Rural historical building protection evaluation system based on big data

Abstract

The application discloses a rural historical building protection evaluation system based on big data, and relates to the technical field of building protection. The system comprises a data processing module, which generates an environment deterioration factor by normalizing environment data, extracts a crack depth parameter and a facade weathering parameter from structural detection data through a feature extraction algorithm, and generates a building life attenuation curve through correlation analysis; a decision module, which establishes a maintenance optimization model according to the building life attenuation curve and historical maintenance data, generates a risk level through threshold comparison, evaluates the expected effect of different maintenance methods on delaying deterioration, and outputs an optimal maintenance suggestion. The application realizes accurate prediction of the deterioration process of rural historical buildings and optimization recommendation of maintenance strategies, and improves the scientificity and efficiency of protection work.

Description

Technical Field

[0001] This invention relates to the field of building conservation technology, specifically a big data-based assessment system for the conservation of historical buildings in rural areas. Background Technology

[0002] The preservation of historical buildings in rural areas is a crucial aspect of cultural heritage transmission, with its core lying in the assessment and proactive maintenance of the building's structural health. In recent years, with the development of big data and sensor technology, preservation solutions for historical buildings in rural areas have gradually evolved towards data-driven and intelligent approaches. Current technologies primarily involve periodically collecting environmental data (such as temperature and humidity) and surface structural data (such as crack width and weathering area) of the building, and then assessing the building's risk level based on threshold judgments or static experience models to formulate maintenance plans. These methods can quantitatively describe the current state of the building and, in most cases, provide a preliminary basis for preservation work, demonstrating a certain degree of applicability.

[0003] However, the deterioration of rural historical buildings is a complex process involving dynamic, continuous, and multi-factor coupling. Current technologies fail to deeply characterize the intrinsic dynamic relationship between environmental stress, material damage, and the time dimension. Specifically, these methods cannot effectively simulate the deep evolution of cracks under alternating temperature and humidity loads, nor can they predict the nonlinear growth trend of surface weathering rates with environmental degradation factors. As a result, existing assessment systems often only reflect the immediate health condition of buildings, lacking the ability to predict the trajectory of building lifespan decline. This leads to repair recommendations being mostly reactive rather than proactive, making it difficult to formulate optimal repair strategies that match the long-term deterioration patterns of buildings. This limitation often results in a reactive, piecemeal approach to conservation efforts, potentially missing the optimal repair window and leading to the misallocation and waste of maintenance resources, failing to meet the need for precise, full-lifecycle protection of precious historical buildings. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a big data-based assessment system for the protection of rural historical buildings.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows:

[0006] This invention discloses a big data-based assessment system for the protection of historical buildings in rural areas, comprising the following steps:

[0007] The data acquisition module is used to acquire environmental data, structural inspection data, and historical maintenance data of the building's location;

[0008] The data processing module is used to normalize the environmental data and generate environmental degradation factors. The crack depth parameter is extracted from the structural detection data using a feature extraction algorithm. and facade weathering parameters The environmental degradation factors The crack depth parameter and the aforementioned facade weathering parameters The data is input into a pre-constructed differential equation prediction model, which is then solved using a numerical iterative method to output the building structural health status. The sequence data varies with time t, and a building lifespan decay curve is generated based on the sequence data. The differential equation prediction model is defined as follows:

[0009] ;

[0010] in, To use the crack depth parameter The crack depth evolution function is defined by the initial conditions. Based on the aforementioned facade weathering parameters Let be the weathering degree evolution function under initial conditions; k is the deterioration rate coefficient. The corresponding weight coefficients, and satisfying ;

[0011] The decision module is used to establish a multi-objective maintenance optimization model based on the building life decay curve and the historical maintenance data, and output the optimal maintenance recommendations through simulation.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0013] 1. This invention can accurately capture the long-term cumulative impact of different environmental conditions (such as temperature and humidity fluctuations and rainfall erosion) on building structures. The building life decay curve constructed by combining time series data can dynamically reflect the degradation trajectory under future environments, making life prediction more in line with actual degradation patterns and providing a basis for judging the timing of protection intervention.

[0014] 2. This invention makes maintenance recommendations specific and traceable. The model can comprehensively consider the actual effect of different maintenance methods on slowing down the degradation rate, avoiding a single maintenance strategy. It reduces the waste of resources caused by over-maintenance and also reduces the risk of accelerated degradation caused by under-maintenance, achieving a dynamic balance between protection costs and protection effects.

[0015] 3. This invention avoids the problems of data fragmentation and isolated analysis in traditional assessments, improves the systematic nature of the assessment process and the reliability of the results, and provides full-process technical support for the refined protection of rural historical buildings. Attached Figure Description

[0016] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. In the drawings, the same reference numerals are used to refer to the same parts. Wherein:

[0017] Figure 1 This is a system module connection diagram of the present invention;

[0018] Figure 2 This is a flowchart illustrating the working principle of the present invention.

[0019] Figure 3 This is a connection diagram of the model update module of the present invention;

[0020] Figure 4 This is a connection diagram of the multi-source data fusion module of the present invention;

[0021] Figure 5 This is a connection diagram of the environmental prediction module of the present invention. Detailed Implementation

[0022] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0023] In existing technologies, the conservation assessment of rural historical buildings largely relies on periodic manual surveys or sensor monitoring, making it difficult to achieve dynamic prediction of damage processes and precise optimization of maintenance plans. Traditional methods lack effective quantitative modeling tools when dealing with the coupled effects of multiple factors such as material aging and environmental erosion, resulting in assessment results that are highly subjective and lack predictability, failing to meet the needs of preventative protection.

[0024] To address these issues, the study found a significant correlation between the degradation rate of building materials and the cumulative effect of environmental stress. A multi-factor driven differential equation model was established to predict building lifespan. The research revealed that the contribution weights of crack depth, weathering area, and environmental load to building health vary with material type. Therefore, an adaptive method based on training model parameters using historical data was proposed.

[0025] After introducing the basic concept of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0026] Example:

[0027] like Figure 1 As shown, the big data-based assessment system for the protection of rural historical buildings includes the following steps:

[0028] The data acquisition module is used to acquire environmental data, structural inspection data, and historical maintenance data of the building's location. Environmental data is acquired through IoT sensor nodes deployed on the building and its surrounding environment, including data on temperature, humidity, precipitation, wind speed, solar radiation intensity, and concentrations of sulfur dioxide and particulate matter in the air. Structural inspection data is acquired through portable acoustic testing equipment and a drone inspection system equipped with high-definition and thermal imaging cameras. Historical maintenance data is extracted from a digital archive and stored in a structured manner according to maintenance time, maintenance location, maintenance materials, maintenance process, and post-maintenance inspection reports.

[0029] The data processing module is used to normalize environmental data and generate environmental degradation factors. Furthermore, crack depth parameters were extracted from the structural inspection data using a feature extraction algorithm. and facade weathering parameters Environmental degrading factors Crack depth parameters and facade weathering parameters The data is input into a pre-constructed differential equation prediction model, which is then solved using a numerical iterative method to output the building structural health status. The sequence data varies with time t, and a building lifespan decay curve is generated based on the sequence data. ;

[0030] The differential equation prediction model is defined as:

[0031] ;

[0032] in, The building structure health status at time t, and the building life decay curve. for A curve that changes over time. Using crack depth parameters The crack depth evolution function is defined by the initial conditions. To use facade weathering parameters Let be the weathering degree evolution function under initial conditions; k is the deterioration rate coefficient. These are environmental degradation factors. Crack depth evolution function Weathering degree evolution function The weight coefficients are obtained by training and fitting historical data using a machine learning regression algorithm. And they satisfy... ,For example( = (0.5, 0.3, 0.2);

[0033] The degradation rate coefficient k is based on the degradation characteristics of common materials used in rural historical buildings (brick, stone, wood, rammed earth, etc.), as shown in the following example:

[0034] Masonry structure: k ∈ [0.015, 0.03] / year (referencing long-term monitoring data of sandstone and granite in historical stone building weathering rates);

[0035] Timber structure: k ∈ [0.025, 0.045] / year (considering the combined effects of wood shrinkage and swelling and biological erosion);

[0036] Rammed earth structure: k ∈ [0.03, 0.05] / year (significantly affected by humidity, with a value slightly higher than that of brick and stone).

[0037] The decision-making module is used to establish a multi-objective maintenance optimization model based on the building life decay curve and historical maintenance data, and output the optimal maintenance recommendations through simulation.

[0038] like Figure 2 As shown, the working principle of this application is as follows: The data acquisition module obtains current and historical status data of the building from multiple dimensions. Through an Internet of Things (IoT) sensor network deployed in the building and its surrounding environment, it continuously monitors and collects key environmental data, including temperature, humidity, precipitation, wind speed, solar radiation intensity, and the concentration of corrosive substances in the air (such as sulfur dioxide and salt particulate matter). Simultaneously, portable acoustic detection equipment is used to detect cracks in the building structure, and a drone inspection system equipped with high-definition and thermal imaging cameras efficiently acquires macroscopic and minute images of damage to the building facade. Furthermore, historical maintenance records are extracted from a digitized archive and structured and stored according to dimensions such as time, location, material, process, and effect, forming a complete building "medical record."

[0039] The data processing module fuses and performs in-depth calculations on the collected raw data to quantify the future health status of buildings. First, real-time environmental data is normalized (e.g., min-max standardization, Z-score), and then an indicator reflecting the intensity of environmental erosion—the environmental degradation factor—is used. Using feature extraction algorithms, crack depth parameters characterizing the degree of crack development are extracted from acoustic detection data. (Unit (mm / cm)), and facade weathering parameters characterizing the degree of surface material degradation were identified and calculated from UAV imagery data. Finally, the module will use these three key parameters. , , It is fed into a pre-built differential equation prediction model.

[0040] The pre-built differential equation prediction model is trained using machine learning methods based on long-term monitoring data of a large number of historical buildings. The core of the differential equation prediction model is to characterize the structural health. Differential equations showing the decay law over time are used to generate building life decay curves that predict future changes in building structural performance by solving them. This allows for the estimation of the building's remaining lifespan. In other embodiments of this application, simplified forms of partial differential equations may also be used, such as treating the building as a homogeneous body and using a one-dimensional diffusion equation to describe the transmission of humidity or stress in the material and its impact on degradation.

[0041] Building structural health The Euler method is used to solve the problem, with a set time step. The iterative formula is:

[0042] ;

[0043] in, It is calculated from the right-hand side function of the differential equation prediction model.

[0044] The decision module will predict the building lifespan degradation curve. Based on this core principle and combined with historical maintenance data provided by the data acquisition module, a multi-objective optimization model is established. Simultaneously considering multiple objectives such as maintenance costs, expected lifespan extension after intervention, and preservation of architectural heritage value, the multi-objective maintenance optimization model uses simulation calculations (such as optimization algorithms) to weigh and select from various possible combinations of maintenance timing, locations, and schemes, ultimately outputting at least one maintenance recommendation that is technically effective, economically reasonable, and optimal in terms of preservation.

[0045] The multi-objective maintenance optimization model can also be solved using various optimization algorithms, including but not limited to: second-generation non-dominated sorting genetic algorithm (NSGA-II), decomposition-based multi-objective evolutionary algorithm (MOEA / D), multi-objective particle swarm optimization algorithm (MOPSO), etc.

[0046] This application, through the aforementioned technical solution, realizes a shift in maintenance model from post-event remediation to pre-event prevention. It not only improves the accuracy of health assessment and early warning capabilities for rural historical buildings, but also effectively reduces long-term maintenance costs by optimizing maintenance strategies while ensuring structural safety, providing core technical support for the digital and intelligent protection of rural historical buildings.

[0047] This application further proposes to extract crack depth parameters from structural inspection data using a feature extraction algorithm. and facade weathering parameters In the middle, crack depth parameter The crack depth is extracted using a crack depth fitting method based on the acoustic echo attenuation curve. This crack depth fitting method specifically includes:

[0048] The system uses acoustic sensors to emit sound waves of a specific frequency into the wall and receives the echo signals.

[0049] The amplitude attenuation characteristics and time delay features of the echo signal are extracted. These characteristics are then matched with a pre-established acoustic feature sample library based on different materials and crack depths to obtain matching results. The crack depth parameters are then calculated by fitting the matching results using a regression algorithm. The matching process involves calculating the similarity between the amplitude attenuation and time delay features extracted from the field (forming a feature vector, such as [ΔA, Δt]) and data in a sample library. Matching is typically based on distance metrics (such as Euclidean distance or cosine similarity) to identify the few entries in the sample library that are closest to the field features. The matching result can be a similarity score or a set of candidate crack depth values. The regression algorithm analyzes the nonlinear relationship between features and depth (e.g., amplitude attenuation and time delay may influence depth in complex ways) and outputs a fitted equation. For example, random forest regression might combine multiple decision trees to provide a predicted crack depth and its confidence level. Ultimately, a continuous numerical value (usually in mm or cm) is calculated, representing the estimated crack depth.

[0050] Amplitude attenuation characteristics, i.e., the degree of energy loss of the echo signal relative to the transmitted signal, reflect the energy dissipation of the sound wave when it passes through the crack interface; time delay characteristics, i.e., the time difference between the echo signal and the transmitted signal, contain information about the length of the sound wave propagation path.

[0051] The acoustic feature sample library was established through extensive laboratory experiments, containing acoustic feature data collected from various known building materials (such as bricks, rammed earth, and stone) and crack samples of different known depths. Regression algorithms (e.g., support vector regression or random forest regression) are used to match and fit the feature vectors measured in the field with the sample library to obtain the matching results. Finally, the optimal estimate of the crack depth, i.e., the accurate crack depth parameter, is calculated. .

[0052] This application transforms abstract acoustic signals into precise physical dimensional parameters. Compared to traditional experience-based visual inspection or probe methods, it achieves non-destructive and quantitative measurement of crack depth, greatly improving the objectivity and accuracy of the detection results.

[0053] This application further proposes to extract crack depth parameters from structural inspection data using a feature extraction algorithm. and facade weathering parameters In the middle, facade weathering parameters The surface weathering region is extracted using a method based on image grayscale change thresholds. This method specifically includes:

[0054] Acquire historical and current images of the building's location on the same facade, and perform registration and pixel-level comparison between the historical and current images to generate grayscale values ​​for multiple areas;

[0055] The rate of change of grayscale values ​​in each area is calculated. When the rate of change exceeds a preset dynamic threshold based on the building material type, the building facade area is determined to be a weathered area. The area percentage of the weathered area is then calculated and used as a facade weathering parameter. .

[0056] The system acquires historical and current images of the same building facade at different times, ensuring temporal comparability and spatial consistency of the image sources. Subsequently, the two sets of images are registered with high precision to eliminate geometric and radiometric distortions caused by differences in shooting angle, distance, or lighting conditions, thereby achieving pixel-level comparison. Based on this, the facade is divided into multiple analysis areas, and the grayscale statistical values ​​of each area are calculated separately.

[0057] The system calculates the grayscale change rate of each region between the current image and historical images, reflecting the degradation of building material surfaces over time. To accurately determine whether weathering has occurred, a dynamic threshold based on building material type is introduced. Different allowable variation ranges are preset according to the characteristics of building materials (such as stone, paint, and fair-faced concrete) to accommodate the differences in the response of different materials to environmental factors. The dynamic threshold is set as the percentile (e.g., 95th percentile) of the normal state data, or the value with the best discrimination ability is determined by ROC curve. For blue bricks, the dynamic threshold may be set at 15%; for stone, it may be set at 20%. If the grayscale change rate of a region exceeds the dynamic threshold for its corresponding material type, the region is determined to be a weathered area.

[0058] Ultimately, the system automatically calculates the percentage of all areas identified as weathered relative to the total facade area, and uses this percentage as a facade weathering parameter. Output.

[0059] In one embodiment, periodic health monitoring was conducted on the eastern exterior wall of a century-old brick-built house in a village to assess its surface weathering. The system retrieved a high-resolution digital image of the wall taken a year ago (as historical imagery) and matched it with the current image collected by the drone during the inspection. An image registration algorithm was used to precisely align the two images, ensuring that every pixel corresponds to the same physical location on the wall. The aligned wall image was divided into several small analysis regions, and the average grayscale value of the pixels within each region was calculated. Comparison revealed that the grayscale values ​​of some areas in the current image showed a significant brightening compared to the historical image from a year ago, with an average grayscale change rate of approximately 18%.

[0060] Since the building's walls are made of blue bricks, the system automatically applied the preset dynamic threshold (15%) for this type of material. The aforementioned 18% average grayscale change rate exceeded the dynamic threshold (15%), therefore the system automatically classified these areas as "weathered areas." The total area of ​​all weathered areas was calculated to be approximately 7.2% of the entire east facade, and this 7.2% was ultimately used as the facade weathering parameter for this assessment. Output.

[0061] This application enables automated, refined identification and quantitative assessment of weathering conditions on building facades. It not only improves the objectivity and accuracy of weathering detection but also helps to detect potential signs of deterioration at an early stage, providing reliable data support for building maintenance decisions, thereby extending the service life of buildings and ensuring safe use.

[0062] This application further proposes that the differential equation prediction model be constructed in the following manner:

[0063] Multiple sets of long-term monitoring data of rural historical buildings during their service life were obtained as training sample sets. The long-term monitoring data included time-series environmental data, structural inspection data, and corresponding measured values ​​of building structural health.

[0064] Environmental degradation factors Crack depth parameters and facade weathering parameters Using the rate of change in building structural health as the input feature and the rate of change as the output label, a machine learning regression algorithm is used to analyze the coefficients k in the differential equation prediction model. Training and fitting are performed until the differential equation prediction model predicts the building life decay curve. The error between the measured value of the building's structural health and the actual change trajectory is less than the preset error threshold.

[0065] In the rural historical building protection assessment system constructed in this application, the differential equation prediction model is a quantitative model trained using machine learning methods based on a large amount of historical data, capable of accurately reflecting the deterioration patterns of specific buildings or similar buildings. Its construction process is as follows:

[0066] The system retrieves multiple sets of long-term monitoring data from databases or historical archives for rural historical buildings with complete service life. Each set of data is a time series, containing environmental data (such as temperature, humidity, and precipitation) and structural inspection data (such as crack depth and weathering area) for the same building at different points in time, as well as the most critical measured value of the building's structural health at that moment. This measured health value is usually derived from periodic manual inspection and rating, precision instrument measurements (such as elastic wave velocity and stiffness tests), or comprehensive indicators evaluated by experts.

[0067] The system uses this historical data as its source material to perform machine learning fitting on the unknown coefficients in the differential equations. Specifically, for each point in time, the system first calculates the environmental degradation factor at that moment. And extract the corresponding crack depth parameters. and facade weathering parameters These are used as an input feature vector. Simultaneously, by calculating the changes in the measured health values ​​at adjacent time points, the rate of change dS(t) / dt of the building's structural health at that moment is obtained and used as the target label for training.

[0068] A machine learning regression algorithm is used to iteratively optimize and train the degradation rate coefficient k and weight coefficients α, β, and γ in the differential equation prediction model. The training objective is to find an optimal set of coefficient combinations that, when a series of continuous inputs are used... , , When data is collected, the predicted curve obtained after numerically solving the differential equation prediction model is the building lifespan decay curve. The error (such as root mean square error RMSE) between the model and the historically measured trajectory of changes in building structural health is less than a preset error threshold, thus ensuring that the model has sufficient prediction accuracy.

[0069] In this embodiment, the machine learning regression algorithm preferably uses the Random Forest Regression algorithm for model training. The specific steps of the training process are as follows:

[0070] 1. Long-term monitoring data obtained from multiple historical buildings will be time-aligned and cleaned. For each time point... ,calculate Environmental degradation factors at all times And obtain the crack depth parameters at the same time. and facade weathering parameters At the same time, based on adjacent time points and , Measured values ​​of building structural health , and Calculated using the central difference method Health change rate at any time This serves as a true label for supervised learning.

[0071] 2. Organize the data at each time point into a sample, and the feature vector at each time point is: , tag as All samples constitute the training set.

[0072] 3. Input the training set into the selected machine learning regression algorithm (using random forest as an example). The task of this algorithm is to learn a mapping function F such that... Let 'a' represent the a-th sample in the dataset (i.e., the a-th time point). Through training, we find a set of optimal model coefficients k, α, β, γ such that the differential equation... The right-hand side function best approximates the actual observed rate of change. During training, cross-validation is used to adjust hyperparameters (such as the number of trees in the random forest, maximum depth, etc.), and joint training is performed on the coefficients k, α, β, and γ in the differential equation model to obtain the final determined coefficient values.

[0073] The loss function used during training is Mean Squared Error (MSE), and the MSE loss function is defined as follows:

[0074] ;

[0075] Where N is the number of training samples; a is the a-th sample in the dataset (i.e., the a-th time point); For a moment The predicted rate of change, representing the change at time t. The corresponding environmental degradation factors Crack depth parameters and facade weathering parameters After inputting the differential equation prediction model, the structural health of the building is calculated by the model. The instantaneous rate of change at this moment; For a moment The measured rate of change, representing the change at time t. The true rate of change in the structural health of a building is calculated based on actual monitoring data (such as periodic inspection results). This is typically measured by comparing adjacent time points (…). and The measured health status values ​​were calculated using numerical differentiation methods (such as the central difference method). The training objective was to minimize the MSE loss function, that is, to make the model's predicted rate of change as close as possible to the actual observed rate of change.

[0076] When training with machine learning regression algorithms (such as random forests), the learning rate is set to 0.05, the batch size to 64, and the number of training epochs to 500. Key hyperparameters can be selected through cross-validation based on the training set size. For example, the number of decision trees in a random forest can be chosen between 100 and 500, and the maximum tree depth can be set to None or no more than 20. The optimizer learning rate during training (e.g., when using gradient boosting trees (GBDT)) can typically be adjusted between 0.01 and 0.1. The batch size can be set to None for full-batch gradient descent and between 32 and 256 for stochastic gradient descent.

[0077] In this embodiment, the preset error threshold refers to the prediction accuracy of the entire building lifespan decay curve. The model training terminates when, on an independent validation set, the model predicts the complete building lifespan decay curve... The root mean square error (RMSE) between the model's predicted structural health value and the actual observed value is less than 0.05 to 0.10 (when the normalized health range is [0,1]). In other words, if the average deviation between the model's predicted structural health value and the actual observed value is controlled within 5% to 10% of the maximum health value, the model is considered to have achieved the required prediction accuracy and can be put into use.

[0078] The differential equation prediction model constructed using the above technical solution can perform personalized modeling for buildings with different materials, environments, and damage states, generating building lifespan degradation curves. This improves the reliability and scientific rigor of the system's predictions of the remaining lifespan and health status of rural historical buildings.

[0079] This application further proposes a crack depth evolution function. Evolution function of weathering degree Based on the construction of a material damage mechanics model, specifically:

[0080] ;

[0081] ;

[0082] in, , Damage coefficient related to building materials; For time Equivalent alternating stress caused by environmental loads; For time The equivalent temperature, taking into account the effect of humidity, is calculated using the following formula: , and They are time Temperature and relative humidity.

[0083] in, (Crack depth damage coefficient) range:

[0084] Masonry / Concrete: ∈ [0.002, 0.005] / (MPa·year) (equivalent alternating stress) Unit: MPa)

[0085] wood: ∈ [0.006, 0.01] / (MPa·year) (wood fibers are more sensitive to breakage).

[0086] Physical meaning: For every 1 MPa·year of accumulated equivalent stress, the crack depth increases. Double the initial value.

[0087] (Weathering Degree Damage Coefficient) Value Range:

[0088] The surface has a protective layer (such as traditional mortar): ∈ [0.001, 0.003] / (℃·%·year) (equivalent temperature T) e 𝑓𝑓 (unit: ℃·%)

[0089] No protective layer: ∈ [0.004, 0.007] / (℃·%·year) (direct exposure to temperature and humidity changes).

[0090] Crack depth evolution function The physical significance lies in the fact that it posits that crack development is not solely determined by the current crack depth parameter. It is not merely a static manifestation, but a dynamic process driven by cumulative fatigue damage. (Integral term) Calculated from the initial time The equivalent alternating stress caused by environmental loads (such as wind loads and temperature stress) up to the current time t. The cumulative amount.

[0091] Among them, equivalent alternating stress The calculation can be simplified to be dominated by the daily temperature difference.

[0092] Specifically, based on the daily maximum temperature of the building's location. and lowest temperature Calculate the daily temperature stress amplitude:

[0093]

[0094] Where E is the elastic modulus of the building material (for example, 10 GPa for blue bricks). The coefficient of linear expansion of the material (e.g., for bricks and stones, it can be taken as...). This calculation equates the daily temperature cycle to an alternating stress amplitude.

[0095] Weathering degree evolution function This study captures the coupling effect between surface weathering processes and the humid and hot environment. The core of the exponential term is time. Equivalent temperature taking into account humidity Time integral ( For temperature, (Relative humidity). Based on the Arrhenius reaction rate theory, higher ambient temperatures ( ) at humidity ( Under the synergistic effect of [various factors], the weathering and deterioration of materials will be accelerated dramatically. Therefore, the proportion of weathered area [is crucial]. The erosion effect on health increases exponentially with the increase of the "humid heat-time" integral.

[0096] For the integral terms in the two evolution functions Discrete calculations are performed using the summation method. Discretizing time in days, the integral is approximated as:

[0097] ;

[0098] in, sky, is the stress or equivalent temperature value on day i.

[0099] By introducing the aforementioned evolution function based on damage mechanics, the differential equation prediction model of this application achieves a key leap from static parameter mapping to dynamic process simulation. It considers not only the current state of the damage (…). , By integrating stress accumulation and equivalent temperature over time, the irreversible driving effect of historical environmental loads on the damage process was quantified. This allows the model to more realistically reflect the essential law of historical buildings "aging in a specific environmental process," improving the accuracy of the building lifespan degradation curve. The accuracy of the predictions provides a solid theoretical basis and quantitative tools for making optimal maintenance decisions over a long service life.

[0100] This application further proposes that, when establishing a multi-objective maintenance optimization model based on building life decay curves and historical maintenance data, it is specifically used for:

[0101] Based on in-depth data mining and structured processing of historical maintenance data, a mapping database of maintenance measures and their effects is constructed. This database records the delaying effect coefficients of different maintenance techniques and materials on specific types of deterioration. The database not only records the building parts targeted in each maintenance, the specific techniques and materials used, but more importantly, it statistically analyzes the actual changes in the building's health after maintenance to derive quantitative delaying effect coefficients for different maintenance schemes on specific types of deterioration. These specific deterioration types include at least cracks, weathering, spalling, and biological erosion. These types of deterioration cover at least the most common problems in rural historical buildings, such as crack propagation, surface weathering, material spalling, and biological erosion. For example, the database explicitly records that for brick wall cracks, the "pressure grouting epoxy resin" scheme has a higher delaying effect coefficient than the "surface grouting and sealing" scheme, but the corresponding cost and level of intervention are also higher.

[0102] The multi-objective maintenance optimization model constructs a multi-objective optimization function based on delaying the degradation rate and maintenance cost, and outputs the optimal maintenance recommendation for the current risk level.

[0103] Multi-objective optimization functions aim to simultaneously optimize two key, often conflicting, objectives: one is the technical objective, which is to maximize the effect of maintenance measures on delaying the building's lifespan decline curve (specifically, to improve the predicted lifespan curve after maintenance). The objectives are: first, to minimize the impact of the maintenance on the service life; and second, to minimize the estimated total cost of the maintenance intervention (including the costs of materials, labor, and equipment).

[0104] The input to the optimization process is the building's current risk level (derived from the building's lifespan decay curve). The system uses a database of current slope or remaining life (as determined by the slope) and maintenance measures and their effects. An optimization algorithm (such as a multi-objective evolutionary algorithm like NSGA-II) searches through all feasible combinations of maintenance options, evaluating each option's position in the "effect-cost" coordinate space. Ultimately, the system provides a set of Pareto optimal solutions, each representing the best technical effect achievable under a given cost constraint, or the lowest cost for the desired technical effect. Decision-makers can then weigh these options and select the optimal maintenance recommendation that best suits their current budget and maintenance strategy.

[0105] Through the aforementioned technical solution, this application transforms the previous qualitative maintenance decision-making, which relied on expert experience, into decision-making based on historical big data and quantitative optimization algorithms. For each building's specific damage state and risk level, the most cost-effective maintenance plan is derived, thereby improving the efficiency of limited maintenance funds while ensuring structural safety and effectively delaying deterioration.

[0106] This application further proposes that the multi-objective maintenance optimization model is solved using a multi-objective optimization function. Represented as:

[0107] ;

[0108] ;

[0109] ;

[0110] in, Indicates the overall degradation rate. The total maintenance cost is represented by i, which represents the i-th type of deterioration or deterioration (e.g., cracks in the load-bearing wall on the east side, peeling off the roof on the west side), and j represents the j-th type of maintenance process or material (e.g., grouting cubic meters, surface treatment square meters). The weighting coefficients for each deteriorated part are determined based on its importance to the overall structural safety of the building (for example, the weight of load-bearing structures is much higher than that of decorative components). The effect of repair scheme x on delaying the i-th type of degradation. Let j be the unit cost of the repair process. The quantity of work required for repair scheme x using the j-th process. Decision variable: Repair scheme (decision variable) x represents a specific repair scheme (e.g., pressure grouting for cracks in the east gable wall, cleaning and water-repellent treatment for weathering on the north wall).

[0111] This represents the predicted degradation rate of type i after the implementation of maintenance plan x. The rate is calculated by querying the maintenance measure-effect mapping database to obtain the mitigation effect coefficient of maintenance plan x, and then substituting it into the updated differential equation prediction model. Minimization This means choosing the solution that can most effectively slow down the decline in the overall health of the building.

[0112] Weighting coefficients for each deteriorated part Example as follows:

[0113] Degradation type Security weight Cultural value weight comprehensive Cracks in load-bearing walls 0.6 0.2 0.45 Facade weathering 0.2 0.5 0.35 Peeling of non-load-bearing components 0.1 0.2 0.15 Biological erosion 0.1 0.1 0.05

[0114] exist In the middle, through Minimizing the overall degradation rate maximizes the technical effectiveness of the maintenance plan by minimizing... The total maintenance cost and the economic efficiency of the control scheme.

[0115] The system utilizes multi-objective optimization algorithms (such as NSGA-II) to solve a multi-objective maintenance optimization model. It searches among several possible maintenance solutions x using a multi-objective optimization function, ultimately outputting a Pareto optimal solution set. Each solution in the Pareto optimal solution set represents a feasible solution that is technically optimal under a specific cost budget, or lowest cost under a specific technical objective. Finally, the system combines this with the current building's risk level (derived from the building's lifespan decay curve). (Determined by the slope), several top-priority solutions are recommended from the Pareto optimal solution set to form the final optimal maintenance recommendation for maintenance personnel to make decisions.

[0116] This application transforms complex engineering decisions into a computable optimization problem. It comprehensively considers the differentiated impact of maintenance plans on the deterioration process of different parts and types of buildings, and quantitatively weighs this impact against actual economic costs within the same framework, thus changing the traditional, experience-based, extensive decision-making model. It can balance technical effectiveness and economic costs while ensuring the core objectives of building structural safety and effectively delaying deterioration, providing a decision-making tool that balances scientific rigor, economic efficiency, and precision for the preservation of rural historical buildings where funding is often limited.

[0117] like Figure 3 As shown, this application further proposes that the big data-based rural historical building protection assessment system also includes a model update module for performing adaptive correction and updating of the differential equation prediction model, specifically including:

[0118] The module initiates the multi-model candidate set initialization process. Based on historical maintenance data, it trains multiple candidate instances of differential equation prediction models by setting different initial values ​​for the deterioration rate coefficient k or weight coefficients α, β, and γ.

[0119] Among them, multiple candidate prediction model instances are derived through one or more of the following methods:

[0120] 1. Based on the optimal coefficients (k, α, β, γ) obtained from training the differential equation prediction model, apply a small random perturbation (e.g., ±5%) to generate a set of model variants with slightly different parameters.

[0121] 2. From the complete historical training dataset, multiple different sub-training sets are extracted using resampling techniques such as Bootstrap, and these are used to train multiple model instances to simulate the local variability of the data distribution.

[0122] 3. Train using different combinations of input features (e.g., using only environmental data, using only structure detection data, or using all data) to obtain candidate models with different sensitivities to various types of data.

[0123] Each candidate instance represents a slightly different mathematical description of the building degradation mechanism, designed to cover the prediction needs of different building types (such as masonry, timber, rammed earth) or different dominant degradation modes (such as weathering or structural cracking).

[0124] Entering the periodic verification and evaluation phase, at the end of each fixed evaluation cycle (such as each quarter or each half year), new environmental data and structural test data are periodically input into the differential equation prediction model of each candidate instance to obtain short-term prediction results, and to obtain the predicted values ​​of crack depth parameters and facade weathering parameters at the end of the next cycle.

[0125] Once the new detection cycle is completed and the actual detection data is available, the module initiates a prediction accuracy assessment, comparing the short-term prediction results with the crack depth parameters and facade weathering parameters detected in the next actual detection, and calculating the prediction error for each candidate instance (usually using indicators such as root mean square error RMSE).

[0126] Based on the prediction error, the weights of the differential equation prediction models for each candidate instance in the final prediction result are dynamically adjusted, or the differential equation prediction model with the smallest prediction error is selected as the currently used differential equation prediction model. Specifically, two strategies are employed:

[0127] Weighted fusion strategy: The contribution weight of each candidate instance in the final prediction result is dynamically adjusted based on their prediction error. Instances with smaller prediction errors have higher weights. The final system prediction result is a weighted average of the prediction values ​​of all candidate instances.

[0128] Optimal instance selection strategy: Directly select the candidate instance with the smallest prediction error as the official differential equation prediction model currently used by the system in the next evaluation period.

[0129] This application achieves continuous learning and self-optimization capabilities through the aforementioned technical solution, effectively overcoming the problem of decreased prediction accuracy caused by initial model parameter setting deviations, the special characteristics of building materials, or the long-term performance evolution of buildings, thus ensuring the building lifespan degradation curve. The system's long-term reliability has been enhanced, improving its applicability and predictive robustness across different rural historical buildings and providing core technical support for achieving precise and long-term protection.

[0130] like Figure 4As shown, this application further proposes that, in order to improve the robustness and reliability of the system evaluation results and overcome the randomness and uncertainty that may exist in a single data source or a single evaluation indicator, the big data-based rural historical building protection evaluation system also includes a multi-source data fusion module, specifically used for:

[0131] Environmental degradation factor Crack depth parameters and facade weathering parameters Multiple degradation levels are defined, and environmental degradation factors are identified. Crack depth parameters and facade weathering parameters Each parameter in the equation constructs a fuzzy membership function belonging to each degradation level;

[0132] Input the current value of each parameter into its corresponding fuzzy membership function to calculate the membership degree of each parameter to each degradation level;

[0133] After normalizing the membership degree, it is directly assigned or proportionally allocated to the basic probability assignment function of the corresponding proposition in the DS evidence theory.

[0134] Applying the synthesis rules of the DS evidence theory, we can analyze the evidence from environmental degradation factors. Crack depth parameters and facade weathering parameters The evidence is synthesized using the basic probability assignment function to obtain a comprehensive basic probability assignment.

[0135] Extract a comprehensive building degradation index for a pre-defined building that is in a state of overall degradation from the comprehensive basic probability distribution.

[0136] By using the comprehensive building deterioration index as a weighting factor and combining it with the deterioration rate coefficient k in the differential equation prediction model, the decay rate of the building life decay curve L(t) is corrected.

[0137] The multi-source data fusion module has three core input parameters: environmental degradation factor. Crack depth parameters and facade weathering parameters This involves defining multiple explicit degradation levels (e.g., "slight," "moderate," "severe"). For each parameter, a fuzzy membership function (typically a trapezoidal or trigonometric function) is constructed to assign its numerical value to each level. These functions transform precise numerical inputs into fuzzy linguistic descriptions, such as a specific... The value may simultaneously belong to the "moderate" level with a membership degree of 0.7 and the "severe" level with a membership degree of 0.3, thus effectively characterizing the inherent uncertainty in the assessment process.

[0138] The proposition set corresponding to the presupposed DS evidence theory can be defined as: {slight degradation, moderate degradation, severe degradation}. The specific implementation method for constructing the Basic Probability Assignment Function (BPA) for each parameter is as follows: Assume the current crack depth parameter... The measured value is 2.1 mm. Calculations using its fuzzy membership function show a membership degree of 0.7 for "moderate degradation" and 0.3 for "severe degradation." After normalizing this membership vector, it can be directly used as the BPA, i.e.: Unassigned probability mass This can be considered as uncertainty. Similarly, after obtaining the BPA of environmental and weathering parameters, Dempster's synthesis rules are applied to fuse the evidence. For example, if the environmental evidence also strongly supports "moderate degradation," the synthesized result will assign a higher confidence level to "moderate degradation."

[0139] Subsequently, the module inputs the current measured values ​​of each parameter into its corresponding fuzzy membership function, calculating the membership degree of each value to each predefined degradation level. This step realizes the conversion from precise values ​​to fuzzy confidence.

[0140] The multi-source data fusion module normalizes these membership degrees, making their sum equal to 1. The normalized results are directly assigned or proportionally allocated to the basic probability assignment function in the DS evidence theory. Assessment information from the environment, cracks, and weathering is treated as an independent "evidence body," and each evidence body provides a different degree of support for the proposition "what is the overall state of deterioration of the building?"

[0141] The multi-source data fusion module applies the synthesis rules of DS evidence theory to fuse the basic probability assignment functions from three independent evidence sources. The synthesis rules effectively reconcile potential conflicts between different pieces of evidence and strengthen their consistent aspects, ultimately outputting a comprehensive basic probability assignment. This comprehensive result reflects a more reliable confidence distribution resulting from the combined effect of all evidence.

[0142] The multi-source data fusion module extracts the confidence level of "the building as a whole is in a deteriorating state" from the integrated basic probability distribution after fusion, and quantifies it into a scalar value - the building comprehensive deterioration index.

[0143] Finally, the multi-source data fusion module uses the derived comprehensive building deterioration index as a dynamic weighting factor, combining it with the inherent deterioration rate coefficient k in the differential equation prediction model, thereby affecting the building life decay curve calculated by the original model. The system dynamically adjusts the rate of decay in real time. When comprehensive evidence indicates a high degree of overall building deterioration, the system dynamically accelerates the decay rate of the predicted curve, and vice versa.

[0144] Specifically, the overall building deterioration index is combined with the deterioration rate coefficient k, and the decay rate of the building life decay curve L(t) is corrected through the following mathematical form:

[0145] Let the overall degradation index be (Its value range is usually [0,1]), then the corrected degradation rate coefficient for:

[0146] ;

[0147] in, This is an adjustable gain factor (typically set from 0.5 to 2.0) used to control the degree to which the overall degradation affects the rate of decay. Substitute the values ​​into the differential equation prediction model and solve for the corrected lifetime decay curve.

[0148] This application upgrades the system from relying on a single model calculation to intelligent assessment based on multi-evidence joint decision-making by introducing a multi-source data fusion module. Through mathematical methods (fuzzy sets and evidence theory), it simulates the thought process of human experts synthesizing information from multiple sources to make comprehensive judgments, enhancing the system's fault tolerance and decision reliability when faced with incomplete, uncertain, or slightly conflicting monitoring data. The final output of the building's comprehensive deterioration index and the corrected lifespan curve incorporates more comprehensive information, making the assessment conclusions more scientific, accurate, and reliable, providing a more solid basis for subsequent maintenance decisions.

[0149] like Figure 5 As shown, this application further proposes that the big data-based rural historical building protection assessment system also includes an environmental prediction module. By predicting future environmental loads, it enables the advance judgment of the building's lifespan decline trend, thereby supporting more strategic preventive maintenance planning and generating environmental degradation factors for future time series. Specifically, it includes:

[0150] Obtain medium- and long-term weather forecast data for the building's location from authoritative meteorological service agencies;

[0151] Meteorological forecast data is input into a pre-trained time-series forecasting model (e.g., a model based on a Long Short-Term Memory (LSTM) network or a Transformer architecture), which outputs a sequence of predicted environmental parameters for a specific future time interval. The predicted environmental parameters include at least temperature, humidity, and precipitation. The pre-trained time-series forecasting model can deeply analyze the complex temporal relationship between the historical variation patterns of parameters such as temperature, humidity, and precipitation and future forecasts, and output a more refined sequence of predicted environmental parameters for a specific future time interval (e.g., the next 30 days or the next quarter), which includes at least the predicted temperature T(t), predicted humidity H(t), and predicted precipitation P(t) for each day of the future.

[0152] The environmental parameter sequences are weighted, fused, and normalized to generate environmental degradation factors for future time series. The calculation formula is:

[0153] ;

[0154] in These are the normalized temperature, humidity, and precipitation, respectively. These are the weighting coefficients for the normalized temperature, humidity, and precipitation. The values ​​are preset based on the sensitivity and degree of damage of various environmental factors to the main building materials, with example values ​​such as 0.4, 0.3, and 0.3.

[0155] The decision module is also used to predict environmental degradation factors. Input a differential equation prediction model to generate future building lifespan decay curves. The model will no longer rely solely on current or historical environmental data for estimations, but will instead solve for future building lifespan degradation curves by predicting environmental load paths. This demonstrates the structural health of buildings under anticipated future weather conditions. The possible evolutionary trajectory.

[0156] This application enables early prediction of risks such as "the upcoming wet rainy season will accelerate wall weathering" or "continuous freeze-thaw cycles will adversely affect crack development." This allows maintenance personnel to take intervention measures before severe weather occurs or in the early stages of accelerated building performance degradation, shifting from passive response to proactive protection. This enhances the foresight and scientific nature of historical building maintenance work, achieving a fundamental shift from corrective maintenance based on the current state to preventive maintenance based on future predictions.

[0157] The following is a specific example of a big data-based assessment system for the protection of historical buildings in rural areas.

[0158] A Qing Dynasty brick-and-wood house (brick and wood structure, with brick walls and wooden beams and columns) in an ancient village in southern China has developed cracks and surface weathering on its east gable wall due to the constant hot and humid climate. To develop a preventative protection plan, the local cultural heritage department has implemented a big data-based assessment system for the protection of rural historical buildings, conducting full-cycle health monitoring and optimizing maintenance strategies.

[0159] The system acquires data through three channels: Environmental data: IoT sensors deployed around the building (sampling frequency 1 time / hour) collect data on temperature (T), relative humidity (H), precipitation (P), and sulfur dioxide concentration in the air over the past year. For example, the highest temperature in the summer of 2024 was 38℃, the humidity during the rainy season reached 90%, and the annual precipitation was 1200mm.

[0160] Structural inspection data: A portable acoustic wave detector (frequency 10-50kHz) was used to detect cracks in the east gable wall. After emitting sound waves, the echo signals were received, and the amplitude attenuation characteristics and time delay features were extracted. The east facade was photographed by a drone equipped with a high-definition camera (50 megapixels), and compared with historical images from one year ago to generate a pixel-level grayscale change map.

[0161] Historical maintenance data: The building’s maintenance records for the past 20 years were retrieved from the digital archive, including structured data such as the 2010 west wall grouting (material: traditional glutinous rice mortar, cost 800 yuan) and the 2018 anti-corrosion treatment of the wooden beams (process: tung oil coating, effect lasts for 5 years).

[0162] The data processing module normalizes temperature, humidity, and precipitation (min-max normalization), and then weights and fuses them to generate E(t):

[0163] Temperature normalization: (Assuming the historical temperature range is 0-40℃);

[0164] Humidity normalization: (Assuming humidity range is 30%-95%)

[0165] Precipitation normalization: (Assuming annual precipitation ranges from 0 to 1500 mm);

[0166] Weighted fusion is used to calculate the environmental degradation factor E(t): (weight) ).

[0167] The acoustic echo signal showed an amplitude attenuation rate of 62% and a time delay of 0.08ms. It achieved a 91% match with the feature "blue brick - crack depth 2mm" in the sample library. The regression algorithm fitted the crack depth parameters. .

[0168] After registering historical and current images, the grayscale change rate of 100 areas on the east facade was calculated. Among them, 12 areas had a change rate exceeding the dynamic threshold of 15% for blue bricks (average change rate 18%). The proportion of weathered area was also calculated. .

[0169] Determine the degradation rate coefficient of masonry structures in the core equation of the differential equation prediction model. , weight α=0.5, β=0.3, γ=0.2;

[0170] Calculate equivalent alternating stress (Temperature difference ΔT = 28℃, elastic modulus E = 10 GPa, coefficient of linear expansion) Cumulative stress integral ; Obtain the crack evolution function

[0171] Calculate equivalent temperature ;get Final weathering evolution function (℃·%·year);

[0172] The initial health score S(0) = 1.0 is substituted into the differential equation prediction model: .

[0173] Health after 1 year Generate lifetime decay curve .

[0174] The decision-making module constructs a "maintenance measures-effects" database based on historical maintenance data, and establishes a multi-objective optimization model for the current risk level (annual health decay of 0.0218). Calculate the total degradation rate Among them, the weight of cracks in load-bearing walls The delaying effect of repair plan x (Pressure grouting); Calculate total maintenance cost Unit cost of pressure grouting Work volume The total cost is 1500 yuan;

[0175] By solving for the Pareto optimal solution using the NSGA-II algorithm, the recommended approach is "local pressure grouting + surface hydrophobic treatment," which reduces the cost of traditional full-scale repair by 40% and decreases the rate of health degradation to 0.012 / year.

[0176] Before the system was deployed, the preservation of this residence relied on manual inspections every three years. Previously, failure to detect cracks spreading during the rainy season led to a 30% overrun in repair costs. After the system was implemented, IoT sensors captured sudden changes in temperature and humidity in real time, drone imagery accurately located 7.2% of the weathered area, and a differential equation model predicted critical health values ​​18 months in advance. The decision-making module, combining historical maintenance data, found a balance between "pressure grouting (high cost, high efficiency)" and "surface grouting (low cost, low efficiency)." The final localized intervention not only extended the estimated remaining lifespan to 25 years but also avoided the risks of traditional "large-scale repairs that damage the original appearance of the building."

[0177] This embodiment recreates the entire process of the system from data collection and dynamic prediction to decision optimization through a specific scenario, verifying its accurate support capability for the protection of rural historical buildings in complex environments, and providing a reusable technical paradigm for the preventive protection of similar buildings.

[0178] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A big data-based assessment system for the protection of rural historical buildings, characterized by: include: The data acquisition module is used to acquire environmental data, structural inspection data, and historical maintenance data of the building's location; The data processing module is used to normalize the environmental data and generate environmental degradation factors. The crack depth parameter is extracted from the structural detection data using a feature extraction algorithm. and facade weathering parameters The environmental degradation factors The crack depth parameter and the aforementioned facade weathering parameters The data is input into a pre-constructed differential equation prediction model, which is then solved using a numerical iterative method to output the building structural health status. The sequence data varies with time t, and a building lifespan decay curve is generated based on the sequence data. The differential equation prediction model is defined as follows: ； in, To use the crack depth parameter The crack depth evolution function is defined by the initial conditions. Based on the aforementioned facade weathering parameters Let be the weathering degree evolution function under initial conditions; k is the deterioration rate coefficient. The corresponding weight coefficients, and satisfying ; The crack depth evolution function and the weathering degree evolution function Based on the construction of a material damage mechanics model, specifically: ； ； in, , Damage coefficient related to building materials; For time Equivalent alternating stress caused by environmental loads; For time The equivalent temperature, taking into account the effect of humidity, is calculated using the following formula: , and They are time Temperature and relative humidity; The decision module is used to establish a multi-objective maintenance optimization model based on the building life decay curve and the historical maintenance data, and output the optimal maintenance suggestions through simulation and solution. The environmental prediction module is used to generate environmental degradation factors for future time series. Specifically, it includes: Obtain medium- and long-term meteorological forecast data for the location of the building; The meteorological forecast data is input into a pre-trained time series prediction model, which outputs a sequence of predicted environmental parameters for a specific future time interval. The predicted environmental parameters include at least temperature, humidity, and precipitation. The environmental parameter sequences are weighted, fused, and normalized to generate environmental degradation factors for future time series. The calculation formula is: ； in These are the normalized temperature, humidity, and precipitation, respectively. These are the corresponding weighting coefficients; The decision module is also used to incorporate the predicted environmental degradation factors. Input the differential equation prediction model to generate future building lifespan decay curves. .

2. The rural historical building protection assessment system based on big data as described in claim 1, characterized in that: Crack depth parameters are extracted from the structural inspection data using a feature extraction algorithm. and facade weathering parameters In the context, the crack depth parameter The crack depth is extracted using a crack depth fitting method based on the acoustic echo attenuation curve. This crack depth fitting method specifically includes: The system uses acoustic sensors to emit sound waves of a specific frequency into the wall and receives the echo signals. The amplitude attenuation characteristics and time delay features of the echo signal are extracted. These characteristics are then matched with a pre-established acoustic feature sample library based on different materials and crack depths to obtain matching results. The crack depth parameters are then calculated using a regression algorithm to fit the matching results. .

3. The rural historical building protection assessment system based on big data as described in claim 1, characterized in that: Crack depth parameters are extracted from the structural inspection data using a feature extraction algorithm. and facade weathering parameters In the context, the facade weathering parameters The surface weathering region is extracted using a method based on image grayscale change thresholds. The specific methods for identifying surface weathering regions include: The historical and current images of the building's location on the same facade are acquired, and the historical and current images are registered and compared at the pixel level to generate multiple regional grayscale values. The rate of change of grayscale values ​​in each area is calculated. When the rate of change exceeds a preset dynamic threshold based on the building material type, the building facade area where the building is located is determined to be a weathered area. The area percentage of the weathered area is then calculated and used as the facade weathering parameter. .

4. The rural historical building protection assessment system based on big data as described in claim 1, characterized in that: The differential equation prediction model is constructed in the following way: Multiple sets of long-term monitoring data of rural historical buildings during their service life were obtained as a training sample set. The long-term monitoring data included time-series environmental data, structural inspection data, and corresponding measured values ​​of building structural health. With the aforementioned environmental degradation factors The crack depth parameter and the aforementioned facade weathering parameters Using the rate of change in building structural health as the input feature and the rate of change as the output label, a machine learning regression algorithm is used to analyze the coefficients k in the differential equation prediction model. The training and fitting process continues until the differential equation prediction model accurately predicts the building lifespan decay curve. The error between the measured value of the building structure health and the actual change trajectory is less than a preset error threshold.

5. The rural historical building protection assessment system based on big data as described in claim 1, characterized in that: When establishing a multi-objective maintenance optimization model based on the building life decay curve and the historical maintenance data, the specific steps are as follows: Based on the historical maintenance data, a mapping database between maintenance measures and maintenance effects is constructed, and the delay effect coefficients of different maintenance processes and materials on specific types of degradation are recorded; the specific types of degradation include at least cracks, weathering, spalling, and biological erosion. The multi-objective maintenance optimization model constructs a multi-objective optimization function based on delaying the deterioration rate and maintenance cost, and outputs the optimal maintenance recommendation for the current risk level.

6. The rural historical building protection assessment system based on big data according to claim 5, characterized in that: The multi-objective maintenance optimization model is solved using a multi-objective optimization function. Represented as: ； ； ； in, Indicates the overall degradation rate. This indicates the total maintenance cost. These are the weighting coefficients for each deteriorated part. The effect of repair scheme x on delaying the i-th type of degradation. Let j be the unit cost of the repair process. The workload for using the j-th process in maintenance scheme x.

7. The rural historical building protection assessment system based on big data as described in claim 1, characterized in that: The big data-based assessment system for the protection of rural historical buildings also includes a model update module, used to perform adaptive correction and updating of the differential equation prediction model, specifically including: Based on the historical maintenance data, by setting different initial values ​​for the deterioration rate coefficient k or the weight coefficients α, β, and γ, multiple candidate instances of the differential equation prediction model are trained. New environmental data and structural inspection data are periodically input into the differential equation prediction model of each candidate instance to obtain short-term prediction results; The short-term prediction results are compared with the crack depth parameters and facade weathering parameters detected in the next actual test, and the prediction error of each candidate instance is calculated. Based on the prediction error, the weights of the differential equation prediction models of each candidate instance in the final prediction result are dynamically adjusted, or the differential equation prediction model with the smallest prediction error is selected as the currently used differential equation prediction model.

8. The rural historical building protection assessment system based on big data according to claim 1, characterized in that: The big data-based assessment system for the protection of rural historical buildings also includes a multi-source data fusion module, specifically used for: The environmental degradation factor The crack depth parameter and the aforementioned facade weathering parameters Multiple degradation levels are defined, and these are referred to as environmental degradation factors. The crack depth parameter and the aforementioned facade weathering parameters Each parameter in the equation constructs a fuzzy membership function belonging to each degradation level; Input the current value of each parameter into its corresponding fuzzy membership function to calculate the membership degree of each parameter to each degradation level; After normalizing the membership degree, it is directly assigned or proportionally allocated to the basic probability allocation function of the corresponding proposition in the DS evidence theory. Applying the synthesis rules of the DS evidence theory, the environmental degradation factors are respectively analyzed. The crack depth parameter and the aforementioned facade weathering parameters The evidence is synthesized using the basic probability assignment function to obtain a comprehensive basic probability assignment. Extract the overall building degradation index for a pre-defined building that is in a state of overall degradation from the basic probability distribution of the comprehensive analysis. The comprehensive building deterioration index is used as a weighting factor and combined with the deterioration rate coefficient k in the differential equation prediction model to correct the decay rate of the building life decay curve L(t).

Images