Method for evaluating development degree of rock painting exfoliation disease

Abstract

The invention provides a method for evaluating the development degree of a rock painting spalling disease, and the method comprises the steps: constructing an evaluation index system according to the existing characteristics of the rock painting spalling disease, and collecting the evaluation data of an evaluation index; determining an evaluation grade and a corresponding evaluation interval; converting the evaluation interval into a digital feature of a cloud model, and generating a standard cloud picture of the evaluation interval by using a forward cloud generator; inputting the evaluation data into a BP neural network weighting model to obtain the weight of each evaluation index; the evaluation data is preprocessed and then substituted into a reverse cloud generator to be converted, and index layer cloud parameters are obtained; calculating a criterion layer cloud parameter according to the weight of each evaluation index and the index layer cloud parameter; calculating a comprehensive cloud parameter by using the criterion layer cloud parameter and the weight of the criterion layer, and inputting the comprehensive cloud parameter into a forward cloud generator to obtain a comprehensive layer cloud picture; and comparing the comprehensive layer cloud picture with the standard cloud picture to obtain an evaluation grade. According to the method, the problems of uncertainty, fuzziness and randomness can be effectively solved, and the method is suitable for a multi-factor interlaced complex evaluation scene of the rock painting exfoliation disease.

Description

Technical Field

[0001] The present invention relates to the technical field of rock painting storage safety and scientific protection, and in particular to a method for evaluating the development degree of rock painting spalling disease. Background Art

[0002] Long-term weathering of stone cultural relics leads to rock degradation. The Helankou rock paintings are located at the mouth of the Helan Mountain Valley in Helan County, Yinchuan City, Ningxia Hui Autonomous Region. However, due to the long history of the carvings, after long-term natural weathering, under the influence of external environmental factors such as rainfall, temperature, humidity changes, and biological erosion, the metamorphic sandstone carrier of the rock paintings has undergone typical stone degradation phenomena, and many of them have developed spalling diseases. Spalling disease is the final form of the development of rock painting diseases, and continued development means the demise of cultural relics. Therefore, in view of the universality, danger, and protection priority of the spalling disease development of the Helankou rock paintings, establishing a method for evaluating the spalling disease of rock paintings is a key step in achieving scientific protection.

[0003] According to the on-site investigation, the flaking of the Helankou rock paintings is the most serious disease among many diseases. In the 570m long rock painting key protection area from the mouth of the Helankou ditch to the south bank of the "Shuiguan", there are 660 individual rock paintings, of which 173 rock paintings have flaked and fallen off, accounting for 26.2% of the rock paintings on the south bank. Most of the rock paintings have also experienced varying degrees of flaking around them, which seriously threatens the survival of the rock paintings. Before the flaking of sandstone layers, hollowing phenomenon will occur, that is, the plate-like bodies of a certain thickness on the surface of the rock will bulge and deform, and a cavity will be formed behind the plate-like bodies. The appearance of hollowing phenomenon indicates the occurrence of flaking disease, and the surface strength of the sandstone carrier of the rock painting has been greatly reduced at this time. Therefore, the assessment of the flaking disease of rock paintings should not only rely on on-site survey data, but also comprehensively evaluate the development degree of its flaking disease by combining on-site survey data and the properties of the metamorphic sandstone carrier of the rock painting.

[0004] At present, the evaluation of rock painting spalling mainly relies on surface observation, lacks a scientific and systematic evaluation method, and does not fully consider the complex relationship between environmental factors and carrier properties. Rock painting protection areas are generally characterized by large areas, large numbers, complex environments, and uneven carrier rock properties. Spalling is affected by multiple factors. Therefore, the evaluation of rock painting diseases should comprehensively consider the complexity of the causes and adopt multiple evaluation indicators. Common cultural relic disease evaluation methods include fuzzy comprehensive evaluation (FCE), gray relationship analysis (GRA), analytic hierarchy process (AHP), topology of ideal solutions similarity (TOPSIS) and principal component analysis (PCA), but these methods may introduce subjective bias and randomness when dealing with nonlinear problems, especially in dealing with the randomness and complexity of stone cultural relic diseases. Therefore, in view of the complexity of spalling diseases and the shortcomings of existing models in dealing with information ambiguity and randomness, a scientific evaluation method is developed to comprehensively evaluate the spalling of rock paintings and provide useful support for the scientific protection of rock paintings. Summary of the invention

[0005] In view of the technical problem that existing cultural relics disease evaluation methods cannot be directly applied to the evaluation of the development degree of rock painting disease, the present invention proposes a method for evaluating the development degree of rock painting spalling disease. Based on parameters such as the physical and mechanical properties of the rock carrier rock, the crack characteristics of the carrier rock and the location of the rock painting, the severity of the rock painting spalling disease is evaluated scientifically, time-saving and economically, providing a decision-making basis for the priority and scientific nature of rock painting protection. At the same time, it can be extended to the evaluation and protection of rock painting spalling diseases in other regions.

[0006] In order to achieve the above object, the technical solution of the present invention is implemented as follows: a method for evaluating the development degree of rock painting spalling disease, the steps of which are as follows:

[0007] Step S1: Based on the existing characteristics of rock painting spalling diseases, an evaluation index system is constructed, and evaluation indicators directly related to the existing characteristics are selected through field investigation and indoor experiments, and evaluation data corresponding to the evaluation indicators of each group of rock paintings at the rock painting site are collected;

[0008] Step S2: determining the evaluation level of the development degree of the spalling disease and the corresponding evaluation interval according to the on-site investigation of the disease and the development characteristics of the disease;

[0009] Step S3: converting the numerical value of the evaluation interval into the digital features of the cloud model, using the forward cloud generator to process the digital features to generate a standard cloud map of the evaluation interval, and obtaining a standard cloud map of the evaluation level;

[0010] Step S4: input the evaluation data into the constructed BP neural network weighting model to obtain the weight of each evaluation index;

[0011] Step S5: The evaluation data is preprocessed and brought into the reverse cloud generator to be converted into a qualitative language represented by three digital features: expectation, entropy, and super entropy, to obtain the index layer cloud parameters of the evaluation index;

[0012] Step S6: Calculate each criterion stratus cloud parameter according to the weight of each evaluation index and the index stratus cloud parameter corresponding to the evaluation index;

[0013] Step S7: Calculate the weight of the criterion layer according to the weight of the evaluation index, calculate the comprehensive cloud parameter using the cloud parameter of each criterion layer and the weight of the corresponding criterion layer, and input the comprehensive cloud parameter into the forward cloud generator to obtain the comprehensive cloud map;

[0014] Step S8: Compare the comprehensive layer cloud map with the evaluation grade standard cloud map, and obtain the final evaluation grade according to the position of the comprehensive layer cloud map in the evaluation grade standard cloud map.

[0015] Preferably, the evaluation index system includes a target layer, a criterion layer and an indicator layer, the target layer is the development of exfoliation disease, and the criterion layer includes the mechanical properties and physical properties of the rock painting carrier, the crack characteristics within 300 mm around the rock painting, and the location of the rock painting;

[0016] The evaluation indicators of the index layer of mechanical properties include rebound strength, compressive strength and longitudinal wave velocity, the evaluation indicators of the index layer of physical properties include porosity, water content and particle density, the evaluation indicators of the index layer of crack characteristics include the closest distance of cracks, crack length and maximum width of cracks, and the evaluation indicators of the index layer of rock painting location include relative elevation and rock surface inclination;

[0017] The positive indicators of the evaluation index include porosity, water content, crack length, and maximum crack width; the negative indicators include rebound strength, compressive strength, longitudinal wave velocity, particle density, the closest distance between cracks, relative elevation, and rock surface inclination.

[0018] Preferably, the evaluation levels of the development degree of the peeling disease are divided into 5 levels, corresponding to 5 evaluation intervals, and the evaluation intervals are [0-0.2], [0.2-0.4], [0.4-0.6], [0.6-0.8], and [0.8-1.0]; the corresponding evaluation levels are very low, low, medium, high, and very high.

[0019] Preferably, the method of converting the numerical value of the evaluation interval into the digital feature of the cloud model is: the digital feature of the cloud model includes three parameters, namely, expectation Ex, entropy En and super entropy He, and the three parameters are:

[0020]

[0021] in, is the minimum and maximum value of the nth evaluation interval, k is a constant; Exn 、En n 、He n are the expectation, entropy and super entropy of the nth evaluation interval respectively.

[0022] Preferably, the method for generating an evaluation grade cloud map by processing digital features using a forward cloud generator is:

[0023] 1) Generate entropy En n is the expected value, He n 2 A normal random number y with variance n =R N (En n ,He n );R N (,) represents the normal random function;

[0024] 2) Generate the expected Ex n is the expected value, y n 2 A normal random number x with variance m =R N (Ex n ,y n ); m is 1-M;

[0025] 3) Calculate the degree of certainty

[0026] 4) With certainty μ(x m ) of the normal random number x m Become a cloud drop in the universe;

[0027] 5) Repeat steps 1)-4) until the required M cloud droplets are generated, and the M cloud droplets constitute a standard cloud image of an evaluation interval;

[0028] Repeat the above steps to obtain the standard cloud map of each evaluation interval, and the standard cloud maps of all evaluation intervals constitute the standard cloud map of the evaluation grade.

[0029] Preferably, the method for obtaining the weight of each evaluation index is:

[0030] 1) Use MATLAB calculation tools to generate a BP neural network weighted model, set the maximum number of iterations and target error, and set the number of hidden layer neurons as the total number of evaluation indicators;

[0031] 2) Import the evaluation data of the evaluation index into the BP neural network weighting model to generate an 18×11 data matrix data;

[0032] 3) Call the calculate weights function to calculate the weighted average weights for the data matrix data.

[0033] Preferably, the method for calculating the weighted average weights of the data matrix data by the calculate weights function is: normalizing the input evaluation data, setting the activation functions of the hidden layer and the output layer, the activation function of the hidden layer is 'logsig', and the activation function of the output layer is 'purelin'; setting training parameters, using the normalized evaluation data to train the BP neural network, extracting the output layer weights of the BP neural network, multiplying the output layer weights by the predicted output of the BP neural network to calculate the weighted average weight, and obtaining the weights of the evaluation indicators, and the weights of all evaluation indicators are non-negative and the sum is 1.

[0034] Preferably, the method for obtaining the index layer cloud parameters of the evaluation index is: the preprocessing of the positive index is:

[0035] Y ij =a+G(X ij -min(X ij ))

[0036] The preprocessing of negative indicators is:

[0037] Y ij =b+G(X ij -max(X ij ))

[0038] Among them, the coefficient G is: b is the maximum value of all evaluation intervals, a is the minimum value of all evaluation intervals, Y ij is the preprocessing result of the evaluation data of the i-th evaluation index and the j-th group of rock paintings, X ij is the evaluation data of the i-th evaluation index and the j-th group of rock paintings, min and max represent the functions of finding the minimum and maximum values ​​respectively;

[0039] The reverse cloud generator converts the J groups of preprocessed evaluation data Y of the i-th evaluation index ij Calculate the expected Ex of the i-th evaluation index i 、Entropy i 、Super entropy He i , as the index layer cloud parameter of the i-th evaluation index.

[0040] Preferably, the method for calculating the cloud parameters of each criterion layer is:

[0041] Expectation of the lth criterion level

[0042] Entropy of the lth criterion layer

[0043] The super entropy of the lth criterion layer

[0044] Among them, Ex l1 、En l1 、He l1 are the expectation, entropy and super entropy of the l1th evaluation index of the lth criterion layer, respectively, l1 is the weight of the l1th evaluation index, L1 is the total number of evaluation indicators in the lth criterion layer;

[0045] The method for calculating the comprehensive cloud parameters is:

[0046] expect

[0047] entropy

[0048] Super Entropy

[0049] Among them, λ l is the weight of the lth criterion layer, and L2 is the total number of criterion layers.

[0050] Preferably, the method of bringing the comprehensive cloud parameters into the forward cloud generator to obtain the comprehensive layer cloud map is:

[0051] 1) Generate entropy En A For the expected value, A normal random number y with variance A =R A (En A ,He A );

[0052] 2) Generate the expected Ex A is the expected value, y A 2 A normal random number x with variance A =R A (Ex A ,y A );

[0053] 3) Calculate the degree of certainty

[0054] 4) With certainty μ(x A ) of the normal random number x A Become a cloud drop in the universe;

[0055] 5) Repeat steps 1)-4) until the required M cloud droplets are generated, and all cloud droplets constitute a comprehensive layer cloud image;

[0056] In the comparison between the comprehensive layer cloud map and the evaluation grade standard cloud map, the similarity between the digital features of the cloud model and the comprehensive cloud parameters is calculated based on the weighted Euclidean distance, the index of the evaluation grade standard cloud map with the maximum similarity and its corresponding index is found, and the evaluation grade of the comprehensive cloud parameters is determined according to the index;

[0057] The similarity is the reciprocal of the weighted Euclidean distance, which is:

[0058]

[0059] Among them, the comprehensive cloud parameter is C A =(Ex A ,En A ,He A ), the cloud parameter C of the nth evaluation interval in the evaluation grade standard cloud map n =(Ex n ,En n ,He n );w Ex 、w En 、w He is the weight coefficient, and w Ex +w En +w He =1;

[0060] The position of the comprehensive layer cloud map in the evaluation grade standard cloud map is the score of the evaluation object, which corresponds to the evaluation interval and determines the evaluation grade of the evaluation object.

[0061] Compared with the prior art, the beneficial effects of the present invention are as follows: selecting evaluation indicators of the development degree of rock painting spalling diseases to construct an evaluation system; determining an evaluation set according to disease characteristics; determining standard cloud digital characteristic values ​​according to the grading standard of the development degree of rock painting spalling diseases and the calculation formula of cloud model theory, and generating a standard cloud map of the cloud model; inputting the evaluation data of the evaluation indicators of the evaluation set into the BP neural network weighting model to weight the evaluation indicators at all levels, and obtaining the weights of the indicators at all levels; preprocessing the evaluation data according to the intervals divided by the evaluation set, using the reverse cloud generator to calculate the evaluation data and the indicator layer weights, and obtaining the indicator layer cloud parameters; bringing the indicator layer cloud parameters and the criterion layer weights into the cloud parameter calculation formula to calculate the criterion layer cloud parameters; using the forward cloud generator to calculate the criterion layer weights and the criterion layer cloud parameters to obtain the comprehensive cloud parameters, and generating a comprehensive cloud map of the cloud model; comparing the comprehensive cloud map with the standard cloud map, calculating the similarity based on the weighted Euclidean distance, finding the maximum similarity and its corresponding standard cloud index, and comparing with the standard cloud map to obtain the final evaluation result. The present invention comprehensively considers various factors that affect the development of rock painting spalling diseases. In an evaluation environment with a large number of spalling diseases and uneven development, a time-saving, scientific and effective disease assessment model is determined through a BP neural network weighting model and a cloud model, so that the assessment of rock painting spalling diseases does not rely solely on surface surveys, but fully considers the existing characteristics of the disease and the characteristics of the carrier rock to make a comprehensive assessment. The present invention provides useful assistance and priority protection suggestions for subsequent protection work and restoration, and is conducive to the rational allocation of protection resources.

[0062] The present invention screens out multiple evaluation factors that may affect the spalling of rock paintings based on the characteristic parameters of rock paintings and their carrier rocks through field investigation, exploration test and indoor sample analysis, and constructs characteristic indicators for evaluating the development degree of rock spalling diseases, aiming to accurately reflect the actual development of rock painting spalling diseases; by comprehensively considering various subjective and objective weighting methods, the BP neural network weighting model is used to weight various evaluation indicators, avoiding the deviation caused by the subjective weighting method, and better reflecting the actual significance and importance of the indicators when dealing with nonlinear problems; and combined with the cloud model for comprehensive evaluation, it can effectively deal with uncertainty, fuzziness and randomness problems, and is suitable for complex evaluation scenarios such as rock painting spalling diseases with multiple factors intertwined, filling the gap in the evaluation method of rock painting spalling diseases, and can use this method to comprehensively analyze the severity of rock painting spalling diseases in the Helankou rock painting area. In addition, the present invention can replace the evaluation indicators, provide convenience for more complex evaluation environments, and can evaluate a large number of rock paintings in a short time; the evaluation results provide a scientific basis for the priority of subsequent protection work, ensure the reasonable allocation of limited resources, and provide data support and decision-making basis for future protection work.

[0063] The present invention comprehensively considers the relationship between the existing diseases of rock paintings, the properties of carrier rocks and the location of rock paintings, selects evaluation indicators that can directly affect the development of spalling diseases, and provides reference and help for subsequent research; due to the special attributes of cultural relics, non-destructive or micro-destructive survey methods are selected under the principle of minimum intervention to obtain evaluation indicators; when selecting samples, rock samples with the same bedding, characteristics, and environmental characteristics as the rock carrier of the rock paintings and a longitudinal wave velocity difference within 10% are selected to restore the original characteristics of the rock carrier of the rock paintings to the maximum extent; an evaluation model for spalling diseases is constructed through MATLAB, and a large number of rock paintings can be evaluated in a short time in an evaluation environment where the conditions of rock paintings and the properties of carrier rocks are different, and the evaluation indicators can be replaced at any time. The present invention realizes a time-saving, economical and effective evaluation process in the process of disease evaluation, provides help and suggestions for subsequent protection and restoration, and is conducive to the rational allocation of protection resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0065] Figure 1 It is a flow chart of the present invention.

[0066] Figure 2 It is a hierarchical model diagram of the evaluation index of the present invention.

[0067] Figure 3 The present invention is a calculation flow chart for evaluating the development degree of rock painting spalling disease.

[0068] Figure 4 It is an evaluation grade standard cloud map of the rock painting spalling disease development degree evaluation set of the present invention, wherein the evaluation interval represented by the pink cloud map is very low, the green cloud map is low, the yellow cloud map is medium, the red interval is high, and the blue cloud map is very high.

[0069] Figure 5 It is the evaluation result diagram of the present invention, wherein (a)-(h) correspond to 8 rock painting sites, and the evaluation results are very low, medium, low, high, medium, medium, high and medium, respectively. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] like Figure 1 As shown, a method for evaluating the development degree of rock painting spalling disease, this embodiment takes 8 groups of rock paintings in the Helankou rock painting area as an example, by investigating the development area of ​​rock painting spalling disease in the rock painting area, the crack characteristics of the rock painting carrier and the geographical location of the rock painting, etc., analyzing the existing disease characteristics of the rock painting, the characteristics of the carrier rock and the geographical location and other factors to quantitatively evaluate the degree of disease development; using the BP neural network weighting model and the cloud model comprehensive evaluation method, the rock paintings in the rock painting area are evaluated in a time-saving, effective, economical and scientific manner, providing suggestions and assistance for subsequent protection work, which is conducive to the rational allocation of protection resources. The specific steps of the present invention are as follows:

[0072] Step S1: According to the existing characteristics of rock painting spalling diseases, evaluation indicators directly related to the existing characteristics are selected through field investigation and indoor experiments, and an evaluation indicator system is constructed.

[0073] The categories of evaluation indicators are: mechanical properties of rock painting carriers (B1), physical properties of rock painting carriers (B2), crack characteristics within 300 mm around the rock painting (B3), and rock painting location (B4).

[0074] The evaluation index selection methods include: non-destructive or micro-destructive testing to select evaluation indexes. In the process of selecting evaluation indexes, the principle of minimum intervention is followed to minimize direct disturbance to the rock paintings. Under the premise of not disturbing the rock painting carrier, when sampling, rock samples with the same bedding, characteristics and occurrence environment as the rock carrier of the rock painting are selected for mechanical and physical experiments, so as to ensure that the selected samples have similar characteristics to the rock painting carrier as much as possible.

[0075] According to the principle of minimum intervention, during the on-site investigation stage, a velocimeter was used to test the longitudinal wave velocity (C3) of the rock carrier of the rock painting as one of the mechanical characteristics for evaluating the spalling of the rock painting. A micro-damage test method was selected, and a rebound strength meter was used to test the rebound strength (C1) of the rock carrier of the rock painting. The test point must be outside the scope of the rock painting. Considering that the compressive strength can express the degree of weathering of the rock carrier of the rock painting, the compressive strength (C2) of the rock carrier of the rock painting was selected as one of the mechanical indicators for evaluating the spalling of the rock painting.

[0076] The evaluation indicators that can directly reflect the rock physical properties (B2) of the rock painting carrier are selected: the porosity (C4), water content (C6) and particle density (C5) of the rock painting carrier rock. Considering that there are often a large number of exfoliation diseases around the cracks of the rock painting carrier rock, the closest distance (C7), crack length (C8) and maximum crack width (C9) within 300mm of the rock painting are selected as evaluation indicators of crack characteristics. The selection range of these characteristics is determined to extend 300mm outward from the rock painting as the center. The height and inclination of the rock painting carrier rock will affect the erosion process of water on the carrier rock. The relative elevation (C10) and rock surface inclination (C11) of the location of the rock painting are selected as evaluation indicators of the rock painting location (B4).

[0077] An evaluation system was constructed based on the selected evaluation indicators, in which the evaluation target was the development degree of spalling disease; the criterion layer was the mechanical and physical properties of the rock carrier rock, the crack characteristics of the rock carrier rock and the location of the rock paintings; the evaluation indicators were rebound strength, compressive strength, longitudinal wave velocity, porosity, water content, particle density, the closest distance of cracks (C7), crack length, maximum crack width, relative elevation and rock surface inclination.

[0078] According to the above evaluation indicators, Figure 2 The evaluation index hierarchical structure diagram is as follows, that is, the mechanical indicators of the carrier rock include rebound strength, compressive strength, and longitudinal wave velocity; the physical indicators include porosity, water content, and particle density; the rock painting location includes relative elevation (C10) and rock surface inclination (C11); the fracture characteristics include the closest distance of the fracture, the fracture length, and the maximum width of the fracture.

[0079] This embodiment selects 8 rock painting sites in the Helankou rock painting area as evaluation objects, each of which contains 18 groups of rock paintings. Due to the large number of rock paintings, this embodiment displays the evaluation data of rock painting No. 1. Table 1 is the evaluation data of the evaluation index of rock painting site No. 1, that is, the evaluation data of the evaluation index mentioned in the present invention. Figure 3 shown.

[0080] Table 1 Data of evaluation indicators of rock painting site No. 1

[0081]

[0082] Step S2: Determine the comment set, i.e., the evaluation interval, and determine the evaluation grade division interval of the spalling disease development degree according to the on-site investigation of the disease and the disease development characteristics.

[0083] The degree of spalling disease development is divided into 5 evaluation intervals, corresponding to 5 levels, and the evaluation intervals are [0-0.2], [0.2-0.4], [0.4-0.6], [0.6-0.8], [0.8-1.0]; the corresponding levels are very low, low, medium, high, and very high. This division is to facilitate the selection of protection priorities, based on previous experience with rock degradation. The very high level means that the rock paintings are completely threatened by the development of spalling diseases, are in an extremely dangerous state, and are close to extinction. The high level means that the rock paintings are seriously threatened by the development of spalling diseases, and relevant protection measures are urgently needed. The medium level means that the spalling disease development of the rock paintings is within the controllable range, and systematic protection measures need to be taken. The low level means that the rock paintings are not seriously threatened by the development of spalling diseases, but it is recommended to take relevant measures to avoid it. The very low level means that the rock paintings are in good condition and are not affected by the development of spalling diseases for the time being.

[0084] Step S3: Determine the evaluation standard cloud map: Convert the comment interval in step S2 into the digital features of the cloud model, and use the forward cloud generator to generate the evaluation level cloud map.

[0085] like Figure 3 As shown, the specific implementation method is: according to the five evaluation intervals [0-0.2], [0.2-0.4], [0.4-0.6], [0.6-0.8], [0.8-1.0] defined in step S2, for each evaluation interval, the digital characteristics of the cloud model are calculated by MATLAB software, including three parameters, namely, expected Ex, entropy En, and super entropy He. The formulas of the three parameters are:

[0086]

[0087] in, is the minimum and maximum value of the nth evaluation interval, k is a constant, generally between 0.001-0.1. In the present invention, the constant k is set to 0.01 based on the fuzziness of the evaluation object itself and previous experience. In the present invention, n=1-5. n 、En n 、He n are the expectation, entropy and super entropy of the nth evaluation interval respectively.

[0088] The expected Ex is the average value of the evaluation interval. For the evaluation interval [0-0.2], Ex = (0 + 0.2) / 2 = 0.1. Entropy En is the length of the evaluation interval divided by 6. For the interval [0-0.2], En = (0.2-0) / 6 = 0.3333. The super entropy He is a fixed value of 0.01. The other intervals are calculated using the above method. Table 2 shows the classification criteria of the evaluation level in the cloud model.

[0089] Table 2 Evaluation level classification standards

[0090]

[0091] The forward cloud generator is used to calculate the digital features of the evaluation interval to generate the evaluation level cloud map, which is the standard cloud map, such as Figure 4 The standard cloud maps of different evaluation intervals are calculated by the forward cloud generator, and the standard cloud maps of all evaluation intervals constitute the standard cloud map of the evaluation level. Figure 4 It is used to compare with the comprehensive cloud map of the evaluation results, so that you can clearly and intuitively see the level of the evaluation results and the evaluation effect.

[0092] The process of the forward cloud generator is as follows:

[0093] Input: Numeric Features: Expected Ex n , Entropy n and super entropy He n , the number of cloud droplets generated is 1000.

[0094] Output: 1000 cloud drops x m and its degree of certainty μ(x m ), m=1,2,…,1000.

[0095] 1) Generate entropy En n is the expected value, He n 2 A normal random number y with variance n =R N (En n ,He n );R N (,) represents the normal random function;

[0096] 2) Generate the expected Ex n is the expected value, y n 2 A normal random number x with variance m =R N (Ex n ,y n );

[0097] 3) Calculate the degree of certainty

[0098] 4) With certainty μ(x m ) of the normal random number x m Become a cloud drop in the domain; with the degree of certainty μ, that is, the degree of membership, a cloud drop is a normal random number x with certainty m , this cloud droplet is a normal random number x on the coordinate axis mAs the horizontal coordinate, the degree of certainty μ is a point on the vertical coordinate. These 1000 cloud droplets are Figure 4 The shape of any interval in the , 5 evaluation intervals can generate 5 shapes, and combined together they form the evaluation grade standard cloud map.

[0099] 5) Repeat steps 1)-4) until the required 1000 cloud droplets are generated, and all cloud droplets constitute a standard cloud map of an evaluation interval.

[0100] Repeat the above steps to get the standard cloud map of each evaluation interval. The standard cloud maps of 5 intervals can constitute the standard cloud map of the evaluation level. Different evaluation intervals have different digital features. The forward cloud generator can repeatedly calculate the specified cloud droplets. The first evaluation interval has 1000 cloud droplets, and the shape formed is Figure 4 The pink shape, 5 shapes in 5 evaluation intervals, and 5 shapes of different colors together form the evaluation level cloud map. Different evaluation intervals have different cloud parameters, and the evaluation level cloud map is also different.

[0101] Step S4: Determine the weights of each level: Construct a BP neural network weighting model to measure the weights of each level of the evaluation index, and obtain the index layer weight and criterion layer weight of the evaluation index. The specific steps are:

[0102] 1) Use MATLAB calculation tools to generate BP neural network weighting model, set the maximum number of iterations and target error, select a specific function calculation, but need to determine the number of hidden layer neurons, hidden layer will have 1-multiple, each hidden layer of the number of neural units can also be set to a random value, the present invention is 1 hidden layer, there are 11 evaluation indicators to select hidden layer has 11 neurons. Adopt BP neural network objective weighting method, when dealing with nonlinear problems can reduce subjective bias and non-randomness caused by low data quality, can better reflect the actual significance and importance of indicators.

[0103] 2) Import the evaluation data of the evaluation indicators of Rock Painting Area No. 1 into the BP neural network weighting model and randomly generate an 18×11 data matrix data.

[0104] 3) Call the calculate weights function to calculate the weighted average weights of the data matrix data. In the calculate weights function, the input evaluation data is first normalized, and then a BP neural network with 1 hidden layer is created, and the number of neurons in the hidden layer and the activation functions of the hidden layer and the output layer are set. The activation function of the hidden layer is 'logsig', and the activation function of the output layer is 'purelin'. Set the training parameters and use the input evaluation data to train the BP neural network. After the training is completed, draw the training performance graph and extract the output layer weights of the BP neural network. Finally, the weight of the output layer weight is multiplied by the predicted output to calculate the weight of the weighted average to obtain the weight of the evaluation index, and ensure that the weight is non-negative and the sum is 1.

[0105] The specific principles are:

[0106]

[0107] Where i is the neuron in the hidden layer of the BP neural network, i = 1…C, C represents the total number of neurons in the hidden layer; f jm1 is the correlation significance coefficient between the input layer neuron j and the output layer neuron m1; F jm1 is the correlation index between the input layer and the output layer; w ij is the weight coefficient between input layer neuron j and hidden layer neuron i; w m1i is the weight coefficient between the output layer neuron m1 and the hidden layer neuron i. These two weight coefficients will be updated and obtained during the training process. j = 1, 2, ..., J, where J represents the number of rock painting groups. C jm1 is the absolute influence coefficient, which is also the weight of the evaluation index. First, the evaluation data of the 11 evaluation indicators are imported into the input layer of the BP neural network, and the output data of the hidden layer is calculated by the logsig function of the hidden layer. Then, the output data of the hidden layer is used as the new input data of the output layer, and the output value of the BP neural network is finally obtained by the purelin function of the output layer. Set the expected error to 10 -4 If the output value does not reach this expected error, the parameters are adjusted through the trainlm training function and iterated repeatedly, setting the maximum number of iterations to 5000 times until the expected error requirement is met.

[0108] The training parameters include the maximum number of iterations and the target error. The maximum number of iterations for training is 50,000 and the target error for training is 1e -8 ,The training stops when the target error or the maximum number of iterations is reached.,Table 3 shows the weights of each indicator extracted by the BP neural network.,The criterion layer weight is the sum of the weights of all subordinate,evaluation indicators.

[0109] Table 3 Weights of evaluation indicators at all levels

[0110]

[0111] Step S5: Determine the index layer cloud parameters of the evaluation index: pre-process the evaluation data of the 18 groups of rock paintings at the rock painting site No. 1 in Table 1 and bring them into the reverse cloud generator to convert them into the expected En c 、EntropyEx c 、Super entropy He c The three numerical features are expressed in qualitative language to obtain the index layer cloud parameters.

[0112] Among them, the positive indicators of the evaluation indexes are: porosity (C4), water content (C6), crack length (C8), and maximum crack width (C9); the negative indicators are: rebound strength (C1), compressive strength (C2), longitudinal wave velocity (C3), particle density (C5), the closest distance of the crack (C7), relative elevation (C10), and rock surface inclination (C11).

[0113] For positive correlation indicators, the preprocessing formula is:

[0114] Y ij =a+G(X ij -min(X ij ))

[0115] For negative indicators, the preprocessing formula is:

[0116] Y ij =b+G(X ij -max(X ij ))

[0117] The calculation formula of coefficient G is:

[0118] Where b is the maximum value of the comment set, and a is the minimum value of the comment set, that is, the maximum and minimum values ​​of all evaluation intervals. In this embodiment, b=1, a=0, Y ij is the preprocessing result of the evaluation data of the i-th evaluation index and the j-th group of rock paintings. ij is the evaluation data of the i-th evaluation index and the j-th group of rock paintings. min and max represent the functions of finding the minimum and maximum values, respectively.

[0119] The reverse cloud generator uses MATLAB software to input the preprocessed data into the reverse cloud generator to obtain the qualitative language of the evaluation index, that is, the digital characteristics. The cloud model brings the preprocessed evaluation data into the reverse cloud generator. The calculation steps of the reverse cloud generator are as follows:

[0120] Input: Sample point x of the i-th evaluation index i There are J preprocessed evaluation data Yij , j = 1, 2, ..., J, i = 1, 2, ..., C. In this embodiment, each evaluation index of rock painting site No. 1 has 18 evaluation samples.

[0121] Output: Numerical characteristics reflecting the qualitative concept of the evaluation index: Expected Ex i 、Entropy i 、Super entropy He i .

[0122] Digital FeaturesEx i 、En i 、He i The calculation steps are:

[0123] 1) According to the sample point x i Calculate the sample mean of this set of evaluation data after preprocessing The first-order sample absolute central moment is: The sample variance is:

[0124] 2) Calculate the expectation:

[0125] 3) Calculate entropy:

[0126] 4) Calculate the super entropy:

[0127] The calculation results of the digital features of each evaluation index obtained by the inverse cloud generator are shown in Table 4.

[0128] Table 4. Evaluation index stratus cloud parameters of rock painting site No. 1

[0129]

[0130] Step S6: Determine the cloud parameters of the criterion layer: Substitute the indicator layer weights of each evaluation indicator in Table 3 in step S4 and the cloud parameters of each indicator layer in Table 4 in step S5 into the cloud parameter calculation formula to obtain the cloud parameters of each criterion layer. The calculation process here is implemented by MATLAB. The criterion layer cloud parameters are calculated by inputting the qualitative language of the evaluation indicators, i.e., the digital features and the corresponding indicator layer weights into the MATLAB software. The calculation formula of the criterion layer cloud parameters is:

[0131]

[0132] Among them, λ C1 , C2 , …, λ C11 The weights of each indicator layer determined by the BP neural network weighting model, Ex1, Ex2, Ex 11 、En1、En2、…、En 11、He1、He2、…、He 11 Index cloud parameters calculated for the inverse cloud generator. Calculation results Table 5.

[0133] Table 5 Criteria stratus cloud parameters

[0134]

[0135] Step S7: Determine comprehensive cloud parameters and generate comprehensive cloud images: Substitute the weights of each criterion layer in step S4 and the cloud parameters of each criterion layer in step S6 into the cloud parameter calculation formula to obtain comprehensive cloud parameters; input the comprehensive layer cloud parameters into the forward cloud generator to obtain a comprehensive layer cloud image.

[0136] The comprehensive stratus cloud parameters are obtained by MATLAB software based on the characteristic parameters of the criterion stratus cloud parameters and the criterion layer weights. The comprehensive stratus cloud parameters are input into the forward cloud generator to generate a comprehensive cloud map. The calculation method of the comprehensive stratus cloud parameters is:

[0137]

[0138] Among them, Ex B1 、Ex B2 、....、Ex B4 Denote the expectations of the criterion layers B1-B4, En B1 、En B2 、....、En B4 Respectively represent the entropy of the criterion layer B1-B4, He B1 、He B2 、....、He B4 denote the super entropy of the criterion layers B1-B4, λ B1 , B2 , ...., λ B4 They represent the criterion layer weights of criterion layers B1-B4 respectively, and the calculation results are shown in Table 6.

[0139] Table 6 Comprehensive stratus cloud parameters

[0140]

[0141] The calculated synthetic cloud parameters are brought into the forward cloud generator to obtain the synthetic layer cloud map. The steps are as follows:

[0142] Input: Digital feature Ex A ,En A , He A , the number of cloud droplets generated is 1000.

[0143] Output: 1000 cloud drops x A and its degree of certainty μ(x A )(A=1,2,…,1000).

[0144] 1) Generate En A For the expected value, A normal random number y with variance A =R A (En A ,He A );

[0145] 2) Generate Ex A is the expected value, y A 2 A normal random number x with variance A =R A (Ex A ,y A );

[0146] 3) Calculate the degree of certainty

[0147] 4) With certainty μ(x A ) of the normal random number x A Become a cloud drop in the universe;

[0148] 5) Repeat steps 1)-4) until the required 1000 cloud droplets are generated, and all cloud droplets constitute a comprehensive cloud image.

[0149] Step S8: Compare the comprehensive cloud map with the evaluation level cloud map, calculate the similarity between the digital features of the cloud model and the comprehensive cloud parameters based on the weighted Euclidean distance similarity calculation formula, and calculate the similarity by the inverse of the distance. The smaller the distance, the higher the similarity. Finally, find the maximum similarity and its corresponding standard cloud index, determine the evaluation level of the comprehensive cloud according to the index, and compare it with the index label of the standard cloud map (that is, the label on the horizontal and vertical axes on the coordinate graph) to obtain the final evaluation result.

[0150] Substitute the comprehensive cloud parameters obtained in step 7 into the weighted Euclidean distance calculation formula. The calculation steps are as follows:

[0151] Set the comprehensive cloud parameters to: C A =(Ex A ,En A ,He A ), the standard cloud parameter C of the nth evaluation interval n =(Ex n ,En n ,He n ). Weighted Euclidean distance d(C A ,C n ) is calculated as:

[0152]

[0153] Among them, n represents the nth evaluation interval, and the weight coefficient w Ex =0.5, w En =0.3,w He = 0.2. The calculation results are shown in Table 7. The synthetic cloud has the highest similarity with the standard cloud (improved distance method), and the similarity value is: 0.96233.

[0154] Table 7 Weighted Euclidean distance calculation results

[0155]

[0156] After calculating the weighted Euclidean distance, the similarity between the evaluation results of rock painting point 1 and the standard cloud was obtained. The results showed that the evaluation result of the development degree of spalling disease at rock painting point 1 was the expected Ex of the comprehensive cloud parameters. A =0.15836, this value belongs to the evaluation interval of [0 0.2], and the evaluation result is very low. The comprehensive cloud map is compared with the standard cloud map generated in step S3, where the position of the comprehensive cloud map in the standard cloud map is the score of the evaluation object, and the corresponding evaluation set can determine the evaluation level of the evaluation object. Repeat the above steps to evaluate the development degree of spalling disease of the remaining 7 rock paintings, and the evaluation results are shown in Table 8. Usually a rock painting site has multiple rock paintings, which are the evaluation samples.

[0157] Table 8 Comprehensive cloud parameters and evaluation results of 8 rock painting points in the Helankou rock painting area

[0158]

[0159] exist Figure 5 The evaluation results and effects can be intuitively seen in the Figure 5 a) in the figure is the index position of the comprehensive cloud map generated in step S7 in the evaluation grade standard cloud map generated in step 3, wherein the black cloud map is the evaluation result comprehensive cloud map generated in step S7, and its position in the evaluation grade standard cloud map represents the evaluation result. The remaining cloud maps are embodiments of the present invention, and only their evaluation results are shown here.

[0160] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for evaluating the development degree of rock painting spalling disease, characterized in that: The steps are as follows: Step S1: Based on the existing characteristics of rock painting spalling diseases, an evaluation index system is constructed, and evaluation indicators directly related to the existing characteristics are selected through field investigation and indoor experiments, and evaluation data corresponding to the evaluation indicators of each group of rock paintings at the rock painting site are collected; Step S2: determining the evaluation level of the development degree of the spalling disease and the corresponding evaluation interval according to the on-site investigation of the disease and the development characteristics of the disease; Step S3: converting the numerical value of the evaluation interval into the digital features of the cloud model, using the forward cloud generator to process the digital features to generate a standard cloud map of the evaluation interval, and obtaining a standard cloud map of the evaluation level; Step S4: input the evaluation data into the constructed BP neural network weighting model to obtain the weight of each evaluation index; Step S5: The evaluation data is preprocessed and brought into the reverse cloud generator to be converted into a qualitative language represented by three digital features: expectation, entropy, and super entropy, to obtain the index layer cloud parameters of the evaluation index; Step S6: Calculate each criterion stratus cloud parameter according to the weight of each evaluation index and the index stratus cloud parameter corresponding to the evaluation index; Step S7: Calculate the weight of the criterion layer according to the weight of the evaluation index, calculate the comprehensive cloud parameter using the cloud parameter of each criterion layer and the weight of the corresponding criterion layer, and input the comprehensive cloud parameter into the forward cloud generator to obtain the comprehensive cloud map; Step S8: Compare the comprehensive layer cloud map with the evaluation grade standard cloud map, and obtain the final evaluation grade according to the position of the comprehensive layer cloud map in the evaluation grade standard cloud map.

2. The method for evaluating the development degree of rock painting spalling disease according to claim 1 is characterized in that: The evaluation index system includes a target layer, a criterion layer and an index layer. The target layer is the development of exfoliation disease, and the criterion layer includes the mechanical properties and physical properties of the rock painting carrier, the crack characteristics within 300 mm around the rock painting, and the location of the rock painting. The evaluation indicators of the index layer of mechanical properties include rebound strength, compressive strength and longitudinal wave velocity, the evaluation indicators of the index layer of physical properties include porosity, water content and particle density, the evaluation indicators of the index layer of crack characteristics include the closest distance of cracks, crack length and maximum width of cracks, and the evaluation indicators of the index layer of rock painting location include relative elevation and rock surface inclination; The positive indicators of the evaluation index include porosity, water content, crack length, and maximum crack width; the negative indicators include rebound strength, compressive strength, longitudinal wave velocity, particle density, the closest distance between cracks, relative elevation, and rock surface inclination.

3. The method for evaluating the development degree of rock painting spalling disease according to claim 1 or 2, characterized in that: The evaluation levels of the development degree of the peeling disease are divided into 5 levels, corresponding to 5 evaluation intervals, namely [0-0.2], [0.2-0.4], [0.4-0.6], [0.6-0.8], and [0.8-1.0]; the corresponding evaluation levels are very low, low, medium, high, and very high.

4. The method for evaluating the development degree of rock painting spalling disease according to claim 3 is characterized in that: The method for converting the numerical value of the evaluation interval into the digital characteristics of the cloud model is as follows: the digital characteristics of the cloud model include three parameters, namely, expectation Ex, entropy En and super entropy He, and the three parameters are: in, is the minimum and maximum value of the nth evaluation interval, k is a constant; Ex n 、En n 、He n are the expectation, entropy and super entropy of the nth evaluation interval respectively.

5. The method for evaluating the development degree of rock painting spalling disease according to claim 4 is characterized in that: The method for generating an evaluation grade cloud map by processing digital features using a forward cloud generator is as follows: 1) Generate entropy En n is the expected value, He n 2 A normal random number y with variance n =R N (En n ,He n );R N (,) represents the normal random function; 2) Generate the expected Ex n is the expected value, y n 2 A normal random number x with variance m =R N (Ex n ,y n ); m is 1-M; 3) Calculate the degree of certainty 4) With certainty μ(x m ) of the normal random number x m Become a cloud drop in the universe; 5) Repeat steps 1)-4) until the required M cloud droplets are generated, and the M cloud droplets constitute a standard cloud image of an evaluation interval; Repeat the above steps to obtain the standard cloud map of each evaluation interval, and the standard cloud maps of all evaluation intervals constitute the standard cloud map of the evaluation grade.

6. The method for evaluating the development degree of rock painting spalling disease according to claim 3 or 5, characterized in that: The method for obtaining the weight of each evaluation index is: 1) Use MATLAB calculation tools to generate a BP neural network weighted model, set the maximum number of iterations and target error, and set the number of hidden layer neurons as the total number of evaluation indicators; 2) Import the evaluation data of the evaluation index into the BP neural network weighting model to generate an 18×11 data matrix data; 3) Call the calculate weights function to calculate the weighted average weights for the data matrix data.

7. The method for evaluating the development degree of rock painting spalling disease according to claim 6, characterized in that: The method of calculating the weighted average weights of the data matrix data by the calculate weights function is as follows: normalizing the input evaluation data, setting the activation functions of the hidden layer and the output layer, the activation function of the hidden layer is 'logsig', and the activation function of the output layer is 'purelin'; Set the training parameters, use the normalized evaluation data to train the BP neural network, extract the output layer weights of the BP neural network, calculate the weighted average weight of the output layer weights and the predicted output of the BP neural network, and obtain the weights of the evaluation indicators. The weights of all evaluation indicators are non-negative and their sum is 1.

8. The method for evaluating the development degree of rock painting spalling disease according to claim 1 or 7, characterized in that: The method for obtaining the index layer cloud parameters of the evaluation index is as follows: the preprocessing of the positive index is: Y ij =a+G(X ij -min(X ij )) The preprocessing of negative indicators is: Y ij =b+G(X ij -max(X ij )) Among them, the coefficient G is: b is the maximum value of all evaluation intervals, a is the minimum value of all evaluation intervals, Y ij is the preprocessing result of the evaluation data of the i-th evaluation index and the j-th group of rock paintings, X ij is the evaluation data of the i-th evaluation index and the j-th group of rock paintings, min and max represent the functions of finding the minimum and maximum values ​​respectively; The reverse cloud generator converts the J groups of preprocessed evaluation data Y of the i-th evaluation index ij Calculate the expected Ex of the i-th evaluation index i 、Entropy i 、Super entropy He i , as the index layer cloud parameter of the i-th evaluation index.

9. The method for evaluating the development degree of rock painting spalling disease according to claim 8, characterized in that: The method for calculating the cloud parameters of each criterion is: Expectation of the lth criterion level Entropy of the lth criterion layer The super entropy of the lth criterion layer Among them, Ex l1 、En l1 、He l1 are the expectation, entropy and super entropy of the l1th evaluation index of the lth criterion layer, respectively, l1 is the weight of the l1th evaluation index, L1 is the total number of evaluation indicators in the lth criterion layer; The method for calculating the comprehensive cloud parameters is: expect entropy Super Entropy Among them, λ l is the weight of the lth criterion layer, and L2 is the total number of criterion layers.

10. The method for evaluating the development degree of rock painting spalling disease according to claim 9, characterized in that: The method of bringing the synthetic cloud parameters into the forward cloud generator to obtain the synthetic layer cloud map is: 1) Generate entropy En A For the expected value, A normal random number y with variance A =R A (En A ,He A ); 2) Generate the expected Ex A is the expected value, y A 2 A normal random number x with variance A =R A (Ex A ,y A ); 3) Calculate the degree of certainty 4) With certainty μ(x A ) of the normal random number x A Become a cloud drop in the universe; 5) Repeat steps 1)-4) until the required M cloud droplets are generated, and all cloud droplets constitute a comprehensive layer cloud image; In the comparison between the comprehensive layer cloud map and the evaluation grade standard cloud map, the similarity between the digital features of the cloud model and the comprehensive cloud parameters is calculated based on the weighted Euclidean distance, the index of the evaluation grade standard cloud map with the maximum similarity and its corresponding index is found, and the evaluation grade of the comprehensive cloud parameters is determined according to the index; The similarity is the reciprocal of the weighted Euclidean distance, which is: Among them, the comprehensive cloud parameter is C A =(Ex A ,En A ,He A ), the cloud parameter C of the nth evaluation interval in the evaluation grade standard cloud map n =(Ex n ,En n ,He n );w Ex 、w En 、w He is the weight coefficient, and w Ex +w En +w He =1; The position of the comprehensive layer cloud map in the evaluation grade standard cloud map is the score of the evaluation object, which corresponds to the evaluation interval and determines the evaluation grade of the evaluation object.

Images