CN-121999900-A - Ancient architecture digital intelligent knowledge base system

CN121999900ACN 121999900 ACN121999900 ACN 121999900ACN-121999900-A

Abstract

The invention belongs to the technical field of ancient architecture digitization, and provides an ancient architecture digitization intelligent knowledge base system which comprises a repair color deviation decomposition module, a formula adjustment dynamic calibration module, a multivariate dynamic optimization module, a knowledge base iteration multiplexing module, a data base integration module, a long-term correction model coefficient, a cosine similarity multiplexing formula, a system improvement repair precision and efficiency and color restoration consistency guarantee, wherein the repair color deviation decomposition module collects pigment purity/impurity data by using a spectrometer, calculates deviation and color value correction to obtain color repair structure data, the formula adjustment dynamic calibration module fits deviation-formula relation coefficient to build a basic model, and combines the structure data to calibrate a proprietary formula adjustment, the multivariate dynamic optimization module establishes a mixed color value prediction model and a multi-objective optimization function, calculates an optimal formula proportion by a genetic algorithm, the optimal formula is verified and calibrated reversely, and the knowledge base iteration multiplexing module integrates the data base, corrects the model coefficient for a long term, and the new batch multiplexes the formulas by cosine similarity.

Inventors

YUAN CHANG
QIAO YU
HAN HUI
ZHANG YUQI
KANG YUHAO
Sui Yuhan
GAO HUA
LIU FENG
YANG TIANHUA
WAN SI
GAO YUE
Ye Maoshen
Zhao Gengchao
KONG XIANGPENG

Assignees

北京首华建设经营有限公司

Dates

Publication Date: 20260508
Application Date: 20251215

Claims (10)

1. The ancient architecture digital intelligent knowledge base system is characterized by comprising: The repair color deviation decomposition module calculates deviation data of pigment ores based on the acquired data and standard data of the pigment ores, fits to obtain color space coefficients, calculates color value correction based on the color space coefficients, calculates repair color deviation through the color value correction, and integrates to obtain current batch color repair structure data; The formula adjustment dynamic calibration module is used for fitting relation coefficients of deviation data and pigment ores, constructing a basic model, and dynamically calibrating the basic model based on color restoration structure data to obtain the special formula adjustment quantity of the pigment ores in the current batch; the multivariate dynamic optimization module is used for constructing a mixed color value prediction model and a multi-objective optimization function, and obtaining the proportion of the exclusive optimal formula through solving by a genetic algorithm; The on-site verification and calibration module is used for verifying the proportion of the exclusive optimal formula and reversely calibrating the basic model; And the knowledge base iteration multiplexing module integrates the collected data and the exclusive optimal formula proportion to form a data set, stores the data set into the ancient building repair pigment formula knowledge base, and corrects the basic model coefficient for a long time based on the data set for rapidly multiplexing the ancient building repair pigment formula.
2. The ancient architecture digital intelligent knowledge base system according to claim 1, wherein the specific process of calculating the color value correction based on the color space coefficient is as follows: Sampling each batch of pigment ore for repairing by using an X-ray fluorescence spectrometer and a near infrared spectrometer, collecting the measurement purity of the pigment ore and the measurement impurity of the pigment ore, obtaining the standard purity of the pigment ore and the standard impurity of the pigment ore, and calculating to obtain the purity deviation of the pigment ore and the impurity deviation of the pigment ore; Fitting to obtain color space coefficients based on the purity deviation of the historical pigment ore and the impurity deviation of the historical pigment ore, wherein the color space coefficients comprise the purity color value deviation coefficients 、、 Coefficient of deviation from impurity color value ; Calculating color value correction according to the color space coefficient obtained by fitting, wherein the color value correction comprises color value correction caused by purity deviation and color value correction caused by impurity deviation; Based on purity deviation Combining the purity color value deviation coefficient 、、 Calculating a color value correction amount due to the purity deviation: ; Based on impurity deviation Combining the impurity color value deviation coefficient Calculating a color value correction amount due to impurity deviation: 。
3. the ancient architecture digital intelligent knowledge base system according to claim 1, wherein the concrete process of calculating the correction color deviation by the color value correction amount is as follows: preparing pigment ore and cementing agent into wet sample by using a spectrocolorimeter under standard light source according to standard process, and collecting CIELAB color value, i.e. actual color value, of the wet sample 、、 Standard colour value corresponding to formula of ancient building digital intelligent knowledge base system 、、 ; Calculating the restoration color deviation of the actual color value and the standard color value: , , The repair color deviation can be decomposed into a purity deviation contribution and an impurity deviation contribution, namely: , 。
4. the ancient architecture digital intelligent knowledge base system according to claim 1, wherein the fitting deviation data and the relation coefficient of pigment ore, and the specific process of constructing the basic model are as follows: Dividing pigment ore into different pigment ore types, wherein the pigment ore types are defined by a pigment type m and a vein region n together, and the purity deviation, impurity deviation and formula adjustment quantity of the pigment ore under each pigment ore type are correspondingly related to each pigment ore type; Preparing a purity gradient sample of pigment ore type and an impurity gradient sample, mixing the purity gradient sample and the impurity gradient sample according to the standard formula proportion, and determining a final color value; If the final color value deviates from the target color value, adjusting the formula proportion of the pigment type m until the final color value is matched with the target color value, and recording the formula adjustment amount required for maintaining the target color value; performing linear fitting on the experimental data to obtain the relation coefficient of purity deviation and formula adjustment quantity under the pigment type m and the ore vein region n And the relation coefficient between impurity deviation and formula adjustment quantity ; Constructing a basic model: , wherein, For deviations in purity of pigment types, when the purity is reduced , For impurity deviation of pigment type, when the impurity is increased , The proportion of the formula corresponding to the purity under unit deviation is adjusted for the pigment m and the ore vein n.
5. The ancient architecture digital intelligent knowledge base system according to claim 1, wherein the specific process of obtaining the specific formula adjustment amount of the pigment ore of the current batch is as follows: Acquiring the measured purity deviation of the current batch and the measured impurity deviation of the current batch in the color restoration structure data, inputting the measured purity deviation of the current batch and the measured impurity deviation of the current batch into a basic model, acquiring the purity deviation, the impurity deviation and the formula adjustment quantity corresponding to the pigment ore type corresponding to the current batch, and calculating the historical average purity deviation Historical average impurity bias ; Introducing calibration coefficients alpha and beta, and for basic model coefficients 、 Carrying out dynamic correction, wherein the coefficient after correction is as follows: ; wherein, the method comprises the steps of, Avoiding the denominator being zero, For the current batch purity deviation, the process is performed, Impurity deviation for the current batch; coefficient after correction 、 Substituting the formula into a basic model to obtain the adjustment quantity of the special formula of the pigment ore in the current batch: 。
6. The ancient architecture digital intelligent knowledge base system according to claim 1, wherein the specific process of obtaining the exclusive optimal formula proportion by solving through genetic algorithm is as follows: Constructing a mixed color value prediction model: Target color value according to ancient architecture color restoration Adopting a CIEDE2000 color difference formula to require the color difference between the mixed color and the target color The formula comprises x pigment types, and the proportion of the ith pigment type is a decision variable The method comprises the following steps: Wherein Is the minimum proportion of the process; actual color value of current batch based on color repair structure data ) Linear superposition yields the predicted CIELAB color values for the mixed colors: ; multi-objective optimization function construction: predicted CIELAB color values and target color values in a mixed color The minimum DE2000 chromatic aberration is taken as a target, an optimization objective function is constructed, and a CIEDE2000 chromatic aberration formula is as follows: Wherein: In order for the brightness deviation to be a function of, For chrominance deviations, in which 、 , Is hue deviation, wherein , By means of trigonometric function derivation, , , Is a weight factor and Can be adjusted according to the visual sensitivity, , , Is the correction factor of brightness, chroma and hue, Is a hue rotation factor; optimization objective is minimization The method comprises the following steps: and solving the exclusive optimal formula proportion meeting the constraint condition by adopting a genetic algorithm.
7. The ancient architecture digital intelligent knowledge base system according to claim 1, wherein the specific process of the reverse calibration basic model is as follows: According to standard process, according to the specific optimal formula proportion, making on-site verification hand sample, using portable spectrocolorimeter, collecting CIELAB color value of hand sample under standard light source Calculating the small sample color value and the target color value by adopting CIEDE2000 color difference formula Is of the color difference of (2) If (if) Judging the passing of the field verification sample, if Entering a basic model coefficient correction flow; Let the adjustment of the original formula be Let the actual needed formula adjustment of the field sample be Model coefficient correction amount The method meets the following conditions: , wherein, For the purity deviation of the current batch of pigment, Impurity deviation for the current batch of pigment; solving by linear least square method And The square sum of errors is minimized, the correction amount is obtained by solving, and after correction, the new model coefficient is as follows: 。
8. The system of claim 1, wherein the integrated collection data and the dedicated optimal formula ratio form a data set, and the specific process of storing the data set in the ancient building repair pigment formula knowledge base is as follows: The collected data of each batch and the exclusive optimal formula proportion form a ancient building repair pigment formula data set, wherein the collected data comprise the measurement purity of pigment ores and the measurement impurities of the pigment ores; The collected data of all batches are classified with the proprietary optimal formula proportion towards the generation D-region R-pigment type M, a data set F is constructed, and the data set F is stored in an ancient building repair pigment formula knowledge base.
9. The ancient architecture digital intelligent knowledge base system according to claim 1, wherein the specific process of correcting the basic model coefficient for a long time based on the data set is as follows: Based on the data set F, the coefficients of the linear regression optimization model are adopted, the optimal coefficients are solved through a least square method, and each time X batches of effective data are added, the coefficients of the basic model are optimized, and the precision of the basic model is improved for a long time.
10. The ancient architecture digital intelligent knowledge base system according to claim 1, wherein the specific process for rapidly multiplexing the ancient architecture repair pigment formula is as follows: And for the new batch of pigments, obtaining the measurement purity of pigment ores of the new batch of pigments and the measurement impurities of the pigment ores, calculating the matching degree with the historical samples through cosine similarity, and screening the high-matching degree sample multiplexing repair rule.

Description

Ancient architecture digital intelligent knowledge base system Technical Field The invention belongs to the technical field of ancient architecture digitization, and particularly relates to an ancient architecture digitization intelligent knowledge base system. Background In the current ancient building restoration field, pigment formula debugging depends on subjective experience of craftsmen for a long time, and a scientific quantification and standardization system support is lacking. On the one hand, pigment ore has obvious fluctuation of purity and impurity content due to the differences of ore vein areas and mining batches, so that the color value of the pigment actually prepared is easy to deviate from the restoration target color value, the traditional method cannot accurately disassemble the influence of the purity deviation and the impurity deviation on the color value, the formula is difficult to adjust in a targeted manner, and the problems of insufficient color reduction degree, obvious color difference and the like often occur; on the other hand, the existing ancient-building digital intelligent color management system generally generates a repair formula based on preset standard pigment parameters, but does not consider batch differences of traditional mineral pigments-the system cannot detect the core characteristics of purity, impurities and the like of actual pigments in each batch in real time, and a mechanism for dynamically adjusting the formula according to the actual pigment characteristics is also lacking. Therefore, when the actual pigment deviates from the standard pigment preset by the system, the pigment color prepared according to the system formula can generate obvious color difference with the original color of ancient architecture or the expected repair color, and the historical reality and visual consistency of repair are seriously affected. Therefore, the invention provides a ancient digital intelligent knowledge base system. Disclosure of Invention In order to overcome the deficiencies of the prior art, at least one technical problem presented in the background art is solved. The technical scheme adopted for solving the technical problems is as follows: The invention provides an ancient architecture digital intelligent knowledge base system, which comprises: The repair color deviation decomposition module calculates deviation data of pigment ores based on the acquired data and standard data of the pigment ores, fits to obtain color space coefficients, calculates color value correction based on the color space coefficients, calculates repair color deviation through the color value correction, and integrates to obtain current batch color repair structure data; The formula adjustment dynamic calibration module is used for fitting relation coefficients of deviation data and pigment ores, constructing a basic model, and dynamically calibrating the basic model based on color restoration structure data to obtain the special formula adjustment quantity of the pigment ores in the current batch; the multivariate dynamic optimization module is used for constructing a mixed color value prediction model and a multi-objective optimization function, and obtaining the proportion of the exclusive optimal formula through solving by a genetic algorithm; The on-site verification and calibration module is used for verifying the proportion of the exclusive optimal formula and reversely calibrating the basic model; And the knowledge base iteration multiplexing module integrates the collected data and the exclusive optimal formula proportion to form a data set, stores the data set into the ancient building repair pigment formula knowledge base, and corrects the basic model coefficient for a long time based on the data set for rapidly multiplexing the ancient building repair pigment formula. In the invention, the specific process of calculating the color value correction based on the color space coefficient is as follows: Sampling each batch of pigment ore for repairing by using an X-ray fluorescence spectrometer and a near infrared spectrometer, collecting the measurement purity of the pigment ore and the measurement impurity of the pigment ore, obtaining the standard purity of the pigment ore and the standard impurity of the pigment ore, and calculating to obtain the purity deviation of the pigment ore and the impurity deviation of the pigment ore; Fitting to obtain color space coefficients based on the purity deviation of the historical pigment ore and the impurity deviation of the historical pigment ore, wherein the color space coefficients comprise the purity color value deviation coefficients 、、Coefficient of deviation from impurity color value; Calculating color value correction according to the color space coefficient obtained by fitting, wherein the color value correction comprises color value correction caused by purity deviation and color value correction caused by impurity deviation; Ba