Crane safety assessment weight dynamic distribution method based on improved analytic hierarchy process

Abstract

The invention discloses a crane safety assessment weight dynamic distribution method based on an improved analytic hierarchy process, which comprises the following steps: when a bridge crane works or is tested, obtaining each characteristic index information, and then effectively fusing the information into weight distribution according to correlation information among indexes; according to the method, the interpretability of AHP consistency inspection and the adaptivity of data driving are both considered, reliable updating of weights is achieved, the scientificity and precision of comprehensive safety evaluation can be remarkably improved, and a reliable basis is provided for real-time monitoring and early warning of the bridge crane; according to the scheme, the dynamic correction factor containing the average relevancy is introduced, and through a Softmax correction mechanism, the robustness of a small sample or an abnormal fluctuation scene can be improved; in the aspects of implementation flexibility and reliability, the scheme only needs common statistics and matrix operation, does not need complex machine learning training or large-scale historical data, is suitable for a bridge crane real-time monitoring system with limited sample size, and is easy to deploy and apply in engineering.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of crane safety monitoring, and particularly relates to a crane safety evaluation weight dynamic distribution method based on an improved analytic hierarchy process. BACKGROUND

[0002] With the development of industrial automation and intelligence, the running safety of a bridge crane, as a key equipment in the fields of ports, metallurgy and mines, is directly related to production efficiency and personnel safety. However, the existing crane safety evaluation methods mainly focus on single index or static weight multi-index evaluation, which has many defects or limitations. For example, in the single index evaluation method, a single index such as critical stress, maximum deformation or safety factor is usually used for evaluation, and this method ignores the coupling influence of multiple source factors (such as structural stress, material stiffness and running stability) on the crane in actual operation, resulting in one-sided evaluation results and insufficient reliability. Some literatures and engineering practices introduce AHP to weight and comprehensively evaluate multiple indexes, but the fixed weight is usually determined by experts before system deployment, and the response capability to data changes in the subsequent operation process is poor. In addition, when the sample size is small or the data fluctuates greatly, the static weight cannot accurately reflect the importance of the indexes, and there are problems such as strong subjectivity, difficulty in ensuring consistency and insufficient evaluation accuracy.

[0003] Therefore, how to evaluate the safety state of the crane by using comprehensive characteristic indexes in the environment of small sample and multiple source detection data is a very positive and practical research topic. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a crane safety evaluation weight dynamic distribution method based on an improved analytic hierarchy process, which is flexible, reliable and has high data fusion reliability.

[0005] In order to achieve the above technical purpose, the technical solution adopted by the present application is:

[0006] A crane safety evaluation weight dynamic distribution method based on an improved analytic hierarchy process, comprising:

[0007] S01, determining characteristic indexes during the working monitoring of the bridge crane, collecting and summarizing the characteristic indexes during the working or testing of the bridge crane, generating an original data matrix X about the characteristic index monitoring, and then performing standardization processing to obtain a dimensionless standardized matrix Z;

[0008] S02, calculating the Pearson correlation coefficient between the characteristic indexes based on the standardized matrix Z to obtain a symmetric parameter correlation matrix R;

[0009] S03, constructing an AHP judgment matrix according to the parameter correlation matrix R, and performing consistency checking processing, solving the characteristic value of the judgment matrix passing the consistency checking to obtain an initial static weight, and calculating a dynamic correction factor of the characteristic index according to the parameter correlation matrix R;

[0010] S04, combining the static weight and the dynamic correction factor, coupling them through a softmax mechanism to obtain a dynamically corrected weight, and then using the dynamically corrected weight for crane safety assessment.

[0011] As a possible implementation, further, the characteristic index described in the scheme includes one or more of stress, stiffness, and stability.

[0012] In the scheme S01, the standardization processing of the original data matrix X to obtain the dimensionless standardized matrix Z includes: calculating the sample mean and sample standard deviation of each characteristic index, then performing Z-score standardization processing on each characteristic index, and then collecting the index data after Z-score standardization processing to obtain the standardized matrix Z.

[0013] Specifically, the scheme records the original data, assuming that under p sampling moments, the original value matrix of m characteristic indexes is collected, which is defined as follows:

[0014]

[0015] Wherein, x ij represents the measurement value of the i-th characteristic index at the j-th sampling time;

[0016] The calculation of the sample mean and sample standard deviation of each characteristic index includes the following:

[0017] The sample mean μ i of the i-th characteristic index is calculated, which is defined as follows:

[0018]

[0019] Wherein, p is the sample number of the i-th characteristic index, x ij is the measurement value of the i-th characteristic index at the j-th sampling time;

[0020] The sample standard deviation σ i of the i-th characteristic index is calculated, which is defined as follows:

[0021]

[0022] Wherein, p is the sample number of the i-th characteristic index, x ij is the measurement value of the i-th characteristic index at the j-th sampling time, and μi is the sample mean of the i-th characteristic index.

[0023] measurement value of the i-th feature index based on each feature index ij , sample mean μ i , sample standard deviation σ i , Z-score standardization processing is performed, which is defined as follows:

[0024]

[0025] wherein Z ij is the value of the i-th feature index measurement value after standardization processing at the j-th sampling; x ij is the measurement value of the i-th feature index at the j-th sampling; μ i is the sample mean of the i-th feature index, σ i is the sample standard deviation of the i-th feature index;

[0026] The data of the m feature indexes after standardization processing is collected to obtain a standardized matrix Z, which is defined as follows:

[0027]

[0028] As a preferred implementation option, preferably, in the scheme S02, the Pearson correlation coefficient between the feature indexes is calculated based on the standardized matrix Z to obtain a symmetric parameter correlation matrix R, including:

[0029] For the standardized matrix Z, the Pearson correlation coefficient r ik between the i-th feature index and the k-th feature index is defined as follows:

[0030]

[0031] wherein the sample number of the i-th feature index and the k-th feature index is p, Z ij , Z kj are the values of the i-th feature index and the k-th feature index measurement value after standardization processing at the j-th sampling, respectively.

[0032] As a preferred implementation option, preferably, in the scheme, the symmetric parameter correlation matrix R is obtained according to the Pearson correlation coefficient between the feature indexes, which is defined as follows:

[0033]

[0034] wherein r ik ∈[-1, 1], which is a linear correlation coefficient; and r ii = 1, r ik = r ki .

[0035] As a preferred implementation option, preferably, the scheme S03, the AHP judgment matrix is constructed according to the parameter correlation matrix R, and the consistency checking process includes:

[0036] The judgment matrix A is constructed, and the Pearson correlation coefficient r in the parameter correlation matrix R is mapped into a scale according to a preset condition ik The value, which is defined as follows:

[0037] A = [a ik ]

[0038]

[0039] Wherein, a ik is the matrix element of the judgment matrix A, s ik is the preset scale score, which is used to reflect the importance of the characteristic index i relative to the characteristic index k;

[0040] The AHP judgment matrix is solved to obtain the maximum eigenvalue λ max , which is defined as follows:

[0041] Aw = λ max w

[0042] Wherein, w = [w i ], which is the vector corresponding to the maximum eigenvalue λ max , A is the judgment matrix;

[0043] The consistency index CI and the consistency ratio CR are used for consistency checking;

[0044] Wherein, the consistency index CI is defined as follows:

[0045]

[0046] Wherein, m is the matrix order, which corresponds to the number of characteristic indexes, and the consistency index CI is used to describe the degree of deviation of the matrix from consistency;

[0047] The consistency ratio CR is defined as follows:

[0048]

[0049] Wherein, RI m is the random consistency index corresponding to the matrix order m, which is queried through a pre-built control table, CI is the consistency index, and if the consistency ratio CR is <0.10, it is considered that the judgment matrix A is logically consistent, otherwise, the value of the preset scale score s ik needs to be further adjusted.

[0050] As a preferred implementation option, preferably, in the scheme S03, solving the eigenvalue of the judgment matrix passing the consistency test to obtain the initial static weight comprises:

[0051] For the eigenvalue, the judgment matrix is solved, which is defined as follows:

[0052] Aw = λw

[0053] where w = [w1, w2, ……wn] is the eigenvector corresponding to the weight of each index without normalization, λ is the eigenroot, which has m solutions, where the largest eigenroot is denoted as λ i , A is the judgment matrix. T max , the eigenroot is denoted as λ i , A is the judgment matrix.

[0054] The preliminary weight vector w is normalized to convert it to percentage form, and the sum of the weights is 1 to satisfy comparability and interpretability, which is defined as follows:

[0055]

[0056] The normalization process is defined as follows:

[0057]

[0058] where i is 1, 2, …, m; let After normalization, the static weight vector w is obtained.

[0059] As a preferred implementation option, preferably, in the scheme S03, the dynamic correction factor includes the average correlation degree φ i and the sensitivity coefficient β; wherein the sensitivity coefficient β is a preset self-defined value greater than 0;

[0060] Calculating the dynamic correction factor of the characteristic index according to the parameter correlation matrix R comprises:

[0061] The average correlation degree φ i is calculated, which is defined as follows:

[0062]

[0063] where φ i ∈ [0, 1], the larger the value, the higher the overall correlation of the i-th index, and the weight sensitivity needs to be moderately amplified; m is the number of characteristic indexes; r ik is the Pearson correlation coefficient of the characteristic index i and the characteristic index k.

[0064] As a preferred implementation option, preferably, in the scheme S04, the static weight and the dynamic correction factor are coupled through the softmax mechanism to obtain the dynamically corrected weight, which includes:

[0065] The weighting amplification based on the static weight and the dynamic correction factor is defined as follows:

[0066]

[0067] wherein, is the amplified temporary weight, β is the sensitivity coefficient, which is a preset self-defined value greater than 0, the greater the value, the more intense the dynamic correction; exp(βφ i ) is the calculation coefficient for exponential amplification of the static weight of the high correlation index;

[0068] The temporary weight is normalized, and the definition is as follows:

[0069]

[0070] wherein, let The weights wi obtained by normalization are collected to obtain the dynamically corrected weight vector w′.

[0071] As a preferred implementation option, preferably, the scheme further includes:

[0072] S05, repeating steps S01-S04 to dynamically update the safety evaluation weight of the characteristic index of the bridge crane in operation.

[0073] Based on the above, the scheme further proposes a crane safety evaluation method, which applies the improved AHP-based crane safety evaluation weight dynamic allocation method described above, which includes:

[0074] Determine the characteristic index of the bridge crane during operation monitoring. During the operation or testing of the bridge crane, the characteristic index is collected and summarized based on the characteristic index;

[0075] According to the characteristic index obtained by summarizing and the weight allocation value corresponding to the characteristic index, the safety state of the bridge crane is evaluated, and the definition is as follows:

[0076]

[0077] wherein, j=1, 2……, p; S j is the comprehensive reliability score of the sample at the jth sampling, m is the number of sampling items; Z ij is the measurement value of the characteristic index i at the jth sampling after standardization processing; w′ i is the dynamically corrected weight value of the characteristic index i.

[0078] According to the comprehensive reliability score S j , the working safety state of the bridge crane is determined.

[0079] Based on the above, the present scheme also proposes a crane safety evaluation system based on improved AHP, which applies the crane safety evaluation weight dynamic allocation method based on improved AHP or the crane safety evaluation method described above, and it includes:

[0080] The data acquisition module is used to determine the characteristic indicators during the working monitoring of the bridge crane, and based on the characteristic indicators, the original data matrix X related to the characteristic indicator monitoring is generated by collecting and summarizing during the working or testing of the bridge crane.

[0081] The data standardization module is used to standardize the original data matrix X to obtain the dimensionless standardized matrix Z.

[0082] The correlation calculation module is used to calculate the Pearson correlation coefficient between the characteristic indicators based on the standardized matrix Z to obtain the symmetric parameter correlation matrix R.

[0083] The weight correction module is used to construct the AHP judgment matrix according to the parameter correlation matrix R, and perform consistency check processing, solve the eigenvalue of the judgment matrix that passes the consistency check to obtain the initial static weight, and also calculate the dynamic correction factor of the characteristic indicators according to the parameter correlation matrix R.

[0084] The weight updating module is used to couple the static weight and the dynamic correction factor through the softmax mechanism to obtain the dynamically corrected weight.

[0085] The safety evaluation module is used to evaluate the safety state of the bridge crane according to the summarized characteristic indicators and the weight distribution value corresponding to the characteristic indicators, and at the same time, determine the working safety state of the bridge crane according to the evaluation result, and finally output the state information of the bridge crane.

[0086] By adopting the technical scheme described above, the present application has the beneficial effects compared with the prior art: the present scheme determines the characteristic indicators during the working monitoring of the bridge crane, obtains the information of each characteristic indicator during the working or testing of the bridge crane, and then effectively fuses the correlation information between the indicators into the weight distribution; it takes into account the explainability of AHP consistency check and the adaptability of data-driven, realizes reliable updating of the weight, and can significantly improve the scientificity and precision of the comprehensive safety evaluation, providing a reliable basis for real-time monitoring and early warning of the bridge crane.

[0087] The scheme constructs a Pearson correlation matrix between indexes, which can quantitatively reflect the coupling relationship of multiple influence factors such as stress, stiffness and stability, avoid one-sided evaluation caused by a single index, improve the comprehensiveness of evaluation, and introduce a dynamic correction factor containing average correlation and a Softmax correction mechanism to directly incorporate real-time data correlation into weight calculation, so as to realize online automatic updating of the weight; when the correlation between indexes changes, the weight of each index can be quickly adjusted, and the robustness to small sample or abnormal fluctuation scene is improved; in terms of implementation flexibility and reliability, the scheme only needs common statistics and matrix operations without complex machine learning training or large-scale historical data, and is suitable for real-time monitoring system of the bridge crane with limited sample size, and is easy to deploy and apply in engineering. BRIEF DESCRIPTION OF DRAWINGS

[0088] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, a brief introduction will be given below to the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0089] Figure 1 is a brief implementation process schematic diagram of the weight dynamic evaluation method of the scheme;

[0090] Figure 2 is a unit module connection schematic diagram of the crane safety evaluation system of the scheme. DETAILED DESCRIPTION

[0091] The present application will be further described in detail below in combination with the drawings and embodiments. It is particularly pointed out that the following embodiments are only used to illustrate the present application, but do not limit the scope of the present application. Similarly, the following embodiments are only some embodiments of the present application, not all embodiments, and all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0092] In combination with Figure 1 shown, the embodiment scheme is a crane safety evaluation weight dynamic allocation method based on an improved analytic hierarchy process, which includes:

[0093] S01, determining the characteristic indexes of the bridge crane during monitoring, collecting and summarizing the characteristic indexes during the operation or test of the bridge crane, generating an original data matrix X about the characteristic index monitoring, and then performing standardization processing to obtain a dimensionless standardized matrix Z;

[0094] S02, based on the standardized matrix Z, the Pearson correlation coefficient between the characteristic indicators is calculated to obtain a symmetric parameter correlation matrix R;

[0095] S03, according to the parameter correlation matrix R, the AHP judgment matrix is constructed, and the consistency check processing is carried out, and the eigenvalue of the judgment matrix passing the consistency check is solved to obtain the initial static weight, and the dynamic correction factor of the characteristic indicator is calculated according to the parameter correlation matrix R;

[0096] S04, combining the static weight and the dynamic correction factor, the dynamic correction weight is obtained through the coupling of the softmax mechanism, and then the dynamic correction weight is used for the crane safety assessment.

[0097] As a possible implementation, further, the characteristic indicators described in the scheme include one or more of stress, stiffness and stability.

[0098] In order to eliminate the influence of dimension and order of magnitude between each characteristic indicator, and provide comparable basic data for subsequent correlation analysis and weight calculation, in S01 of the scheme, the original data matrix X is standardized to obtain the dimensionless standardized matrix Z, which includes: calculating the sample mean and sample standard deviation of each characteristic indicator, then performing Z-score standardization processing on each characteristic indicator, and then collecting the indicator data after Z-score standardization processing to obtain the standardized matrix Z.

[0099] Specifically, the scheme records the original data, assuming that under p sampling time, the original value matrix of m characteristic indicators is collected, which is defined as follows:

[0100]

[0101] Wherein, x ij represents the measurement value of the i-th characteristic indicator at the j-th sampling time;

[0102] The calculation of the sample mean and sample standard deviation of each characteristic indicator includes the following:

[0103] The sample mean μ i of the i-th characteristic indicator is calculated, which is defined as follows:

[0104]

[0105] Wherein, p is the sample number of the i-th characteristic indicator, x ij is the measurement value of the i-th characteristic indicator at the j-th sampling time.

[0106] The sample standard deviation σ i of the i-th characteristic indicator is calculated, which is defined as follows:

[0107]

[0108] wherein p is the sample number of the ith feature index, x ij is the measurement value of the ith feature index at the jth sampling, μ i is the sample mean of the ith feature index.

[0109] In the present scheme, the sample mean μ i is used to represent the central position of the corresponding index, and the sample standard deviation σ i is used to measure the dispersion degree of the index.

[0110] On the basis of the above, based on the measurement value x ij of each feature index, the sample mean μ i , and the sample standard deviation σ i , Z-score standardization processing is performed, which is defined as follows:

[0111]

[0112] wherein Z ij is the value of the measurement value of the ith feature index at the jth sampling after standardization processing; x ij is the measurement value of the ith feature index at the jth sampling; μ i is the sample mean of the ith feature index, σ i is the sample standard deviation of the ith feature index.

[0113] The data of the m feature indexes after standardization processing is collected to obtain a standardized matrix Z, which is defined as follows:

[0114]

[0115] Through the standardized data, each index can be converted into dimensionless data with a mean of 0 and a variance of 1, which is used to eliminate the differences in dimension and magnitude.

[0116] In order to quantify the linear correlation between each index, and to provide a basis for judgment matrix initialization and dynamic factor involvement, as a preferred implementation choice, preferably, in the present scheme S02, the Pearson correlation coefficient between the feature indexes is calculated based on the standardized matrix Z to obtain a symmetric parameter correlation matrix R, including:

[0117] For the standardized matrix Z, the Pearson correlation coefficient r ik between the feature index i and the feature index k is defined as follows:

[0118]

[0119] wherein the sample number of the characteristic index i and the characteristic index k are both p, and Z ij , Z kj are the values of the measurement of the characteristic index i and the characteristic index k at the jth sampling after standardization, respectively.

[0120] As a preferred implementation option, preferably, the scheme obtains a symmetric parameter correlation matrix R according to the Pearson correlation coefficients between the characteristic indexes, which is defined as follows:

[0121]

[0122] wherein r ik ∈ [-1, 1] is a linear correlation coefficient; and r ii = 1, r ik = r ki .

[0123] By constructing the symmetric parameter correlation matrix R, highly correlated and redundant indexes can be identified, and data reference can be provided for constructing the initial judgment matrix.

[0124] Therefore, as a preferred implementation option, preferably, the scheme S03 constructs the AHP judgment matrix according to the parameter correlation matrix R, and performs consistency check processing, which includes:

[0125] The judgment matrix A is constructed, and the Pearson correlation coefficient r ik in the parameter correlation matrix R is mapped into a scale (e.g., saaty 1-9 level) according to a preset condition, and the judgment matrix A is defined as follows:

[0126] A = [a ik ]

[0127]

[0128] wherein a ik is a matrix element of the judgment matrix A, s ik is a preset scale score, which is used to reflect the importance of the characteristic index i relative to the characteristic index k; wherein s ik may also be given a corresponding numerical value by expert evaluation method. As an example, if r ik ≈ 0.8, then a ik is mapped to 5.

[0129] As an example of mapping the Pearson correlation coefficient r ik in the parameter correlation matrix R into a scale, it can be represented as follows:

[0130] Without considering the influence of the sign on the "importance" measurement, the absolute correlation coefficient |r ik| to do mapping, threshold mapping table as follows:

[0131]

[0132]

[0133] In the judgment matrix A generation, first, define the diagonal elements, let a ii = 1, wherein, i = 1, 2 … m; for non-diagonal elements, for any i ≠ good case, first defined ρ = | r ik |, then the table ρ mapping into the corresponding scale s, namely,

[0134] a ik = s ik At the same time, meet AHP reciprocal, which is defined as

[0135] On the basis of the above, AHP judgment matrix Eigenvalue solution, get the largest eigenvalue λ max , defined as follows:

[0136] Aw= λ max w

[0137] Where, w = [w i ], which is the corresponding maximum eigenvalue λ max Vector, A is the judgment matrix;

[0138] By consistency index CI and consistency ratio CR consistency test;

[0139] Wherein, the definition of consistency index CI as follows:

[0140]

[0141] Where, m is the order of the matrix, which corresponds to the number of characteristic index, consistency index CI for describing the degree of deviation from the consistency of the matrix;

[0142] The definition of consistency ratio CR as follows:

[0143]

[0144] Where, RI m For the random consistency index corresponding to the order of the matrix m, through the pre-built control table for inquiry, CI is the consistency index, if the consistency ratio CR < 0.10, then the judgment matrix A is considered to be logically consistent, otherwise, the need to further adjust the preset scale value s ik .

[0145] In terms of adjusting weights and weight influencing factors, as a preferred implementation option, preferably, in this solution S03, solving the eigenvalues ​​of the judgment matrix that passes the consistency test to obtain the initial static weights includes:

[0146] For eigenvalues, based on the judgment matrix Solve the eigenvalue, which is defined as follows:

[0147] Aw=λw

[0148] Where w=[w1, w2, ... w i ] T , which is the eigenvector corresponding to the unnormalized weight of each indicator, λ is the eigenroot, which has m solutions, of which the largest eigenroot is denoted as λ max , A is the judgment matrix;

[0149] The initial weight vector w is normalized and converted into a percentage form, so that the sum of the weights is 1 to ensure comparability and interpretability. It is defined as follows:

[0150]

[0151] The normalization process is defined as follows:

[0152]

[0153] Among them, i is 1, 2...m; let After normalization, the static weight vector w is obtained.

[0154] In this solution S03, the dynamic correction factor includes the average correlation φ i and a sensitivity coefficient β; wherein the sensitivity coefficient β is a preset custom value greater than 0;

[0155] The dynamic correction factors of characteristic indicators calculated according to the parameter correlation matrix R include:

[0156] Calculate the average correlation φ i , which is defined as follows:

[0157]

[0158] Among them, φ i ∈[0,1], the larger its value is, the higher the overall correlation of the i-th indicator is, and its weight sensitivity needs to be moderately amplified; m is the number of characteristic indicators; r ik is the Pearson correlation coefficient between feature index i and feature index k.

[0159] In the weight correction aspect, as a preferred implementation option, in the scheme S04, the static weight and the dynamic correction factor are coupled through the softmax mechanism to obtain the dynamically corrected weight, including:

[0160] The weighted amplification based on the static weight and the dynamic correction factor is defined as follows:

[0161]

[0162] wherein, is the temporary weight after amplification, β is the sensitivity coefficient, which is a preset self-defined value greater than 0, the larger the value, the more intense the dynamic correction; exp(βφ i ) is the calculation coefficient for exponential amplification of the static weight of the high correlation index;

[0163] The temporary weight is normalized, and the definition is as follows:

[0164]

[0165] wherein, let The weights w' i obtained after normalization are collected to obtain the dynamically corrected weight vector w.

[0166] Based on the above, in order to facilitate dynamic adjustment of the weight of the feature index, as a preferred implementation option, the scheme further includes:

[0167] S05, repeating steps S01-S04 to realize dynamic updating of the safety evaluation weight of the feature index of the bridge crane during operation.

[0168] Based on the above, the scheme further proposes a crane safety evaluation method, which applies the crane safety evaluation weight dynamic distribution method based on the improved analytic hierarchy process described above, which includes:

[0169] Determine the feature index during the operation monitoring of the bridge crane, and collect and summarize the feature index during the operation or test of the bridge crane based on the feature index;

[0170] According to the feature index and the weight distribution value corresponding to the feature index obtained by summarizing, the safety state of the bridge crane is evaluated, and the definition is as follows:

[0171]

[0172] wherein, j = 1, 2 …, p; S j is the comprehensive reliability score of the sample at the jth sampling, and m is the number of sampling items; Z ijis the value of the measurement of the feature index i at the jth sampling after standardization; w i is the dynamically corrected weight value of the feature index i;

[0173] According to the comprehensive reliability score S j , the working safety state of the bridge crane is determined.

[0174] As an embodiment, the present scheme takes structural stress, material stiffness, and stability index as feature indexes, which are substituted into the following examples to illustrate the method of the present scheme:

[0175] 1. Raw data

[0176]

[0177]

[0178] Stress is the internal force per unit area of the main beam of the bridge crane metal structure under the action of load, which reflects the load strength limit of the material. When the stress exceeds the standard, it may cause plastic deformation, cracks, or even rupture of the structure. In the present scheme, strain gauges are pasted on the dangerous cross sections of the main beam span and end, the surface is polished and cleaned with acetone, then the strain is measured through a half-bridge / full-bridge circuit, and the stress is converted.

[0179] Correspondingly, the present scheme also collects stiffness and stability index through existing technical devices, which will not be described here.

[0180] The sample mean and sample standard deviation of each feature index are calculated, which include the following:

[0181] The sample mean μ of the ith feature index is calculated i , which is defined as follows:

[0182]

[0183] where p is the sample size of the ith feature index, in the present example, p = 5, x ij is the measurement value of the ith feature index at the jth sampling.

[0184] The sample standard deviation σ of the ith feature index is calculated i , which is defined as follows:

[0185]

[0186] where p is the sample size of the ith feature index, in the present example, p = 5, x ij is the measurement value of the ith feature index at the jth sampling, μ iThe sample mean of the ith feature index.

[0187] In this scheme, the sample mean μ i is used for the central position of the reaction corresponding index, and the sample standard deviation σ i is used to measure the dispersion of the index.

[0188] In this example, the sample mean and sample standard deviation corresponding to the stress, stiffness, and stability index are shown in the following table:

[0189]

[0190] On the basis of the above, based on the measured value x ij of each feature index, the sample mean μ i , and the sample standard deviation σ i , Z-score standardization processing is performed, which is defined as follows:

[0191]

[0192] where Z ij is the value of the ith feature index after standardization processing at the jth sampling; x ij is the measured value of the ith feature index at the jth sampling; μ i is the sample mean of the ith feature index, and σ i is the sample standard deviation of the ith feature index.

[0193] 2. After standardization processing of the original data, the following standardized data is obtained:

[0194]

[0195]

[0196] Based on the standardized matrix Z, the Pearson correlation coefficient between the feature indexes is calculated, and a symmetric parameter correlation matrix R is obtained, including:

[0197] For the standardized matrix Z, the Pearson correlation coefficient between the ith feature index and the kth feature index is

[0198] r ik is defined as follows:

[0199]

[0200] where the sample size of the ith feature index and the kth feature index is p, and in this scheme, p = 5, Z ij and Z kj are the values of the ith feature index and the kth feature index after standardization processing at the jth sampling, respectively.

[0201] Pearson correlation coefficient r as pressure characteristic index and rigidity characteristic index 压力-刚度 One calculation example is calculated as follows:

[0202]

[0203] Similarly, the Pearson correlation coefficients between the pressure characteristic index, the rigidity characteristic index, and the stability index characteristic index are calculated, and then fitted into the following parameter correlation matrix R, which is defined as follows:

[0204]

[0205] Without considering the influence of positive and negative signs on "importance" measurement, use the absolute correlation coefficient |r ik |to do mapping, and the threshold mapping table is as follows:

[0206]

[0207] In terms of the generation of judgment matrix A, first, define the diagonal elements, let a ii =1, where i=1, 2, …, m; for non-diagonal elements, for any i≠k, first define ρ=|r ik |, and then map the ρ in the table above to the corresponding scale s, that is,

[0208] a ik =s ik , at the same time, satisfy the AHP reciprocal, which is defined as

[0209] Then the final judgment matrix A is represented as follows:

[0210]

[0211] For the eigenvalue, based on the judgment matrix , the eigenvalue is solved, which is defined as follows:

[0212] Aw=λw

[0213] Where λ max =3.3276, w=[2.5316, 0.6365, 0.1600] T .

[0214] The preliminary weight vector w is normalized to convert it to percentage form, and the sum of the weights is 1 to satisfy comparability and interpretability. The normalization process is defined as follows:

[0215]

[0216] wherein i is 1, 2, 3; let

[0217] After normalization processing, the static weight vector

[0218] The consistency test is performed through the consistency index CI and the consistency ratio CR.

[0219] wherein the consistency index CI is defined as follows:

[0220]

[0221] In this example, the random consistency index RI corresponding to the matrix order m is m = 0.58, in which case the consistency ratio CR is defined as follows:

[0222]

[0223] wherein CR = 0.2824 > 0.1, so it is considered that the logical consistency of the judgment matrix A is not ideal and needs to be manually corrected.

[0224] In the calculation of the average correlation degree φ i , the definition is as follows:

[0225]

[0226] In this example,

[0227]

[0228] wherein in this example, the sensitivity coefficient β takes a value of 1; in this scheme S04, the static weight and the dynamic correction factor are coupled through the softmax mechanism, and the weight after dynamic correction is obtained, including:

[0229] Based on the static weight and the dynamic correction factor for weighted amplification, the definition is as follows:

[0230]

[0231] wherein, is the temporary weight after amplification, β is the sensitivity coefficient, which is a preset self-defined value greater than 0, the larger the value, the more intense the dynamic correction; exp(βφ i ) is a calculation coefficient for exponential amplification of the static weight of the high correlation index.

[0232] Through the static weight vector , w 压力 = 0.7628,

[0233] w 刚度 = 0.1902, w 稳定性指标 = 0.0480.

[0234] wherein, the temporary weight is calculated as follows:

[0235]

[0236] The temporary weight is normalized, which is defined as follows:

[0237]

[0238] The following dynamic corrected weight is calculated:

[0239]

[0240] On the basis of the above, the following standardized matrix data obtained from the original data is taken as an example, and its comprehensive reliability score is calculated.

[0241] For the data at the jth sampling, its standardized value Z j = [Z 1j , Z 2j , Z 3j ] T , specifically, Z1 = (-0.7483, -0.4472, 1.2584).

[0242] The relevant specific data are shown in the following table

[0243]

[0244] The comprehensive reliability score calculation definition for evaluating the safety state of the bridge crane is as follows:

[0245]

[0246] S1 = 0.7767 * (-0.7483) + 0.1741 * (-0.4472) + 0.0492 * 1.2584 = -0.5971.

[0247] The data of other groups are calculated in the same way as above, and will not be described here. After obtaining the comprehensive reliability score, the safety condition of the bridge crane can be further judged by combining the preset threshold value.

[0248] As an example, the following table definition can be made for the comprehensive reliability score:

[0249]

[0250] According to the above table, it can be known that in the working condition of Z1, immediate stop inspection or maintenance needs to be performed on the bridge crane to eliminate safety hazards.

[0251] In combination Figure 2 Based on the above, the present application also proposes a crane safety evaluation system based on an improved analytic hierarchy process, which applies the crane safety evaluation weight dynamic distribution method based on the improved analytic hierarchy process described above or the crane safety evaluation method described above, and includes:

[0252] A data acquisition module is configured to determine characteristic indexes during monitoring of the bridge crane, and collect and aggregate the characteristic indexes during operation or testing of the bridge crane to generate an original data matrix X related to the characteristic index monitoring.

[0253] A data standardization module is configured to perform standardization processing on the original data matrix X to obtain a dimensionless standardized matrix Z.

[0254] A correlation calculation module is configured to calculate Pearson correlation coefficients between the characteristic indexes based on the standardized matrix Z to obtain a symmetric parameter correlation matrix R.

[0255] A weight correction module is configured to construct an AHP judgment matrix according to the parameter correlation matrix R, perform consistency check processing, solve the characteristic values of the judgment matrix that passes the consistency check to obtain initial static weights, and calculate a dynamic correction factor of the characteristic indexes according to the parameter correlation matrix R.

[0256] A weight updating module is configured to couple the static weights and the dynamic correction factor through a softmax mechanism to obtain the dynamically corrected weights.

[0257] A safety evaluation module is configured to evaluate the safety state of the bridge crane according to the aggregated characteristic indexes and the weight distribution values corresponding to the characteristic indexes, determine the working safety state of the bridge crane according to the evaluation result, and finally output the state information of the bridge crane.

[0258] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0259] If the integrated unit is implemented in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0260] The above only describes some embodiments of the present application, and does not limit the protection scope of the present application. Any equivalent device or equivalent process transformation, or direct or indirect application in other related technical fields based on the content of the present application specification and drawings are also included in the patent protection scope of the present application.

Claims

1. A crane safety assessment weight dynamic allocation method based on improved analytic hierarchy process, characterized in that: It includes: S01. Determine characteristic indicators for monitoring the operation of a bridge crane. When the bridge crane is operating or being tested, collect and summarize the characteristic indicators to generate a raw data matrix X for monitoring the characteristic indicators. Then, perform normalization on the raw data matrix X to obtain a dimensionless normalized matrix Z. S02. Based on the standardized matrix Z, calculate the Pearson correlation coefficient between the characteristic indicators to obtain a symmetrical parameter correlation matrix R; S03. Constructing an AHP judgment matrix based on the parameter correlation matrix R and performing a consistency check, solving the eigenvalues ​​of the judgment matrix that passes the consistency check to obtain the initial static weights, and also calculating the dynamic correction factors of the characteristic indicators based on the parameter correlation matrix R; S04. Combine the static weight and the dynamic correction factor, couple them through the softmax mechanism, obtain the dynamically corrected weight, and then use the dynamically corrected weight for crane safety assessment.

2. The crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 1 is characterized in that: The characteristic index includes one or more of stress, stiffness, and stability; In S01, the original data matrix X is normalized to obtain a dimensionless normalized matrix Z, including: calculating the sample mean and sample standard deviation of each characteristic indicator, performing Z-score normalization on each characteristic indicator, and then aggregating the indicator data after Z-score normalization to obtain the normalized matrix Z; Among them, for the record of original data, it is assumed that at p sampling moments, a total of m original value matrices of characteristic indicators are collected, which are defined as follows: Among them, x ij It represents the measured value of the i-th characteristic index at the j-th sampling time; The sample mean and sample standard deviation of each characteristic index are calculated as follows: Calculate the sample mean μ of the i-th feature index i , which is defined as follows: Among them, p is the number of samples of the i-th feature index, x ij is the measured value of the i-th characteristic index at the j-th sampling time; Calculate the sample standard deviation σ of the i-th characteristic index i , which is defined as follows: Among them, p is the number of samples of the i-th feature index, x ij is the measured value of the i-th characteristic index at the j-th sampling, μ i is the sample mean of the i-th feature index; Based on the measurement value x of each characteristic indicator ij , sample mean μ i , sample standard deviation σ i , perform Z-score standardization, which is defined as follows: Among them, Z ij is the value of the measured value of the i-th characteristic index after normalization at the j-th sampling; x ij is the measured value of the i-th characteristic index at the j-th sampling; μ i is the sample mean of the i-th characteristic index, σ i is the sample standard deviation of the i-th characteristic indicator; The standardized data of m characteristic indicators are collected to obtain the standardized matrix Z, which is defined as follows:

3. The crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 1 or 2, characterized in that: In S02, based on the standardized matrix Z, the Pearson correlation coefficient between the characteristic indicators is calculated to obtain the symmetrical parameter correlation matrix R including: For the standardized matrix Z, the Pearson correlation coefficient r of the characteristic index i and the characteristic index k ik The definition is as follows: Among them, the number of samples of characteristic index i and characteristic index k is p, Z ij , Z kj are the standardized values ​​of the characteristic index i and characteristic index k at the jth sampling time; According to the Pearson correlation coefficient between the characteristic indicators, the symmetrical parameter correlation matrix R is obtained, which is defined as follows: Among them, r ik ∈[-1,1], which is the linear correlation coefficient; and r ii =1, r ik =r ki .

4. The crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 3 is characterized in that: S03. Constructing an AHP judgment matrix based on the parameter correlation matrix R and performing consistency check processing includes: Construct the judgment matrix A, according to the preset conditions based on the Pearson correlation coefficient r in the parameter correlation matrix R ik The values ​​are mapped to scales, which are defined as follows: A=[a ik ] Among them, a ik is the matrix element of the judgment matrix A, s ik is a preset scale score, which is used to reflect the importance of feature index i relative to feature index k; AHP judgment matrix Solve the eigenvalue and obtain the maximum eigenvalue λ max , which is defined as follows: Aw = λ max w Where w=[w i ], which corresponds to the maximum eigenvalue λ max A is the judgment matrix; The consistency test is carried out through the consistency index CI and consistency ratio CR; Among them, the definition of consistency index CI is as follows: Where m is the matrix order, which corresponds to the number of terms in the characteristic index, and the consistency index CI is used to describe the degree to which the matrix deviates from consistency; The definition of the consistency ratio CR is as follows: Among them, RI m It is a random consistency index corresponding to the matrix order m, which is queried through a pre-built comparison table. CI is a consistency index. If the consistency ratio CR is less than 0.10, it is considered that the logic of the judgment matrix A is consistent. Otherwise, it is necessary to further adjust the preset scale score s. ik The numerical value of .

5. The crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 4 is characterized in that: In S03, solving the eigenvalue of the judgment matrix that passes the consistency test to obtain the initial static weight includes: For eigenvalues, based on the judgment matrix Solve the eigenvalue, which is defined as follows: Aw=λw Where w=[w1, w2, ... w i ] T , which is the eigenvector corresponding to the unnormalized weight of each indicator, λ is the eigenroot, which has m solutions, of which the largest eigenroot is denoted as λx ax , A is the judgment matrix; The initial weight vector w is normalized and converted into a percentage form, so that the sum of the weights is 1 to ensure comparability and interpretability. It is defined as follows: The normalization process is defined as follows: Where i is 1, 2...m; let After normalization, the static weight vector w is obtained.

6. The crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 5 is characterized in that: In S03, the dynamic correction factor includes the average correlation φ i and a sensitivity coefficient β; wherein the sensitivity coefficient β is a preset custom value greater than 0; The dynamic correction factors of characteristic indicators calculated according to the parameter correlation matrix R include: Calculate the average correlation φ i , which is defined as follows: Among them, φ i ∈[0,1], the larger its value is, the higher the overall correlation of the i-th indicator is, and its weight sensitivity needs to be moderately amplified; m is the number of characteristic indicators; r ik is the Pearson correlation coefficient between feature index i and feature index k.

7. The crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 6 is characterized in that: In S04, the static weights and dynamic correction factors are combined and coupled through the softmax mechanism to obtain the dynamically corrected weights: Weighted amplification is performed based on static weights and dynamic correction factors, which are defined as follows: in, is the amplified temporary weight, β is the sensitivity coefficient, which is a preset custom value greater than 0. The larger the value, the more severe the dynamic correction; exp(βφ i ) is the calculation coefficient for exponentially amplifying the static weight of the highly correlated index; The temporary weights are normalized and are defined as follows: Among them, The weight w′ obtained by normalization i Aggregate and obtain the dynamically corrected weight vector w′.

8. The crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 7 is characterized in that: It also includes: S05. Repeat steps S01-S04 to dynamically update the safety assessment weights of the characteristic indicators of the bridge crane during operation.

9. A crane safety assessment method, which uses the crane safety assessment weight dynamic allocation method based on the improved analytic hierarchy process according to claim 8, characterized in that: It includes: Determine the characteristic indicators when monitoring the operation of the bridge crane, and collect and summarize them based on the characteristic indicators when the bridge crane is working or testing; The safety status of the bridge crane is evaluated based on the summarized characteristic indicators and the weight distribution values ​​corresponding to the characteristic indicators, which are defined as follows: Where, j = 1, 2, ..., p; S j is the comprehensive reliability score of the sample at the jth sampling time, m is the number of sampled items; Z ij is the value of the characteristic index i measured at the jth sampling after normalization; w′ i is the weight value of characteristic index i after dynamic correction; According to the comprehensive reliability score S j , determine the working safety status of the bridge crane.

10. A crane safety assessment system based on an improved analytic hierarchy process, which uses the crane safety assessment weight dynamic allocation method based on an improved analytic hierarchy process according to any one of claims 1 to 8 or the crane safety assessment method according to claim 9, characterized in that: It includes: The data acquisition module is used to determine the characteristic indicators of the bridge crane during operation monitoring. When the bridge crane is working or testing, the characteristic indicators are collected and summarized to generate the original data matrix X for the characteristic indicator monitoring; The data normalization module is used to normalize the original data matrix X to obtain a dimensionless normalized matrix Z; The correlation calculation module is used to calculate the Pearson correlation coefficient between the characteristic indicators based on the standardized matrix Z to obtain the symmetrical parameter correlation matrix R; The weight correction module is used to construct the AHP judgment matrix according to the parameter correlation matrix R, and perform consistency test processing, solve the eigenvalue of the judgment matrix that passes the consistency test to obtain the initial static weight, and calculate the dynamic correction factor of the characteristic index according to the parameter correlation matrix R; The weight update module is used to combine the static weight and the dynamic correction factor, couple them through the softmax mechanism, and obtain the dynamically corrected weight; The safety assessment module is used to evaluate the safety status of the bridge crane based on the summarized characteristic indicators and the weight distribution values ​​corresponding to the characteristic indicators. At the same time, based on the assessment results, the working safety status of the bridge crane is determined and finally the bridge crane status information is output.

Images