Real-time monitoring method for humidity and thermal conductivity of roof thermal insulation layer

Abstract

The invention relates to the technical field of roof heat preservation monitoring, and discloses a real-time monitoring method for humidity and heat conductivity of a roof heat preservation layer. The method comprises the following steps: collecting historical humidity data and historical temperature data of a roof thermal insulation layer in real time, identifying a dynamic change mode of humidity and temperature from the historical humidity data and the historical temperature data, and constructing a monitoring track based on the dynamic change mode; extracting a plurality of humidity hysteresis effect features and converting the humidity hysteresis effect features into hysteresis influence quantities; real-time temperature and humidity data streams are collected through an environment sensor, the dynamic interference degree of environment factors on humidity is determined, and a humidity stability index during temperature fluctuation is derived in combination with temperature and humidity correlation changes; and predicting the thermal conductivity in real time by using the hysteresis influence quantity and the humidity stability index, calculating a prediction deviation, and when the deviation exceeds a preset threshold value, increasing the data sampling frequency of the sensor. According to the method, historical data characteristics, dynamic interference factors and a hysteresis effect are comprehensively considered, accurate prediction of the thermal conductivity and adaptive adjustment of a monitoring strategy are realized, and the real-time performance and accuracy of monitoring are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of roof insulation monitoring, in particular to a roof insulation layer humidity and thermal conductivity real-time monitoring method. BACKGROUND

[0002] As an important part of building envelope structure, the thermal performance of roof insulation layer directly affects the energy consumption and indoor thermal environment stability of buildings. In actual use, the humidity state of the insulation layer is the key factor leading to changes in its thermal conductivity. The increase in humidity will significantly reduce the thermal insulation performance of the insulation material, and even cause material aging, structural damage and other problems. Therefore, effective monitoring of the humidity and thermal conductivity of the roof insulation layer is an important means to ensure building energy saving effect and structural safety.

[0003] Traditional fixed-point sampling or periodic detection methods are often used for monitoring the roof insulation layer. However, these methods have obvious limitations. On the one hand, the sampling period is long, and it is difficult to capture the dynamic change process of humidity and temperature, especially in the case of frequent climate change or extreme weather conditions, which may miss critical state change information. On the other hand, existing monitoring methods often consider the influence of humidity or temperature on thermal conductivity separately, ignoring the interaction between the two and the hysteresis effect of humidity changes, resulting in a large deviation between the predicted thermal conductivity and the actual situation. In addition, environmental factors such as precipitation, wind speed, and sunlight have random and complex interference on the humidity of the insulation layer, and traditional monitoring methods are difficult to quantify these dynamic disturbances, further reducing the reliability of the monitoring data.

[0004] With the increasing demand for building energy saving, higher requirements are put forward for the real-time, accuracy and comprehensiveness of roof insulation layer monitoring. The existing technology fails to effectively integrate historical data features, dynamic interference factors and hysteresis effects, making it difficult to accurately predict thermal conductivity and adaptively adjust the sampling strategy based on the monitoring state, resulting in waste of monitoring resources or omission of key information. Therefore, developing a monitoring method that can consider the influence of multiple factors and real-time capture the change law of humidity and thermal conductivity has become a problem to be solved in the field of building energy saving monitoring. SUMMARY

[0005] The purpose of the present application is to provide a roof insulation layer humidity and thermal conductivity real-time monitoring method to solve the problems raised in the background art.

[0006] To achieve the above purpose, the present application provides a roof insulation layer humidity and thermal conductivity real-time monitoring method, which comprises:

[0007] real-time collection of historical humidity data and historical temperature data of the roof insulation layer;

[0008] identify a dynamic change pattern of the humidity and the temperature in the monitoring process from the historical humidity data and the historical temperature data;

[0009] construct a monitoring track of the humidity and the thermal conductivity of the roof insulation layer based on the dynamic change pattern, extract a plurality of humidity hysteresis effect features of the roof insulation layer according to a change rate of the monitoring track, and convert all the humidity hysteresis effect features into a hysteresis influence quantity of the roof insulation layer in the monitoring process;

[0010] start an environmental sensor of the roof insulation layer, collect a real-time humidity and temperature data stream, determine a dynamic interference degree of the environmental factor on the humidity based on an interaction characteristic of the humidity and the temperature in the real-time humidity and temperature data stream;

[0011] deduce a humidity stability index of the roof insulation layer in temperature fluctuation from a correlation change between the dynamic interference degree and the humidity and the temperature;

[0012] perform real-time prediction calculation on the thermal conductivity of the roof insulation layer through the hysteresis influence quantity and the humidity stability index, and further obtain a thermal conductivity prediction deviation of the roof insulation layer;

[0013] when the thermal conductivity prediction deviation exceeds a preset monitoring threshold, increase a data sampling frequency of the environmental sensor of the roof insulation layer.

[0014] Preferably, the identifying the dynamic change pattern of the humidity and the temperature in the monitoring process from the historical humidity data and the historical temperature data specifically comprises: calculating a change rate set of adjacent data points in the historical humidity data and the historical temperature data;

[0015] deduce a dynamic range boundary of the historical data from the change rate set;

[0016] extract a dynamic change pattern of the humidity and the temperature in the monitoring process based on the dynamic range boundary, and take the dynamic change pattern as a basic input for subsequent monitoring track construction.

[0017] Preferably, the constructing the monitoring track of the humidity and the thermal conductivity of the roof insulation layer based on the dynamic change pattern specifically comprises: defining an acceptable monitoring range of the humidity and the thermal conductivity of the roof insulation layer;

[0018] generating a monitoring track of the humidity and the thermal conductivity of the roof insulation layer from the dynamic change pattern and the acceptable monitoring range, and applying the monitoring track to a screening process of the hysteresis effect features.

[0019] Preferably, the determining the dynamic interference degree of the environmental factor on the humidity based on the interaction characteristic of the humidity and the temperature in the real-time humidity and temperature data stream specifically comprises: analyzing the interaction characteristic of the humidity and the temperature in the real-time humidity and temperature data stream;

[0020] calculating a correlation strength between the humidity and the temperature from the interaction characteristics;

[0021] determining a dynamic disturbance degree of the environmental factor to the humidity from the correlation strength, and using the dynamic disturbance degree for derivation of the humidity stability index.

[0022] Preferably, the derivation of the humidity stability index of the roof insulation layer under temperature fluctuation from the dynamic disturbance degree and the correlation change between the humidity and the temperature specifically comprises: identifying a correlation change pattern between the humidity and the temperature;

[0023] establishing a humidity feedback mechanism based on the dynamic disturbance degree and the correlation change pattern; deriving the humidity stability index of the roof insulation layer under temperature fluctuation from the humidity feedback mechanism, and delivering the humidity stability index to the thermal conductivity prediction calculation step.

[0024] Preferably, the method further comprises: when the thermal conductivity prediction deviation is equal to a preset monitoring threshold, maintaining the current data sampling frequency of the roof insulation layer environmental sensor; and when the thermal conductivity prediction deviation is less than the preset monitoring threshold, reducing the data sampling frequency of the roof insulation layer environmental sensor, and feeding back the adjusted sampling frequency data to the real-time collection step.

[0025] Preferably, the historical humidity data and the historical temperature data are defined as a complete set of all humidity values and temperature values within a specified time period, and the complete set is used in the identification process of the dynamic change pattern.

[0026] Preferably, after starting the environmental sensor of the roof insulation layer, the collected real-time humidity and temperature data stream includes environmental temperature fluctuation data and surface humidity distribution data.

[0027] The environmental temperature fluctuation data and the surface humidity distribution data are integrated into a unified data stream through a data fusion technology, used in the analysis process of the interaction characteristics, and the analysis result is applied to the determination of the dynamic disturbance degree.

[0028] Preferably, the real-time prediction calculation of the thermal conductivity of the roof insulation layer from the hysteresis influence quantity and the humidity stability index specifically comprises: constructing a thermal conductivity prediction model, and taking the hysteresis influence quantity and the humidity stability index as model inputs.

[0029] An input data is processed by using a machine learning algorithm to generate a thermal conductivity prediction value.

[0030] A thermal conductivity prediction deviation between the prediction value and an actual measurement value is calculated, and the thermal conductivity prediction deviation is output to the monitoring threshold comparison step.

[0031] Preferably, a cycle iteration framework is adopted, in which the output data of the real-time acquisition step is used for dynamic change pattern recognition, the recognition result is used for trajectory construction monitoring, the construction result is used for lag effect feature extraction, the extraction result is used for lag influence quantity conversion, the conversion result is used for dynamic interference degree determination, the determination result is used for humidity stability index derivation, the derivation result is used for real-time prediction calculation of thermal conductivity, the prediction result is used for prediction deviation acquisition, the acquisition result is used for data sampling frequency adjustment, and the adjustment result is fed back to the real-time acquisition step to form a closed-loop data interaction.

[0032] Compared with the prior art, the present application has the following beneficial effects:

[0033] The roof insulation layer humidity and thermal conductivity real-time monitoring method integrates historical data and real-time monitoring information to build a more comprehensive monitoring system. Based on historical humidity and temperature data, it identifies dynamic change patterns, deeply mines the internal correlation rules between them, and provides a solid foundation for subsequent thermal conductivity prediction. By extracting humidity lag effect features and converting them into lag influence quantities, the time accumulation effect of humidity changes on thermal conductivity is fully considered, prediction deviations caused by neglecting hysteresis are avoided, and the monitoring results are more in line with the actual situation.

[0034] In the real-time monitoring link, the method collects real-time data streams through environmental sensors, determines the dynamic interference degree of environmental factors based on the interaction characteristics of humidity and temperature, and can accurately quantify the influence of complex environmental factors such as precipitation and sunshine on the humidity of the insulation layer, so as to include these dynamic variables in the thermal conductivity prediction, improving the adaptability and accuracy of the prediction. At the same time, combined with the dynamic interference degree and the correlation change of temperature and humidity, the humidity stability index is derived, which can effectively reflect the humidity retention ability of the insulation layer under temperature fluctuations, and provides an important basis for evaluating the performance stability of the insulation layer.

[0035] The lag influence quantity and the humidity stability index are used for real-time prediction calculation of thermal conductivity, and the sampling frequency is adjusted according to the prediction deviation, realizing the adaptive optimization of the monitoring process. When the prediction deviation exceeds the preset threshold, the data sampling frequency is increased, which can capture more detailed information in the key state change stage, ensuring that important state transitions are not missed; while in the stable state, the normal sampling frequency can be maintained, avoiding unnecessary resource consumption, and realizing the balance between monitoring efficiency and data quality. In addition, the method uses real-time data processing and feedback mechanism throughout the process, which can timely discover the abnormal change of thermal conductivity of the insulation layer, provide timely information support for the maintenance and adjustment of the roof insulation system, help to prolong the service life of the insulation layer, and reduce the energy waste caused by performance degradation.

[0036] The method breaks through the limitation of single factor consideration and static analysis in traditional monitoring, and comprehensively improves the real-time performance, accuracy and intelligent level of the monitoring through multi-dimensional data integration and dynamic adjustment strategy, so that the high standard requirement of modern building on the insulation layer monitoring can be better met, and the method has obvious practicability and superiority in practical application. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 A working principle diagram of the roof insulation layer humidity and thermal conductivity real-time monitoring method is provided.

[0038] Figure 2 A working principle diagram of dynamic change pattern recognition is provided.

[0039] Figure 3 A working principle diagram of dynamic interference degree determination is provided.

[0040] Figure 4 A working principle diagram of real-time data stream integration is provided. DETAILED DESCRIPTION

[0041] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0042] Please refer to Figure 1 The present application provides a roof insulation layer humidity and thermal conductivity real-time monitoring method, which comprises:

[0043] The historical humidity and temperature data are collected by the environmental sensor, the dynamic change pattern is identified therefrom, and the monitoring track is constructed. The humidity hysteresis effect feature is extracted based on the change rate of the monitoring track, and is converted into a hysteresis influence quantity. The real-time data stream determines the dynamic interference degree of the environmental factors by analyzing the interaction characteristics of humidity and temperature, and deduces the humidity stability index in combination with the associated changes of humidity and temperature. The hysteresis influence quantity and the humidity stability index are used for thermal conductivity prediction calculation, and the sampling frequency is adjusted when the prediction deviation exceeds the threshold value. The method forms a closed-loop data interaction, and realizes dynamic monitoring optimization.

[0044] Embodiment 1: Please refer to Figure 2The identification process of dynamic change patterns starts with the preprocessing of historical humidity and temperature data. Raw data may contain noise or missing values, so first data cleaning is performed to remove outliers and fill in gaps using interpolation methods. The cleaned data is arranged in chronological order to form continuous time series. The change rate of adjacent data points is obtained by difference calculation, that is, the difference between the current humidity or temperature value and the previous time interval divided by the time interval, to obtain the instantaneous change rate. This process covers all historical data points to generate a change rate set, reflecting the fluctuation characteristics of humidity and temperature in the monitoring period.

[0045] The determination of dynamic range boundaries is based on the statistical properties of the change rate set. The extreme values of the change rate, i.e. the maximum rising rate and the maximum falling rate, are calculated as the upper and lower boundaries of dynamic change. These boundaries are used to constrain the range of subsequent pattern recognition, avoiding false judgments due to extreme fluctuations. The extraction of dynamic change patterns uses sliding window analysis, with window size set according to data sampling frequency and thermal response time of the insulation material. Within each window, the change trend of humidity and temperature is modeled by fitting algorithms to identify typical patterns such as linear growth, exponential decay or periodic oscillation. If the data in the window shows nonlinear characteristics, piecewise fitting method is used to ensure the accuracy of pattern recognition.

[0046] The construction of the monitoring trajectory combines dynamic change patterns and pre-set acceptable monitoring ranges. The acceptable monitoring range is set according to the physical properties of the insulation material, including humidity upper limit, thermal conductivity threshold and other parameters. Dynamic change patterns provide the evolution trend of actual monitoring data, while the acceptable range serves as a reference benchmark. The monitoring trajectory is generated between the two by interpolation algorithms to form a continuous curve, reflecting the dynamic relationship between humidity and thermal conductivity in ideal and actual states. The interpolation process considers the smoothness of data to avoid distortion of the trajectory due to sudden changes.

[0047] The extraction of hysteresis effect features is based on the rate analysis of the monitoring trajectory. The slope of each point on the trajectory curve is calculated to identify regions where the change rate changes significantly. These regions usually correspond to the delayed response of humidity to temperature changes, i.e. hysteresis effect. The concavity of the trajectory is calculated by the second derivative to further determine the inflection point position as a key feature point of hysteresis effect. The time difference and amplitude difference of these feature points are used to quantify the influence degree of hysteresis effect, finally converted into hysteresis impact quantity. The conversion process uses normalization to make hysteresis effects of different time scales comparable.

[0048] The whole implementation process adopts modular design, and the steps such as data preprocessing, pattern recognition, trajectory construction and feature extraction are independent of each other, which is convenient for adjustment and optimization. The data stream is efficiently transmitted between modules to ensure real-time requirements. The identification result of the dynamically changing mode directly affects the accuracy of the monitoring trajectory, and the accuracy of the lag effect quantity depends on the smoothness of the trajectory and the effective extraction of the feature points. This method can adapt to the monitoring needs under different environmental conditions, and improve the robustness and adaptability of the system by dynamically adjusting the analysis parameters.

[0049] In specific applications, the sampling frequency and window size of historical data need to be selected in combination with the actual monitoring scene. High-frequency sampling can capture more subtle changes, but the computational complexity increases; low-frequency sampling may miss key fluctuations. The setting of the window size needs to balance the response speed and the stability of pattern recognition, and a too small window may cause noise interference, and a too large window may mask short-term changes. The setting of the dynamic range boundary needs to refer to long-term monitoring data to avoid unreasonable contraction or expansion of the boundary due to occasional extreme values.

[0050] The selection of the interpolation algorithm of the monitoring trajectory affects the extraction effect of the lag effect feature. Linear interpolation is simple but may lose details, and spline interpolation is smoother but has larger computational complexity. In practical applications, the system resources and the requirements for accuracy can be balanced. The normalization processing of the lag effect quantity ensures that the monitoring results of different time periods are comparable, avoiding evaluation bias due to different monitoring periods.

[0051] The core of this method is the accurate identification of the dynamically changing mode and the reasonable construction of the monitoring trajectory. By combining the statistical characteristics of historical data and the physical limitations of the material itself, a scientific and reasonable monitoring standard is formed. The extraction of the lag effect feature further quantifies the influence of environmental factors on the performance of the insulation layer, providing key input for subsequent thermal conductivity prediction. The whole process forms a closed loop, and real-time data continuously update the historical database, dynamically optimize the monitoring strategy, and realize long-term stable performance evaluation.

[0052] Example 2: refer to Figure 3 The interaction characteristics of real-time humidity and temperature data streams are analyzed using a time-synchronized multivariate statistical method. The original data collected by the environmental sensor is first time-aligned to ensure that the timestamps of the humidity and temperature readings are strictly matched. Time alignment compensates for minor time deviations through interpolation, avoiding analysis errors caused by sensor response delays. The aligned data stream forms a two-dimensional time series, with each time point containing paired observations of humidity and temperature. The quantification of interaction characteristics is based on statistical correlation calculations, which measure the degree of linear association between the two in the time domain. Correlation analysis uses a sliding window technique, with the window width set according to the typical time scale of environmental changes to balance the ability to capture short-term fluctuations and long-term trends.

[0053] The calculation of dynamic interference degree introduces multivariate data analysis techniques. Humidity and temperature data form a covariance matrix, reflecting the statistical characteristics of their joint variation. Eigenvalue decomposition of the matrix reveals the main variation direction of the data, and the eigenvector corresponding to the largest eigenvalue indicates the strongest coupling variation mode of humidity and temperature. The dynamic interference degree is represented by the proportion of the variance explained by the first principal component, i.e., the ratio of the variance explained by the first principal component to the total variance. The larger this ratio, the more significant the driving effect of environmental factors on humidity variation. The calculation process uses an incremental update algorithm to adapt to the continuous input of real-time data stream, avoiding the computational burden of repeatedly processing historical data.

[0054] The derivation of the humidity stability index is based on the identification of the associated variation mode. The associated variation mode is found through time-lag correlation analysis, which calculates the correlation between humidity and temperature at different time offsets and finds the time difference corresponding to the correlation peak. This time difference reflects the lag time of humidity response to temperature changes and is an important parameter for evaluating the dynamic characteristics of the system. The dynamic interference degree is used as a weight factor in stability evaluation, with more emphasis on short-term fluctuation suppression in high-interference environments and more emphasis on long-term stability maintenance in low-interference environments. The calculation of the stability index uses a dynamic system analysis method based on differential equations to quantify the system's self-regulation ability through the ratio of humidity change rate to temperature disturbance.

[0055] The modeling of the humidity feedback mechanism uses a discrete-time system description. The humidity change is decomposed into a temperature-driven term and a self-regulation term, with the former reflecting the direct impact of environmental temperature and the latter representing the moisture absorption and release characteristics of the insulation material. The feedback coefficient is determined through regression analysis of historical data and is updated regularly to adapt to slow changes in material performance. The results of the stability index calculation are used to dynamically adjust the monitoring strategy, triggering a high-frequency monitoring mode when the index falls below a critical value to capture possible signs of performance degradation. The feedback mechanism operates independently of precise physical models and instead learns the dynamic characteristics of the system through data-driven adaptation.

[0056] Dynamic assessment of the impact of environmental factors on humidity needs to consider periodic disturbances such as daily temperature differences and seasonal changes. The monitoring system automatically identifies the periodic components of the data and extracts the main periodic features through Fourier analysis. These features are used to establish a baseline comparison model to compare real-time data with historical data of the same period, distinguishing between normal fluctuations and abnormal changes. The calculation of the dynamic interference degree accordingly introduces a periodic correction factor to assess the strength of sudden disturbances after eliminating known periodic effects. This method can effectively distinguish between predictable environmental changes and random disturbances, improving the relevance of interference assessment.

[0057] The interaction analysis also involves the treatment of spatial heterogeneity. When there are local temperature differences on the roof, the humidity response may be different at different locations. The monitoring system uses a distributed sensor network to obtain spatial data, and identifies areas with similar response characteristics through cluster analysis. The interaction characteristics of each area are analyzed independently, and the degree of dynamic interference is calculated separately by area. The derivation of the stability index uses a zoning modeling strategy to reflect the local performance differences of the roof insulation layer. The results of spatial analysis are used to guide the optimization of sensor placement, and the monitoring density is increased in areas with complex response characteristics.

[0058] The computational efficiency of real-time data analysis is ensured by the stream processing framework. The reception, alignment, analysis, and storage of data streams use a pipeline architecture, and each processing stage is executed in parallel. The calculation of the degree of dynamic interference uses sliding window statistics, which only requires the data within the window to be updated. The derivation of the stability index uses a recursive algorithm, and the current result only depends on the calculation value of the previous step and the latest input data. This design enables the system to process high-frequency sampling data with limited computing resources, meeting the real-time requirements. The output of the analysis results uses an event-driven mechanism, which only triggers the subsequent processing flow when significant changes are detected.

[0059] The adaptive capability of the monitoring system is reflected in the dynamic adjustment of parameters. The size of the sliding window for interaction analysis is automatically optimized according to the data fluctuation characteristics. A wider window is used during the smooth period to improve statistical stability, and a narrower window is used during the mutation period to enhance sensitivity. The threshold for determining the degree of dynamic interference fluctuates with environmental conditions, and the standard is appropriately relaxed during seasonal transitions to avoid frequent false alarms. The calculation method of the stability index also has a multi-mode switching function, and different evaluation algorithms are used for different types of roof insulation materials. These adaptive mechanisms are implemented through online learning techniques, and the system continuously tracks the reliability indicators of the analysis results to automatically adjust the parameter settings.

[0060] The influence of sensor accuracy needs to be considered when deploying this method in practice. The measurement error of the humidity sensor may mask the true interaction characteristics, and the response delay of the temperature sensor may distort the time delay correlation analysis. The monitoring system uses redundant sensor placement and reading cross-validation techniques to reduce the impact of measurement error. An error propagation model is introduced in the data analysis stage to quantify the uncertainty contribution of sensor accuracy to the final stability index. When the uncertainty exceeds the acceptable range, the system automatically triggers the sensor calibration process to ensure that the data quality meets the analysis requirements.

[0061] The actual use environment of the roof insulation layer has various interference sources, such as precipitation, changes in the angle of sunlight, etc. The monitoring system distinguishes different interference types through multi-source data fusion technology. Precipitation data come from weather stations or rain sensors, which are used to identify the real cause of humidity mutation. Sunlight data are obtained through solar position calculation or light sensors, which help to explain the driving factors of temperature changes. These auxiliary information participates in the interpretation of interaction characteristics, improving the physical interpretability of dynamic interference degree evaluation. The derivation process of the stability index introduces environmental context information, making the evaluation results more in line with the actual working conditions.

[0062] The accumulation of long-term monitoring data supports method optimization. The system regularly replays historical data to examine the change law of interaction characteristics in different periods. The evaluation standard of dynamic interference degree is gradually refined with data accumulation, establishing reference benchmarks under different seasons and weather conditions. The calculation method of the stability index is continuously adjusted through historical data verification, eliminating detection logic with high false positive rate and retaining evaluation paths with high reliability. This long-term data-based method evolution enables the monitoring system to adapt to the aging process of building materials and continuously provide accurate performance evaluation.

[0063] The technical implementation of this method focuses on the balance between computational complexity and analysis depth. Interaction analysis avoids complex mathematical models and uses intuitive statistics to capture main features. The calculation of dynamic interference degree considers both physical meaning and computational efficiency, choosing the most concentrated representation of information. The derivation of the stability index focuses on key influencing factors and does not excessively pursue refined parameter estimation. This design concept enables the system to efficiently run on embedded devices, meeting the strict requirements of roof monitoring scenarios on power consumption and cost.

[0064] The distributed computing architecture supports the deployment of large-scale roof monitoring networks. Data analysis of each monitoring node is completed locally, and only feature extraction results and abnormal events are reported to the central node. The central node is responsible for cross-regional data correlation analysis and identification of systematic risk factors. This edge computing mode reduces data transmission volume and improves system response speed. The evaluation of dynamic interference degree includes both local real-time calculation and the fusion of regional environmental background values, forming a multi-scale interference evaluation system. The final judgment of the stability index integrates local readings and reference data from neighboring nodes, reducing the probability of isolated false positives.

[0065] Real-time monitoring of roof insulation humidity and thermal conductivity requires coordination of multiple technical elements. The implementation of Example 2 builds a reliable environmental interaction analysis system through a rigorous data processing flow, adaptive analysis algorithms, and multi-level quality control. The quantitative assessment of dynamic interference degree provides a basis for understanding the humidity change mechanism, and the accurate derivation of stability indicators supports the reliability of thermal conductivity prediction. This method forms a closed-loop optimization in data processing, feature extraction, and state evaluation, achieving a synergistic improvement in monitoring accuracy and computational efficiency.

[0066] Example 3: Adjustment strategy for sampling frequency is based on the comparison result of thermal conductivity prediction deviation and preset monitoring threshold. When the prediction deviation exceeds the threshold, the system enters high sensitivity monitoring mode, and the sampling frequency is increased exponentially. The frequency adjustment algorithm uses an adaptive step mechanism, and the step size is proportional to the magnitude of the deviation exceeding the threshold. The specific frequency adjustment amount is calculated by the following formula:

[0067]

[0068] where, represents the frequency adjustment amount, is the baseline sampling frequency, is the material characteristic coefficient, is the current thermal conductivity prediction deviation, is the monitoring threshold. This formula ensures that the monitoring density is quickly increased when the deviation significantly increases, while minor fluctuations only trigger moderate adjustments. The frequency up-regulation process sets an upper limit to prevent the sampling frequency from being too high due to continuous abnormalities, which would exhaust system resources. When the prediction deviation falls below the threshold, the system enters the frequency decay phase, and the sampling frequency slowly decreases according to a logarithmic curve, avoiding monitoring discontinuity caused by frequent switching.

[0069] The complete set of historical data is managed using a rolling time window strategy. The window length is determined according to the thermal relaxation time of the insulation material, usually set to 3-5 times the time required for the material to reach thermal equilibrium. The data points in the window are sorted by collection time to form a paired sequence of humidity and temperature. The data preprocessing stage includes three levels of filtering: first, outlier detection based on standard deviation is applied to remove abnormal readings that are significantly deviated from the group; second, median filtering is used to smooth short-term fluctuations, and the filter window width matches the sensor response time; finally, data normalization is performed to map humidity and temperature values to the [0,1] interval, eliminating the influence of dimension differences on subsequent analysis. The preprocessed data set retains the original timestamp information for time reference in dynamic change pattern recognition.

[0070] The recognition process of dynamic change patterns adopts a multi-scale analysis method. On the complete data set, first, coarse-grained trend extraction is applied to fit the overall change trend line of humidity and temperature by the least square method. Then, fine-grained fluctuation analysis is performed on the local time segment, and the data stream is divided into paragraphs with consistent characteristics by using an adaptive segmentation algorithm. The pattern type of each paragraph is identified by shape matching technology, compared with the preset typical pattern library, and classified into basic patterns such as linear growth, exponential decay or oscillation. The pattern recognition result is accompanied by a confidence score, reflecting the reliability of the classification. The low-confidence paragraph triggers the data review process, and if necessary, the time window is expanded for reanalysis.

[0071] The attenuation algorithm of sampling frequency is designed considering the stability requirements of the system. When the prediction deviation is lower than the threshold , the frequency adjustment amount calculation introduces an attenuation factor :

[0072]

[0073] wherein represents the frequency attenuation amount, is the current sampling frequency, is the attenuation rate coefficient. The algorithm ensures that the frequency reduction process is smooth, avoiding premature reduction of monitoring density due to short-term improvement. The attenuation process sets a minimum frequency limit to ensure that the basic monitoring requirements are not interrupted. The frequency adjustment decision integrates the current reading and historical trends. When continuous improvement of deviation is detected for multiple periods, the attenuation process is started to prevent single accidental fluctuations from misleading system behavior.

[0074] The maintenance of data integrity is realized by a double-checking mechanism. Each newly collected data point needs to pass the timeliness verification and the reasonableness verification. The timeliness verification checks the synchronization state of the data timestamp and the system clock, and eliminates expired data with serious lag. The reasonableness verification refers to the material physical property parameters, and rejects abnormal values that are obviously beyond the physical possibility. The data that passes the verification is immediately added to the rolling time window, and the window data reorganization is triggered at the same time. The window reorganization algorithm optimizes the data storage structure, uses a time-indexed binary tree to accelerate query operations, and ensures efficient access to large-capacity historical data.

[0075] The dynamic optimization of monitoring thresholds is a key step for the continuous improvement of the system. The initial thresholds are set according to the standard parameters of the material, and are gradually optimized during operation through machine learning techniques. The system records all prediction deviation events and their subsequent developments, and establishes a correlation model between deviation amplitude and actual impact degree. The threshold adjustment algorithm analyzes this model and seeks the optimal balance point between false alarm rate and missed alarm rate. The optimization process adopts a conservative strategy, with each adjustment not exceeding 10% of the original value, and requires verification and confirmation through multiple monitoring periods. Dynamic thresholds and fixed safety thresholds are used in parallel, with the former guiding regular monitoring strategies and the latter serving as the last line of defense to prevent major oversights.

[0076] The improvement of environmental adaptability is achieved through the background learning module. The system continuously monitors the baseline changes of environmental parameters, such as the average diurnal temperature difference and seasonal humidity fluctuation range, and establishes an environmental background feature library. Newly collected data are first compared with the background features to distinguish between normal environmental fluctuations and abnormal states. This mechanism effectively reduces the probability of false triggering caused by regular climate changes, allowing the system's resources to focus on addressing real abnormal situations. Background learning uses incremental updating, with new data continuously fine-tuning feature parameters to track slow environmental changes.

[0077] The system's robustness is enhanced through multiple redundancy designs. The core monitoring module uses a primary and backup dual-machine architecture, automatically switching to the backup node in case of primary node failure. The data transmission channel is equipped with redundant routing, with critical data being uploaded simultaneously through wired and wireless methods. The historical database implements real-time mirror backup, with any write operation being synchronized to the backup storage. The self-diagnosis program periodically checks the system's health status, including sensor sensitivity, computing unit load, storage space, and other key indicators. Abnormal states trigger warning signals, prompting maintenance personnel to intervene.

[0078] Real-time performance assurance relies on efficient task scheduling strategies. Data collection, processing, and storage tasks are divided into different priorities, with core monitoring tasks enjoying the highest scheduling privileges. Time-slice rotation is used for computationally intensive analysis tasks to avoid long-term blocking of real-time data streams. Object pool technology is used for memory management to reduce the delay caused by dynamic memory allocation. The multi-thread synchronization mechanism is specially optimized, with a lock-free access design used in critical data areas to maximize concurrent performance.

[0079] Long-term running stability is maintained through resource recycling mechanisms. The system periodically cleans up expired historical data, retaining only feature extraction results and key statistics. Temporary files are automatically recycled, with storage space being released immediately after task completion. The memory cache implements the least recently used eviction algorithm, keeping available memory at a safe level. Network connections are set with heartbeat detection, automatically reconnecting and resuming transmission after an abnormal interruption. These mechanisms collectively ensure the long-term stable operation of the system in unattended environments.

[0080] The user interface design follows the principle of functional visibility. The monitoring status is visually displayed through color coding, with green indicating normal operation, yellow indicating attention, and red indicating abnormality. The operation buttons are arranged in layers according to usage frequency, with frequently used functions accessible at one click. The data visualization supports multiple time scale switching, allowing seamless browsing from macro trends to micro fluctuations. The alarm information is presented in a structured manner, including key elements such as abnormal description, occurrence time, and suggested measures. The interface response is specially optimized, with any operation feedback delay controlled within an acceptable range.

[0081] The system deployment scheme takes into account actual engineering constraints. The sensor nodes adopt a low-power design, with battery-powered versions capable of continuous operation for more than 6 months. The wired version supports standard industrial interfaces, facilitating integration into existing building management systems. The wireless networking scheme supports self-organizing networks, automatically adapting to changes in the physical environment of the roof. The gateway device has edge computing capabilities, allowing basic monitoring tasks to be completed locally and reducing the burden on the central server. The installation bracket design adapts to various roof structures, ensuring accurate and reliable positioning of the sensors.

[0082] The technical implementation of this method focuses on the unity of engineering practicality and theoretical rigor. The sampling frequency adjustment algorithm balances response speed and system stability, and the historical data analysis considers both computational efficiency and result reliability. The system architecture design adapts to various deployment environments, from small-scale monitoring points running independently to large-scale building groups connected for monitoring. All technical choices are guided by actual monitoring needs, avoiding the complexity brought by overdesign. This pragmatic design philosophy enables the implementation to run reliably in real engineering scenarios, providing effective technical support for roof insulation layer performance monitoring.

[0083] Example 4: Refer to Figure 4 The real-time data collected by the roof insulation layer environmental sensor includes multiple types, among which environmental temperature fluctuation data and surface humidity distribution data are the core monitoring indicators. Environmental temperature fluctuation data is collected by a temperature sensor array arranged above the insulation layer, with each sensor recording the temperature value at its location to form a spatially distributed temperature field. Surface humidity distribution data is obtained through embedded humidity sensors arranged in a grid pattern, with each node measuring the local humidity state. These two types of data are collected synchronously in the time dimension, but differ in spatial resolution and physical meaning, requiring data fusion techniques to integrate them into a unified data stream.

[0084] The data fusion process starts with time alignment. Since different types of sensors may have different sampling periods, the system uses interpolation methods to unify the data to the same time reference. For example, the temperature sensor collects data every minute, and the humidity sensor collects data every two minutes. The fusion algorithm generates a complete set of temperature and humidity data at every minute. For the time without direct humidity readings, linear interpolation is used to supplement the intermediate values. After time alignment, the data is matched according to the spatial position, and each temperature measurement point is associated with the nearest humidity measurement point to form a position-paired observation value.

[0085] Table 1 shows a typical data set segment after space-time alignment processing. The data contains five fields: time stamp, temperature sensor number, temperature value, nearest humidity sensor number, and humidity value. This structured representation facilitates subsequent analysis algorithm processing while preserving the original data's space-time correlation characteristics.

[0086] Table 1: Typical data set segment after space-time alignment processing.

[0087]

[0088] Interaction characteristic analysis is based on the fused data stream. The system calculates the temperature and humidity cooperative change characteristics of each location point, and identifies the time lag relationship between them. For example, after the temperature in a certain area rises, does the humidity change show a delay pattern? The analysis process uses sliding time window technology, and the window width is dynamically adjusted according to the material response characteristics. For fast-responding insulation materials, the window width is set to 15-30 minutes; for slow-responding materials, it is extended to 1-2 hours. The data points in the window calculate the statistical correlation to evaluate the influence strength of temperature change on humidity.

[0089] The determination of dynamic disturbance degree considers the variation characteristics of both spatial and temporal dimensions. In the spatial dimension, the system analyzes the response differences of different regions to disturbance. Some locations may be more sensitive to temperature fluctuations, and the humidity change amplitude is significantly greater than that in other regions. This spatial heterogeneity is identified by region clustering algorithm, which divides the roof into several sub-regions with similar response characteristics. In the time dimension, the disturbance degree evaluation focuses on the duration of abnormal events. Short-term spikes may be caused by measurement noise, while sustained fluctuations often reflect real environmental disturbances. The system establishes a filtering mechanism based on the duration, ignoring transient disturbances and focusing on processing long-term changes that have actual impact.

[0090] The input features of the thermal conductivity prediction model are carefully designed. The lag effect is converted into three specific features: maximum lag time, average lag amplitude, and lag response area. The maximum lag time refers to the time difference between temperature change and humidity response reaching the peak, reflecting the slowest response speed of the material. The average lag amplitude statistics the typical fluctuation range of humidity following temperature change in historical data. The lag response area is the integral of the lag curve, which comprehensively considers the response time and amplitude. The humidity stability index is also decomposed into multiple feature components, including short-term fluctuation rate, long-term drift amount, and recovery speed, etc. These features together constitute the input vector of the prediction model, comprehensively describing the dynamic thermal and humidity characteristics of the insulation layer.

[0091] The training of the prediction model uses historical monitoring data sets. The training samples contain temperature and humidity change records under various seasons and weather conditions, ensuring that the model has wide environmental adaptability. The feature engineering stage standardizes the original data, eliminating the scale difference caused by different physical dimensions. The model structure uses an ensemble learning method, combining the advantages of multiple basic predictors. Each predictor focuses on a feature subspace, and the final prediction result is the weighted combination of the outputs of each predictor. The training process uses cross-validation techniques to evaluate model performance to prevent overfitting problems. The trained model is deployed to the real-time monitoring system, and the model parameters are updated regularly with new data to track the slow changes in material performance.

[0092] The calculation of prediction bias uses a double verification mechanism. The system records both the model prediction value and the actual measurement value, and the bias calculation considers not only the instantaneous difference but also the trend consistency. The instantaneous difference reflects the prediction accuracy at the current point, and the trend consistency evaluates the model's ability to capture the change direction. When both of them deviate significantly, a prediction bias alarm is triggered. The bias evaluation results guide the sampling frequency adjustment and also feedback to the model training link for improving subsequent prediction accuracy. This closed-loop learning mechanism enables the system to continuously optimize prediction performance and adapt to changing monitoring environments.

[0093] The distributed computing architecture supports data processing for large-scale sensor networks. Each sensor node has local computing capability and can complete basic feature extraction and simple prediction. The regional gateway is responsible for aggregating data from multiple nodes and performing more complex analysis and model prediction. The central server mainly undertakes long-term trend analysis and system management functions. This hierarchical processing architecture effectively shares computing load, reduces network transmission data volume, and improves system response speed. Key prediction tasks use redundant computing strategies, with multiple nodes independently computing and comparing results to ensure the reliability of important decisions.

[0094] The system deployment takes into account actual engineering constraints. The sensor nodes are designed with industrial-grade protection, capable of withstanding harsh outdoor environments. The power supply scheme is flexible and diverse, supporting wired power supply, battery power supply, and solar power supply, among other modes. The wireless communication module has an adaptive frequency modulation function, maintaining stable connections in complex building environments. The installation process follows standardized procedures to ensure accurate sensor placement and consistent orientation. Comprehensive functional testing is conducted during system debugging, including sensor calibration, communication quality detection, and algorithm verification, to ensure monitoring quality after commissioning.

[0095] The maintenance and management functions are designed with practicality in mind. The system automatically generates device health reports, prompting the replacement of batteries or the cleaning of sensors. Remote diagnostic tools support online troubleshooting, reducing the number of on-site maintenance visits. Software upgrades use differential update technology, only transmitting changed parts to reduce network bandwidth requirements. User permissions are managed hierarchically, with different roles having corresponding operational permissions. Data backup strategies combine local storage and cloud synchronization to prevent accidental data loss. These design details collectively improve system maintainability and reduce long-term operating costs.

[0096] The actual operation of the roof insulation layer monitoring system requires handling various abnormal situations. Sensors may produce abnormal readings due to dust accumulation, aging, or extreme weather. The system uses multi-dimensional verification methods to identify faulty sensors, including historical data comparison, adjacent sensor reference, and physical model verification. Faulty sensors are automatically marked as unusable, and their monitoring area is covered by neighboring sensors until maintenance is complete. Data during communication interruptions is temporarily stored in local cache, and automatically resumed after connection is restored, ensuring the integrity of monitoring data. These fault-tolerant mechanisms ensure that the system can maintain basic monitoring functions in various unexpected situations.

[0097] The visualization of data analysis results focuses on balancing information density and readability. The monitoring interface uses a hierarchical display design, with the homepage displaying an overall status overview and secondary pages providing detailed data for each zone. Trend charts support interactive zooming, making it easy to view changes at different time scales. Exceptional events are highlighted with prominent markers, associated with relevant analysis data and disposal recommendations. Report generation functions support customized output to meet the viewing needs of different users. The visualization scheme has been optimized for human engineering, reducing visual fatigue during long-term monitoring.

[0098] The application of this method in a typical roof project demonstrates its technical advantages. A commercial building with a roof area of about 2000 square meters was equipped with 36 temperature sensors and 24 humidity sensors, forming a complete monitoring network. After the system was put into operation, it successfully identified abnormal humidity fluctuations in the southeast corner of the roof area. The humidity in this area decreased significantly slower than in other areas when the temperature rose in the afternoon. Further inspection found that the insulation layer in this area had construction defects, causing moisture accumulation. After repair, the temperature and humidity response characteristics of this area returned to normal, verifying the effectiveness of the monitoring method. The accumulation of similar cases provides practical reference for optimizing system parameters, making the monitoring algorithm closer to the needs of engineering practice.

[0099] The continuous improvement of the technical solution is based on the analysis of operational data. The system collects all monitoring events and their processing results to form a case library. Regular analysis of these cases extracts effective feature patterns and decision rules for optimizing algorithm parameters. The improvement process adopts a gradual strategy, with each update first tested in a small range and then promoted to the entire network after verification. This empirical evolution approach ensures that system improvements are always in the right direction, avoiding performance fluctuations caused by subjective assumptions. The case library also provides a reference for new project deployment, shortening the system debugging period.

[0100] The technical implementation of roof insulation layer humidity and thermal conductivity monitoring integrates knowledge from multiple disciplines. Sensor technology provides accurate environmental perception, data science methods build effective analysis models, and engineering experience guides the practical design of the system. The implementation of Example 4 demonstrates how to organically combine these elements to form a reliable and efficient monitoring solution. The complete process from data collection to analysis and prediction is carefully designed, with each step optimized for actual engineering needs. This systematic technical approach provides valuable monitoring tools for building maintenance, helping to improve the long-term performance of roof insulation systems.

[0101] Example 5: The design of the cyclic iteration framework uses a state machine model to realize the coordinated operation of each functional module. The system initialization stage loads historical data and establishes an initial state, and the real-time acquisition module is activated as the entry point. The raw data collected after format conversion and validity check triggers the state transition of pattern recognition. The pattern recognition module receives the standardized data stream, extracts dynamic change characteristics through sliding window analysis, and generates pattern descriptors after processing and passes them to the monitoring trajectory construction module. The trajectory construction algorithm generates continuous monitoring curves based on the current pattern and historical benchmarks, and the output results include lag effect feature identification points.

[0102] The hysteresis influence quantity conversion module receives the trajectory construction result, calculates the quantitative index through feature point analysis and normalization processing. The conversion process considers material characteristic parameters and environmental background values to ensure the comparability of the calculation results under different working conditions. The generated hysteresis influence quantity is used as a key parameter to trigger the start of the dynamic interference analysis module. The module synchronously receives real-time data streams and evaluates the interference strength of environmental factors through multivariate statistical methods. The interference degree calculation result and the temperature fluctuation data are jointly input into the humidity stability derivation module, which uses time domain analysis method to establish a humidity response model and outputs a stability index value.

[0103] The thermal conductivity prediction calculation module, as the core processing unit, integrates the hysteresis influence quantity and stability index for comprehensive evaluation. The prediction algorithm uses incremental updating, processing only the latest input parameters each time to maintain efficient operation. The prediction result is compared with the real-time measurement value to generate deviation data, and the deviation analysis module determines the sampling frequency adjustment strategy according to the preset logic. The frequency adjustment instruction is fed back to the real-time acquisition module through the control bus, forming a closed-loop control circuit. The entire state conversion process records detailed logs, including activation time, processing duration, and data flow of each module, which are used for system performance optimization.

[0104] The data flow transfer mechanism adopts a publish-subscribe mode to realize a loosely coupled architecture. Each functional module runs as an independent service and exchanges data through a message middleware. The message format uses a standardized protocol definition, including timestamp, data type, and payload content. Data transmission is set up with a priority queue, and critical state transition messages have the highest transmission priority. The exception handling mechanism considers scenarios such as message loss and repetition, ensuring data reliability through sequence number verification and retransmission confirmation. The service discovery mechanism supports dynamic expansion of modules, and new functional components can be added to the system at any time without affecting existing processes.

[0105] Time sequence consistency is maintained through a global clock synchronization protocol. Distributed nodes regularly calibrate with a time server to maintain millisecond-level synchronization accuracy. The data processing flow strictly maintains the event order, and subsequent dependent modules always wait for the complete output of the predecessor module. For time-sensitive calculation tasks, a timeout mechanism is set to avoid infinite waiting. Delayed data processing uses special markers to distinguish, preventing outdated information from interfering with current analysis. The system continuously monitors the processing delay of each module during operation, dynamically adjusting resource allocation to ensure the timeliness of the critical path.

[0106] The resource management strategy adopts a dynamic quota allocation approach. CPU resources are preferentially scheduled for computationally intensive modules such as pattern recognition and thermal conductivity prediction. Memory allocation preloads data based on historical usage patterns of each module, reducing real-time processing waiting time. Storage resources are managed hierarchically by data type, with high-frequency access real-time data saved in high-speed storage media and historical archive data transferred to large-capacity low-speed storage. Network bandwidth allocation considers the urgency of data transmission, with control commands and alarm information being transmitted preferentially. The resource monitoring component tracks the resource utilization of each module in real time, and triggers an alarm and records detailed snapshots for abnormal usage patterns.

[0107] The fault tolerance recovery mechanism covers hardware failure and software exception scenarios. Key modules are deployed with primary and backup redundancy, automatically switching to the backup instance in the event of failure. The data processing flow sets checkpoints, which can be restored from the nearest checkpoint after an unexpected interruption. The message system implements persistent storage to ensure that critical data is not lost. The self-repair function periodically detects the system health status, automatically restarts and recovers the context of abnormal components. The maintenance interface provides manual intervention capabilities, allowing administrators to take control when necessary. Fault events record detailed diagnostic information, including stack traces, memory dumps, and associated data states, facilitating subsequent analysis and improvement.

[0108] Performance optimization adopts a multi-dimensional tuning strategy. The calculation algorithm selects the optimal implementation based on hardware characteristics, such as GPU acceleration for parallel computation-intensive tasks. The data caching strategy considers temporal and spatial locality, prefetching adjacent data that may be used. Memory access patterns are optimized to reduce cache invalidation and memory fragmentation. Network communication uses compression and batch processing techniques to reduce transmission overhead. Disk I / O improves throughput through merged writing and asynchronous flushing. The system periodically performs performance profiling to identify bottlenecks and optimize them accordingly. Configuration parameters support runtime adjustments, allowing optimized settings to be applied without restarting.

[0109] The security protection system is implemented throughout the system. End-to-end encryption is used for data transmission to prevent man-in-the-middle attacks. An identity authentication mechanism ensures that only authorized components can access the system. Access control policies limit the minimum necessary permissions for each module. Operation audit logs are complete, supporting post-trace analysis. Firmware and software are updated regularly to patch known vulnerabilities. Physical security measures protect critical infrastructure from unauthorized access. The security monitoring component detects abnormal behavior patterns in real time, such as frequent login attempts or abnormal data access, and triggers a defense response in a timely manner.

[0110] The user interaction interface is decoupled from the core system design. The presentation layer obtains processing result data through a dedicated interface and does not directly participate in the calculation process. The visualization solution supports access by multiple terminal devices, including desktop browsers and mobile applications. The interface layout takes into account the usage habits of different roles, with system status monitoring being the focus for operations personnel and comprehensive reports being the focus for management personnel. Operation instructions are strictly verified before being forwarded to the core system for execution to prevent misoperation from affecting the system. The user behavior analysis function identifies common operation paths to optimize interface element layout and improve efficiency. The help system integrates context-sensitive prompts to reduce learning costs.

[0111] The deployment scheme supports a flexible and scalable architecture. Small systems can be deployed with all components on a single machine, while large deployments use a distributed cluster architecture. Cloud-native design enables the system to flexibly expand computing resources to handle load fluctuations. Edge computing nodes process local data to reduce the pressure on central nodes. Hybrid deployment mode allows sensitive data to be processed locally, while non-sensitive computing tasks are offloaded to the cloud. Containerization simplifies the deployment process, supports rapid expansion, and version rollback. Infrastructure as Code enables automated configuration management to ensure environment consistency.

[0112] The system maintenance function design takes into account long-term operation needs. Remote diagnosis interface supports online troubleshooting by technical personnel. Automated testing framework covers core function modules, and complete regression testing is performed before major updates. Configuration management system records all parameter change history, supporting quick rollback of error settings. Capacity planning tool predicts resource demand trends, preparing expansion solutions in advance. Knowledge base accumulates solutions to common problems, speeding up the fault handling process. Maintenance tasks are scheduled during low-load periods to minimize the impact on monitoring business.

[0113] The technology evolution path keeps the system continuously improving. The architecture design reserves extension interfaces to support future function enhancement. Modular decomposition allows components to be upgraded independently without affecting other parts. Technology selection tends to open standards to avoid vendor lock-in. Experimental functions are first verified in a test environment before being introduced into the production system. User feedback channels collect actual use experience to guide improvement direction. The technology radar mechanism tracks industry trends and introduces valuable new methods in a timely manner. Regular architecture review assesses system technical debt and develops a reasonable repayment plan.

[0114] The roof insulation layer monitoring system's iterative framework connects various functional modules through a state machine model. The complete process from data collection to analysis and prediction forms a closed-loop control, providing real-time feedback to optimize monitoring strategies. The loosely coupled architecture design ensures the flexibility and scalability of the system, allowing it to adapt to different deployment sizes. Fault-tolerant mechanisms and security protections ensure reliable system operation, maintaining the continuity and accuracy of building monitoring data.

[0115] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and illustrative figures, it should be apparent that the scope of the present application is not limited to these specific embodiments.

[0116] While the embodiments of the application have been shown and described herein, it will be understood by those skilled in the art that many changes, modifications, substitutions and alterations to these embodiments can be made without departing from the principles and spirits of the application, and it is intended that the scope of the application be limited solely by the scope of the appended claims and the equivalents thereof.

Claims

1. A real-time monitoring method for humidity and thermal conductivity of a roof insulation layer, characterized in that: include: Real-time collection of historical humidity and temperature data of the roof insulation layer; Identifying dynamic change patterns of humidity and temperature during the monitoring process from the historical humidity data and the historical temperature data; Based on the dynamic change pattern, a monitoring trajectory of the humidity and thermal conductivity of the roof insulation layer is constructed; multiple humidity hysteresis effect features of the roof insulation layer are extracted according to the change rate of the monitoring trajectory, and all humidity hysteresis effect features are converted into hysteresis influence quantities of the roof insulation layer during the monitoring process; activating an environmental sensor of the roof insulation layer to collect real-time humidity and temperature data streams, and determining a degree of dynamic interference of environmental factors on humidity based on the interactive characteristics of humidity and temperature in the real-time humidity and temperature data streams; Derived from the dynamic interference degree and the correlation between humidity and temperature, the humidity stability index of the roof insulation layer during temperature fluctuations; The thermal conductivity of the roof insulation layer is predicted and calculated in real time by using the hysteresis effect amount and the humidity stability index, thereby obtaining the prediction deviation of the thermal conductivity of the roof insulation layer; When the thermal conductivity prediction deviation exceeds a preset monitoring threshold, the data sampling frequency of the roof insulation layer environmental sensor is increased.

2. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: The identifying the dynamic change pattern of humidity and temperature during the monitoring process from the historical humidity data and the historical temperature data specifically includes: calculating a set of change rates of adjacent data points in the historical humidity data and the historical temperature data; Derived from the set of change rates a dynamic range boundary of the historical data; The dynamic change pattern of humidity and temperature during the monitoring process is extracted based on the dynamic range boundary, and the dynamic change pattern is used as the basic input for constructing the subsequent monitoring trajectory.

3. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: The constructing of a monitoring trajectory of the humidity and thermal conductivity of the roof insulation layer based on the dynamic change pattern specifically includes: defining an acceptable monitoring range of the humidity and thermal conductivity of the roof insulation layer; A monitoring trajectory of the humidity and thermal conductivity of the roof insulation layer is generated based on the dynamic change pattern and the acceptable monitoring range, and the monitoring trajectory is applied to a screening process of hysteresis effect characteristics.

4. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: The determining the degree of dynamic interference of environmental factors on humidity based on the interactive characteristics of humidity and temperature in the real-time humidity and temperature data stream specifically includes: analyzing the interactive characteristics of humidity and temperature in the real-time humidity and temperature data stream; Calculating the correlation strength between humidity and temperature based on the interaction characteristics; The dynamic interference degree of environmental factors on humidity is determined by the correlation strength, and the dynamic interference degree is used to derive the humidity stability index.

5. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: The derivation of the humidity stability index of the roof insulation layer during temperature fluctuations from the dynamic interference degree and the correlation change between humidity and temperature specifically includes: identifying the correlation change pattern between humidity and temperature; A humidity feedback mechanism is established based on the dynamic interference degree and the associated change pattern; a humidity stability index of the roof insulation layer during temperature fluctuations is derived from the humidity feedback mechanism, and the humidity stability index is transferred to the thermal conductivity prediction calculation step.

6. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: Also includes: When the thermal conductivity prediction deviation is equal to the preset monitoring threshold, the current data sampling frequency of the roof insulation layer environmental sensor is maintained; When the thermal conductivity prediction deviation is less than a preset monitoring threshold, the data sampling frequency of the roof insulation layer environmental sensor is reduced, and the adjusted sampling frequency data is fed back to the real-time acquisition step.

7. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: The historical humidity data and the historical temperature data are defined as a complete set of all humidity values ​​and temperature values ​​within a specified time period, and the complete set is used in the process of identifying the dynamic change pattern.

8. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: After the environmental sensor of the roof insulation layer is activated, the real-time humidity and temperature data stream collected includes ambient temperature fluctuation data and surface humidity distribution data; The ambient temperature fluctuation data and the surface humidity distribution data are integrated into a unified data stream through data fusion technology, which is used in the analysis process of the interaction characteristics, and the analysis results are applied to the determination of the dynamic interference degree.

9. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: The real-time prediction calculation of the thermal conductivity of the roof insulation layer by using the hysteresis effect amount and the humidity stability index specifically includes: constructing a thermal conductivity prediction model, and using the hysteresis effect amount and the humidity stability index as model inputs; Use machine learning algorithms to process input data and generate thermal conductivity predictions; The thermal conductivity prediction deviation between the predicted value and the actual measured value is calculated, and the thermal conductivity prediction deviation is output to the monitoring threshold comparison step.

10. The method for real-time monitoring of humidity and thermal conductivity of a roof insulation layer according to claim 1, wherein: A cyclic iterative framework is adopted, in which the output data of the real-time acquisition step is used to identify the dynamic change pattern, the identification result is used to construct the monitoring trajectory, the construction result is used to extract the hysteresis effect characteristics, the extraction result is used to convert the hysteresis influence quantity, the conversion result is used to determine the degree of dynamic interference, the determination result is used to derive the humidity stability index, the derivation result is used for real-time prediction and calculation of thermal conductivity, the prediction result is used to obtain the prediction deviation, the acquisition result is used to adjust the data sampling frequency, and the adjustment result is fed back to the real-time acquisition step to form a closed-loop data interaction.

Images