of the present invention; Figure 11 This is a two-dimensional phase plane state trajectory mapping diagram of the present invention; Figure 12 This is a time-series diagram comparing the condensation risk index and system response of the present invention; Figure 13 This is a statistical comparison chart of the early warning lead time and false alarm rate under multiple different operating conditions of the present invention.

[0022] Among them, 10 is the processing module; 20 is the acquisition module; 21 is the ambient temperature and humidity sensor; 22 is the boundary layer temperature and humidity sensor; 23 is the infrared temperature sensor; 30 is the execution module; 31 is the water supply temperature sensor; 32 is the water valve; 40 is the computer equipment; 41 is the processor; 42 is the memory; and 43 is the storage medium. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] See attached document Figure 1 The radiant cooling surface condensation monitoring system includes a processing module 10, a data acquisition module 20, and an execution module 30.

[0025] The data acquisition module 20 includes an ambient temperature and humidity sensor 21, a boundary layer temperature and humidity sensor 22, and an infrared temperature sensor 23.

[0026] The ambient temperature and humidity sensor 21 is placed in an area of ​​the room that is not directly affected by hot or cold sources, and is used to obtain the dry-bulb temperature and relative humidity of the air.

[0027] The boundary layer temperature and humidity sensor 22 is positioned 3 to 5 centimeters away from the radiant cooling surface to acquire the boundary layer air temperature and relative humidity.

[0028] Infrared temperature sensor 23 is positioned facing the radiant cooling surface to acquire transient temperature data of the radiant surface. Processing module 10 performs emissivity compensation on the transient temperature data acquired by infrared temperature sensor 23 based on the emissivity parameters of the radiant cooling surface material; alternatively, it performs on-site calibration of infrared temperature sensor 23 during system debugging to reduce the impact of surface material differences on the temperature measurement results.

[0029] The execution module 30 is located at the mixing center and includes a water supply temperature sensor 31 and a water valve 32.

[0030] Water supply temperature sensor 31 is installed on the water supply pipeline to obtain the water supply temperature.

[0031] Water valve 32 is a three-way mixing regulating valve or a water supply regulating valve with equivalent regulating function. It is used to receive analog control commands output by processing module 10 and change the water supply temperature and cooling input entering the radiant pipe network by adjusting the mixing ratio of chilled water and return water and / or the flow rate of cold water.

[0032] In one embodiment, the water valve 32 is an electrically controlled regulating valve with opening feedback function, and the processing module 10 obtains the valve opening feedback sequence of the water valve 32 through an analog feedback signal or a digital communication interface. In another embodiment, when the water valve 32 does not have an opening feedback function, the processing module 10 uses the active thermal disturbance command sequence as the water-side input characterization sequence.

[0033] The execution module 30 may also include a flow sensor installed in the water supply or return pipeline to acquire water flow data. When the system is not equipped with a flow sensor, the processing module 10 estimates the water flow based on the opening command or opening feedback value of the water valve 32 and in conjunction with the pre-stored valve flow characteristic curve.

[0034] The processing module 10 is communicatively connected to both the acquisition module 20 and the execution module 30. It receives data uploaded by the acquisition module 20 and the execution module 30, and performs data processing, feature extraction, state mapping, and control command generation. The processing module 10 is equipped with a floating-point arithmetic unit to meet the accuracy requirements of temperature and humidity calculation, dynamic feature extraction, and phase plane mapping operations.

[0035] See attached document Figure 2 The method for monitoring condensation on radiant cooling surfaces can be performed according to the following steps: S10, after the processing module 10 completes the power-on initialization, it sends a step test command to the water valve 32 and simultaneously receives the transient temperature collected by the infrared temperature sensor 23. It calculates the step response time based on the response process of the radiating surface to the step disturbance in order to obtain the thermal time constant. S20, the processing module 10 determines the form and parameters of the active thermal disturbance command according to the thermal time constant, and outputs the active thermal disturbance command to the water valve 32 after superimposing it on the reference control command; wherein, the active thermal disturbance command is a low-frequency sinusoidal micro-disturbance signal or a low-frequency sparse step pulse signal; S30, the acquisition module 20 and the water supply temperature sensor 31 synchronously acquire data according to a preset discrete sampling period. The processing module 10 receives the dry-bulb temperature, relative humidity, boundary layer air temperature, boundary layer relative humidity, and transient temperature uploaded by the acquisition module 20, and receives the water supply temperature uploaded by the water supply temperature sensor 31; at the same time, the processing module 10 synchronously receives or calculates the valve opening feedback value, the measured water flow value, the estimated water flow value, and / or the estimated water-side input disturbance value, and performs low-pass digital filtering processing on the above data; S40, the processing module 10 extracts dynamic thermal response features based on the water-side input characterization sequence and the transient temperature sequence. The water-side input characterization sequence includes at least one of the following: an active thermal disturbance command sequence, a valve opening feedback sequence, a water supply temperature sequence, or a water-side input disturbance estimation sequence estimated based on the water supply temperature, water flow rate, and a preset reference water temperature. The processing module 10 calculates the phase delay and amplitude attenuation ratio in the frequency domain, or calculates the peak response time lag and the response recovery segment attenuation time constant in the time domain, and constructs dynamic thermal impedance weighting coefficients based on the extracted features. S50, the processing module 10 calculates the water vapor migration driving characteristic quantity based on the received temperature and humidity data at each point, and simultaneously extracts the time domain change rate of the water vapor migration driving characteristic quantity and the cooling rate of the radiating surface through the sliding time window differential algorithm. S60, the processing module 10 performs dimensionless normalization on the time-domain change rate and cooling rate, and maps the normalization result to a preset two-dimensional phase plane, thereby calculating the polar coordinate vector angle and magnitude of the current state coordinates in the two-dimensional phase plane. ; S70, the processing module 10 multiplies the polar coordinate vector angle by the dynamic thermal impedance weighting coefficient to obtain the coupling phase transition precursor index. The processing module 10 compares the coupling phase transition precursor index with the safety judgment threshold and records the over-limit status within the continuous confirmation filtering window; S80, when the coupling phase change precursor index does not exceed the safety judgment threshold, the processing module 10 maintains the current active thermal disturbance command and continues to perform cyclic monitoring; when the coupling phase change precursor index exceeds the safety judgment threshold and meets the confirmation conditions, the processing module 10 determines that the radiant cooling surface has entered the phase change precursor stage, generates an anti-condensation intervention command and sends it to the water valve 32, and reduces the cold input of the radiant terminal by reducing the opening degree of the cold water side, increasing the mixing ratio of the return water side, or increasing the supply water temperature setting value of the mixing center.

[0036] See attached document Figure 3 When implementing the radiant cooling surface condensation monitoring method, step S10 is used to complete the initial calibration of the thermal time constant of the radiant cooling system. Specifically, the processing module 10 can extract the thermal time constant according to the following process.

[0037] S101, after the system is powered on and initialized, the processing module 10 detects the system's operating status and sends a step test command with a fixed amplitude to the water valve 32 of the execution module 30.

[0038] This step test command is used to adjust the valve opening of water valve 32 from a preset initial opening or current reference opening to a preset constant opening percentage, and maintain this fixed opening. As a possible implementation, the preset constant opening percentage can be set to 30% to 50% of the maximum opening stroke of water valve 32.

[0039] After water valve 32 performs a step adjustment, cooling water with a set supply temperature enters the radiant pipe network, or the flow rate of cooling water entering the radiant pipe network changes, thus creating a water-side thermal flow disturbance within the system. Because different building radiant terminals differ in structural thermal resistance and material heat capacity—for example, thin metal radiant panels and concrete embedded pipe slabs have different surface temperature hysteresis response characteristics to internal water flow disturbances—the thermal response hysteresis characteristics of the current radiant terminal under field conditions can be measured by applying a step test command, providing physical calibration parameters for the matching and generation of subsequent active thermal disturbance commands.

[0040] S102, while issuing the step test command, the processing module 10 opens the data interception time window and schedules the infrared temperature sensor 23 to acquire the transient temperature data of the radiation surface in discrete sampling mode.

[0041] Within the data capture time window, the processing module 10 continuously reads and monitors the time series of transient temperature data. When the rate of change of the transient temperature data approaches zero and remains within the preset error band, for example, when the temperature fluctuation amplitude is less than 0.1 degrees Celsius for ten consecutive sampling cycles, or when the continuous running time of the data capture time window reaches the preset maximum calibration time limit, the processing module 10 determines the corresponding moment as the end point of the data capture time window.

[0042] By setting a maximum calibration time limit, it is possible to avoid the system failing to reach thermodynamic steady state for an extended period due to external environmental thermal disturbances, thus preventing the calibration process from continuously waiting. The data capture time window covers the dynamic process of the radiating surface responding to the initial temperature change and the step disturbance of the water-side heat flow, gradually approaching a new thermal equilibrium state.

[0043] S103, the processing module 10 extracts the transient temperature discrete data points within the data interception time window and calculates the thermal time constant of the radiative cooling system based on the first-order thermal response model.

[0044] Processing module 10 substitutes the temperature characteristic values ​​of the start and end stages of the data sequence into the system response equation, which is as follows: ; In the formula, This indicates the time elapsed within the intercept time window by the infrared temperature sensor 23. The collected transient temperature; This represents the initial steady-state temperature of the radiating surface before the step test command is issued; This represents the final steady-state temperature of the radiating surface at the end of the intercepted time window; A variable representing the duration from the date the instruction was issued; This represents the thermal time constant of the radiative cooling system to be solved.

[0045] Processing module 10 matches the actually acquired discrete temperature sequence with the aforementioned system response equation. In one specific embodiment, when the temperature change of the radiating surface reaches 63.2% of the total temperature difference between the initial steady-state temperature and the final steady-state temperature, processing module 10 uses the elapsed time corresponding to that moment as the thermal time constant. The preliminary estimate.

[0046] To improve parameter extraction accuracy and reduce the impact of single-point noise, the processing module 10 can also substitute the extracted discrete temperature data into the exponential response equation and obtain the thermal time constant through least-squares curve fitting. The obtained thermal time constant is used to characterize the inherent thermal inertia of the radiating terminals in the current building room and serves as the basic parameter for generating adaptive active thermal disturbance commands.

[0047] See attached document Figure 4 In implementing the radiant cooling surface condensation monitoring method, step S20 is used to generate a matching active thermal disturbance command based on the thermal inertia characteristics of the radiant terminal. In actual engineering, the response characteristics of water-side disturbances transmitted to the radiant surface vary among different buildings due to differences in terminal structure, covering layer material, and heat capacity. Therefore, the processing module 10 determines the active thermal disturbance command based on the thermal time constant obtained in the aforementioned steps in the following manner.

[0048] S201, the processing module 10 compares the thermal time constant with the preset calibration threshold.

[0049] In one specific embodiment, the calibration threshold is determined based on prior physical statistical data of the radiant terminal type. For example, the calibration threshold is relatively small for lightweight metal radiant panel systems and relatively large for concrete pipe-embedded floor slab systems; the calibration threshold can be set in the range of twenty to thirty minutes.

[0050] When the acquired thermal time constant is less than the calibration threshold, the processing module 10 classifies the current radiant cooling system into a low thermal inertia mode; when the thermal time constant is greater than or equal to the calibration threshold, the processing module 10 classifies the current radiant cooling system into a high thermal inertia mode. Through the above mode classification, the processing module 10 can avoid adapting to different thermal inertia terminals with a single fixed perturbation form, thereby reducing the risk of perturbation identification failure in high heat capacity terminals.

[0051] S202, in low thermal inertia mode, the processing module 10 generates a continuous low-frequency sinusoidal micro-perturbation signal.

[0052] The processing module 10 superimposes the low-frequency sinusoidal micro-disturbance signal onto the system's reference control command to form an active thermal disturbance command, and outputs this active thermal disturbance command to the water valve 32 of the execution module 30. The reference control command is the conventional control signal corresponding to maintaining the indoor base set temperature. In this mode, the total control command can be expressed as: ; In the formula, This indicates the current moment's overall control command output by processing module 10; Indicates the reference control command at the current moment; The amplitude parameter represents the low-frequency sinusoidal micro-perturbation signal; This indicates the set characteristic frequency; Represents a time variable.

[0053] Processing module 10 for amplitude parameters Physical boundary constraints are imposed. Amplitude parameters serve as feasible technical implementation parameters. The opening degree is limited to 5% to 10% of the valve opening corresponding to the baseline control command. This limitation ensures that heat flow fluctuations in the water supply pipeline do not cause significant macroscopic air temperature fluctuations when transferred to the indoor space. For example, the change in indoor air temperature can be controlled to no more than 0.2 degrees Celsius, thereby maintaining the original thermal comfort level while performing condensation precursor detection.

[0054] In addition, characteristic frequencies Set to low-frequency parameters in the millihertz range, such as 0.1 millihertz to 5 millihertz, to match the mechanical response speed of the HVAC valve and the thermal response hysteresis characteristics of the radiant terminals.

[0055] S203, in the high thermal inertia mode, due to the obvious low-pass filtering effect of the building structure layer, continuous small-amplitude sinusoidal disturbances are easily weakened by the structural thermal capacity, and the water valve 32 has difficulty in stably following continuous small-amplitude signals.

[0056] In this mode, processing module 10 stops generating continuous sinusoidal micro-perturbation signals and instead generates low-frequency sparse step pulse signals, which are then superimposed on the reference control command. The total control command in this mode can be expressed as: ; In the formula, This represents a discrete rectangular pulse function, or a sparse rectangular pulse function containing a forward excitation segment and a reverse compensation segment, which has a set pulse generation period. Single pulse width and step amplitude .

[0057] Processing module 10 outputs a discrete rectangular pulse function, causing water valve 32 to operate within a single pulse width. The valve performs a relatively large-scale abrupt change in opening. This type of large-stroke abrupt change helps to overcome the mechanical friction dead zone of the water valve 32 and generates a strong transient thermal excitation on the water side.

[0058] As one specific implementation method, the single pulse width The step amplitude can be set to 1.5 to 2 times the mechanical stroke time required for water valve 32 to go from fully closed to fully open; The opening increment can be set to 30% to 50%.

[0059] Short-duration, strong water flow changes can create transient thermal flow disturbances containing multiple frequency components. These transient thermal flow disturbances penetrate high thermal resistance building coverings more easily than continuous micro-amplitude disturbances, and generate transient temperature fluctuations on radiating surfaces that can be captured by infrared thermometer 23.

[0060] Processing module 10 will generate pulse cycles Set to be significantly larger than the single pulse width The numerical value. For example, the pulse generation period. The single pulse width can be set. More than ten times that of the pulse signal, which keeps the pulse signal sparsely distributed on the time axis, thereby reducing the average disturbance of the water channel injected into the system per unit time.

[0061] when When using a sparse rectangular pulse function that includes a positive excitation segment and a reverse compensation segment, the processing module 10 makes the integral value of the disturbance command within one pulse cycle approach zero, so as to reduce the impact of pulse excitation on the indoor macroscopic temperature and reduce the risk of the indoor temperature deviating from the set range.

[0062] See attached document Figure 5In implementing the radiant cooling surface condensation monitoring method, step S30 is used to uniformly collect, time-series align, and pre-process data from different sources and sampling frequencies. Because various sensors in a building HVAC system differ in response speed, communication interface, and upload frequency, the processing module 10 performs time-series unification and noise reduction on the heterogeneous data according to the following process.

[0063] S301, the processing module 10 sets a reference discrete sampling period and uses this reference discrete sampling period as a unified time scale for multi-source data synchronous processing.

[0064] As a possible implementation method, the reference discrete sampling period is determined according to the hardware refresh rate of the infrared temperature sensor 23, for example, it is set to one to five seconds.

[0065] The ambient temperature and humidity sensor 21, boundary layer temperature and humidity sensor 22, and infrared temperature sensor 23 in the acquisition module 20, and the water supply temperature sensor 31 in the execution module 30, have different data upload frequencies due to differences in hardware interfaces and communication cycles. The processing module 10 receives raw data with local timestamps uploaded by each sensor in real time. The raw data includes dry-bulb air temperature, relative humidity, boundary layer air temperature, boundary layer relative humidity, transient temperature, and water supply temperature. Simultaneously, the processing module 10 receives or calculates valve opening feedback values, measured water flow rates, estimated water flow rates, and / or estimated water-side input disturbance values. The above data is sequentially written into a circular buffer in memory according to the reception time or data timestamp.

[0066] Before performing multi-source data fusion, processing module 10 performs time alignment on the data in the circular buffer using the benchmark discrete sampling period as the time scale to reduce the impact of sampling time offset on subsequent time-domain difference calculations. Specifically, at the current benchmark sampling trigger time, processing module 10 searches the circular buffer and selects the valid historical sample value in each data channel whose timestamp is closest to the trigger time as the synchronization data for the current benchmark sampling time, thereby forming a multi-source synchronized data matrix under the same time benchmark.

[0067] To prevent sensor communication anomalies from causing subsequent algorithms to use expired data or enter an invalid waiting state, the processing module 10 calculates the time difference between the timestamp of a valid historical sample value and the current reference sampling trigger time when extracting valid historical sample values. When the time difference exceeds a preset data validity period limit, for example, if no new data from the corresponding sensor is obtained within ten consecutive reference sampling periods, the processing module 10 determines that the sensor channel data is invalid and sets the corresponding communication anomaly status flag.

[0068] S302 In actual engineering sites, the raw data collected by the sensor is easily affected by electrical interference, communication jitter and local airflow disturbances in the room, and manifests as high-frequency numerical spikes in the time domain.

[0069] Processing module 10 performs low-pass digital filtering on the time-aligned multi-source synchronized data matrix to reduce random disturbances and preserve the gradual trends of temperature, humidity, and surface temperature. Taking any environmental variable time series in the multi-source synchronized data matrix as an example, processing module 10 employs a first-order lag filtering algorithm in the discrete time domain, and its filtering process can be expressed as follows: ; In the formula, This represents the effective data value after filtering at the current discrete sampling time. This represents the original input data value extracted from the multi-source synchronous data matrix at the current discrete sampling time; This represents the filtered output data value stored at the previous discrete sampling time. A sequence index variable representing the reference discrete sampling period; This represents the filter coefficients.

[0070] The processing module 10 sets corresponding filtering coefficients based on the changing characteristics of different physical quantities. For example, temperature parameters are affected by the thermal inertia of the radiating terminals and the building envelope, and their changes are relatively gradual; their filtering coefficients can be set to 0.05 to 0.1. Relative humidity parameters are more sensitive to fluctuations in water vapor in the local microenvironment; their filtering coefficients can be set to 0.1 to 0.2. Through these settings, high-frequency noise spikes can be suppressed while retaining the dynamic change information required for condensation precursors.

[0071] In other embodiments, the processing module 10 may also use other low-pass digital filters to replace the first-order hysteresis filtering algorithm described above, based on the characteristics of the on-site noise spectrum. After filtering is completed, the processing module 10 writes the filtering results of each data channel into the corresponding register or buffer, serving as the data basis for subsequent dynamic thermal impedance feature extraction and water vapor migration-driven feature calculation.

[0072] See attached document Figure 6 When implementing the surface condensation monitoring method for radiant cooling, step S40 is used to extract dynamic thermal impedance characteristics from the surface temperature response caused by active thermal disturbance, and construct dynamic thermal impedance weighting coefficients accordingly.

[0073] When the surface temperature of the radiant cooling system gradually approaches the air dew point temperature and exhibits a microscopic phase change trend, the latent heat of phase change released during the condensation process will offset part of the cooling capacity, causing a change in the response characteristics of the radiant surface to water-side input disturbances. The processing module 10 extracts the thermal impedance change characteristics using either frequency domain or time domain methods, based on the system's current thermal inertia mode and the type of active thermal disturbance command.

[0074] S401, when the system is in low thermal inertia mode and the active thermal disturbance command is a continuous sinusoidal micro-disturbance signal, the processing module 10 adopts the frequency domain feature extraction method.

[0075] The processing module 10 selects a data window from memory containing at least three complete disturbance cycles. This data window includes a water-side input characterization sequence and a transient temperature sequence. The water-side input characterization sequence can be at least one of the following: an active thermal disturbance command sequence, a valve opening feedback sequence, a water supply temperature sequence, or a water-side input disturbance estimation sequence estimated based on water supply temperature, water flow rate, and a preset reference water temperature.

[0076] Processing module 10 applies a Hanning window to the data window to reduce the impact of spectral leakage on feature frequency extraction, and then performs a fast Fourier transform to convert the time-domain data into frequency-domain data. Processing module 10 locates the feature frequency point corresponding to the active thermal disturbance signal in the frequency-domain data, extracts the input amplitude and input phase of the water-side input characterization sequence at that frequency point, and simultaneously extracts the output amplitude and output phase of the transient temperature sequence at that frequency point.

[0077] When the input amplitude is lower than the preset minimum effective amplitude, the processing module 10 determines that the input excitation in the current analysis window is insufficient to support effective thermal impedance identification, and uses the previous effective dynamic thermal impedance weighting coefficient, or switches to a degraded calculation mode.

[0078] As a specific implementation method, the preset minimum effective amplitude is determined based on the equivalent noise temperature difference of the infrared temperature sensor 23 or the sampling resolution of the system analog-to-digital converter, so as to ensure that the signal-to-noise ratio of the extracted signal meets the algorithm identification requirements.

[0079] When the input amplitude meets the validity condition, the processing module 10 uses the difference between the output phase and the input phase as the phase delay, and the ratio between the output amplitude and the input amplitude as the amplitude attenuation ratio. The phase delay and amplitude attenuation ratio are used to characterize the degree of transmission hysteresis and attenuation of water-side thermal disturbances by the radiation cover layer.

[0080] S402, when the system is in high thermal inertia mode and the active thermal disturbance command is a sparse step pulse signal, since the pulse response is not suitable for stable periodic frequency domain analysis, the processing module 10 switches to time domain feature extraction mode.

[0081] The processing module 10 uses the rising edge of the pulse signal as the zero point for calculation and searches backward in the transient temperature sequence for the response extremum point with the largest deviation from the reference temperature before the pulse. When the pulse causes the radiation surface temperature to drop, this response extremum point corresponds to a local minimum point; when the absolute amplitude of the temperature disturbance is used as the analysis object, this response extremum point corresponds to the absolute amplitude peak point.

[0082] To reduce the impact of residual noise on extreme value identification, the processing module 10 performs a time-domain neighborhood consistency check after detecting the extreme value point, that is, confirms that the data change trend within a preset number of adjacent sampling periods before and after the point meets preset conditions. As a feasible approach, the number of adjacent sampling periods can be set to 3 to 5 periods.

[0083] After verification, the processing module 10 determines the extreme point of the response as the peak point of the response, and calculates the time difference between the corresponding time of the peak point and the calculation zero point, which is used as the peak response time lag. Subsequently, the processing module 10 extracts the transient temperature recovery segment data after the peak point of the response, and uses a first-order exponential decay model to fit the curve of the recovery segment to obtain the decay time constant of the response recovery segment.

[0084] If no valid response peak is detected within a complete pulse generation cycle, the processing module 10 determines that the transient heat flux has not formed a valid penetration response within that cycle, and assigns the response peak time lag and the response recovery segment decay time constant to a preset safety upper limit value to avoid the algorithm entering an invalid waiting state. As a specific implementation, the preset safety upper limit value can be set to two to three times the initial thermal time constant of the system obtained in step S10.

[0085] S403, the processing module 10 constructs a dynamic thermal impedance weighting coefficient based on the above-mentioned characteristic parameters.

[0086] Before constructing the dynamic thermal impedance weighting coefficient, the processing module 10 first calibrates the reference thermal impedance characteristic parameters. Specifically, during the initial system commissioning phase, the processing module 10 controls the water supply temperature to be higher than the current indoor air dew point temperature by a preset safety margin, and confirms that the radiant cooling surface is in a dry condition. Under this dry condition, the processing module 10 continuously executes multiple active thermal disturbance cycles, calculates at least one characteristic parameter among amplitude attenuation ratio, phase delay, response peak time lag, and response recovery segment attenuation time constant, and stores their statistical mean or median as reference thermal impedance characteristic parameters.

[0087] The reference thermal impedance characteristic parameters include the reference amplitude attenuation ratio. The decay time constant of the reference response recovery segment and / or the peak time lag of the reference response .

[0088] When the radiating surface does not undergo a phase change, the extracted characteristic parameters are close to the baseline level under dry conditions. When the radiating surface is in the pre-condensation stage, the latent heat of phase change will weaken the surface temperature's ability to follow water-side input disturbances, which manifests as a decrease in amplitude attenuation ratio or an increase in the attenuation time constant of the response recovery segment.

[0089] In low thermal inertia mode, the processing module 10 can construct an exponential dynamic thermal impedance weighting coefficient based on the amplitude attenuation ratio, and its calculation formula is as follows: ; In the formula, This represents the dynamic thermal impedance weighting coefficient corresponding to the current sampling time; This represents the amplitude attenuation ratio calculated at the current moment; This represents the reference amplitude attenuation ratio obtained under normal dry operating conditions (such as when the water supply temperature is more than three degrees Celsius higher than the indoor dew point temperature); Indicates the system correction coefficient; Represented by natural constant An exponential function with base 0.

[0090] Processing module 10 uses the aforementioned exponential relationship to nonlinearly amplify the deviation of thermal impedance from the reference state. As a possible implementation, the system correction coefficient... The value can be set according to the thermal conductivity of the radiant terminal material, the thickness of the covering layer, or the results of on-site calibration, for example, a value of 2.0 to 4.0.

[0091] In the high thermal inertia mode, processing module 10 constructs dynamic thermal impedance weighting coefficients based on time-domain response characteristics. When the response recovery segment decay time constant is used as the input variable, the dynamic thermal impedance weighting coefficients can be expressed as: ; In the formula, This represents the decay time constant of the response recovery segment extracted within the current pulse period. This represents the decay time constant of the baseline response recovery segment obtained during system calibration under dry operating conditions.

[0092] Optionally, the processing module 10 can also construct weighting coefficients based on the peak response time lag, and the calculation formula is as follows: ; In the formula, This indicates the peak response time lag extracted within the current pulse period; This represents the peak time lag of the baseline response obtained by calibrating the system under dry operating conditions.

[0093] In another embodiment, the processing module 10 can perform a weighted fusion of the weighting coefficients obtained based on the decay time constant of the response recovery segment and the weighting coefficients obtained based on the response peak time lag, and use the fusion result as the dynamic thermal impedance weighting coefficients.

[0094] The dynamic thermal impedance weighting coefficient generated by the processing module 10 is used to correct the subsequent condensation precursor judgment results, so that the thermal response changes caused by surface micro-phase change can participate in the risk assessment.

[0095] See attached document Figure 7 When implementing the condensation monitoring method for radiant cooling surfaces, step S50 is used to calculate the characteristic quantity of water vapor migration near the radiant surface and extract its time-domain change rate and the cooling rate of the radiant surface.

[0096] The process of water vapor migrating from ambient air to the radiant cooling surface and undergoing microscopic condensation is closely related to the local water vapor partial pressure distribution and surface temperature changes. Specifically, the water vapor partial pressure difference or gradient characterizes the driving force for water vapor migration to the boundary layer, while the temperature change rate of the radiant surface reflects the accumulation of surface cooling. Therefore, the processing module 10 quantifies these kinetic parameters through the following steps.

[0097] S501, the processing module 10 calculates the water vapor migration driving characteristic quantity in the microenvironment space based on the temperature and humidity data obtained by the acquisition module 20.

[0098] The processing module 10 reads the ambient air dry-bulb temperature, ambient relative humidity, boundary layer air temperature, and boundary layer relative humidity at the current discrete sampling time from the synchronized multi-source synchronization data matrix.

[0099] To obtain the water vapor migration driving characteristic quantity, processing module 10 first calculates the corresponding saturated water vapor pressure based on the air temperature, and then combines it with relative humidity to obtain the actual water vapor partial pressure at the environmental measuring point and the boundary layer measuring point. The difference between the two actual water vapor partial pressures is obtained as the water vapor partial pressure difference. Given the equivalent diffusion distance L between the environmental temperature and humidity measuring point and the boundary layer temperature and humidity measuring point, processing module 10 can also divide the water vapor partial pressure difference by the equivalent diffusion distance L to obtain the water vapor partial pressure gradient characteristic quantity. When the equivalent diffusion distance L is difficult to accurately determine on-site, processing module 10 uses the water vapor partial pressure difference as the equivalent water vapor migration driving characteristic quantity in subsequent calculations. The aforementioned water vapor partial pressure difference, water vapor partial pressure gradient characteristic quantity, or equivalent water vapor migration driving characteristic quantity are uniformly referred to as the water vapor migration driving characteristic quantity in subsequent steps.

[0100] The specific calculation formula is as follows: ; ; In the formula, Indicates the dry bulb temperature The saturated water vapor pressure of air at that time; The independent variable is temperature, which is substituted into the formula. The values ​​611.2, 17.67, and 243.5 in the formula are all Magnus empirical constants. The saturated water vapor pressure calculated based on these constants is in Pascals (Pa). This represents the water vapor partial pressure difference calculated at the current sampling time, or the water vapor partial pressure gradient characteristic after normalization by the equivalent diffusion distance; This indicates the dry-bulb temperature of the ambient air obtained by the ambient temperature and humidity sensor 21; This indicates the ambient relative humidity obtained by the ambient temperature and humidity sensor 21; This indicates the boundary layer air temperature acquired by the boundary layer temperature and humidity sensor 22; This indicates the relative humidity of the boundary layer acquired by the boundary layer temperature and humidity sensor 22.

[0101] In other embodiments, the processing module 10 may replace the above-mentioned saturated water vapor pressure calculation formula with the Goff-Gratch formula or other equivalent humid air state calculation formula, depending on the controller's computing power or measurement accuracy requirements.

[0102] S502, the processing module 10 calculates the time first derivative parameter based on the water vapor migration-driven characteristic quantity and the transient temperature of the radiation surface using a sliding time window difference method.

[0103] The processing module 10 sets a sliding time window with a fixed data depth in memory to store historical data within multiple consecutive reference sampling periods. As a possible implementation, the sliding time window can contain ten to thirty historical sampling points to balance noise resistance in differential calculations and transient response retention.

[0104] Processing module 10 extracts data from the beginning and end of the sliding time window and calculates the time-domain rate of change using a difference approximation method. The specific calculation formula is as follows: ; ; In the formula, This represents the rate of change in the time domain of the water vapor migration-driven characteristic quantity obtained at the current moment; This represents the cooling rate of the radiating surface at the current moment. This represents the water vapor migration driving feature at the current sampling moment; This indicates the transient temperature of the radiating surface acquired by the infrared temperature sensor 23 at the current sampling moment; and These represent the water vapor migration-driven characteristic and transient temperature recorded at the beginning of the sliding time window, respectively; This indicates the number of valid historical sampling points included within the sliding time window; This represents the time span value of the system's reference discrete sampling period. In a preferred implementation, to overcome timing jitter caused by sensor communication packet loss, the processing module 10 replaces the denominator term in the above formula with the actual timestamp difference between the data at the beginning and end of the sliding time window. This is to ensure the absolute physical time accuracy of the rate of change calculation.

[0105] During the initial power-on startup of the system or after a program reset, a sufficient amount of historical sampling data has not yet accumulated in memory. To avoid affecting the continuity of calculation due to the waiting window filling up, the processing module 10 adopts a dynamic window scaling mechanism: when the number of valid historical data written to the buffer is less than the set window depth, the processing module 10 replaces N in the formula with the current actual number of valid historical data for differential calculation until the sliding time window reaches the set data depth.

[0106] As an anomaly protection measure, when the actual number of valid historical data N is equal to zero, the processing module 10 skips the differential operation of the current cycle and assigns the corresponding derivative parameter to a default value of zero to avoid division by zero anomalies. Through the above processing, the processing module 10 converts the static sampled values ​​of temperature, humidity and surface temperature into dynamic rate characteristics that reflect the development trend of condensation.

[0107] See attached document Figure 8 When implementing the radiative cooling surface condensation monitoring method, step S60 is used to convert the cooling rate and the rate of change of water vapor migration-driven characteristic quantities into phase plane state characteristics under a unified dimension.

[0108] In actual operation, the physical dimensions and numerical ranges corresponding to the cooling rate and the rate of change of characteristic quantities driven by water vapor migration are not the same. To facilitate subsequent unified judgment, the processing module 10 performs normalization and phase plane mapping according to the following process.

[0109] S601, the processing module 10 performs dimensionless normalization processing on the extracted first-order derivative parameters.

[0110] The processing module 10 acquires the time-domain rate of change of the water vapor migration-driven characteristic quantity and the cooling rate of the radiating surface calculated in the previous steps, and extracts the effective components of both in the condensation risk direction. Specifically, when the radiating surface is in a heating state, the processing module 10 sets the risk direction component corresponding to the cooling rate to zero; when the water vapor migration-driven characteristic quantity shows a decreasing trend, the processing module 10 sets the risk direction component corresponding to the rate of change of the water vapor migration-driven characteristic quantity to zero.

[0111] Processing module 10 performs ratio calculations between the aforementioned risk direction components and their corresponding historical extreme value scaling factors to obtain normalized dimensionless parameters. The specific calculation formula is as follows: ; ; In the formula, This represents a dimensionless parameter indicating the normalized cooling rate at the current sampling time. This represents a dimensionless parameter indicating the rate of change of the water vapor migration-driven characteristic quantity after normalization at the current sampling time. This represents the rate of temperature change at the current moment with the radiative surface cooling direction as the positive direction; This represents the rate of change in the time domain at the current moment, with the increase of the characteristic quantity driven by water vapor migration being the positive direction. This represents the scaling factor for the historical extreme values ​​of the set cooling rate; This represents the historical extreme value scaling factor for the rate of change of the water vapor migration-driven characteristic quantity.

[0112] As a specific implementation, the two historical extreme value scaling factors mentioned above can be determined by the maximum value of the corresponding risk direction component recorded by the system during the initial debugging phase. To adapt to climate change during long-term operation, the processing module 10 is configured with an extreme value sliding update mechanism. When the system is in a dry operating condition confirmation state and has not triggered an early warning state, physical intervention state, or sensor abnormal state, if the risk direction component calculated in real time is greater than the currently stored historical extreme value scaling factor for multiple consecutive sampling periods, the processing module 10 updates the corresponding historical extreme value scaling factor with the new risk direction component. In addition, to prevent the historical extreme value scaling factor from being excessively expanded due to occasional extreme operating conditions, which would cause the normalization result of the system to remain too small and lose monitoring sensitivity in subsequent long-term operation, the processing module 10 is configured with extreme value forgetting and recovery logic. The processing module 10 multiplies the historical extreme value scaling factor by a forgetting coefficient less than one (such as 0.9) according to a set long-term time period (e.g., every thirty days of operation or at the beginning of each cooling season), or resets it to the baseline debugging extreme value according to the current seasonal climate characteristics.

[0113] The dry operating condition confirmation state refers to the operating state in which the processing module 10 confirms, based on the transient temperature of the radiating surface, the ambient air dew point temperature, or the boundary layer air dew point temperature, that the radiating surface temperature is higher than the preset safety margin of the dew point temperature, and the coupling phase change precursor index is continuously lower than the safety judgment threshold.

[0114] When the system is in an early warning state, a physical intervention state, or a sensor abnormal state, the processing module 10 pauses updating the historical extreme value scaling factor to avoid the extreme values ​​under abnormal operating conditions affecting the subsequent normalization scale.

[0115] Furthermore, to prevent division by zero errors caused by the scaling factor of historical extreme values ​​not yet being recorded or abnormally returning to zero, the processing module 10 performs a numerical verification of the historical extreme value scaling factor before performing the normalization operation. When the historical extreme value scaling factor is less than a preset minimum positive threshold (e.g., 1 × 10⁻⁶), the scaling factor will be checked. -6When the scaling factor is set to the minimum positive threshold, the processing module 10 forces the scaling factor to be assigned to the minimum positive threshold value, thereby ensuring the continuity and security of the algorithm operation.

[0116] S602, the processing module 10 constructs a two-dimensional phase plane based on the normalization result and calculates the polar coordinate vector angle and magnitude.

[0117] Processing module 10 uses a dimensionless parameter of the normalized cooling rate. The horizontal axis represents the dimensionless parameter representing the rate of change of characteristic quantities driven by the normalized water vapor migration. Using the vertical axis as the coordinate axis, a two-dimensional phase plane state point is formed in memory. Since both of the above parameters are non-negative components in the risk direction, the state point extracted by the current system is located in the first quadrant of the two-dimensional phase plane.

[0118] Processing module 10 further calculates the magnitude of the state point. This is used to characterize the current dynamic change magnitude. Specifically, It can be calculated using the following formula: ; Processing module 10 uses the arctangent function to map the one-dimensional rate of change state into polar coordinate vector angles. The specific mathematical formula is as follows: ; In the formula, This represents the polar coordinate vector angle mapped at the current sampling time, with its numerical unit being radians; This represents the arctangent function.

[0119] In the arithmetic logic implementation of a control system, to avoid the denominator in the input term of the arctangent function... The computational dead zone error caused by a value of zero is handled by singularity bypass logic in module 10. equal to zero and When the value is greater than zero, processing module 10 skips the division and arctangent operations and directly converts the polar coordinate vector angle to zero. The value is assigned as half of pi (i.e., (Radians). When both are zero, it is directly assigned zero radians.

[0120] Through phase plane mapping, processing module 10 transforms the dynamic process of condensation into a geometric rotation of a vector angle in the first quadrant. Those skilled in the art will understand that this polar coordinate vector angle... The numerical value represents the relative proportion between the driving force of moisture migration and the surface cooling and heat storage force. Vector angle approximation. The accumulation rate of water vapor condensation potential energy in the microenvironment is significantly increasing relative to the surface cooling rate. This geometric feature is used by the processing module 10 for subsequent phase change hazard assessment.

[0121] See attached document Figure 9 When implementing the condensation monitoring method for radiant cooling surfaces, step S70 is used to generate a condensation precursor evaluation quantity based on the phase plane mapping result and the dynamic thermal impedance weighting coefficient, and to continuously confirm and judge it.

[0122] The processing module 10 integrates the polar coordinate vector angle obtained in the previous steps with the dynamic thermal impedance weighting coefficient to quantify the condensation risk of the radiative cooling surface.

[0123] S701, Processing module 10 generates the coupling phase transition precursor index.

[0124] Processing module 10 reads the polar coordinate vector angle mapped at the current discrete sampling time and simultaneously reads the corresponding dynamic thermal impedance weighting coefficient. To avoid low-amplitude random fluctuations entering the risk assessment process, processing module 10 first determines the magnitude of the state points in the two-dimensional phase plane before generating the coupling phase transition precursor index. Whether the minimum effective radius threshold has been reached.

[0125] when When the value is less than the minimum effective radius threshold, the processing module 10 does not perform the over-limit judgment for this cycle, or sets the coupling phase transition precursor index for the current cycle to the default safe value.

[0126] when When the minimum effective radius threshold is reached or exceeded, processing module 10 multiplies the polar coordinate vector angle by the dynamic thermal impedance weighting coefficient to obtain the coupling phase transition precursor index. Its calculation formula is as follows: ; In the formula, This represents the coupled phase transition precursor index calculated at the current discrete sampling moment; This represents the dynamic thermal impedance weighting coefficient at the current moment; This represents the polar vector angle at the current moment.

[0127] The minimum effective radius threshold can be determined based on the modulus of the two-dimensional phase plane state point under dry operating conditions. The historical mean, noise upper limit, or standard deviation are determined. For example, processing module 10 will determine the module length under dry operating conditions. The minimum effective radius threshold is set to two to three times the standard deviation to avoid low-amplitude random fluctuations being misjudged as precursors to condensation.

[0128] In the above multiplicative relationship, the polar coordinate vector angle mainly characterizes the proportional change of the driving force of water vapor migration in the microenvironment relative to the surface cooling and cold storage effect, while the dynamic thermal impedance weighting coefficient mainly characterizes the degree of deviation of the thermal response characteristics when the radiating surface approaches the phase change state. Under normal operating conditions, the dynamic thermal impedance weighting coefficient is close to one, and the coupled phase change precursor index mainly changes with meteorological dynamic parameters; when the radiating surface approaches the dew point and shows a tendency for micro-condensation, the dynamic thermal impedance weighting coefficient increases with the deviation of the thermal impedance characteristics, thereby amplifying the precursor information corresponding to the polar coordinate vector angle.

[0129] S702, the processing module 10 performs continuous confirmation filtering and over-limit judgment.

[0130] The processing module 10 internally presets a safety judgment threshold and compares the coupled phase transition precursor index obtained in the current cycle with this safety judgment threshold. As one implementation, the safety judgment threshold can be determined based on historical statistical results of the coupled phase transition precursor index under dry operating conditions, for example, by taking 2 to 3 times the historical statistical average under dry operating conditions. Combining the range of polar coordinate vector angles and the gain of the system correction coefficient, the safety judgment threshold can also be set within the range of 2.5 to 4.0 according to the specific warning sensitivity requirements of the radiation terminal.

[0131] During operation, electrical interference, communication packet loss, or local airflow disturbances may cause isolated spikes in the coupled phase transition precursor index. To reduce false alarms caused by such spikes, the processing module 10 sets up a continuous acknowledgment filtering window. This continuous acknowledgment filtering window contains multiple consecutive valid sampling periods, such as 3 to 5 consecutive valid sampling periods.

[0132] When processing module 10 determines that the coupled phase transition precursor index of the current period is greater than or equal to the safety judgment threshold, it does not immediately output a warning command, but instead sets the over-limit status corresponding to the current sampling period. Processing module 10 continues to count the over-limit status within the continuous confirmation filter window; only when the coupled phase transition precursor index is continuously greater than or equal to the safety judgment threshold within the continuous confirmation filter window, processing module 10 confirms that there is a risk of condensation and generates a warning command.

[0133] If, within the continuous confirmation filtering window, the coupled phase transition precursor index falls below the safety judgment threshold, processing module 10 clears the over-limit count or the corresponding status register and returns to the normal monitoring process. Through this continuous confirmation mechanism, the probability of false alarms caused by occasional noise can be reduced while maintaining the early warning response speed.

[0134] After the warning command is output, the processing module 10 continues to collect data and update the status judgment. To avoid frequent switching between the warning state and the safe state, the processing module 10 is equipped with a delayed release logic. When the coupled phase transition precursor index falls below the preset release threshold, for example, below 80% to 90% of the safe judgment threshold, and the duration meets the continuous confirmation filter window requirements, the processing module 10 cancels the warning command, clears the relevant status registers, and resumes normal monitoring. This release logic is used to prevent the warning state from being repeatedly triggered or released due to short-term fluctuations.

[0135] See attached document Figure 10 When implementing the condensation monitoring method for radiant cooling surfaces, step S80 is used to control the execution module 30 based on the condensation risk assessment result, and to restore normal operation after the risk is eliminated.

[0136] Based on the coupling phase change precursor index and its status confirmation results obtained in the previous steps, the processing module 10 controls the execution module 30 to adjust the cooling input of the radiant terminal to reduce the risk of condensation on the radiant cooling surface.

[0137] S801, the processing module 10 performs routine dry operating condition maintenance control.

[0138] When the coupling phase change precursor index is lower than the preset safety threshold and the system does not trigger an early warning command, the processing module 10 determines that the radiant cooling surface is in a safe operating condition without phase change precursors. Under this condition, the processing module 10 controls the execution module 30 to maintain the basic room temperature regulation logic.

[0139] In practice, the processing module 10 calculates a basic control quantity based on the deviation between the actual indoor temperature and the set target temperature, and adjusts the opening of the water supply regulating valve accordingly. This basic control quantity can be obtained using a proportional-integral-derivative algorithm or other conventional room temperature control algorithms.

[0140] While maintaining basic room temperature control, the processing module 10 continues to send active thermal disturbance commands to the water supply end according to a preset cycle. The processing module 10 synchronously records the baseline control command and the active thermal disturbance command, and during the dynamic thermal impedance feature extraction process, it uses the characteristic frequency or pulse triggering time of the active thermal disturbance command as a synchronization reference to reduce the impact of changes in the basic room temperature control command on the feature extraction results.

[0141] Therefore, while maintaining indoor thermal comfort, the system continuously acquires the dynamic thermal impedance characteristics of the radiant terminals. The parameter tuning for the proportional-integral-differential algorithm can be conventionally configured based on the specific building's room heat load, terminal type, and control response requirements.

[0142] S802, the processing module 10 performs intervention and recovery control for condensation precursors.

[0143] When the processing module 10 receives the warning command generated in step S70, it determines that there is a risk of microscopic condensation on the radiant cooling surface and puts the system into a physical intervention state. The processing module 10 sends a restriction command with a higher priority than the basic room temperature control to the execution module 30.

[0144] As a specific intervention method, the processing module 10 controls the execution module 30 to reduce the upper limit of the cold water side opening, increase the mixing ratio of the return water side, or increase the supply water temperature setting value of the mixing center, so as to reduce the cold energy input to the radiant terminal and prevent the local surface temperature from continuing to approach the dew point.

[0145] During physical intervention, processing module 10 suspends the issuance of active thermal disturbance commands to avoid additional disturbances that could exacerbate the current condensation risk. Since processing module 10 can no longer update the amplitude attenuation ratio or the attenuation time constant of the response recovery segment in real time through time-domain or frequency-domain response analysis after the active thermal disturbance is suspended, processing module 10 switches to degraded calculation mode.

[0146] In the degraded calculation mode, the processing module 10 locks the dynamic thermal impedance weighting coefficient to the last valid calculated value before the warning is triggered. At this time, the coupled phase change precursor index is mainly obtained based on the polar coordinate vector angle calculated from real-time temperature and humidity data, combined with the locked dynamic thermal impedance weighting coefficient, thereby maintaining the continuity of the risk assessment process.

[0147] When the duration of the downgraded calculation mode exceeds the preset maximum intervention time, and the current coupled phase change precursor index is already below the safety judgment threshold, the processing module 10 can issue a safety detection disturbance with an amplitude lower than that of the normal active thermal disturbance, reacquire the dynamic thermal impedance characteristics, and update the dynamic thermal impedance weight coefficient.

[0148] During physical intervention, processing module 10 continuously monitors the coupled phase transition precursor index in the degradation calculation mode. As the cold input decreases, the cooling rate of the radiative surface decreases, or the characteristic quantity driven by water vapor migration in the microenvironment decreases, and the polar coordinate vector angle decreases accordingly.

[0149] When processing module 10 determines that the coupling phase change precursor index has fallen below the preset release threshold and the duration meets the continuous confirmation filter window requirement, processing module 10 confirms that the condensation risk has been suppressed. Subsequently, processing module 10 cancels the restriction instructions on execution module 30, restores the basic room temperature control logic, and restarts the cyclic issuance of active thermal disturbance instructions.

[0150] As a protective measure during the recovery phase, the processing module 10 does not immediately exit the degraded calculation mode after restarting the active thermal disturbance command. Since time-domain or frequency-domain feature extraction requires accumulating a data window of a preset length, the processing module 10 continues to use the locked historical dynamic thermal impedance weighting coefficients until the data window is filled.

[0151] When the system collects sufficient transient response data and successfully calculates the new dynamic thermal impedance weighting coefficient, the processing module 10 releases the weighting coefficient lock and resumes normal monitoring calculation logic. If the new dynamic thermal impedance weighting coefficient is not successfully calculated within the preset recovery time, the processing module 10 continues to maintain the degraded calculation mode and maintains the priority of anti-condensation safety control. The preset recovery time mentioned here is determined based on the length of the feature extraction data window, for example, set to the coverage time of three to five complete data windows, thereby avoiding indefinite suspension and waiting caused by sensor failure.

[0152] Through the above closed-loop intervention and recovery control, the system can reduce the cooling input of the radiant terminal when the risk of condensation occurs, and smoothly return to normal operation after the risk is eliminated.

[0153] In a specific engineering embodiment, the present invention is applied to a concrete-embedded radiant floor cooling system in an office building. This type of system typically exhibits characteristics of high heat capacity and high thermal inertia.

[0154] System initialization and calibration (S10-S20): After the system is powered on, the processing module issues a step test command (the water valve opening jumps abruptly from 30% to 70%). An infrared temperature sensor records the temperature response of the radiating surface, and the thermal time constant of the system is obtained through fitting. =35 minutes. Because this time constant exceeds the preset high thermal inertia calibration threshold (25 minutes), the system automatically enters high thermal inertia mode. The processing module generates an active thermal disturbance command as a sparse step pulse signal, setting the single pulse width. =2 minutes, step amplitude = +40% opening increment, pulse generation cycle = 30 minutes.

[0155] Steady-state reference calibration (dry operating conditions): Under the initial absolutely dry operating conditions (water supply temperature 18℃, indoor dew point 12℃), the system was injected with the above pulses multiple times, and the statistical mean of the decay time constant of the reference response recovery section was measured. =15 minutes, and set the system correction coefficient. =3.0.

[0156] The scaling factor for historical extreme values ​​is initially calibrated as: the maximum cooling rate. =0.005℃ / s, the maximum rate of change of water vapor partial pressure gradient in the time domain =0.1Pa / s.

[0157] Dynamic calculation of sudden wet load during operation (S30-S70) One afternoon, due to the dense crowd and the short-term opening of the windows, the indoor wet load surged.

[0158] Temperature, humidity, and rate calculations: At the current sampling time, the ambient temperature and humidity are 26℃ and 60%, respectively. The calculated saturated water vapor pressure is approximately 3360 Pa. The actual ambient water vapor partial pressure... ≈2016 Pa; Boundary layer temperature and humidity are 22℃ and 75%, actual boundary layer water vapor partial pressure ≈1981 Pa. Water vapor partial pressure difference. =35 Pa. The system calculates the rate of change of the sharp increase in the current water vapor partial pressure difference using a sliding time window differential. =0.046 Pa / s, the radiating surface is under continuous cooling, and the cooling rate is... =0.0016℃ / s Normalization and phase plane mapping: Calculate the dimensionless parameter of the cooling rate =0.0016 / 0.005=0.32; Calculate the dimensionless parameter for the change in the driving force of water migration. =0.046 / 0.1=0.46; The system maps it to a two-dimensional phase plane and calculates the polar coordinate vector angle. (0.46 / 0.32)≈0.963 radians. At this point, the vector angle is biased towards the Y-axis, indicating that the accumulation of wet load dominates the current dynamics.

[0159] Dynamic thermal impedance characteristic update: During the most recent thermal pulse perturbation, due to the invisible microscopic water vapor adsorption on the surface (latent heat of phase change offsets part of the cooling), the decay time constant of the current response recovery segment identified by the system has been extended to [a later value]. =18 minutes. Generate dynamic thermal resistance weighting coefficients: ≈1.822.

[0160] Generate early warning command: Calculate the coupling phase transition precursor index =1.822 × 0.963 ≈ 1.75. The system's preset safety judgment threshold is 1.2. Since 1.75 ≥ 1.2, and continues to exceed the limit within the subsequent continuous confirmation filtering window (e.g., 3 consecutive samples), the processing module accurately determines the precursor to phase change and triggers the anti-condensation intervention command.

[0161] Closed-loop intervention and recovery (S80): The system forcibly limits the opening of the water supply mixing valve to 50% of its current operating opening, reducing the cooling capacity input. During this period, active disturbances are suspended. Locked at 1.822, relying on polar coordinate vector angles. The real-time temperature drop was monitored. Approximately 15 minutes later, the surface cooling trend was halted. Reduced to 0.4 radians, When the value drops back to 0.72 (below the release threshold of 0.96), the system automatically releases the restriction and resumes normal PID room temperature control, effectively preventing the formation of macroscopic condensation droplets.

[0162] Experimental verification and effect comparison: To verify the technical effectiveness of this invention, a full-size radiant floor test platform was built in an artificial environment test chamber. The proposed method was compared with the traditional "dew point temperature approximation method (static margin control, such as setting a constant surface temperature margin of 1°C from the dew point)" in a comparative experiment. The experiment included a step-load wet disturbance condition (simulating sudden window opening), and a high-definition macro camera was used to monitor the droplet formation initiation point on the radiant surface.

[0163] Appendix Figure 11 The figure visually illustrates the first quadrant state space, composed of the dimensionless parameter of cooling rate (X-axis) and the dimensionless parameter of water vapor migration rate (Y-axis). The scatter plot trajectory in the figure reflects the physical process of evolution from a safe dry condition to the precursor of condensation through polynomial smoothing. It can be seen that after the introduction of a sudden wet load, the system state point rapidly transitions to the high elevation angle region (a sharp increase in vector angle), demonstrating the extremely high sensitivity of the phase plane mapping method to sudden changes in wet load.

[0164] Appendix Figure 12 Employing a dual Y-axis structure, the upper half compares the changes in the "coupled phase change precursor index" generated by this invention with the "dew point margin" of the traditional method on the time axis; the lower half plots the water valve control opening curve. Data shows that, 8 minutes after the occurrence of wet disturbance, the coupled phase change precursor index of this invention exceeds the threshold and triggers the action, beginning to reduce the water valve opening; while at this time, the surface temperature of the traditional method has not yet reached the static alarm boundary. The high-definition camera only captures the growth of microscopic droplets at 26 minutes. That is, this invention achieves a lead time warning of up to 18 minutes.

[0165] Appendix Figure 13 Presented in the form of grouped bar charts. In 30 test conditions covering gradual high humidity, sudden high humidity, and step load, the average warning lead time of the algorithm of this invention reached 15.4 minutes. Moreover, thanks to the addition of "dynamic thermal impedance characteristics" and "continuous confirmation filtering mechanism", its false alarm rate was 0% (it did not misjudge simple airflow temperature fluctuations as condensation). In contrast, the average warning lead time of traditional methods was only 2.1 minutes, and it generated a false alarm rate as high as 16.7% when the ambient temperature scale fluctuated at high frequencies.

[0166] Verification conclusion: Experimental data fully demonstrate that this invention does not simply rely on the comparison of static temperature thresholds, but rather, through the coupling of dynamic thermal impedance weighting coefficients and phase plane geometric vector angles in the time-frequency domain, it accurately captures the heat transfer hysteresis phenomenon caused by the latent heat of microscopic phase change and the dynamic trend of water vapor before macroscopic condensation droplets accumulate. It balances high early warning sensitivity with strong noise resistance, possessing significant technological advancements and engineering application value.

[0167] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for monitoring condensation on surfaces in radiative cooling systems based on multi-source sensor fusion, characterized in that, Includes the following steps: The operating status parameters of the radiant cooling system are obtained, including transient temperature of the radiant surface, ambient temperature and humidity, and boundary layer temperature and humidity. Based on the ambient temperature and humidity and the boundary layer temperature and humidity, calculate the ambient water vapor partial pressure and the boundary layer water vapor partial pressure respectively. The surface cooling rate is calculated based on the transient temperature of the radiating surface, and the rate of change of water vapor migration-driven characteristic quantities is calculated based on the difference between the ambient water vapor partial pressure and the boundary layer water vapor partial pressure. The surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity are normalized, and the normalized result is mapped to a two-dimensional phase plane to calculate the polar coordinate vector angle. An active thermal disturbance is applied to the radiant cooling system, and a dynamic thermal impedance weighting coefficient is calculated based on the transient response of the radiant surface to the active thermal disturbance, including: Obtain the thermal time constant of the radiant cooling system; When the thermal time constant is less than a preset calibration threshold, the system is confirmed to enter a low thermal inertia mode, and a continuous sinusoidal micro-perturbation signal is generated as the active thermal perturbation; the input amplitude corresponding to the sinusoidal micro-perturbation signal is extracted; when the input amplitude is lower than a preset minimum effective amplitude, the previous effective dynamic thermal impedance weighting coefficient is used, or the system is switched to a degraded calculation mode; wherein, the preset minimum effective amplitude is determined based on the equivalent noise temperature difference of the infrared temperature sensor or the sampling resolution of the system analog-to-digital converter; when the input amplitude meets the preset minimum effective amplitude, the amplitude attenuation ratio is extracted, and the dynamic thermal impedance weighting coefficient is constructed based on the amplitude attenuation ratio; When the thermal time constant is greater than or equal to a preset calibration threshold, the system is confirmed to enter a high thermal inertia mode, and a sparse step pulse signal is generated as the active thermal disturbance; the transient temperature recovery segment data caused by the sparse step pulse signal is extracted, and the response recovery segment decay time constant at the current sampling time is fitted to obtain the reference response recovery segment decay time constant obtained by the calibration of the radiant cooling system under initial dry conditions; the dynamic thermal impedance weighting coefficient is calculated with the natural constant as the base and the target product as the exponent; the target product is obtained by multiplying the system correction coefficient by the difference between the response recovery segment decay time constant and the reference response recovery segment decay time constant, and then dividing by the reference response recovery segment decay time constant; wherein, the pulse generation period of the sparse step pulse signal is greater than the single pulse width; Based on the product of the dynamic thermal impedance weighting coefficient and the polar coordinate vector angle, the coupling phase change precursor index is calculated, and when the coupling phase change precursor index is greater than the preset safety judgment threshold, the anti-condensation control signal is output.

2. The method for monitoring condensation on a radiative cooling surface based on multi-source sensor fusion according to claim 1, characterized in that, In the steps of calculating the surface cooling rate based on the transient temperature of the radiating surface and calculating the rate of change of water vapor migration-driven characteristic quantities based on the difference between the ambient water vapor partial pressure and the boundary layer water vapor partial pressure: Extract the historical sampling sequence of the transient temperature of the radiating surface and the historical sampling sequence of the water vapor migration-driven characteristic quantity within a sliding time window of a set length; Obtain the actual timestamp difference corresponding to the first and last valid data within the sliding time window; Using the actual timestamp difference as the time denominator, and combining it with the difference between the first and last data of the historical sampling sequence, the surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity with absolute physical time accuracy are calculated respectively.

3. The method for monitoring condensation on a radiative cooling surface based on multi-source sensor fusion according to claim 1, characterized in that, In the step of normalizing the surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity: Extract the maximum surface cooling rate and the maximum rate of change of water vapor migration-driven characteristic quantity recorded in the historical operation data of the system, and use them as the first historical extreme value scaling factor and the second historical extreme value scaling factor, respectively. Normalization is achieved by dividing the currently calculated surface cooling rate and the rate of change of water vapor migration-driven characteristic quantity by using the first historical extreme value scaling factor and the second historical extreme value scaling factor as normalization denominators respectively. Specifically, according to the set long-term time period, the first historical extreme value scaling factor and the second historical extreme value scaling factor are multiplied by a forgetting coefficient less than one, or the extreme values ​​are reset based on the current seasonal climate characteristics to perform extreme value forgetting recovery logic.

4. The method for monitoring condensation on a radiative cooling surface based on multi-source sensor fusion according to claim 1, characterized in that, The steps following the output of the anti-condensation control signal also include: The maximum allowable opening of the cold source water valve is limited according to the anti-condensation control signal, and the application of the active thermal disturbance is suspended. At the same time, the dynamic thermal impedance weighting coefficient is locked to the current calculated value when the early warning is triggered. During the duration of the anti-condensation control signal, the coupling phase change precursor index is updated based on the real-time calculated polar coordinate vector angle. If the coupling phase change precursor index falls below the safety release threshold, the restriction on the maximum allowable opening of the cold source water valve is released. If sufficient transient response data is not collected within the preset recovery time to update the dynamic thermal impedance weighting coefficient, the degraded calculation mode is maintained and the priority of the anti-condensation control signal is preserved; wherein, the duration limit of the preset recovery time is determined based on the length of the sliding time window for extracting feature data.

5. The method for monitoring condensation on a radiative cooling surface based on multi-source sensor fusion according to claim 1, characterized in that, In the step of obtaining the operating status parameters of the radiant cooling system: The transient temperature of the radiating surface is obtained non-contactly using an infrared temperature sensor. The ambient temperature and humidity data are collected by an ambient temperature and humidity sensor installed in an indoor reference area, and then the ambient water vapor partial pressure is calculated. The boundary layer temperature and humidity data are collected by a boundary layer temperature and humidity sensor placed close to the radiating surface, and then the partial pressure of water vapor in the boundary layer is calculated.

6. A radiant cooling surface condensation monitoring system based on multi-source sensor fusion, used to execute the radiant cooling surface condensation monitoring method based on multi-source sensor fusion as described in any one of claims 1-5, characterized in that, include: The data acquisition module is used to acquire the operating status parameters of the radiant cooling system, including the transient temperature of the radiant surface, ambient temperature and humidity, and boundary layer temperature and humidity. The rate calculation module is used to calculate the ambient water vapor partial pressure and the boundary layer water vapor partial pressure based on the ambient temperature and humidity and the boundary layer temperature and humidity, respectively. The surface cooling rate is calculated based on the transient temperature of the radiating surface, and the rate of change of water vapor migration-driven characteristic quantities is calculated based on the difference between the ambient water vapor partial pressure and the boundary layer water vapor partial pressure. The phase plane mapping module is used to normalize the surface cooling rate and the rate of change of the water vapor migration-driven characteristic quantity, and to map the normalized result to a two-dimensional phase plane to calculate the polar coordinate vector angle. The impedance assessment module is used to apply an active thermal disturbance to the radiant cooling system and calculate the dynamic thermal impedance weighting coefficient based on the transient response of the radiant surface to the active thermal disturbance. The early warning control module is used to calculate the coupling phase change precursor index based on the product of the dynamic thermal impedance weighting coefficient and the polar coordinate vector angle, and output an anti-condensation control signal when the coupling phase change precursor index is greater than a preset safety judgment threshold.

7. The radiative cooling surface condensation monitoring system based on multi-source sensor fusion according to claim 6, characterized in that, The system further includes a sensing detection end communicatively connected to the data acquisition module, the sensing detection end comprising: An infrared temperature sensor is used for non-contact acquisition of the transient temperature of the radiating surface; An ambient temperature and humidity sensor is installed in an indoor reference area to collect ambient temperature and humidity data in order to calculate the ambient water vapor partial pressure. A boundary layer temperature and humidity sensor is installed close to the radiant surface of the radiant cooling system to collect boundary layer temperature and humidity data in order to calculate the partial pressure of water vapor in the boundary layer.

Images