Cable hoisting falling risk early warning optimization method based on precision positioning

Abstract

The invention belongs to the field of cable hoisting, and particularly discloses a cable hoisting falling risk early warning optimization method based on precision positioning, and the method comprises the steps: installing positioning equipment, constructing a cable falling risk early warning model, and carrying out real-time early warning on the cable falling risk. According to the scheme, high-precision positioning equipment and sensors are installed at key monitoring points of a cable hoisting system, a regularization neighborhood component analysis method and a regularization support vector machine are combined, and feature weights are adjusted for multiple times by calculating the weighted distance between samples, a feature probability function and the optimal balance between regression precision and complexity of a regression model; the cable falling risk early warning model can better focus on key influence factors, Gaussian process prior and likelihood functions are used to construct posterior distribution, Gaussian kernel parameters and penalty parameters are continuously optimized, and the prediction precision of the cable falling risk early warning model is improved.

Description

Technical Field

[0001] The present invention relates to the field of cable hoisting, and specifically refers to an optimized method for early warning of cable hoisting fall risks based on precise positioning. Background Art

[0002] Cable hoisting systems are widely used in engineering construction such as bridges and large buildings. Monitoring the cable hoisting system can detect potential hazards in advance, enabling workers to have sufficient time to take protective measures, ensuring the smooth progress of the hoisting process, and helping to ensure the accurate installation of each component, thereby improving the overall quality of the project. However, it is difficult to monitor the cable hoisting system in a complex environment. Traditional monitoring methods cannot accurately locate key risk points, making it difficult to accurately predict the risk of cable fall in advance. Existing cable fall risk early warning models are not efficient enough in data processing. The models are easily affected by noise and missing values during training, resulting in poor reliability of the early warning results. When selecting model parameters, it usually relies on subjective experience or simple trial and error, and it is impossible to obtain optimal parameters, which affects the model performance. Summary of the Invention

[0003] In view of the above situation, to overcome the defects of the existing technology, the present invention provides an optimized method for early warning of cable hoisting fall risks based on precise positioning. Aiming at the technical problem that it is difficult to monitor the cable hoisting system in a complex environment, and traditional monitoring methods cannot accurately locate key risk points, making it difficult to accurately predict the risk of cable fall in advance, this solution installs high-precision positioning devices and sensors at key monitoring points of the cable hoisting system, and combines the regularized neighborhood component analysis method and the regularized support vector machine to build a cable fall risk early warning model, which can obtain the state information of the cable hoisting system in real time and accurately, capture potential fall risks more sensitively, and reduce the probability of cable fall accidents. Aiming at the technical problem that existing cable fall risk early warning models are not efficient enough in data processing, and the models are easily affected by noise and missing values during training, resulting in poor reliability of the early warning results, this solution adjusts the feature weights multiple times by calculating the weighted distance between samples, the feature probability function, and the best balance between the regression accuracy and complexity of the regression model, so that the cable fall risk early warning model can better focus on key influencing factors and improve the adaptability and learning ability of the cable fall risk early warning model to complex data. Aiming at the technical problem that when selecting model parameters, it usually relies on subjective experience or simple trial and error, and it is impossible to obtain optimal parameters, which affects the model performance, this solution uses the Gaussian process prior and likelihood function to construct the posterior distribution, and continuously optimizes the Gaussian kernel parameter and the penalty parameter, so that the cable fall risk early warning model can more accurately capture the complex relationship between the features of the training data when mapping to a high-dimensional space, and improve the prediction accuracy of the cable fall risk early warning model for fall risks.

[0004] The technical solution adopted by the present invention is as follows: The present invention provides an optimized method for warning of cable hoisting fall risks based on precise positioning. The optimized method for warning of cable hoisting fall risks based on precise positioning specifically includes the following steps:

[0005] Step S1: Install positioning equipment, determine the key monitoring points in the cable hoisting system, install high-precision positioning equipment and sensors at the key monitoring points, collect positioning data and sensing data. The sensing data includes environmental data and the state data of the hoisting equipment. Calibrate the positioning data according to the local geodetic coordinate reference, and align the sensing data on the time axis for time synchronization;

[0006] Step S2: Build a cable fall risk warning model. Through the regularized neighborhood component analysis method, build a cable fall risk warning model based on the regression model of the regularized support vector machine;

[0007] Step S3: Real-time warning of cable fall risks. Input the sensing data into the cable fall risk warning model. The cable fall risk warning model outputs the cable hoisting fall risk coefficient. The cable hoisting fall risk coefficient is an integer in the range of [0, 3]. The greater the cable hoisting fall risk coefficient, the higher the cable hoisting fall risk. When the cable hoisting fall risk coefficient is greater than or equal to 2, the system issues an alarm and sends the positioning data. Relevant personnel organize the crowd to stay away from the cable hoisting system according to the positioning data.

[0008] Further, in step S2, building a cable fall risk warning model specifically includes the following steps:

[0009] Step S21: Collect the cable fall risk data set. The cable fall risk data set includes a feature set and corresponding labels. And one piece of data in the cable fall risk data set is used as a sample. The feature set includes historical environmental data and historical hoisting equipment state data. The corresponding label is the fall risk coefficient corresponding to the historical environmental data and historical hoisting equipment state data;

[0010] Step S22: Preprocess the cable fall risk data set, remove the noise data and error data in the cable fall risk data set. For the missing values in the cable fall risk data set, use the linear interpolation method to supplement them. Divide the cable fall risk data set into a training set and a test set according to the ratio of 7:3;

[0011] Step S23: Perform regularized neighborhood component analysis on the feature set, initialize the initial weights of each feature, calculate the weighted distance between samples, remove the feature with the smallest weighted distance, and obtain the selected feature set. The formula used is as follows:

[0012]

[0013] In the formula, xi and x j are two different features, where i and j are the indices of the features, and d(x i , x j ) is the weighted distance between the two features, k is the index of the sample, n is the total number of samples, and c k is the distance weight of the k-th sample, and x ki and x kj are the i-th feature and the j-th feature of the k-th sample, respectively;

[0014] Step S24: Calculate the probability function of each feature and adjust the feature weights. The formula used is as follows:

[0015]

[0016] In the formula, p(x i ) is the probability of the i-th feature, and σ k is the scaling parameter of the feature of the k-th sample;

[0017] Step S25: Create and initialize a regression model based on a regularized support vector machine as the cable fall risk warning model, define the convex loss function of the regression model, and minimize the convex loss function. The formula used is as follows:

[0018]

[0019] In the formula, L is the loss function, b is the coefficient value of the regression variable, b T is the transpose of b, γ is the regularization parameter, and δ i are the non-negative slack parameters of the i-th feature exceeding the upper limit of the region and falling below the lower limit of the region, respectively;

[0020] Step S26: Input the samples in the selected feature set into the regression model, calculate the regression accuracy and complexity of each feature, calculate the minimized regularized loss function of the regression model, find the best balance between the regression accuracy and complexity, and adjust the feature weights again. The formula used is as follows:

[0021]

[0022] In the formula, L1 is the regularized loss function, λ is the regularization value, y k is the corresponding label in the cable fall risk dataset, is the predicted label of the regression model, p is the number of features in the selected feature set, and l is the index of the features in the selected feature set;

[0023] Step S27: Calculate the optimal Gaussian kernel parameter and penalty parameter, map the samples in the training set to a high-dimensional space using the Gaussian kernel function, and calculate the two Lagrange multipliers in the regression model. The used formulas are as follows:

[0024]

[0025] In the formula, F(r, h) is the Lagrangian function, r and h are the Lagrange multipliers, K is the Gaussian kernel function, and ε is the penalty parameter of the regression model;

[0026] Step S28: Use the Lagrange multipliers to adjust the regression model, output the final predicted label, that is, the falling risk coefficient, and obtain the trained cable falling risk warning model;

[0027] Further, in step S27, calculating the optimal Gaussian kernel parameter and penalty parameter specifically includes the following steps:

[0028] Step S271: Preset the ranges of the Gaussian kernel parameter and penalty parameter, randomly generate w combinations of the Gaussian kernel parameter and penalty parameter within the ranges, calculate the cross-validation error corresponding to each combination of the Gaussian kernel parameter and penalty parameter, and use the combination of the Gaussian kernel parameter and penalty parameter and the corresponding cross-validation error to construct an initial data set;

[0029] Step S272: Assume that the cross-validation error follows a Gaussian process prior, combine the Gaussian process prior distribution and the likelihood function to construct a posterior distribution, and calculate the posterior distribution of the new combination of the Gaussian kernel parameter and penalty parameter. The used formula is as follows:

[0030]

[0031] In the formula, P(e|D, {P}) is the posterior distribution, e is the cross-validation error, D is the initial data set, P is the new combination of the Gaussian kernel parameter and penalty parameter, N represents the Gaussian process prior, K is the covariance matrix of the initial data set, K * is the covariance matrix of all combinations of the Gaussian kernel parameter and penalty parameter, K ** is the autocovariance matrix of the new combination of the Gaussian kernel parameter and penalty parameter, I is the identity matrix, τ is the Gaussian noise, and m is the mean function of the Gaussian process;

[0032] Step S273: Randomly sample at the boundaries of the ranges of the Gaussian kernel parameter and penalty parameter, calculate the minimum value of the average of the posterior distribution, and optimize the combination of the Gaussian kernel parameter and penalty parameter in the initial data set through local search to obtain a new combination of the Gaussian kernel parameter and penalty parameter;

[0033] Step S274: Calculate the cross - validation error value of the new combination of Gaussian kernel parameters and penalty parameters, add the new combination of Gaussian kernel parameters and penalty parameters to the initial dataset, preset the maximum number of iterations, and repeat steps S272 to S274 until the maximum number of iterations is reached to obtain the optimal Gaussian kernel parameters and penalty parameters.

[0034] The beneficial effects achieved by the present invention using the above - mentioned solution are as follows:

[0035] (1) Aiming at the technical problem that it is difficult to monitor the cable - hoisting system in a complex environment, and the traditional monitoring method cannot accurately locate the key risk points, resulting in difficulty in accurately predicting the cable - falling risk in advance. In this solution, high - precision positioning devices and sensors are installed at the key monitoring points of the cable - hoisting system, and a cable - falling risk warning model is constructed by combining the regularized neighborhood component analysis method and the regularized support vector machine, which can obtain the state information of the cable - hoisting system in real - time and accurately, capture potential falling risks more sensitively, and reduce the occurrence probability of cable - falling accidents.

[0036] (2) Aiming at the technical problem that the existing cable - falling risk warning model is not efficient enough in data processing, and the model is easily affected by noise and missing values during training, resulting in poor reliability of the warning results. In this solution, the weighted distance between samples, the feature probability function, and the best balance between the regression accuracy and complexity of the regression model are calculated to adjust the feature weights multiple times, so that the cable - falling risk warning model can better focus on the key influencing factors and improve the adaptability and learning ability of the cable - falling risk warning model to complex data.

[0037] (3) Aiming at the technical problem that when selecting model parameters, it usually depends on subjective experience or simple trial - and - error, and the optimal parameters cannot be obtained, affecting the model performance. In this solution, the Gaussian process prior and likelihood function are used to construct the posterior distribution, and the Gaussian kernel parameters and penalty parameters are continuously optimized, so that the cable - falling risk warning model can more accurately capture the complex relationships between the features of the training data when mapping to a high - dimensional space, and improve the prediction accuracy of the cable - falling risk warning model for falling risks. Description of the Drawings

[0038] Figure 1 It is a step connection diagram of an optimized method for cable - hoisting falling risk warning based on precision positioning provided by the present invention;

[0039] Figure 2 It is a step connection diagram of constructing a cable - falling risk warning model provided in step S2 of the present invention.

[0040] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, but do not constitute a limitation to the present invention. Detailed Embodiments

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0042] Embodiment 1: Refer to Figure 1 , this embodiment provides an optimized method for warning of cable hoisting fall risks based on precision positioning. The optimized method for warning of cable hoisting fall risks based on precision positioning specifically includes the following steps:

[0043] Step S1: Install positioning devices, determine the key monitoring points in the cable hoisting system, install high-precision positioning devices and sensors at the key monitoring points, collect positioning data and sensing data. The sensing data includes environmental data and the state data of the hoisting equipment. Calibrate the positioning data according to the local geodetic coordinate reference, and align the sensing data on the time axis for time synchronization.

[0044] Step S2: Construct a cable fall risk warning model. Through the regularized neighborhood component analysis method, construct a cable fall risk warning model based on the regression model of the regularized support vector machine.

[0045] Step S3: Real-time warning of cable fall risks. Input the sensing data into the cable fall risk warning model. The cable fall risk warning model outputs the cable hoisting fall risk coefficient. The cable hoisting fall risk coefficient is an integer in the range of [0, 3]. The larger the cable hoisting fall risk coefficient, the higher the cable hoisting fall risk. When the cable hoisting fall risk coefficient is greater than or equal to 2, the system issues an alarm and sends the positioning data. Relevant personnel organize the crowd to stay away from the cable hoisting system according to the positioning data.

[0046] Through the above embodiments, aiming at the technical problems that it is difficult to monitor the cable hoisting system in a complex environment, the traditional monitoring method cannot accurately locate the key risk points, resulting in difficulty in accurately warning the cable fall risk in advance. This solution installs high-precision positioning devices and sensors at the key monitoring points of the cable hoisting system, constructs a cable fall risk warning model by combining the regularized neighborhood component analysis method and the regularized support vector machine, can obtain the state information of the cable hoisting system in real time and accurately, capture potential fall risks more sensitively, and reduce the occurrence probability of cable fall accidents.

[0047] Embodiment 2: Based on the above embodiment, when selecting a high-precision positioning device, it is selected according to actual applications. For cable hoisting in an outdoor open environment, a high-precision GPS locator can be selected. In scenarios with obstructions or high-precision requirements, indoor positioning devices based on ultra-wideband technology or optical positioning devices such as total stations can be used;

[0048] This solution can be applied to bridge hoisting. In a bridge cable hoisting system, the key monitoring points include cable connection points, the mid-span part of the cable, the anchor system, and the cable-bridge connection points;

[0049] The relevant characteristics affecting the fall of cable hoisting include: cable strength, cable tension, cable coating erosion, cable anchor, meteorological conditions, and geological conditions;

[0050] Correspondingly, the sensors used are fiber Bragg grating sensors, pressure sensors and tension sensors, eddy current sensors, displacement sensors and tilt sensors, anemometers and wind vanes, and temperature and humidity sensors, and ground stress sensors.

[0051] Embodiment 3: Refer to Figure 1 and Figure 2 , this embodiment is based on the above embodiment. In step S2, a cable fall risk warning model is constructed, which specifically includes the following steps:

[0052] Step S21: Collect a cable fall risk data set. The cable fall risk data set includes a feature set and corresponding labels, and one piece of data in the cable fall risk data set is used as a sample. The feature set includes historical environmental data and historical hoisting equipment status data, and the corresponding label is the fall risk coefficient corresponding to the historical environmental data and historical hoisting equipment status data;

[0053] Step S22: Preprocess the cable fall risk data set, remove the noise data and error data in the cable fall risk data set. For the missing values in the cable fall risk data set, linear interpolation is used for supplementation, and the cable fall risk data set is divided into a training set and a test set according to a ratio of 7:3;

[0054] Step S23: Perform regularized neighborhood component analysis on the feature set, initialize the initial weights of each feature, calculate the weighted distance between samples, remove the feature with the smallest weighted distance, and obtain the selected feature set. The formula used is as follows:

[0055]

[0056] In the formula, x i and x j are two different features, i and j are the indices of the features, d(x i ,x j) is the weighted distance between two features, k is the index of the sample, n is the total number of samples, and c k is the distance weight of the k-th sample, and x ki and x kj are the i-th feature and the j-th feature of the k-th sample respectively;

[0057] Step S24: Calculate the probability function of each feature and adjust the feature weights. The formula used is as follows:

[0058]

[0059] In the formula, p(x i ) is the probability of the i-th feature, and σ k is the scaling parameter of the feature of the k-th sample;

[0060] Step S25: Create and initialize a regression model based on a regularized support vector machine as the cable fall risk warning model, define the convex loss function of the regression model, and minimize the convex loss function. The formula used is as follows:

[0061]

[0062] In the formula, L is the loss function, b is the coefficient value of the regression variable, b T is the transpose of b, γ is the regularization parameter, and δ i are the non-negative slack parameters of the i-th feature exceeding the upper limit of the region and falling below the lower limit of the region respectively;

[0063] Step S26: Input the samples in the selected feature set into the regression model, calculate the regression accuracy and complexity of each feature, calculate the minimized regularized loss function of the regression model, find the best balance between the regression accuracy and complexity, and adjust the feature weights again. The formula used is as follows:

[0064]

[0065] In the formula, L1 is the regularized loss function, λ is the regularization value, y k is the corresponding label in the cable fall risk dataset, is the predicted label of the regression model, p is the number of features in the selected feature set, and l is the index of the features in the selected feature set;

[0066] Step S27: Calculate the optimal Gaussian kernel parameter and penalty parameter, map the samples in the training set to a high-dimensional space using the Gaussian kernel function, and calculate the two Lagrange multipliers in the regression model. The formula used is as follows:

[0067]

[0068] In the formula, F(r, h) is the Lagrangian function, r and h are the Lagrange multipliers, K is the Gaussian kernel function, and ε is the penalty parameter of the regression model;

[0069] Step S28: Use the Lagrange multipliers to adjust the regression model, output the final predicted label, that is, the falling risk coefficient, and obtain the trained cable falling risk warning model.

[0070] Through the above embodiments, aiming at the technical problems that the existing cable falling risk warning model is not efficient enough in data processing, the model is easily affected by noise and missing values during training, resulting in poor reliability of the warning results. This solution adjusts the feature weights multiple times by calculating the weighted distance between samples, the feature probability function, and the best balance between the regression accuracy and complexity of the regression model, so that the cable falling risk warning model can better focus on the key influencing factors and improve the adaptability and learning ability of the cable falling risk warning model to complex data.

[0071] Example 4: Refer to Figure 1 and Figure 2 , based on the above Example 3, the regression model outputs the final predicted label, and the formula used is as follows:

[0072]

[0073] In the formula, is the final predicted label output by the regression model, x k is the feature of the kth sample, is the feature of the test set.

[0074] Example 5: Refer to Figure 1 and Figure 2 , based on the above embodiments, in step S27, calculate the optimal Gaussian kernel parameter and penalty parameter, which specifically includes the following steps:

[0075] Step S271: Preset the ranges of the Gaussian kernel parameter and the penalty parameter, randomly generate w combinations of the Gaussian kernel parameter and the penalty parameter within the ranges, calculate the cross-validation error corresponding to each combination of the Gaussian kernel parameter and the penalty parameter, and use the combination of the Gaussian kernel parameter and the penalty parameter and the corresponding cross-validation error to construct an initial data set;

[0076] Step S272: Assume that the cross-validation error follows a Gaussian process prior, combine the Gaussian process prior distribution and the likelihood function to construct a posterior distribution, and calculate the posterior distribution of the new combination of the Gaussian kernel parameter and the penalty parameter. The formula used is as follows:

[0077]

[0078] Wherein, P(e|D,{P}) is the posterior distribution, e is the cross-validation error, D is the initial data set, P is the combination of new Gaussian kernel parameters and penalty parameters, N represents the Gaussian process prior, K is the covariance matrix of the initial data set, K * is the covariance matrix of all combinations of Gaussian kernel parameters and penalty parameters, K ** is the autocovariance matrix of the new combination of Gaussian kernel parameters and penalty parameters, I is the identity matrix, τ is Gaussian noise, and m is the mean function of the Gaussian process;

[0079] Step S273: Randomly sample at the boundaries of the Gaussian kernel parameter and penalty parameter ranges, calculate the minimum value of the average of the posterior distribution, and optimize the combination of Gaussian kernel parameters and penalty parameters in the initial data set through local search to obtain a new combination of Gaussian kernel parameters and penalty parameters;

[0080] Step S274: Calculate the cross-validation error value of the new combination of Gaussian kernel parameters and penalty parameters, add the new combination of Gaussian kernel parameters and penalty parameters to the initial data set, preset the maximum number of iterations, and repeat Steps S272 to S274 until the maximum number of iterations is reached to obtain the optimal Gaussian kernel parameters and penalty parameters.

[0081] Through the above embodiments, for the technical problem that when selecting model parameters, it usually depends on subjective experience or simple trial and error, and the optimal parameters cannot be obtained, affecting the model performance, this solution uses the Gaussian process prior and the likelihood function to construct the posterior distribution, and continuously optimizes the Gaussian kernel parameters and penalty parameters, so that the cable fall risk warning model can more accurately capture the complex relationships between the features of the training data when mapping in the high-dimensional space, and improve the prediction accuracy of the cable fall risk warning model for the fall risk.

[0082] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0083] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

[0084] The above describes the present invention and its implementation manners. Such description is not restrictive. What is shown in the drawings is only one of the implementation manners of the present invention, and the actual structure is not limited thereto. All in all, if those of ordinary skill in the art are inspired by it and, without departing from the gist of the present invention, creatively design structural manners and embodiments similar to the technical solution, they shall fall within the protection scope of the present invention.

Claims

1. A cable hoisting fall risk warning optimization method based on precision positioning, characterized in that: The specific steps include: Step S1: Install positioning equipment, determine key monitoring points in the cable hoisting system, install high-precision positioning equipment and sensors at the key monitoring points, and collect positioning data and sensor data; Step S2: constructing a cable fall risk warning model, and constructing a cable fall risk warning model based on a regularized support vector machine regression model by using a regularized neighborhood component analysis method; Step S3: Real-time warning of cable falling risk. Input the sensor data into the cable falling risk warning model, and output the cable hoisting falling risk coefficient. The cable hoisting falling risk coefficient is an integer of [0, 3]. When the cable hoisting falling risk coefficient is greater than or equal to 2, the system will issue an alarm and send positioning data.

2. According to the method for optimizing the early warning of cable hoisting falling risk based on precision positioning according to claim 1, it is characterized in that: Step S2, constructing a cable fall risk warning model, specifically comprising the following steps: Step S21: collecting a cable fall risk data set, wherein the cable fall risk data set includes a feature set and a corresponding label, and a piece of data in the cable fall risk data set is used as a sample; Step S22: preprocessing the cable fall risk data set, removing noise data and erroneous data in the cable fall risk data set, supplementing the missing values ​​in the cable fall risk data set by linear interpolation, and dividing the cable fall risk data set into a training set and a test set in a ratio of 7:3; Step S23: Perform regularized neighborhood component analysis on the feature set, initialize the initial weight of each feature, calculate the weighted distance between samples, remove the feature with the smallest weighted distance, and obtain the selected feature set. The formula used is as follows: In the formula, x i and x j are two different features, i and j are the indexes of the features, d(x i ,x j ) is the weighted distance between two features, k is the index of the sample, n is the total number of samples, c k is the distance weight of the kth sample, x ki and x kj They are the i-th feature and j-th feature of the k-th sample respectively; Step S24: Calculate the probability function of each feature and adjust the feature weight; Step S25: creating and initializing a regression model based on a regularized support vector machine as a cable fall risk warning model, defining a convex loss function of the regression model, and minimizing the convex loss function; Step S26: Input the samples in the selected feature set into the regression model, calculate the regression accuracy and complexity of each feature, calculate the minimum regularized loss function of the regression model, find the best balance between regression accuracy and complexity, and adjust the feature weights again. The formula used is as follows: In the formula, L1 is the regularized loss function, λ is the regularization value, and y k is the corresponding label in the cable fall risk dataset, is the predicted label of the regression model, p is the number of features in the selected feature set, and l is the index of the feature in the selected feature set; Step S27: Calculate the optimal Gaussian kernel parameters and penalty parameters, use the Gaussian kernel function to map the samples in the training set to a high-dimensional space, and calculate two Lagrange multipliers in the regression model; Step S28: Use Lagrange multipliers to adjust the regression model, output the final prediction label, that is, the fall risk coefficient, and obtain a trained cable fall risk warning model.

3. The cable hoisting fall risk early warning optimization method based on precision positioning according to claim 2 is characterized in that: Step S27, calculating the optimal Gaussian kernel parameters and penalty parameters, specifically includes the following steps: Step S271: Preset the range of Gaussian kernel parameters and penalty parameters, randomly generate w combinations of Gaussian kernel parameters and penalty parameters within the range, calculate the cross-validation error corresponding to each combination of Gaussian kernel parameters and penalty parameters, and use the Gaussian kernel parameters and penalty parameter combinations and the corresponding cross-validation errors to construct an initial data set; Step S272: Assuming that the cross-validation error obeys the Gaussian process prior, the Gaussian process prior distribution and the likelihood function are combined to construct the posterior distribution, and the posterior distribution of the new Gaussian kernel parameter and penalty parameter combination is calculated. The formula used is as follows: Where P(e|D,{P}) is the posterior distribution, e is the cross-validation error, D is the initial data set, P is the new Gaussian kernel parameter and penalty parameter combination, N represents the Gaussian process prior, K is the covariance matrix of the initial data set, and K * is the covariance matrix of all Gaussian kernel parameter and penalty parameter combinations, K ** is the autocovariance matrix of the new Gaussian kernel parameter and penalty parameter combination, I is the identity matrix, τ is the Gaussian noise, and m is the mean function of the Gaussian process; Step S273: randomly sampling at the boundary of the Gaussian kernel parameter and penalty parameter range, calculating the minimum value of the mean value of the posterior distribution, optimizing the Gaussian kernel parameter and penalty parameter combination in the initial data set through local search, and obtaining a new Gaussian kernel parameter and penalty parameter combination; Step S274: Calculate the cross-validation error value of the new Gaussian kernel parameter and penalty parameter combination, add the new Gaussian kernel parameter and penalty parameter combination to the initial data set, preset the maximum number of iterations, repeat steps S272 to S274 until the maximum number of iterations is reached, and obtain the optimal Gaussian kernel parameters and penalty parameters.

Images