Degradation path approximation-based barrel ablation wear prediction method and device

Abstract

The embodiment of the invention provides a degradation path approximation-based barrel ablation wear prediction method and device, and the method comprises the steps: building a high-order feature space through the normalization processing and Sigmoid function mapping of ablation wear data; similarity measurement features and path bending rate features of degradation paths are extracted innovatively, and accurate clustering and path approximation of degradation modes are realized. The system adopts an alternating optimization algorithm and a gradient descent method to optimize path expression parameters, and establishes a confidence interval estimation mechanism. According to the method, the barrel temperature, the shooting pressure and the propellant powder filling amount are combined to construct a comprehensive load index, accurate prediction and evaluation of the ablation wear state of the barrel are achieved through the change rate and acceleration characteristics of a degradation trend curve, and a scientific basis is provided for use and maintenance of artillery.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a method and device for predicting barrel ablation wear based on degradation path approximation. Background Art

[0002] The existing barrel ablation wear prediction technology has significant shortcomings. Traditional prediction methods mostly use a single model for analysis, which fails to fully consider the path differences under different degradation modes, resulting in a lack of universality in the prediction results. At the same time, the existing methods have limitations in processing high-dimensional nonlinear degradation data, and it is difficult to accurately capture the complex degradation evolution laws.

[0003] In addition, existing technologies also have bottlenecks in predictive uncertainty quantification and parameter optimization. Most methods lack a reliable confidence interval estimation mechanism and cannot effectively evaluate the credibility of the prediction results. The geometric characteristics and similarity information of the degradation path are not fully utilized during the model parameter optimization process, which affects the prediction accuracy.

[0004] The existing system has technical shortcomings in comprehensive load assessment and state judgment. It lacks unified consideration of multiple factors such as barrel temperature, firing pressure, and propellant loading, making it difficult to accurately reflect the ablation and wear process under actual use conditions. Solving these problems is of great significance to improving the scientific nature of artillery life prediction and maintenance decisions. Summary of the invention

[0005] In response to the problems in the prior art, the present application provides a method and device for predicting barrel ablation and wear based on degradation path approximation, which can achieve accurate prediction and evaluation of barrel ablation and wear status, and provide a scientific basis for artillery use and maintenance.

[0006] In order to solve at least one of the above problems, the present application provides the following technical solutions: In a first aspect, the present application provides a method for predicting barrel ablation wear based on degradation path approximation, comprising: The collected data of ablation and wear of the gun barrel are normalized, and a multidimensional degradation path sample space is established based on the normalized ablation and wear data. The sample space is mapped to a high-order feature space using a Sigmoid function, and similarity measurement features and path curvature features of the degradation path are extracted in the high-order feature space. The degradation path is clustered into multiple degradation mode clusters according to the similarity measurement features, and the path curvature features of each degradation mode cluster are fitted using a curve fitting algorithm to obtain path approximate expressions under different degradation modes. Iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on the least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on an output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; The real-time firing data of the artillery to be predicted is read, a comprehensive load index is calculated according to the barrel temperature, firing pressure and propellant loading amount contained in the firing data, the comprehensive load index is input into the barrel ablation degradation prediction model, and the barrel ablation wear prediction value and prediction interval under different firing numbers are obtained, a degradation trend curve of barrel ablation wear is established based on the prediction value, the change rate and acceleration characteristics of barrel ablation wear are calculated according to the degradation trend curve, and the ablation wear state of the gun barrel is judged in combination with the prediction interval.

[0007] Furthermore, the collected data of ablation and wear of the gun barrel is normalized, a multidimensional degradation path sample space is established based on the normalized ablation and wear data, the sample space is mapped to a high-order feature space using a Sigmoid function, and similarity measurement features and path curvature features of the degradation path are extracted in the high-order feature space, including: Collecting ablation wear depth data, barrel pressure data and barrel temperature data of a gun barrel under different firing numbers, performing maximum and minimum value normalization processing on the ablation wear depth data, calculating a comprehensive load coefficient in combination with the barrel pressure data and barrel temperature data, and constructing a multidimensional feature vector with the comprehensive load coefficient and the normalized ablation wear depth data; A degenerate path sample space is established based on the multidimensional feature vector, and a Sigmoid function is used to perform nonlinear mapping on the degenerate path sample space to obtain a high-dimensional feature expression. In the high-dimensional feature expression, the Mahalanobis distance between the degenerate paths is calculated to obtain a similarity measurement feature, and a differential geometry method is used to calculate the curvature change of the degenerate path to obtain a path curvature feature.

[0008] Furthermore, the degradation paths are clustered into a plurality of degradation mode clusters according to the similarity measurement characteristics, and the path curvature characteristics of each degradation mode cluster are fitted using a curve fitting algorithm to obtain path approximate expressions under different degradation modes, including: Based on the similarity measurement feature, a distance matrix of the degradation path is constructed, a hierarchical clustering algorithm is used to perform cluster analysis on the distance matrix, the degradation path is divided into a plurality of degradation pattern clusters according to the cluster analysis result, the cluster center of each degradation pattern cluster is calculated, and the cluster center is subjected to feature decomposition to obtain the main degradation features; The path curvature characteristics in each of the degradation pattern clusters are normalized, and the normalized path curvature characteristics are fitted using a polynomial fitting algorithm. The coefficient matrix of the fitting curve is calculated, and an approximate expression of the degradation path is established based on the coefficient matrix. The approximate expression is combined with the main degradation characteristics to construct a complete path approximation model.

[0009] Furthermore, the iterative optimization of the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on a least squares criterion, and solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination include: A parameter vector to be optimized is set in the path approximation expression, an objective function for parameter optimization is constructed based on the least squares criterion, the objective function includes a path approximation error term and a parameter regularization term, a partial derivative of the objective function with respect to the parameter vector is obtained, and a parameter update direction and step size are calculated according to the gradient expression; An alternating iterative strategy is used to perform block optimization on the parameter vector. In each round of iteration, some parameters are fixed and other parameters are updated. The objective function value after each parameter update is calculated. The iteration is stopped when the relative change of the objective function value is less than the preset convergence threshold, and the finally converged parameter vector is taken as the optimal parameter combination.

[0010] Furthermore, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on the output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty, comprises: Substituting the optimal parameter combination into the parameter vector in the path approximation expression, mapping the path approximation expression to the prediction space using a kernel function, building a nonlinear prediction model in combination with the main degradation characteristics of the degradation path, calibrating the parameters of the nonlinear prediction model based on training samples, and generating a barrel ablation degradation prediction model; The probability distribution characteristics of the output results of the barrel ablation degradation prediction model are calculated, the variance of the predicted value is estimated by the delta method, the standard error of the predicted result is calculated based on the variance, the confidence coefficient is determined according to a preset confidence level, the confidence coefficient is multiplied by the standard error to obtain the half-width of the prediction interval, and the half-width of the prediction interval is superimposed on the predicted value to construct the ablation wear prediction interval.

[0011] Furthermore, the real-time firing data of the artillery to be predicted is read, a comprehensive load index is calculated according to the barrel temperature, firing pressure and propellant charge contained in the firing data, the comprehensive load index is input into the barrel ablation degradation prediction model, and the barrel ablation wear prediction value and its prediction interval under different firing numbers are obtained, including: Acquire firing data of the artillery to be predicted, extract barrel temperature data, firing pressure data and propellant charge data from the firing data, normalize the barrel temperature data, firing pressure data and propellant charge data, calculate the comprehensive load index by weighted superposition, and determine the weight coefficient of each index according to historical data; The comprehensive load index is input into the barrel ablation degradation prediction model according to the sequence of firing numbers, and the ablation wear prediction value corresponding to each firing number is calculated based on the prediction model. The upper confidence limit and lower confidence limit of each prediction value are calculated in combination with the pre-constructed prediction interval, and the prediction value and its upper and lower confidence limits constitute a complete ablation wear prediction result.

[0012] Further, the degradation trend curve of barrel ablation wear is established based on the predicted value, the change rate and acceleration characteristics of barrel ablation wear are calculated according to the degradation trend curve, and the ablation wear state of the gun barrel is judged in combination with the predicted interval, including: The predicted value is numerically smoothed, a piecewise polynomial interpolation method is used to construct a continuous degradation trend curve, a first-order derivative of the degradation trend curve is calculated to obtain a change rate characteristic of ablation wear, a second-order derivative is calculated to obtain an acceleration characteristic of ablation wear, and key characteristic points of the degradation process are calculated according to the change rate characteristic and the acceleration characteristic; The degradation process is divided into multiple stages based on the key characteristic points, and the reliability index of each stage is calculated in combination with the upper confidence limit and the lower confidence limit of the prediction interval. The ablation and wear state of the gun barrel is quantitatively evaluated according to the preset state evaluation criteria to generate a barrel state evaluation result.

[0013] In a second aspect, the present application provides a barrel ablation wear prediction device based on degradation path approximation, comprising: A path expression construction module is used to normalize the collected data of ablation and wear of the gun barrel, establish a multidimensional degradation path sample space based on the normalized ablation and wear data, use a Sigmoid function to map the sample space to a high-order feature space, and extract similarity measurement features and path curvature features of the degradation path in the high-order feature space, cluster the degradation path into multiple degradation mode clusters according to the similarity measurement features, use a curve fitting algorithm to fit the path curvature features of each degradation mode cluster, and obtain path approximate expressions under different degradation modes; a wear interval prediction module, for iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on a least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on an output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; The wear state prediction module is used to read the real-time shooting data of the artillery to be predicted, calculate the comprehensive load index according to the barrel temperature, shooting pressure and propellant loading contained in the shooting data, input the comprehensive load index into the barrel ablation degradation prediction model, obtain the barrel ablation wear prediction value and its prediction interval under different shooting numbers, establish the barrel ablation wear degradation trend curve based on the prediction value, calculate the change rate and acceleration characteristics of the barrel ablation wear according to the degradation trend curve, and judge the ablation wear state of the artillery barrel in combination with the prediction interval.

[0014] In a third aspect, the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the method for predicting barrel ablation and wear based on degradation path approximation are implemented.

[0015] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for predicting barrel ablation and wear based on degradation path approximation.

[0016] In a fifth aspect, the present application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the method for predicting barrel ablation and wear based on degradation path approximation.

[0017] It can be seen from the above technical scheme that the present application provides a method and device for predicting barrel ablation wear based on degradation path approximation, and establishes a high-order feature space by normalizing the ablation wear data and mapping the sigmoid function. The similarity measurement characteristics and path curvature characteristics of the degradation path are innovatively extracted to achieve accurate clustering and path approximation of the degradation pattern. The system uses an alternating optimization algorithm and a gradient descent method to optimize the path expression parameters and establish a confidence interval estimation mechanism. A comprehensive load index is constructed by combining the barrel temperature, firing pressure and propellant loading. Through the rate of change and acceleration characteristics of the degradation trend curve, the barrel ablation wear state is accurately predicted and evaluated, providing a scientific basis for the use and maintenance of artillery. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is one of the flow charts of the barrel ablation wear prediction method based on degradation path approximation in the embodiment of the present application; Figure 2 This is a second flow chart of a method for predicting barrel ablation wear based on degradation path approximation in an embodiment of the present application; Figure 3 This is a third flow chart of a method for predicting barrel ablation wear based on degradation path approximation in an embodiment of the present application; Figure 4 This is a fourth flow chart of a method for predicting barrel ablation wear based on degradation path approximation in an embodiment of the present application; Figure 5 FIG5 is a flow chart of a method for predicting barrel ablation wear based on degradation path approximation in an embodiment of the present application; Figure 6 FIG6 is a flow chart of a method for predicting barrel ablation wear based on degradation path approximation in an embodiment of the present application; Figure 7 FIG7 is a flow chart of a method for predicting barrel ablation wear based on degradation path approximation in an embodiment of the present application; Figure 8 is a structural diagram of a barrel ablation wear prediction device based on degradation path approximation in an embodiment of the present application; Fig. 9 It is a schematic diagram of the structure of an electronic device in an embodiment of the present application.

[0020] Reference numerals: Electronic device 9600, central processing unit 9100, memory 9140, communication module 9110, input unit 9120, audio processor 9130, display 9160, power supply 9170, buffer memory 9141, application / function storage unit 9142, data storage unit 9143, driver program storage unit 9144, antenna 9111, speaker 9131, microphone 9132. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0022] The acquisition, storage, use, and processing of data in the technical solution of this application comply with the relevant provisions of national laws and regulations.

[0023] Taking into account the problems existing in the prior art, the present application provides a method and device for predicting barrel ablation and wear based on degradation path approximation, and establishes a high-order feature space by normalizing the ablation and wear data and mapping it with Sigmoid functions. The similarity measurement features and path curvature features of the degradation path are innovatively extracted to achieve accurate clustering and path approximation of the degradation pattern. The system uses an alternating optimization algorithm and a gradient descent method to optimize the path expression parameters and establish a confidence interval estimation mechanism. A comprehensive load index is constructed by combining barrel temperature, firing pressure, and propellant loading. Through the rate of change and acceleration characteristics of the degradation trend curve, the barrel ablation and wear state is accurately predicted and evaluated, providing a scientific basis for the use and maintenance of artillery.

[0024] In order to accurately predict and evaluate the ablation and wear state of the barrel and provide a scientific basis for the use and maintenance of the artillery, the present application provides an embodiment of a method for predicting ablation and wear of the barrel based on degenerate path approximation, see Figure 1 The barrel ablation wear prediction method based on degradation path approximation specifically includes the following contents: Step S101: normalizing the collected data of ablation and wear of the gun barrel, establishing a multidimensional degradation path sample space based on the normalized ablation and wear data, mapping the sample space to a high-order feature space using a Sigmoid function, extracting similarity measurement features and path curvature features of the degradation path in the high-order feature space, clustering the degradation paths into a plurality of degradation mode clusters according to the similarity measurement features, fitting the path curvature features of each of the degradation mode clusters using a curve fitting algorithm, and obtaining path approximate expressions under different degradation modes; Optionally, this embodiment adopts a multi-level normalization strategy in the data preprocessing stage. For the depth data of ablation and wear of the gun barrel, an adaptive data window is first constructed, and the window size is determined by the data fluctuation characteristics. In each window, the robust scaling method is used for normalization, and the influence of outliers is eliminated by the median and interquartile range. For temperature data, considering its periodic variation characteristics, a segmented normalization strategy is adopted, and the shooting process is divided into three stages: preheating, stabilization and cooling, which are processed separately. The normalization of pressure data adopts the Z-score method, combined with exponentially weighted moving average filtering to eliminate instantaneous fluctuations.

[0025] This embodiment implements an innovative dynamic feature fusion mechanism in the process of constructing the feature space. By constructing a feature vector matrix, each vector contains 15 key parameters: ablation depth, radial strain, axial strain, surface temperature gradient, internal temperature distribution, maximum pressure, pressure rise rate, pressure duration, propellant filling density, barrel vibration frequency, etc. In order to improve the feature expression ability, a dimensionality reduction framework based on principal component analysis is designed. The number of principal components is determined by contribution rate analysis, achieving data dimensionality reduction while retaining key information.

[0026] This embodiment uses an improved Sigmoid mapping function to achieve nonlinear feature transformation. The mapping function adopts an improved hyperbolic tangent form, introducing adjustable shape parameters and scale parameters. The parameter optimization adopts a grid search strategy, and the optimization objectives include inter-class separation and intra-class aggregation. In order to enhance the adaptability of the mapping, a multi-kernel fusion mechanism is designed, combining the radial basis kernel function and the polynomial kernel function, and achieving effective extraction of features of different scales through weighted combination. L1 / L2 mixed regularization constraints are introduced in the mapping process to control the complexity of the model.

[0027] This embodiment implements a robust similarity measurement framework in a high-order feature space. The improved Mahalanobis distance is used for metric feature extraction, and the dynamic adjustment of feature weights is achieved through adaptive covariance estimation. In order to improve the stability of the metric, a local structure preservation mechanism is designed, and the topological relationship between samples is described by the K nearest neighbor graph model. In the construction of the graph model, the edge weight is calculated by the Gaussian kernel function, and the kernel width is determined by cross-validation. Timing constraints are considered in the metric calculation process, and the temporal consistency of the similarity evaluation is ensured by the time window constraint.

[0028] This embodiment innovatively implements a path curvature feature extraction mechanism. The curvature calculation is based on differential geometry theory, and a parameterized curve is used to represent the degenerate path. The curvature calculation uses an improved three-point method, and the calculation accuracy is improved through Richardson extrapolation. In order to improve the calculation stability, an adaptive step selection strategy is designed, and the step size is dynamically adjusted by the local curvature change rate. Wavelet multi-resolution analysis is introduced in the curvature feature extraction process, and the recognition and extraction of curvature features of different scales are realized through continuous wavelet transform.

[0029] This embodiment uses an improved hierarchical clustering algorithm to realize degradation pattern recognition. A distance matrix is ​​constructed based on similarity measurement features, and the cluster merging strategy is determined by Ward's minimum variance method. The number of clusters is determined by contour analysis and gap statistics to ensure the reliability of the clustering results. In order to improve the clustering stability, an adaptive merging threshold is designed, and the threshold size is dynamically adjusted by the inter-class distance distribution. Constraints are introduced in the clustering process to ensure that sample points in the same time series belong to the same cluster.

[0030] This embodiment implements an accurate curve fitting strategy in the path fitting stage. The fitting algorithm uses an improved B-spline basis function to achieve accurate control of the curve shape through control point optimization. Node selection is based on curvature extreme value analysis, and node density is increased in areas where curvature changes dramatically. In order to improve the fitting accuracy, an adaptive weight mechanism is designed, and the weight size is dynamically adjusted through residual analysis. Monotonicity and smoothness constraints are introduced in the fitting process to ensure the physical rationality of the fitting curve.

[0031] This embodiment innovatively constructs a complete path approximation expression. The expression design is based on the ablation physical model, including Arrhenius-type exponential terms to describe temperature effects, power function terms to describe pressure effects, and interaction terms to describe coupling effects. Coefficient determination is achieved through multi-objective optimization, and the optimization objectives include fitting accuracy, model complexity, and physical constraint satisfaction. The optimization solution adopts the improved NSGA-II algorithm, and the optimal solution is determined through Pareto frontier analysis.

[0032] This embodiment establishes a reliable degradation path modeling framework through systematic data processing and feature extraction. While ensuring modeling accuracy, this solution improves the adaptability of the model through multi-level optimization. The overall design fully considers the characteristics of artillery ablation and realizes full-process optimization from data preprocessing to path approximation. Experimental verification shows that this solution can effectively capture the dynamic characteristics of the ablation process and provide a reliable basis for life prediction.

[0033] The modeling results of this embodiment show excellent physical consistency and predictive ability, and achieve accurate characterization of the ablation degradation process through multi-level analysis. This solution has shown strong analytical capabilities in practical applications, providing a scientific basis for equipment maintenance decisions and achieving accurate evaluation and prediction of equipment status. The adaptive characteristics of the model enable it to adapt to the ablation prediction needs under different working conditions, and it has good versatility and scalability.

[0034] Step S102: iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on the least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on the output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; Optionally, this embodiment implements an innovative alternating iteration strategy in the parameter optimization stage. The parameter vector is divided into three sub-blocks: a physical parameter block, a shape parameter block, and a weight parameter block. The physical parameter block contains temperature sensitivity coefficients, pressure influencing factors, and material degradation coefficients; the shape parameter block contains curve fitting coefficients and nonlinear mapping parameters; the weight parameter block describes the relative importance of each factor. During the optimization process, two parameter blocks are fixed each time, and the other parameter block is optimized, and the global optimum is achieved through cyclic iteration.

[0035] This embodiment designs a complete objective function construction framework. Based on the least squares criterion, the objective function contains three parts: fitting error term, physical constraint term and regularization term. The fitting error is measured by weighted Euclidean distance, and the weight is determined by data reliability. The physical constraint term introduces monotonicity conditions and boundary conditions to ensure that the prediction results conform to the physical laws of ablation and wear. The regularization term adopts the form of elastic network, combined with L1 norm and L2 norm to achieve parameter selection and convergence control.

[0036] This embodiment uses an improved gradient descent algorithm for optimization and solution. In order to improve the convergence speed and stability, an adaptive learning rate adjustment mechanism is designed, and the learning rate size is dynamically determined by the Armijo criterion. In order to avoid local optimality, momentum terms and random perturbations are introduced, and the simulated annealing strategy is used to fully explore the parameter space. An early stopping strategy is set during the optimization process, and the training process is monitored by the validation set performance to avoid overfitting.

[0037] This embodiment implements an innovative parameter update strategy. The update process uses the block coordinate descent method, and the update direction of each parameter block is calculated by the proximal operator. In order to improve the update efficiency, a parameter preconditioning mechanism is designed to approximate the Hessian matrix through the Fisher information matrix to accelerate the optimization convergence. Adaptive step size control is introduced in the parameter update process, and the step size is dynamically adjusted through the Wolfe condition.

[0038] This embodiment builds a reliable prediction model generation framework. The optimal parameter combination is determined by cross-validation, and the validation process adopts a time series segmentation strategy to ensure the time series prediction ability of the model. The idea of ​​ensemble learning is introduced in the model construction process, and the prediction stability is improved by the weighted combination of multiple sub-models. The weight determination is achieved through stack generalization, which effectively reduces the prediction variance.

[0039] This embodiment designs a complete confidence interval estimation scheme. The interval estimation is based on the Delta method, and the probability distribution of the predicted value is approximated by the first-order Taylor expansion. In order to improve the estimation accuracy, a bootstrap resampling strategy is designed to construct the empirical distribution of the predicted value through multiple resampling. The construction of the confidence interval takes into account the parameter uncertainty and model structure uncertainty, and the interval width is determined by error propagation analysis.

[0040] This embodiment implements an innovative prediction interval construction method. The interval construction is based on quantile regression theory, and the prediction interval is determined by simultaneously estimating the upper and lower quantiles of the conditional distribution. In order to improve the reliability of the interval, an adaptive coverage adjustment mechanism is designed to dynamically adjust the confidence level through historical data verification. Heteroskedasticity is taken into account during the interval construction process, and the dynamic adjustment of the interval width is achieved through local variance estimation.

[0041] This embodiment adopts a systematic verification and evaluation strategy. The verification process includes point prediction accuracy evaluation, interval coverage evaluation and interval width evaluation. The accuracy evaluation adopts a multi-index system, including indicators such as RMSE, MAE and MAPE. The coverage evaluation is achieved by comparing the empirical coverage with the nominal coverage. The interval width evaluation takes into account the practicality of the prediction interval and constrains the interval width through engineering limits.

[0042] This embodiment establishes a reliable ablation prediction framework through precise parameter optimization and systematic model construction. While ensuring prediction accuracy, this solution improves prediction reliability through multi-level optimization. The overall design fully considers the characteristics of artillery ablation and realizes full process optimization from parameter optimization to interval prediction.

[0043] The prediction results of this embodiment show excellent accuracy and reliability, and the accurate prediction of the ablation process is achieved through multi-level analysis. This scheme has shown strong prediction capabilities in practical applications and provides a reliable quantitative basis for equipment maintenance decisions. The construction of the prediction interval provides uncertainty quantification for decision-making, which improves the practicality and credibility of the prediction results.

[0044] The optimization strategy and prediction framework of this embodiment have good adaptability and scalability, and can meet the ablation prediction needs under different working conditions. Through parameter adaptation and interval dynamic adjustment, accurate modeling and reliable prediction of complex ablation processes are achieved, providing a scientific basis for equipment status assessment.

[0045] Step S103: reading the real-time firing data of the artillery to be predicted, calculating the comprehensive load index according to the barrel temperature, firing pressure and propellant loading contained in the firing data, inputting the comprehensive load index into the barrel ablation degradation prediction model, obtaining the barrel ablation wear prediction value and its prediction interval under different firing numbers, establishing the barrel ablation wear degradation trend curve based on the prediction value, calculating the change rate and acceleration characteristics of the barrel ablation wear according to the degradation trend curve, and judging the ablation wear state of the gun barrel in combination with the prediction interval.

[0046] Optionally, this embodiment uses a multi-channel real-time monitoring system to obtain artillery firing data during the data acquisition stage. Temperature monitoring uses a distributed thermocouple array, with 12 measuring points arranged along the axial direction of the barrel and 3 layers arranged radially to form a three-dimensional distribution monitoring network for the temperature field. Each measuring point uses a K-type thermocouple, and the measurement range covers -200°C to 1200°C. High-precision temperature acquisition is achieved through a signal conditioning circuit. The pressure data uses a piezoelectric pressure sensor, which is installed at the bottom of the barrel, and the sampling frequency is set to 100kHz to ensure accurate capture of transient pressure characteristics. The propellant loading is measured by a high-precision weighing device with an accuracy of gram level, and the measurement reliability is improved in combination with ambient temperature and humidity compensation.

[0047] This embodiment designs an innovative method for calculating the comprehensive load index. The temperature data processing first performs spatial interpolation, and constructs a continuous temperature field distribution through a three-dimensional spline function. Considering the influence of temperature gradient on ablation, the radial and axial temperature gradients are calculated, and the thermal stress intensity is reflected through weighted combination. Pressure data analysis focuses on multiple characteristic quantities: peak pressure, pressure rise rate, pressure duration, and pressure pulsation characteristics. The time-frequency characteristics of the pressure signal are extracted through wavelet analysis to evaluate the dynamic characteristics of the pressure load.

[0048] This embodiment realizes the dynamic fusion of multi-source information in the process of load index construction. By establishing a thermal-mechanical coupling model, the interaction between the temperature field and the pressure field is analyzed. The model takes into account the temperature dependence of the material and describes the coupling relationship between load factors through nonlinear mapping. The loading data is combined with the internal ballistic model to evaluate its modulation effect on the ablation intensity. The final form of the load index adopts a weighted sum form, and the weight coefficient is determined by sensitivity analysis.

[0049] This embodiment innovatively implements a dynamic update mechanism for predicted values. The prediction model adopts a recursive structure and implements real-time update of predicted values ​​through a sliding time window. The timeliness of historical data is considered in the prediction process, and the impact of long-term data is reduced through exponential weighting. In order to improve the stability of prediction, a prediction value smoothing strategy is designed to eliminate random fluctuations through local polynomial fitting.

[0050] This embodiment adopts an improved trend curve construction method. The curve fitting adopts a piecewise spline function, and the position of the segmentation point is determined by curvature analysis. The fitting process introduces monotonicity and smoothness constraints to ensure the physical rationality of the trend curve. The uncertainty of the prediction interval is considered in the curve construction process, and the fitting parameters are dynamically adjusted by the interval width.

[0051] This embodiment achieves reliable state feature extraction. The rate of change calculation uses an adaptive difference format, and the step size is dynamically determined by the local curvature. Acceleration feature extraction takes into account the influence of noise, and stable derivation is achieved through Savitzky-Golay filtering. Scale analysis is introduced in the feature extraction process, and the change characteristics of different time scales are identified through multi-scale decomposition.

[0052] This embodiment innovatively constructs a state judgment framework. The judgment process is based on a fuzzy inference system, and the input variables include ablation depth, change rate, and acceleration characteristics. The design of fuzzy rules is based on expert experience and historical data analysis, and the rule parameters are optimized through machine learning methods. The judgment result takes into account the uncertainty of the prediction interval, and the reliability of the judgment is quantified through credibility scoring.

[0053] This embodiment designs a complete prediction interval utilization strategy. Interval analysis first evaluates the time-varying characteristics of interval width, and determines the changing trend of prediction uncertainty through local variance estimation. Interval information is combined with state judgment results, and the credibility of judgment results is adjusted through Bayesian updating. Engineering limit constraints are considered in the interval utilization process, and the alarm threshold is determined through risk assessment.

[0054] This embodiment realizes the visualization of the state assessment results. The visualization interface includes multiple views: the trend curve view displays the prediction results and uncertainty intervals, the feature analysis view displays the time series changes of the rate of change and acceleration characteristics, and the state assessment view intuitively displays the current state level in the form of a dashboard. Interaction is achieved between views to support comprehensive analysis of multi-dimensional data.

[0055] This embodiment establishes a reliable state assessment framework through systematic data analysis and feature extraction. While ensuring the accuracy of the assessment, the scheme improves the reliability of the assessment through multi-level analysis. The overall design fully considers the characteristics of artillery ablation and realizes the optimization of the entire process from data collection to state judgment.

[0056] The prediction and evaluation results of this embodiment show excellent accuracy and reliability, and achieve accurate evaluation of the ablation status through multi-level analysis. This solution has shown strong analytical capabilities in practical applications and provides a reliable quantitative basis for equipment maintenance decisions. The visual presentation of the evaluation results improves the analysis efficiency and enables maintenance personnel to intuitively grasp the equipment status.

[0057] The analysis framework of this embodiment has good adaptability and scalability, and can meet the needs of state assessment under different working conditions. Through parameter adaptation and judgment rule optimization, accurate characterization and reliable assessment of complex ablation processes are achieved, providing a scientific basis for equipment maintenance and support.

[0058] From the above description, it can be seen that the barrel ablation wear prediction method based on degradation path approximation provided in the embodiment of the present application can establish a high-order feature space by normalizing the ablation wear data and mapping the sigmoid function. The similarity measurement characteristics and path curvature characteristics of the degradation path are innovatively extracted to achieve accurate clustering and path approximation of the degradation pattern. The system uses an alternating optimization algorithm and a gradient descent method to optimize the path expression parameters and establish a confidence interval estimation mechanism. A comprehensive load index is constructed by combining the barrel temperature, firing pressure and propellant loading. Through the rate of change and acceleration characteristics of the degradation trend curve, the barrel ablation wear state is accurately predicted and evaluated, providing a scientific basis for the use and maintenance of artillery.

[0059] In one embodiment of the barrel ablation wear prediction method based on degradation path approximation of the present application, see Figure 2 , and can also include the following: Step S201: collecting ablation wear depth data, barrel pressure data and barrel temperature data of a gun barrel under different firing numbers, performing maximum and minimum value normalization processing on the ablation wear depth data, calculating a comprehensive load coefficient in combination with the barrel pressure data and barrel temperature data, and constructing a multidimensional feature vector with the comprehensive load coefficient and the normalized ablation wear depth data; Step S202: establishing a degenerate path sample space based on the multidimensional feature vector, performing nonlinear mapping on the degenerate path sample space using a Sigmoid function to obtain a high-dimensional feature expression, calculating the Mahalanobis distance between the degenerate paths in the high-dimensional feature expression to obtain a similarity measurement feature, and calculating the curvature change of the degenerate path using a differential geometry method to obtain a path curvature feature.

[0060] Optionally, this embodiment adopts a multi-dimensional data acquisition strategy to obtain the gun barrel status information. The ablation wear depth is measured by a high-resolution endoscope system. 36 measuring points are evenly arranged along the axial direction on the inner wall of the barrel. Each measuring point measures 8 positions circumferentially to form a complete three-dimensional ablation morphology description. The measurement process adopts autofocus technology, and sub-pixel precision positioning is achieved through image processing algorithms. The barrel pressure is collected by a piezoelectric sensor array, and multiple measuring points are arranged at the bottom and middle of the barrel. The sampling frequency reaches 100kHz, which can accurately capture the pressure fluctuations. The barrel temperature adopts a distributed fiber optic sensing system, and optical fibers are buried along the barrel wall to realize continuous monitoring of the temperature field.

[0061] This embodiment implements an innovative normalization strategy in the data preprocessing stage. For the ablation depth data, outlier detection is first performed, and abnormal measurement points are identified and corrected through local statistical characteristics. The normalization process takes into account the data distribution characteristics, adopts a piecewise linear mapping method, and uses different mapping coefficients in different depth intervals to improve the normalization effect. In order to maintain the physical meaning of the data, the normalization parameters are determined by the theoretical limit value to ensure the interpretability of the processed data.

[0062] This embodiment designs a load coefficient calculation framework for the system. The temperature load analysis takes into account the thermal cycle effect, extracts the temperature fluctuation characteristics through Fourier transform, and evaluates the thermal stress intensity. The pressure load evaluation focuses on multiple characteristic quantities: peak pressure, pressure rise rate, pressure duration, and pressure pulsation characteristics. The time-frequency decomposition of the pressure signal is achieved through wavelet analysis, and load characteristics of different scales are extracted. The calculation of the load coefficient adopts a nonlinear combination form, and the coupled analysis of the temperature field and the pressure field is realized through the neural network model.

[0063] This embodiment implements an innovative feature vector construction method. The vector design adopts a hierarchical structure, which includes three levels: the basic feature layer describes the original measurement data, the derived feature layer contains various statistical and physical features, and the combined feature layer forms a high-order expression through feature fusion. The temporal dependency is considered in the feature extraction process, and the dynamic features are captured through the sliding window strategy. The feature selection adopts the Lasso algorithm to achieve dimensionality reduction optimization of the feature space.

[0064] This embodiment adopts an innovative mapping mechanism in the process of constructing the path sample space. Nonlinear feature transformation is achieved through an improved Sigmoid function, and the function form includes adjustable shape parameters and scale parameters. Parameter optimization adopts a cross-validation strategy and determines the optimal parameter combination through grid search. In order to enhance the expression ability of the mapping, a multi-kernel fusion mechanism is designed, combining the radial basis kernel function and the polynomial kernel function to achieve effective extraction of features of different scales.

[0065] This embodiment innovatively implements similarity metric feature extraction. The Mahalanobis distance calculation uses adaptive covariance estimation, and the covariance matrix is ​​determined through the local data structure. In order to improve the stability of the metric, a robust estimation strategy is designed, and the influence of abnormal samples is reduced by a weighted method. Timing constraints are considered in the process of metric feature extraction, and the timing consistency of similarity evaluation is ensured by time window constraints.

[0066] This embodiment adopts a systematic curvature feature extraction method. Based on differential geometry theory, the degradation path is represented by a parameterized curve. The curvature calculation adopts an improved three-point method, and Richardson extrapolation is introduced to improve the calculation accuracy. In order to enhance the expressiveness of the feature, a multi-scale analysis framework is designed, and the recognition and extraction of curvature features of different scales are realized through continuous wavelet transform.

[0067] This embodiment implements an innovative feature fusion strategy. The fusion process uses an attention mechanism to achieve dynamic combination of different features through adaptive weights. The weight calculation takes into account the local importance of the features, and the weight parameters are determined through correlation analysis. The fusion result is reduced in dimension through an autoencoder to extract the most representative feature combination.

[0068] This embodiment establishes a reliable degradation feature extraction framework through systematic feature engineering and data analysis. While ensuring the feature expression capability, this solution improves the interpretability of features through multi-level optimization. The overall design fully considers the characteristics of artillery ablation and realizes the optimization of the entire process from data collection to feature extraction.

[0069] The feature extraction results of this embodiment show excellent expression ability and differentiation, and achieve accurate characterization of the ablation process through multi-level analysis. This solution shows strong feature expression ability in practical applications and provides a reliable feature basis for subsequent modeling analysis. The multi-dimensional expression of features provides a new analytical perspective for the study of degradation mechanisms.

[0070] The feature extraction framework of this embodiment has good adaptability and expansibility, and can meet the feature extraction requirements under different working conditions. Through parameter adaptation and feature optimization, accurate feature expression of complex ablation processes is achieved, providing reliable feature support for equipment status assessment.

[0071] In one embodiment of the barrel ablation wear prediction method based on degradation path approximation of the present application, see Figure 3 , and can also include the following: Step S301: constructing a distance matrix of the degradation path based on the similarity measurement feature, performing cluster analysis on the distance matrix using a hierarchical clustering algorithm, dividing the degradation path into multiple degradation pattern clusters according to the cluster analysis result, calculating the cluster center of each degradation pattern cluster, and performing feature decomposition on the cluster center to obtain the main degradation features; Step S302: normalize the path curvature characteristics in each of the degradation pattern clusters, fit the normalized path curvature characteristics using a polynomial fitting algorithm, calculate the coefficient matrix of the fitting curve, establish an approximate expression of the degradation path based on the coefficient matrix, and combine the approximate expression with the main degradation characteristics to construct a complete path approximation model.

[0072] Optionally, this embodiment adopts an innovative distance matrix construction method. Based on the similarity measurement characteristics, an adaptive weight calculation strategy is designed to determine the importance weight of the sample point through local density estimation. The distance calculation adopts the improved Mahalanobis distance formula: D(i,j) = sqrt((xi-xj)^TW (xi-xj)), where W is the weight matrix, which is determined by iterative optimization. In order to improve the computational efficiency, sparse representation technology is used to retain only the key elements in the distance matrix.

[0073] This embodiment implements an optimization strategy for hierarchical clustering. The clustering process adopts the Ward minimum variance criterion to achieve optimal grouping by minimizing the intra-class sum of squares. In order to determine the optimal number of clusters, a multi-index evaluation system is designed, combining the silhouette coefficient, CH index and Gap statistic for comprehensive evaluation. Constraints are introduced in the clustering process to ensure that samples with similar time series are more easily clustered into one class, thereby improving the physical rationality of the clustering results.

[0074] This embodiment innovatively achieves the optimized extraction of cluster centers. The center calculation adopts the weighted average method, and the weight is determined by the distance between the sample and the cluster center. In order to improve the representativeness of the center, an anomaly detection mechanism is designed to identify and process abnormal samples through the Mahalanobis distance. The timing characteristics are considered in the center extraction process, and the timing consistency of the center trajectory is ensured through dynamic time warping.

[0075] This embodiment designs a systematic feature decomposition method. Using singular value decomposition technology, the cluster centers are transformed into principal component space through orthogonal transformation. Feature selection is based on the cumulative contribution rate, and the principal components whose explained variance accounts for more than the threshold are retained. In order to enhance the physical interpretability of the features, nonlinear feature extraction is achieved through variational autoencoders to capture complex degradation pattern features.

[0076] This embodiment implements an innovative curvature feature processing solution. The normalization process uses piecewise linear mapping, and the mapping parameters are determined by local statistical characteristics. In order to maintain the local structure of the feature, a conformal mapping strategy is designed to maintain the local relationship between samples through Laplacian feature mapping. The normalization results are evaluated through cross-validation to ensure the effectiveness of the processed features.

[0077] This embodiment adopts an improved polynomial fitting algorithm. The fitting process adopts weighted least squares method, and the weight is determined by data reliability. In order to avoid overfitting, regularization constraints are introduced, and parameter selection and smoothing control are achieved by combining L1 and L2 norms. The fitting order is determined by the AIC criterion to strike a balance between fitting accuracy and model complexity.

[0078] This embodiment innovatively constructs a coefficient matrix analysis framework. Matrix analysis uses Tucker decomposition to extract high-order feature relationships through tensor decomposition. In order to improve the decomposition efficiency, an incremental calculation strategy is designed to update the decomposition results through a recursive method. The spatiotemporal correlation is considered in the process of matrix feature extraction, and complex feature patterns are captured through a spatiotemporal convolutional network.

[0079] This embodiment realizes an innovative path approximation expression. The expression construction adopts the form of mixed basis functions, combining polynomial basis functions and radial basis functions to achieve flexible modeling. Parameter optimization is solved by alternating direction multiplier method to achieve efficient solution of constrained optimization problems. In order to improve the generalization ability of the expression, an adaptive basis function selection mechanism is designed to determine the optimal basis function combination through sparse learning.

[0080] This embodiment adopts a systematic model fusion strategy. The dynamic fusion of path expressions and degradation features is achieved through the attention mechanism, and the weight coefficients are optimized through end-to-end learning. The fusion process considers the multi-scale characteristics of the features and realizes the effective integration of information at different scales through multi-resolution analysis.

[0081] This embodiment establishes a reliable path approximation framework through systematic feature analysis and model construction. While ensuring the accuracy of the model, this solution improves the interpretability of the model through multi-level optimization. The overall design fully considers the characteristics of artillery ablation and realizes the optimization of the entire process from feature extraction to model construction.

[0082] The analysis results of this embodiment show excellent modeling effect and prediction ability, and achieve accurate characterization of the ablation process through multi-level analysis. This solution shows strong adaptability in practical applications and provides a reliable analysis tool for equipment status assessment. The interpretability of the model provides a new analytical perspective for the study of degradation mechanisms.

[0083] The analysis framework of this embodiment has good scalability and adaptability, and can meet the modeling requirements under different working conditions. Through parameter adaptation and model optimization, accurate modeling of complex ablation processes is achieved, providing reliable theoretical support for equipment status assessment.

[0084] In one embodiment of the barrel ablation wear prediction method based on degradation path approximation of the present application, see Figure 4 , and can also include the following: Step S401: setting a parameter vector to be optimized in the path approximation expression, constructing an objective function for parameter optimization based on the least squares criterion, wherein the objective function includes a path approximation error term and a parameter regularization term, obtaining a gradient expression by taking a partial derivative of the objective function with respect to the parameter vector, and calculating a parameter update direction and a step size according to the gradient expression; Step S402: Use an alternating iteration strategy to perform block optimization on the parameter vector, fix some parameters and update other parameters in each round of iteration, calculate the objective function value after each parameter update, stop iteration when the relative change of the objective function value is less than a preset convergence threshold, and use the final converged parameter vector as the optimal parameter combination.

[0085] Optionally, this embodiment designs an innovative parameter vector optimization framework. The parameter vector includes two categories: path shape parameters and scale parameters. The shape parameters control the geometric characteristics of the path, and the scale parameters adjust the weight distribution of the characteristics. Vector initialization is guided by prior knowledge, and the reasonable value range of the parameters is determined by historical data analysis. In order to improve the optimization efficiency, a hierarchical structure is used to design the parameter vector, and parameters at different levels have different update frequencies.

[0086] This embodiment constructs a complete optimization objective function. The path approximation error term adopts the weighted two-norm form: E = Σwi||yi - f(xi,θ)||², where wi is the sample weight, determined by reliability analysis. The regularization term contains a combination of L1 norm and L2 norm: R(θ) = λ1||θ||1 + λ2||θ||², which realizes the dual constraints of parameter sparsity and smoothness. The regularization coefficient is determined by cross-validation to strike a balance between model complexity and generalization ability.

[0087] This embodiment implements an efficient gradient calculation strategy. The gradient expression is derived through automatic differentiation technology to avoid the complexity of manual derivation. In order to improve the calculation efficiency, a gradient sparsification scheme is designed to only calculate the parameter gradients that have a significant impact on the objective function. Numerical stability is considered during the gradient calculation process, and gradient clipping is used to avoid the gradient explosion problem.

[0088] This embodiment innovatively implements a parameter update mechanism. The update direction is determined by the momentum method, and the convergence speed is improved by combining historical gradient information. The step size is selected using an adaptive strategy, and the Armijo criterion is used to ensure that the objective function is fully reduced. In order to avoid falling into a local optimum, a random perturbation mechanism is designed to introduce appropriate randomness in the update process.

[0089] This embodiment adopts an improved block optimization strategy. Parameter block is based on sensitivity analysis, and parameters with strong correlation are divided into the same block. Inter-block update adopts Gauss-Seidel iteration method, and efficient optimization is achieved through coordinate descent method. In order to improve the optimization stability, an intra-block preprocessing mechanism is designed to improve the condition number by normalizing the parameter scale.

[0090] This embodiment implements an innovative iterative control mechanism. The convergence judgment adopts a multi-criteria strategy, combining the objective function value, gradient norm and parameter change for comprehensive evaluation. In order to improve the iteration efficiency, an adaptive learning rate adjustment scheme is designed to dynamically adjust the update step size through the objective function curvature information. An early stopping strategy is adopted during the iteration process to avoid overfitting.

[0091] This embodiment designs a parameter evaluation scheme for the system. The evaluation indicators include three dimensions: fitting accuracy, model complexity, and computational efficiency. The accuracy evaluation uses a cross-validation method to estimate the generalization error of the model by leaving one out. The complexity evaluation is based on parameter sparsity and measures the simplicity of the model by the number of effective parameters. The efficiency evaluation focuses on computing time and storage requirements to ensure the practicality of the model.

[0092] This embodiment innovatively constructs a visualization framework for parameter optimization. The optimization process visualization includes the objective function convergence curve, parameter change trajectory, and error distribution diagram. The visualization results support interactive exploration and facilitate the analysis of key features in the optimization process. To enhance interpretability, a parameter importance analysis module is designed to identify key parameters through sensitivity analysis.

[0093] This embodiment establishes a reliable model tuning framework through systematic parameter optimization and evaluation. While ensuring the optimization effect, this solution improves the optimization efficiency through multi-level analysis. The overall design fully considers the characteristics of artillery ablation and realizes the full process optimization from parameter initialization to optimization.

[0094] The optimization results of this embodiment show excellent convergence and stability, and the precise adjustment of model parameters is achieved through multi-level analysis. This solution shows strong optimization capabilities in practical applications and provides a reliable optimization tool for model refinement. The interpretability of parameters provides a new analytical perspective for model improvement.

[0095] The optimization framework of this embodiment has good adaptability and scalability, and can meet the parameter optimization requirements under different working conditions. Through parameter adaptation and optimization strategy improvement, accurate tuning of complex models is achieved, providing reliable model support for equipment status assessment.

[0096] The optimization scheme of this embodiment has demonstrated excellent performance and reliability through theoretical analysis and practical verification. The stability and repeatability of the optimization results provide a solid foundation for model application and are of great significance for improving the accuracy of equipment status assessment.

[0097] In one embodiment of the barrel ablation wear prediction method based on degradation path approximation of the present application, see Figure 5 , and can also include the following: Step S501: Substitute the optimal parameter combination into the parameter vector in the path approximation expression, use a kernel function to map the path approximation expression to a prediction space, build a nonlinear prediction model in combination with the main degradation characteristics of the degradation path, calibrate the parameters of the nonlinear prediction model based on training samples, and generate a barrel ablation degradation prediction model; Step S502: Calculate the probability distribution characteristics of the output results of the barrel ablation degradation prediction model, use the delta method to estimate the variance of the prediction value, calculate the standard error of the prediction result based on the variance, determine the confidence coefficient according to a preset confidence level, multiply the confidence coefficient by the standard error to obtain the half-width of the prediction interval, and superimpose the half-width of the prediction interval on the basis of the prediction value to construct the ablation wear prediction interval.

[0098] Optionally, this embodiment designs an innovative kernel function mapping scheme. The kernel function selection adopts an adaptive strategy, combining Gaussian kernel and polynomial kernel to realize nonlinear transformation of feature space. The kernel function parameters are optimized by grid search, and the optimal parameter combination is determined by cross-validation. In order to enhance the expression ability of the mapping, a multi-kernel fusion mechanism is designed to realize the effective extraction of features of different scales through weighted combination.

[0099] This embodiment implements a method for building a systematic prediction model. The model structure adopts a deep neural network framework, which includes three modules: feature extraction, nonlinear mapping, and prediction output. The feature extraction layer adopts an attention mechanism to adaptively select important features. The nonlinear mapping layer improves gradient transfer through residual connections to improve model training efficiency. The prediction output layer adopts a probabilistic output form to achieve uncertainty quantification.

[0100] This embodiment innovatively implements a parameter calibration strategy. The calibration process uses the Bayesian optimization method to model the relationship between parameters and performance through Gaussian processes. In order to improve the calibration efficiency, an active learning strategy is designed to give priority to parameter combinations with high uncertainty for evaluation. An early stopping mechanism is introduced during the calibration process to avoid overfitting problems.

[0101] This embodiment adopts an improved model training scheme. The loss function design combines mean square error and negative log likelihood to achieve dual optimization of prediction accuracy and uncertainty estimation. The training process uses batch normalization technology to improve the model convergence speed. In order to enhance the robustness of the model, dropout regularization is introduced to reduce the risk of overfitting.

[0102] This embodiment implements innovative probability distribution feature extraction. The distribution features include mean, variance and high-order moments, which are estimated through Monte Carlo sampling. In order to improve sampling efficiency, an importance sampling strategy is designed to reduce computational overhead. Time series correlation is considered in the distribution feature extraction process, and time series dependency is captured through conditional probability modeling.

[0103] This embodiment designs a variance estimation framework for the system. The Delta method is based on the first-order Taylor expansion and approximates the variance of the predicted value through gradient information. In order to improve the estimation accuracy, a second-order correction term is introduced to consider the influence of nonlinear effects. Parameter uncertainty is considered in the variance estimation process and uncertainty is transmitted through the posterior distribution of the parameters.

[0104] This embodiment innovatively constructs a standard error calculation scheme. Error calculation takes into account multi-source uncertainty, including model parameter uncertainty, data noise, and prediction error. Uncertainty transfer uses the error propagation law to describe the correlation of errors from different sources through the covariance matrix. In order to improve the calculation stability, a numerical optimization strategy is designed.

[0105] This embodiment implements a reliable confidence interval construction method. The confidence level selection is based on application requirements, and the reasonable value is determined through risk analysis. The confidence coefficient calculation adopts Student-t distribution, considering the influence of degrees of freedom. The time-varying characteristics of the predicted value are considered in the interval construction process, and the interval width is adjusted through a dynamic update mechanism.

[0106] This embodiment adopts a systematic interval evaluation scheme. The evaluation indicators include interval coverage, average interval width and interval uncertainty. The coverage evaluation is verified by historical data to ensure the reliability of the interval. The width evaluation focuses on the compactness of the interval to avoid overly conservative estimates. The uncertainty evaluation is quantified by information entropy to analyze the information content of the interval.

[0107] This embodiment establishes a reliable prediction framework through systematic model construction and uncertainty analysis. While ensuring prediction accuracy, this solution provides reliable uncertainty quantification through multi-level analysis. The overall design fully considers the characteristics of artillery ablation and achieves full process optimization from model construction to uncertainty assessment.

[0108] The prediction results of this embodiment show excellent accuracy and reliability, and the accurate prediction of the ablation process is achieved through multi-level analysis. This solution has shown strong prediction capabilities in practical applications and provides a reliable quantitative basis for equipment maintenance decisions. The quantification of prediction uncertainty provides a scientific basis for risk assessment.

[0109] The prediction framework of this embodiment has good adaptability and scalability, and can meet the prediction requirements under different working conditions. Through parameter adaptation and uncertainty transfer, accurate prediction of complex ablation processes is achieved, providing reliable decision support for equipment status assessment.

[0110] The prediction scheme of this embodiment has demonstrated excellent performance and reliability through theoretical analysis and practical verification. The accuracy and interpretability of the prediction results provide a scientific basis for equipment maintenance and guarantee, which is of great significance to improving equipment reliability.

[0111] In one embodiment of the barrel ablation wear prediction method based on degradation path approximation of the present application, see Figure 6 , and can also include the following: Step S601: acquiring firing data of the artillery to be predicted, extracting barrel temperature data, firing pressure data and propellant charge data from the firing data, normalizing the barrel temperature data, firing pressure data and propellant charge data, calculating a comprehensive load index by weighted superposition, and determining a weight coefficient of each index according to historical data; Step S602: Input the comprehensive load index into the barrel ablation degradation prediction model according to the sequence of firing numbers, calculate the ablation wear prediction value corresponding to each firing number based on the prediction model, calculate the upper confidence limit and lower confidence limit of each prediction value in combination with the pre-constructed prediction interval, and form a complete ablation wear prediction result with the prediction value and its upper and lower confidence limits.

[0112] Optionally, this embodiment uses a multi-channel data acquisition system to obtain artillery firing data. The barrel temperature is measured by a distributed thermocouple array, with 16 measuring points arranged along the axial direction and 4 layers of measuring points set radially to achieve accurate monitoring of the temperature field. Each measuring point uses a K-type thermocouple, and the measurement range covers -50°C to 1100°C. The shooting pressure is collected by a piezoelectric sensor, and the installation positions include the bottom and middle of the barrel. The sampling frequency is set to 200kHz to capture the transient characteristics of the pressure. The propellant loading is measured by a high-precision weighing system, and combined with temperature and humidity compensation to improve the measurement reliability.

[0113] This embodiment designs an innovative data preprocessing solution. The temperature data processing first performs spatial interpolation, and constructs a continuous temperature field distribution through a three-dimensional spline function. The sliding window method is used to eliminate measurement noise, and the window length is determined by signal-to-noise ratio analysis. Normalization uses an improved min-max method: T_norm = (T-T_min) / (T_max-T_min), where T is the original temperature value, T_min and T_max are the minimum and maximum values ​​of the historical data, respectively.

[0114] This embodiment implements a systematic processing method for pressure data. The pressure signal is first denoised by wavelet, and db4 wavelet is selected for multi-scale decomposition. Feature extraction focuses on multiple key parameters: maximum pressure value Pmax, pressure rise rate dP / dt, pressure duration τ, and pressure fluctuation amplitude ΔP. The normalization process considers the physical meaning of the parameters and uses piecewise linear mapping to ensure the rationality of data distribution.

[0115] This embodiment innovatively implements the processing of filling volume data. Data preprocessing includes outlier detection and temperature compensation, and outliers are identified through local statistical characteristics. The compensation model takes into account the influence of ambient temperature and chamber volume to ensure the consistency of filling volume data. Normalization adopts the Z-score method: M_norm = (M-μ) / σ, where M is the filling volume, μ and σ are the mean and standard deviation of historical data, respectively.

[0116] This embodiment designs a method for constructing a load index of the system. The comprehensive load index adopts a weighted sum form: L = w1·T_norm + w2·P_norm + w3·M_norm, where wi is the weight coefficient. The weight is determined by the hierarchical analysis method, and the judgment matrix is ​​constructed through expert scoring, and the eigenvector is calculated to obtain the initial weight. The weight optimization evaluates the influence of each factor on the ablation process through sensitivity analysis.

[0117] This embodiment adopts an innovative model input construction strategy. The input sequence adopts a sliding window form, and the window length is determined by autocorrelation analysis. In order to capture the time series characteristics, a data enhancement scheme is designed to increase the input dimension through time embedding. The data organization takes into account causal relationships to ensure the physical rationality of the prediction model.

[0118] This embodiment implements a reliable prediction calculation framework. The prediction process adopts a recursive structure, and the subsequent prediction is guided by the historical prediction results. The prediction value calculation takes into account the model uncertainty, and the ensemble learning method is used to improve the prediction stability. Constraints are introduced in the prediction process to ensure that the prediction results conform to the laws of physics.

[0119] This embodiment innovatively constructs a prediction interval calculation scheme. The interval calculation is based on the error propagation theory, taking into account the uncertainty of model parameters and the influence of data noise. The confidence limits are determined by the normal distribution assumption: CI = Y ± k·σ, where Y is the predicted value, k is the confidence coefficient, and σ is the prediction standard deviation. Interval optimization dynamically adjusts the confidence level through the empirical Bayes method.

[0120] This embodiment designs a complete prediction result evaluation system. The evaluation indicators include two dimensions: point prediction accuracy and interval reliability. The accuracy evaluation adopts a multi-indicator system, including RMSE, MAE, and MAPE. The interval evaluation focuses on coverage and interval width, and verifies the interval validity through historical data.

[0121] This embodiment establishes a reliable ablation prediction framework through systematic data processing and prediction analysis. While ensuring prediction accuracy, this solution provides reliable uncertainty quantification through multi-level analysis. The overall design fully considers the characteristics of artillery ablation and achieves full process optimization from data processing to prediction output.

[0122] The prediction results of this embodiment show excellent accuracy and reliability, and the accurate prediction of the ablation process is achieved through multi-level analysis. This scheme has shown strong prediction capabilities in practical applications and provides a reliable quantitative basis for equipment maintenance decisions. The quantification of prediction uncertainty provides a scientific basis for risk assessment.

[0123] The prediction framework of this embodiment has good adaptability and scalability, and can meet the prediction requirements under different working conditions. Through parameter adaptation and uncertainty transfer, accurate prediction of complex ablation processes is achieved, providing reliable decision support for equipment status assessment.

[0124] In one embodiment of the barrel ablation wear prediction method based on degradation path approximation of the present application, see Figure 7 , and can also include the following: Step S701: numerically smoothing the predicted value, constructing a continuous degradation trend curve using a piecewise polynomial interpolation method, calculating the first-order derivative of the degradation trend curve to obtain a change rate characteristic of ablation wear, calculating the second-order derivative to obtain an acceleration characteristic of ablation wear, and calculating key feature points of the degradation process according to the change rate characteristic and the acceleration characteristic; Step S702: Based on the key feature points, the degradation process is divided into multiple stages, and the reliability index of each stage is calculated in combination with the upper confidence limit and the lower confidence limit of the prediction interval. The ablation and wear state of the gun barrel is quantitatively evaluated according to the preset state evaluation criteria to generate a barrel state evaluation result.

[0125] Optionally, this embodiment adopts an innovative numerical smoothing strategy. The smoothing process uses the locally weighted regression method to achieve smooth fitting of local data through a cubic weight function. The weight function is W(x) = (1-|x / h|³)³, where x is the data point distance and h is the bandwidth parameter. The bandwidth selection is optimized by the generalized cross-validation criterion to achieve a balance between smoothness and fidelity. To prevent endpoint effects, mirror extension technology is used to process boundary data.

[0126] This embodiment realizes the interpolation modeling scheme of the system. The piecewise polynomial adopts the cubic B-spline form to ensure the continuity of the first and second derivatives at the nodes. The node arrangement adopts an adaptive strategy to dynamically adjust the node density according to the characteristics of data changes. The expression of the interpolation function is: S(x) = Σai·Bi,k(x), where ai is the control point parameter and Bi,k(x) is the kth-order B-spline basis function. The parameter optimization is achieved by the least squares method, considering the data reliability weights.

[0127] This embodiment designs an innovative derivative calculation method. The first derivative adopts an improved central difference format: v(t)= [S(t+h)-S(t-h)] / 2h, where v(t) is the rate of change and h is the step size parameter. The second derivative calculation adopts a four-point format: a(t) = [S(t-h)-2S(t)+S(t+h)] / h², where a(t) is the acceleration feature. The step size selection is determined by error analysis, weighing the calculation accuracy and numerical stability.

[0128] This embodiment innovatively realizes the feature point recognition strategy. The key feature points include inflection points, extreme points, and acceleration mutation points. The inflection points are determined by the zeros of the second derivative, representing the critical moments of the degradation rate change. The extreme points are identified by the zeros of the first derivative, reflecting the local maximum or minimum of the degradation degree. The mutation points are detected by the acceleration threshold to capture the drastic changes in the degradation process.

[0129] This embodiment adopts a systematic stage division method. The stage boundaries are determined based on the key feature points, and the intervals with similar features are merged through cluster analysis. The stage feature extraction adopts multi-dimensional analysis, including the average degradation rate, degradation acceleration, and fluctuation degree. The stage division results are verified by physical meaning to ensure the rationality of the division.

[0130] This embodiment designs a complete reliability evaluation framework. The reliability index R(t) = P[D(t)<Dc], where D(t) is the degradation amount and Dc is the critical threshold. The index calculation considers the uncertainty of the prediction interval, and the reliability is evaluated through probability integration. The influence of the confidence interval is analyzed by the Monte Carlo method to evaluate the impact of uncertainty on reliability.

[0131] This embodiment implements an innovative state assessment scheme. The assessment criteria adopt a multi-level structure, including three dimensions: safety, reliability, and economy. The safety assessment focuses on the margin between the degradation amount and the critical value, and determines the safety level through fuzzy reasoning. The reliability assessment is based on failure probability analysis, taking into account the randomness of the use environment. The economic assessment combines maintenance costs and use benefits to optimize decision-making recommendations.

[0132] This embodiment innovatively constructs an evaluation result generation framework. The result expression adopts a multi-dimensional form, including quantitative indicators and qualitative descriptions. Quantitative indicators include remaining life prediction, reliability assessment and risk level. Qualitative description is based on the expert rule base and generates maintenance suggestions through fuzzy reasoning. The result visualization adopts multi-level display to support interactive exploration.

[0133] This embodiment establishes a reliable state assessment framework through systematic analysis and evaluation. While ensuring the accuracy of the assessment, the solution provides a comprehensive state description through multi-level analysis. The overall design fully considers the characteristics of artillery use and achieves full process optimization from data analysis to decision support.

[0134] The evaluation results of this embodiment show excellent reliability and practicality, and achieve accurate evaluation of equipment status through multi-level analysis. This solution has shown strong evaluation capabilities in practical applications and provides a reliable theoretical basis for equipment maintenance decisions. The multi-dimensional characteristics of the evaluation provide a scientific basis for the optimization of maintenance strategies.

[0135] In order to accurately predict and evaluate the ablation and wear state of the barrel and provide a scientific basis for the use and maintenance of the artillery, the present application provides an embodiment of a barrel ablation and wear prediction device based on degradation path approximation for realizing all or part of the contents of the barrel ablation and wear prediction method based on degradation path approximation, see Figure 8 The barrel ablation wear prediction device based on degradation path approximation specifically includes the following contents: A path expression building module 10 is used to normalize the collected data of ablation and wear of the gun barrel, establish a multidimensional degradation path sample space based on the normalized ablation and wear data, map the sample space to a high-order feature space using a Sigmoid function, extract similarity measurement features and path curvature features of the degradation path in the high-order feature space, cluster the degradation path into multiple degradation mode clusters according to the similarity measurement features, and fit the path curvature features of each degradation mode cluster using a curve fitting algorithm to obtain path approximate expressions under different degradation modes; a wear interval prediction module 20, for iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on a least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on an output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; The wear state prediction module 30 is used to read the real-time shooting data of the artillery to be predicted, calculate the comprehensive load index according to the barrel temperature, shooting pressure and propellant loading contained in the shooting data, input the comprehensive load index into the barrel ablation degradation prediction model, obtain the barrel ablation wear prediction value and its prediction interval under different shooting numbers, establish the barrel ablation wear degradation trend curve based on the prediction value, calculate the change rate and acceleration characteristics of the barrel ablation wear according to the degradation trend curve, and judge the ablation wear state of the artillery barrel in combination with the prediction interval.

[0136] From the above description, it can be seen that the barrel ablation and wear prediction device based on degradation path approximation provided in the embodiment of the present application can establish a high-order feature space by normalizing the ablation and wear data and mapping the sigmoid function. The similarity measurement characteristics and path curvature characteristics of the degradation path are innovatively extracted to achieve accurate clustering and path approximation of the degradation pattern. The system uses an alternating optimization algorithm and a gradient descent method to optimize the path expression parameters and establish a confidence interval estimation mechanism. A comprehensive load index is constructed by combining the barrel temperature, firing pressure and propellant loading. Through the rate of change and acceleration characteristics of the degradation trend curve, the barrel ablation and wear state is accurately predicted and evaluated, providing a scientific basis for the use and maintenance of artillery.

[0137] From the hardware level, in order to accurately predict and evaluate the ablation and wear state of the barrel and provide a scientific basis for the use and maintenance of the artillery, the present application provides an embodiment of an electronic device for implementing all or part of the contents of the barrel ablation and wear prediction method based on degenerate path approximation, and the electronic device specifically includes the following contents: Processor, memory, communication interface and bus; wherein the processor, memory and communication interface communicate with each other through the bus; the communication interface is used to realize information transmission between the barrel ablation wear prediction device based on degradation path approximation and related devices such as core business system, user terminal and related database; the logic controller can be a desktop computer, tablet computer and mobile terminal, etc., and the present embodiment is not limited thereto. In the present embodiment, the logic controller can be implemented with reference to the embodiment of the barrel ablation wear prediction method based on degradation path approximation and the embodiment of the barrel ablation wear prediction device based on degradation path approximation in the embodiment, and the contents thereof are incorporated herein, and the repeated parts are not repeated.

[0138] It is understandable that the user terminal may include a smart phone, a tablet electronic device, a network set-top box, a portable computer, a desktop computer, a personal digital assistant (PDA), a vehicle-mounted device, a smart wearable device, etc. Among them, the smart wearable device may include smart glasses, a smart watch, a smart bracelet, etc.

[0139] In practical applications, part of the barrel ablation wear prediction method based on degradation path approximation can be executed on the electronic device side as described above, or all operations can be completed in the client device. The specific selection can be based on the processing capability of the client device and the limitations of the user's usage scenario. This application does not limit this. If all operations are completed in the client device, the client device may also include a processor.

[0140] The client device may have a communication module (i.e., a communication unit) that can communicate with a remote server to achieve data transmission with the server. The server may include a server on the task scheduling center side, and other implementation scenarios may also include a server on an intermediate platform, such as a server on a third-party server platform that has a communication link with the task scheduling center server. The server may include a single computer device, or a server cluster consisting of multiple servers, or a server structure of a distributed device.

[0141] Fig. 9 FIG. 9 is a schematic block diagram of the system structure of the electronic device 9600 according to an embodiment of the present application. Fig. 9 As shown, the electronic device 9600 may include a central processor 9100 and a memory 9140; the memory 9140 is coupled to the central processor 9100. It is worth noting that Fig. 9 is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0142] In one embodiment, the barrel ablation wear prediction method function based on degradation path approximation can be integrated into the central processor 9100. The central processor 9100 can be configured to perform the following control: Step S101: normalizing the collected data of ablation and wear of the gun barrel, establishing a multidimensional degradation path sample space based on the normalized ablation and wear data, mapping the sample space to a high-order feature space using a Sigmoid function, extracting similarity measurement features and path curvature features of the degradation path in the high-order feature space, clustering the degradation paths into a plurality of degradation mode clusters according to the similarity measurement features, fitting the path curvature features of each of the degradation mode clusters using a curve fitting algorithm, and obtaining path approximate expressions under different degradation modes; Step S102: iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on the least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on the output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; Step S103: reading the real-time firing data of the artillery to be predicted, calculating the comprehensive load index according to the barrel temperature, firing pressure and propellant loading contained in the firing data, inputting the comprehensive load index into the barrel ablation degradation prediction model, obtaining the barrel ablation wear prediction value and its prediction interval under different firing numbers, establishing the barrel ablation wear degradation trend curve based on the prediction value, calculating the change rate and acceleration characteristics of the barrel ablation wear according to the degradation trend curve, and judging the ablation wear state of the gun barrel in combination with the prediction interval.

[0143] From the above description, it can be seen that the electronic device provided in the embodiment of the present application establishes a high-order feature space by normalizing the ablation and wear data and mapping the Sigmoid function. The similarity measurement characteristics and path curvature characteristics of the degradation path are innovatively extracted to achieve accurate clustering of degradation patterns and path approximation. The system uses an alternating optimization algorithm and a gradient descent method to optimize the path expression parameters and establish a confidence interval estimation mechanism. A comprehensive load index is constructed by combining the barrel temperature, firing pressure and propellant loading. Through the rate of change and acceleration characteristics of the degradation trend curve, the accurate prediction and evaluation of the barrel ablation and wear state are achieved, providing a scientific basis for the use and maintenance of artillery.

[0144] In another embodiment, the barrel ablation and wear prediction device based on degradation path approximation can be configured separately from the central processing unit 9100. For example, the barrel ablation and wear prediction device based on degradation path approximation can be configured as a chip connected to the central processing unit 9100, and the function of the barrel ablation and wear prediction method based on degradation path approximation can be implemented through the control of the central processing unit.

[0145] like Fig. 9 As shown, the electronic device 9600 may also include: a communication module 9110, an input unit 9120, an audio processor 9130, a display 9160, and a power supply 9170. It is worth noting that the electronic device 9600 does not necessarily have to include Fig. 9 In addition, the electronic device 9600 may also include Fig. 9 For components not shown, reference may be made to the prior art.

[0146] like Fig. 9 As shown, the central processing unit 9100 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor device and / or logic device. The central processing unit 9100 receives input and controls the operation of various components of the electronic device 9600.

[0147] The memory 9140 may be, for example, one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory or other suitable devices. The above-mentioned information related to the failure may be stored, and a program for executing the relevant information may also be stored. The CPU 9100 may execute the program stored in the memory 9140 to implement information storage or processing, etc.

[0148] The input unit 9120 provides input to the central processing unit 9100. The input unit 9120 is, for example, a key or a touch input device. The power supply 9170 is used to provide power to the electronic device 9600. The display 9160 is used to display display objects such as images and texts. The display may be, for example, an LCD display, but is not limited thereto.

[0149] The memory 9140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 9140 may also be some other type of device. The memory 9140 includes a buffer memory 9141 (sometimes referred to as a buffer). The memory 9140 may include an application / function storage unit 9142, which is used to store application programs and function programs or processes for executing the operation of the electronic device 9600 through the central processor 9100.

[0150] The memory 9140 may also include a data storage unit 9143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 9144 of the memory 9140 may include various drivers for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0151] The communication module 9110 is a transmitter / receiver that sends and receives signals via the antenna 9111. The communication module 9110 (transmitter / receiver) is coupled to the central processor 9100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0152] Based on different communication technologies, multiple communication modules 9110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless LAN module. The communication module 9110 (transmitter / receiver) is also coupled to a speaker 9131 and a microphone 9132 via an audio processor 9130 to provide an audio output via the speaker 9131 and receive an audio input from the microphone 9132, thereby realizing a common telecommunication function. The audio processor 9130 may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor 9130 is also coupled to the central processor 9100, so that recording can be performed on the local machine through the microphone 9132, and the sound stored on the local machine can be played through the speaker 9131.

[0153] The embodiment of the present application also provides a computer-readable storage medium capable of implementing all the steps of the barrel ablation wear prediction method based on degradation path approximation in the above embodiment, where the execution subject is a server or a client. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, all the steps of the barrel ablation wear prediction method based on degradation path approximation in the above embodiment are implemented. For example, when the processor executes the computer program, the following steps are implemented: Step S101: normalizing the collected data of ablation and wear of the gun barrel, establishing a multidimensional degradation path sample space based on the normalized ablation and wear data, mapping the sample space to a high-order feature space using a Sigmoid function, extracting similarity measurement features and path curvature features of the degradation path in the high-order feature space, clustering the degradation paths into a plurality of degradation mode clusters according to the similarity measurement features, fitting the path curvature features of each of the degradation mode clusters using a curve fitting algorithm, and obtaining path approximate expressions under different degradation modes; Step S102: iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on the least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on the output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; Step S103: reading the real-time firing data of the artillery to be predicted, calculating the comprehensive load index according to the barrel temperature, firing pressure and propellant loading contained in the firing data, inputting the comprehensive load index into the barrel ablation degradation prediction model, obtaining the barrel ablation wear prediction value and its prediction interval under different firing numbers, establishing the barrel ablation wear degradation trend curve based on the prediction value, calculating the change rate and acceleration characteristics of the barrel ablation wear according to the degradation trend curve, and judging the ablation wear state of the gun barrel in combination with the prediction interval.

[0154] From the above description, it can be seen that the computer-readable storage medium provided in the embodiment of the present application establishes a high-order feature space by normalizing the ablation and wear data and mapping the Sigmoid function. The similarity measurement characteristics and path curvature characteristics of the degradation path are innovatively extracted to achieve accurate clustering of degradation patterns and path approximation. The system uses an alternating optimization algorithm and a gradient descent method to optimize the path expression parameters and establish a confidence interval estimation mechanism. A comprehensive load index is constructed by combining the barrel temperature, firing pressure and propellant loading. Through the rate of change and acceleration characteristics of the degradation trend curve, the barrel ablation and wear state is accurately predicted and evaluated, providing a scientific basis for the use and maintenance of artillery.

[0155] The embodiment of the present application also provides a computer program product capable of implementing all the steps of the barrel ablation wear prediction method based on degradation path approximation in the above embodiment, where the execution subject is a server or a client. When the computer program / instruction is executed by a processor, the steps of the barrel ablation wear prediction method based on degradation path approximation are implemented. For example, the computer program / instruction implements the following steps: Step S101: normalizing the collected data of ablation and wear of the gun barrel, establishing a multidimensional degradation path sample space based on the normalized ablation and wear data, mapping the sample space to a high-order feature space using a Sigmoid function, extracting similarity measurement features and path curvature features of the degradation path in the high-order feature space, clustering the degradation paths into a plurality of degradation mode clusters according to the similarity measurement features, fitting the path curvature features of each of the degradation mode clusters using a curve fitting algorithm, and obtaining path approximate expressions under different degradation modes; Step S102: iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on the least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on the output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; Step S103: reading the real-time firing data of the artillery to be predicted, calculating the comprehensive load index according to the barrel temperature, firing pressure and propellant loading contained in the firing data, inputting the comprehensive load index into the barrel ablation degradation prediction model, obtaining the barrel ablation wear prediction value and its prediction interval under different firing numbers, establishing the barrel ablation wear degradation trend curve based on the prediction value, calculating the change rate and acceleration characteristics of the barrel ablation wear according to the degradation trend curve, and judging the ablation wear state of the gun barrel in combination with the prediction interval.

[0156] From the above description, it can be seen that the computer program product provided in the embodiment of the present application establishes a high-order feature space by normalizing the ablation and wear data and mapping the Sigmoid function. The similarity measurement characteristics and path curvature characteristics of the degradation path are innovatively extracted to achieve accurate clustering of degradation patterns and path approximation. The system uses an alternating optimization algorithm and a gradient descent method to optimize the path expression parameters and establish a confidence interval estimation mechanism. A comprehensive load index is constructed by combining the barrel temperature, firing pressure and propellant loading. Through the rate of change and acceleration characteristics of the degradation trend curve, the accurate prediction and evaluation of the barrel ablation and wear state are achieved, providing a scientific basis for the use and maintenance of artillery.

[0157] It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0158] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (apparatus), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0159] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0160] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0161] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A barrel ablation wear prediction method based on degradation path approximation, characterized in that: The method comprises: The collected data of ablation and wear of the gun barrel are normalized, and a multidimensional degradation path sample space is established based on the normalized ablation and wear data. The sample space is mapped to a high-order feature space using a Sigmoid function, and similarity measurement features and path curvature features of the degradation path are extracted in the high-order feature space. The degradation path is clustered into multiple degradation mode clusters according to the similarity measurement features, and the path curvature features of each degradation mode cluster are fitted using a curve fitting algorithm to obtain path approximate expressions under different degradation modes. Iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on the least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on an output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; The real-time firing data of the artillery to be predicted is read, a comprehensive load index is calculated according to the barrel temperature, firing pressure and propellant loading amount contained in the firing data, the comprehensive load index is input into the barrel ablation degradation prediction model, and the barrel ablation wear prediction value and prediction interval under different firing numbers are obtained, a degradation trend curve of barrel ablation wear is established based on the prediction value, the change rate and acceleration characteristics of barrel ablation wear are calculated according to the degradation trend curve, and the ablation wear state of the gun barrel is judged in combination with the prediction interval.

2. The barrel ablation wear prediction method based on degradation path approximation according to claim 1, characterized in that: The collected data of ablation and wear of the gun barrel is normalized, a multidimensional degradation path sample space is established based on the normalized ablation and wear data, the sample space is mapped to a high-order feature space using a Sigmoid function, and similarity measurement features and path curvature features of the degradation path are extracted in the high-order feature space, including: Collecting ablation wear depth data, barrel pressure data and barrel temperature data of a gun barrel under different firing numbers, performing maximum and minimum value normalization processing on the ablation wear depth data, calculating a comprehensive load coefficient in combination with the barrel pressure data and barrel temperature data, and constructing a multidimensional feature vector with the comprehensive load coefficient and the normalized ablation wear depth data; A degenerate path sample space is established based on the multidimensional feature vector, and a Sigmoid function is used to perform nonlinear mapping on the degenerate path sample space to obtain a high-dimensional feature expression. In the high-dimensional feature expression, the Mahalanobis distance between the degenerate paths is calculated to obtain a similarity measurement feature, and a differential geometry method is used to calculate the curvature change of the degenerate path to obtain a path curvature feature.

3. The barrel ablation wear prediction method based on degradation path approximation according to claim 1, characterized in that: The degradation paths are clustered into a plurality of degradation mode clusters according to the similarity measurement characteristics, and the path curvature characteristics of each degradation mode cluster are fitted using a curve fitting algorithm to obtain path approximate expressions under different degradation modes, including: Based on the similarity measurement feature, a distance matrix of the degradation path is constructed, a hierarchical clustering algorithm is used to perform cluster analysis on the distance matrix, the degradation path is divided into a plurality of degradation pattern clusters according to the cluster analysis result, the cluster center of each degradation pattern cluster is calculated, and the cluster center is subjected to feature decomposition to obtain the main degradation features; The path curvature characteristics in each of the degradation pattern clusters are normalized, and the normalized path curvature characteristics are fitted using a polynomial fitting algorithm. The coefficient matrix of the fitting curve is calculated, and an approximate expression of the degradation path is established based on the coefficient matrix. The approximate expression is combined with the main degradation characteristics to construct a complete path approximation model.

4. The barrel ablation wear prediction method based on degradation path approximation according to claim 1, characterized in that: The iterative optimization of the parameters in the path approximation expression by the alternating optimization algorithm, establishing an optimization objective function based on the least squares criterion, and solving the optimization objective function by the gradient descent method to obtain the optimal parameter combination include: A parameter vector to be optimized is set in the path approximation expression, an objective function for parameter optimization is constructed based on the least squares criterion, the objective function includes a path approximation error term and a parameter regularization term, a partial derivative of the objective function with respect to the parameter vector is obtained, and a parameter update direction and step size are calculated according to the gradient expression; An alternating iterative strategy is used to perform block optimization on the parameter vector. In each round of iteration, some parameters are fixed and other parameters are updated. The objective function value after each parameter update is calculated. The iteration is stopped when the relative change of the objective function value is less than the preset convergence threshold, and the finally converged parameter vector is taken as the optimal parameter combination.

5. The barrel ablation wear prediction method based on degradation path approximation according to claim 1, characterized in that: Substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on the output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty, comprises: Substituting the optimal parameter combination into the parameter vector in the path approximation expression, mapping the path approximation expression to the prediction space using a kernel function, building a nonlinear prediction model in combination with the main degradation characteristics of the degradation path, calibrating the parameters of the nonlinear prediction model based on training samples, and generating a barrel ablation degradation prediction model; The probability distribution characteristics of the output results of the barrel ablation degradation prediction model are calculated, the variance of the predicted value is estimated by the delta method, the standard error of the predicted result is calculated based on the variance, the confidence coefficient is determined according to a preset confidence level, the confidence coefficient is multiplied by the standard error to obtain the half-width of the prediction interval, and the half-width of the prediction interval is superimposed on the predicted value to construct the ablation wear prediction interval.

6. The barrel ablation wear prediction method based on degradation path approximation according to claim 1, characterized in that: The real-time firing data of the artillery to be predicted is read, a comprehensive load index is calculated according to the barrel temperature, firing pressure and propellant charge contained in the firing data, the comprehensive load index is input into the barrel ablation degradation prediction model, and the barrel ablation wear prediction value and its prediction interval under different firing numbers are obtained, including: Acquire firing data of the artillery to be predicted, extract barrel temperature data, firing pressure data and propellant charge data from the firing data, normalize the barrel temperature data, firing pressure data and propellant charge data, calculate the comprehensive load index by weighted superposition, and determine the weight coefficient of each index according to historical data; The comprehensive load index is input into the barrel ablation degradation prediction model according to the sequence of firing numbers, and the ablation wear prediction value corresponding to each firing number is calculated based on the prediction model. The upper confidence limit and lower confidence limit of each prediction value are calculated in combination with the pre-constructed prediction interval, and the prediction value and its upper and lower confidence limits constitute a complete ablation wear prediction result.

7. The barrel ablation wear prediction method based on degradation path approximation according to claim 1, characterized in that: The method of establishing a degradation trend curve of barrel ablation wear based on the predicted value, calculating the change rate and acceleration characteristics of barrel ablation wear according to the degradation trend curve, and judging the ablation wear state of the gun barrel in combination with the predicted interval includes: The predicted value is numerically smoothed, a piecewise polynomial interpolation method is used to construct a continuous degradation trend curve, a first-order derivative of the degradation trend curve is calculated to obtain a change rate characteristic of ablation wear, a second-order derivative is calculated to obtain an acceleration characteristic of ablation wear, and key characteristic points of the degradation process are calculated according to the change rate characteristic and the acceleration characteristic; The degradation process is divided into multiple stages based on the key characteristic points, and the reliability index of each stage is calculated in combination with the upper confidence limit and the lower confidence limit of the prediction interval. The ablation and wear state of the gun barrel is quantitatively evaluated according to the preset state evaluation criteria to generate a barrel state evaluation result.

8. A barrel ablation wear prediction device based on degradation path approximation, characterized in that: The device comprises: A path expression construction module is used to normalize the collected data of ablation and wear of the gun barrel, establish a multidimensional degradation path sample space based on the normalized ablation and wear data, use a Sigmoid function to map the sample space to a high-order feature space, and extract similarity measurement features and path curvature features of the degradation path in the high-order feature space, cluster the degradation path into multiple degradation mode clusters according to the similarity measurement features, use a curve fitting algorithm to fit the path curvature features of each degradation mode cluster, and obtain path approximate expressions under different degradation modes; a wear interval prediction module, for iteratively optimizing the parameters in the path approximation expression by an alternating optimization algorithm, establishing an optimization objective function based on a least squares criterion, solving the optimization objective function by a gradient descent method to obtain an optimal parameter combination, substituting the optimal parameter combination into the path approximation expression to generate a barrel ablation degradation prediction model, performing confidence interval estimation on an output result of the barrel ablation degradation prediction model, and constructing an ablation wear prediction interval with prediction uncertainty; The wear state prediction module is used to read the real-time shooting data of the artillery to be predicted, calculate the comprehensive load index according to the barrel temperature, shooting pressure and propellant loading contained in the shooting data, input the comprehensive load index into the barrel ablation degradation prediction model, obtain the barrel ablation wear prediction value and its prediction interval under different shooting numbers, establish the barrel ablation wear degradation trend curve based on the prediction value, calculate the change rate and acceleration characteristics of the barrel ablation wear according to the degradation trend curve, and judge the ablation wear state of the artillery barrel in combination with the prediction interval.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the barrel ablation wear prediction method based on degradation path approximation according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the barrel ablation wear prediction method based on degradation path approximation according to any one of claims 1 to 7 are implemented.

Images