Method for measuring space angle of dredger rake pipe

Abstract

The invention relates to the technical field of ship engineering, and particularly discloses a dredger rake pipe space angle measurement method, which comprises the following steps of: collecting motion data of a dredger body and a rake pipe in real time, removing high-frequency noise by adopting median filtering, extracting features, constructing a comprehensive feature vector, and separating nonlinear influence of ship body shaking and rake pipe attitude by utilizing a fuzzy reasoning system, so as to measure the rake pipe space angle of the dredger. The attitude estimation result is corrected in combination with real-time error feedback, the estimation result is dynamically optimized and the system deviation is corrected through Bayesian estimation and a particle filter algorithm, so that the measurement precision is improved, and the system also evaluates the overall measurement accuracy in combination with a machine learning model, automatically adjusts the measurement parameters and flow, and improves the measurement accuracy. The adaptability and robustness of the system are improved, and for an inaccurate measurement result, the system can analyze reasons and re-measure the space angle so as to ensure long-term stable operation.

Description

Technical Field

[0001] The present invention relates to the technical field of ship engineering, and particularly to a method for measuring the spatial angle of a dredger's rake pipe. Background Art

[0002] As an important equipment for water area dredging, port construction and river regulation, dredgers are widely used in complex underwater operations. The rake pipe system of a dredger is one of its core components, responsible for excavating and transporting sediment. To ensure the accuracy and efficiency of operations, it is necessary to accurately measure and real-time monitor the spatial angle of the rake pipe. The accurate measurement of the spatial angle of the rake pipe not only helps to optimize the operation angle and improve the dredging effect, but also reduces the impact on the water environment and avoids unnecessary resource waste. With the development of modern computing technology, the angle measurement method based on sensor data has gradually become the mainstream, and with the continuous progress of signal processing, data fusion and artificial intelligence technologies, the measurement accuracy and real-time performance of the spatial angle of the dredger's rake pipe have been significantly improved.

[0003] The existing technology has the following deficiencies:

[0004] Due to the dynamic changes of the ship itself, such as the hull's swaying and the influence of waves, there may be non-linear errors in attitude calculation, resulting in inaccurate angle calculation. During the operation of the rake pipe, complex non-linear motions may occur due to sediment resistance factors, which makes it difficult for sensors to accurately track its spatial angle. Summary of the Invention

[0005] The purpose of the present invention is to provide a method for measuring the spatial angle of a dredger's rake pipe to solve the problems in the above background.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A method for measuring the spatial angle of a dredger's rake pipe includes the following steps:

[0008] S1: Real-time collect the motion data of the hull and the rake pipe, including acceleration, angular velocity and attitude information;

[0009] S2: According to the motion data of the hull and the rake pipe, separate the non-linear influence and optimize the attitude estimation;

[0010] S3: According to the optimized attitude estimation, through real-time error feedback calculation, evaluate the difference between the estimation result and the actual measurement value;

[0011] S4: Based on the evaluation result, use the error feedback information to real-time correct the estimation result in the filtering process;

[0012] S5: Based on the real-time correction results, the overall accuracy of the dredger rake tube spatial angle measurement system is evaluated, and for inaccurate identification results, the causes of inaccurate identification are analyzed, and the spatial angle measurement is re-performed.

[0013] As a further solution of the present invention: acquiring the motion data of the hull and the rake tube, and using median filtering to remove high-frequency noise;

[0014] The acceleration, angular velocity and attitude information of the hull and rake tube are extracted and a comprehensive feature vector is constructed.

[0015] As a further solution of the present invention: the construction process of the comprehensive feature vector is:

[0016] Obtain the motion data of the hull and rake tube, and extract time domain features from the acceleration, angular velocity and attitude data, including mean, standard deviation, maximum, minimum, peak value and variance;

[0017] Use fast Fourier transform to perform frequency domain analysis on acceleration and angular velocity data and extract spectrum features such as frequency distribution, spectrum density and frequency components;

[0018] Combined with time-frequency analysis technology, time-frequency domain features are extracted to capture the time-varying characteristics in the data;

[0019] The extracted time domain, frequency domain and time-frequency domain features are combined to form a preliminary feature vector;

[0020] The gradient boosting tree algorithm is used to train the preliminary feature vector and automatically select the features most relevant to the target prediction. During the gradient boosting tree training process, new feature combinations and cross-features are automatically generated to improve the feature expression capability. The influence of each feature on the final goal is evaluated, irrelevant features are removed, the most qualified features are retained, and the optimized features are combined to generate the final comprehensive feature vector.

[0021] As a further solution of the present invention: the separation of nonlinear influences and optimization of posture estimation specifically includes:

[0022] Obtain comprehensive feature vectors, and use singular spectrum decomposition to convert the comprehensive feature vectors of the hull and the rake tube into input features for fuzzy inference system processing; formulate fuzzy rules to describe the relationship between the hull sway and the rake tube posture; use adaptive methods to adjust the membership function and rule weight of fuzzy rules based on real-time data to dynamically optimize the performance of the inference system;

[0023] The collected motion data is input into the fuzzy inference system, and the nonlinear effects between the hull sway and the rake tube attitude are separated based on the output of the fuzzy inference system;

[0024] Through the membership value of fuzzy reasoning, the influence of hull sway and the estimated value of rake tube posture are calculated to achieve dynamic separation of nonlinear influences;

[0025] According to the real-time feedback data, the fuzzy rules and membership functions in the fuzzy reasoning system are dynamically updated. Through the output of the fuzzy reasoning system, combined with the real-time data and the optimized model, accurate posture estimation results of the hull and the rake tube are provided.

[0026] As a further solution of the present invention: the difference between the evaluation result and the actual measurement value specifically includes:

[0027] According to the optimized posture estimation results, the actual measured value is compared with the estimated value, and the estimated deviation value is calculated. The calculation process is:

[0028] The difference between the estimated value and the actual measured value is calculated to obtain the estimated deviation. The estimated deviation is combined with the prior distribution and the likelihood function to obtain the posterior distribution, and the deviation estimate is updated. The calculation expression is:

[0029] ;

[0030] In the formula, represents the actual measured value, represents the estimation bias, Indicates the actual measured value given Then, estimate the deviation Update, Indicates the given estimated deviation In the case of The probability of represents the probability distribution of the estimation deviation, represents the normalization constant, ensuring that the total probability of the posterior distribution is 1;

[0031] Estimation bias It follows a normal distribution, and the prior distribution is:

[0032] ;

[0033] In the formula, represents the prior variance of the estimated bias, represents the covariance matrix of the estimates, represents the transpose operation, Represents the dimension of the data;

[0034] Substituting the prior distribution and likelihood function into the Bayesian formula, we get the posterior distribution, and the calculation expression is:

[0035] ;

[0036] In the formula, Indicates an estimated value;

[0037] By maximizing the posterior distribution, the maximum posterior estimate of the estimated deviation is obtained, and the calculation expression is:

[0038] ;

[0039] In the formula, represents the maximum a posteriori estimate, represents the maximum value function;

[0040] By taking the derivative of the posterior distribution and setting it to 0, we get the estimated deviation value. The calculation expression is:

[0041] ;

[0042] In the formula, represents the estimated deviation value, represents the precision matrix of the actual measurement data, A precision matrix representing the a priori estimates.

[0043] As a further solution of the present invention: the use of error feedback information to correct the estimation result in the filtering process in real time specifically includes:

[0044] Obtain the estimated deviation value, calculate the correction coefficient based on the estimated deviation value in combination with the particle filter algorithm, adjust the estimation result at the current moment according to the correction coefficient, and obtain the corrected estimation state.

[0045] As a further solution of the present invention: the process of obtaining the correction coefficient is:

[0046] Get the estimated deviation value and calculate the correction coefficient based on the estimated deviation value. The calculation expression is:

[0047] ;

[0048] In the formula, represents the correction factor, represents the estimated deviation value, represents the transpose operation, Indicates the first Particles, represents the total number of particles, represents the time step, represents the covariance matrix of the state estimate, Particle weight.

[0049] As a further solution of the present invention: the overall accuracy of the dredger rake tube spatial angle measurement system is evaluated, specifically including:

[0050] According to the real-time correction result, the estimated deviation value and correction coefficient in the process of measuring the spatial angle of the dredger rake pipe are used to construct a combined comprehensive feature vector as the input of the machine learning model. The machine learning model is trained in combination with historical data. According to the training result, the accuracy evaluation result of the whole system is output. The machine learning model is a deep learning model. The deep learning model includes an input layer: for receiving the combined comprehensive feature vector, a hidden layer: composed of a multi-layer fully connected neural network, for extracting nonlinear features, an output layer: outputting the accuracy evaluation result of the whole system, and using historical data to perform supervised learning training on the deep learning model. The accuracy label in the historical data is the accuracy score of the whole system.

[0051] As a further solution of the present invention: according to the accuracy evaluation result of the whole system, the quality recognition result of the system is divided into accurate recognition result and inaccurate recognition result, which specifically includes:

[0052] Obtain the accuracy score of the dredger rake tube spatial angle measurement system, and determine whether the accuracy score is greater than or equal to a preset threshold. If so, it is classified as an accurate recognition result; if not, it is classified as an inaccurate recognition result.

[0053] As a further solution of the present invention: the inaccurate recognition result, analyzing the cause of the inaccurate recognition, and re-measuring the spatial angle specifically includes:

[0054] By comparing the similarity between the inaccurate recognition result and the historical measurement data, if the similarity is lower than the preset threshold, it is determined that the current measurement value is inconsistent with the historical data characteristics; according to the cause of the inaccurate recognition, the measurement parameters and processes are dynamically adjusted, including: recalibrating the sensor and adjusting the parameters of the sampling frequency; re-measuring the rake tube space angle to obtain a new measurement value, and correcting the measurement value based on the correction coefficient, calculating a new corrected estimate, reconstructing the comprehensive feature vector, combining the corrected estimate and measurement value with the historical data, and updating the overall accuracy evaluation result of the system.

[0055] Beneficial effects of the present invention:

[0056] (1) The present invention optimizes the measurement process of the spatial angle of the dredger's rake pipe by combining various advanced technical means. In terms of real-time optimizing attitude estimation, the system collects the acceleration, angular velocity, and attitude information of the hull and the rake pipe, and uses median filtering to remove high-frequency noise to ensure the smoothness and accuracy of the input data. In the error feedback correction process, the system uses Bayesian estimation and particle filtering algorithms to dynamically adjust the estimation results, and corrects the estimation deviation in real time, thereby eliminating the interference of external environmental changes (such as hull shaking, waves) on the measurement results. Through continuous error feedback and adaptive correction, the system can accurately separate the non-linear effects and ensure that the measurement results are highly consistent with the actual state. In addition, by combining a fuzzy inference system to model the complex relationship between the hull and the rake pipe attitudes, and dynamically updating through the optimized model, the robustness and accuracy of the angle measurement are further improved, enabling this method to provide stable and reliable measurement results in complex and changeable environments.

[0057] (2) The present invention enables the dredger's rake pipe spatial angle measurement system to have a high degree of adaptability by integrating the mechanism of real-time correction and machine learning model feedback. The system first analyzes the real-time correction results and continuously updates the estimation model to ensure the accuracy and stability of the measurement results. When inaccurate recognition results occur during the measurement process, the system analyzes the historical data through a deep learning model, automatically identifies the error sources, and adjusts the measurement parameters and processes. For example, the system can recalibrate the sensors, adjust the sampling frequency, or change the filtering strategy to cope with the impacts brought by environmental changes. Through continuous optimization and adaptive adjustment, the system can maintain the stability of the measurement accuracy under different working environments, operating conditions, and complex dynamic scenarios. In addition, the system can also dynamically evaluate its overall accuracy, timely feedback and correct inaccurate results, thereby ensuring long-term stable operation, avoiding the performance degradation or accuracy decline problems that are prone to occur in traditional methods under complex conditions, and enhancing the robustness and reliability of the system. Description of the Drawings

[0058] The present invention will be further described below in conjunction with the drawings.

[0059] Figure 1 It is the specific step flow block diagram of the dredger's rake pipe spatial angle measurement method of the present invention;

[0060] Figure 2 It is the calculation flow block diagram of the estimated deviation value in the present invention. Detailed Embodiments

[0061] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] See also Figure 1 As shown, the present invention is a method for measuring the spatial angle of a dredger rake pipe, comprising the following steps:

[0063] S1: Real-time collection of motion data of the hull and rake tube, including acceleration, angular velocity and attitude information;

[0064] S2: Separate nonlinear effects and optimize attitude estimation based on hull and rake motion data;

[0065] S3: Based on the optimized posture estimation, the difference between the estimation result and the actual measurement value is evaluated through real-time error feedback calculation;

[0066] S4: Based on the evaluation results, the estimation results in the filtering process are corrected in real time using the error feedback information;

[0067] S5: Based on the real-time correction results, the overall accuracy of the dredger rake tube spatial angle measurement system is evaluated, and for inaccurate identification results, the causes of inaccurate identification are analyzed, and the spatial angle measurement is re-performed.

[0068] In S1, the motion data of the hull and the rake tube are collected in real time, including acceleration, angular velocity and attitude information, including:

[0069] A variety of high-precision inertial measurement units (IMUs), including accelerometers, gyroscopes, and attitude sensors, are installed on the hull and rake tube of the dredger. The IMU of the hull is arranged near the center of gravity to ensure that the collected data can accurately reflect the overall motion state; the IMU of the rake tube is installed at key moving parts, such as joint nodes and ends, to facilitate the capture of local motion characteristics. Each sensor is in close contact with the measurement surface through an industrial-grade fixture to avoid data deviation caused by loose installation.

[0070] The sensor collects the 3D acceleration, 3D angular velocity and attitude (such as Euler angle) information of the hull and rake tube. The data acquisition frequency is designed to be set to 100Hz, for example, to ensure the capture of high-frequency details of dynamic motion. The collected raw data is transmitted to the central processing unit (CPU) in real time through a high-speed communication protocol (such as wireless network) for unified processing, and the data streams of each sensor are synchronized using timestamps to ensure the alignment of the motion information of the hull and rake tube.

[0071] The raw data received in real time is preprocessed, including noise filtering, zero bias correction and drift compensation. The acceleration and angular velocity data are fused through the Kalman filter to eliminate the influence of random noise and obtain more stable attitude information. At the same time, according to the known sensor installation position and reference coordinate system, the collected data is uniformly converted to the global coordinate system to provide accurate input data for subsequent spatial angle measurement and correction.

[0072] In S2, based on the hull and rake motion data, nonlinear effects are separated and attitude estimation is optimized, including:

[0073] Get the motion data of the hull and rake tube, and use median filtering to remove high-frequency noise;

[0074] Extract the acceleration, angular velocity and attitude information of the hull and rake tube and construct a comprehensive feature vector;

[0075] The construction process of the comprehensive feature vector is:

[0076] First, the motion data of the hull and the rake tube are collected in real time through IMU, accelerometer, and gyroscope sensors. The collected data include acceleration (Ax, Ay, Az), angular velocity (ωx, ωy, ωz), and attitude angle (pitch angle, roll angle, yaw angle).

[0077] The collected raw data is processed by median filtering to remove high-frequency noise. Median filtering can effectively filter out instantaneous spike noise, maintain the smoothness of the signal, and avoid the interference of high-frequency noise on subsequent feature extraction.

[0078] Feature extraction includes:

[0079] Time domain features: Extract basic statistical features such as mean, standard deviation, maximum, minimum, peak, and variance from acceleration, angular velocity, and attitude data. By calculating these features, the basic laws and fluctuation characteristics of motion can be captured from the original data.

[0080] Frequency domain features: Use Fast Fourier Transform (FFT) to perform frequency domain analysis on acceleration and angular velocity data to extract spectrum features, such as frequency distribution, spectrum density, and main frequency components. Frequency domain features can reveal the motion patterns of the hull and rake tube at different frequencies and help identify the laws of nonlinear motion.

[0081] Combined features of time domain and frequency domain: Through time-frequency analysis techniques (such as short-time Fourier transform and waveform transformation), the combined features of time domain and frequency domain are extracted to capture the time-varying characteristics of the hull and rake tube under dynamic changes.

[0082] Constructing a comprehensive feature vector includes:

[0083] The time domain features, frequency domain features and time-frequency domain features extracted from the acceleration, angular velocity and attitude data are combined to form a preliminary feature set. Each feature vector will include:

[0084] Acceleration characteristics (such as mean, standard deviation, maximum value)

[0085] Angular velocity characteristics (such as frequency components, variance)

[0086] Attitude characteristics (such as statistics of pitch angle, roll angle, and yaw angle)

[0087] These multi-dimensional features are concatenated into a large feature vector that contains all the important dynamic information;

[0088] Use the gradient boosting tree to select features and evaluate the influence of each feature on the prediction target. The gradient boosting tree algorithm automatically selects the most informative features, removes irrelevant features, and improves the expressiveness of features.

[0089] During the training process of the gradient boosting tree, the cross-effects of different features are considered. The gradient boosting tree adaptively handles the cross-effects of features through the tree structure and generates new composite features. For example, different axis components of acceleration are combined with angular velocity to create new features to enhance the ability to capture complex dynamic patterns.

[0090] The specific steps of separating nonlinear effects are:

[0091] Real-time collection of motion data of the hull and the rake tube, including acceleration (Ax, Ay, Az), angular velocity (ωx, ωy, ωz) and attitude angle (pitch angle, roll angle, yaw angle) information;

[0092] Preprocess the collected raw data, use median filtering to remove high-frequency noise, and normalize the data to ensure the uniform scale of the input data;

[0093] Using the singular spectrum decomposition method in "Nonlinear Dynamics Analysis", potential nonlinear dynamic features are identified from acceleration, angular velocity and attitude data;

[0094] Decompose the signal by SSD method, identify the main periodic components and mutation points, and understand the behavior patterns of the hull and rake tube in a nonlinear environment;

[0095] The acceleration, angular velocity and attitude data of the hull and rake tube are mapped to the dynamic nonlinear feature space to extract the dynamic features in the time domain and frequency domain. These features will be used for analysis in the subsequent fuzzy inference system.

[0096] Design an adaptive fuzzy inference system (FIS) to handle non-linear dynamic characteristics in the form of a fuzzy rule base. The inputs to the fuzzy system are acceleration, angular velocity, attitude angle, and features extracted from non-linear dynamic analysis.

[0097] Based on the non-linear relationships observed in the data, formulate appropriate fuzzy rules to describe the interaction between the hull and the rake pipe. These include: "When the hull acceleration is large and the angular velocity is small, the attitude change of the rake pipe is slow."

[0098] Use the adaptive fuzzy inference method to automatically adjust the membership functions and rule weights of the fuzzy rules. These adjustments are based on real-time data feedback to optimize the system's response ability to non-linear effects.

[0099] Utilize the fuzzy inference system to automatically identify the non-linear effects between hull sway and rake pipe attitude and perform dynamic separation. Based on real-time data input, the system calculates the output of the fuzzy rules, gives the membership degree values for each state, and thus estimates the non-linear dynamics of the hull and the rake pipe respectively.

[0100] Through the output of the fuzzy inference, obtain the influence value of hull sway and the estimated value of rake pipe attitude, and perform the separation of non-linear effects. This process does not rely on traditional non-linear models, but rather the rules of the fuzzy inference system dynamically adapt to changing motion states.

[0101] According to the real-time collected data, the system automatically updates the fuzzy rules and membership functions. Through the real-time feedback mechanism, the fuzzy inference system can adaptively adjust the inference process according to changes in the environment (such as wave changes, hull load changes), avoiding error accumulation caused by improper initial settings.

[0102] Use real-time control algorithms to optimize the performance of the fuzzy system, gradually improving the system's recognition and separation accuracy of the non-linear relationship between hull sway and rake pipe attitude.

[0103] The output of the fuzzy inference system will be used as the input for the final attitude estimation. Combining real-time feedback information, give the accurate attitude angles of the hull and the rake pipe. The final attitude estimation result is output through the data fusion method and used for the dynamic control and positioning of the hull and the rake pipe.

[0104] Please refer to Figure 2 As shown, in S3, according to the optimized attitude estimation, through real-time error feedback calculation, evaluate the difference between the estimation result and the actual measurement value, specifically including:

[0105] According to the optimized attitude estimation result, compare and analyze the actual measurement value with the estimated value, calculate the estimation deviation value, and the calculation process is:

[0106] The difference between the estimated value and the actual measured value is calculated to obtain the estimated deviation. The estimated deviation is combined with the prior distribution and the likelihood function to obtain the posterior distribution, and the deviation estimate is updated. The calculation expression is:

[0107] ;

[0108] In the formula, represents the actual measured value, represents the estimation bias, Indicates the actual measured value given Then, estimate the deviation Update, Indicates the given estimated deviation In the case of The probability of represents the probability distribution of the estimation deviation, represents the normalization constant, ensuring that the total probability of the posterior distribution is 1;

[0109] Estimation bias It follows a normal distribution, and the prior distribution is:

[0110] ;

[0111] In the formula, represents the prior variance of the estimated bias, represents the covariance matrix of the estimates, represents the transpose operation, Represents the dimension of the data;

[0112] Substituting the prior distribution and likelihood function into the Bayesian formula, we get the posterior distribution, and the calculation expression is:

[0113] ;

[0114] In the formula, Indicates an estimated value;

[0115] By maximizing the posterior distribution, the maximum posterior estimate of the estimated deviation is obtained, and the calculation expression is:

[0116] ;

[0117] In the formula, represents the maximum a posteriori estimate, represents the maximum value function;

[0118] By taking the derivative of the posterior distribution and setting it to 0, we get the estimated deviation value. The calculation expression is:

[0119] ;

[0120] In the formula, represents the estimated deviation value, represents the precision matrix of the actual measurement data, A precision matrix representing the a priori estimates.

[0121] It should be noted that: this embodiment corrects the deviation of the optimized posture estimation result based on Bayesian theory, and calculates the estimated deviation value by combining the comparative analysis of the actual measured value and the estimated value. First, the initial deviation is obtained by difference calculation, and the posterior distribution is constructed by combining the prior distribution and the likelihood function. The prior distribution describes the initial belief of the estimated deviation, which is usually assumed to obey the normal distribution, and the variance is calculated based on the covariance matrix of the estimated value; the likelihood function represents the probability distribution of the actual measured value under a given estimated deviation. By substituting the prior distribution and the likelihood function into the Bayesian formula, the posterior distribution of the estimated deviation is obtained, and the maximum posterior estimate is obtained by maximizing the posterior distribution. By taking the derivative of the posterior distribution and setting it to zero, an accurate estimated deviation value is obtained, in which the measurement data is combined with the prior estimate by the definition of the precision matrix to ensure that the corrected deviation value is more in line with the actual situation, thereby improving the accuracy and robustness of the estimate.

[0122] In S4, based on the evaluation result, the error feedback information is used to correct the estimation result in the filtering process in real time, specifically including:

[0123] Obtain the estimated deviation value, calculate the correction coefficient based on the estimated deviation value and the particle filter algorithm, adjust the current estimation result according to the correction coefficient, and obtain the corrected estimation state;

[0124] The process of obtaining the correction coefficient is as follows:

[0125] Get the estimated deviation value and calculate the correction coefficient based on the estimated deviation value. The calculation expression is:

[0126] ;

[0127] In the formula, represents the correction factor, represents the estimated deviation value, represents the transpose operation, Indicates the first Particles, represents the total number of particles, represents the time step, represents the covariance matrix of the state estimate, Particle weight.

[0128] The correction process is: adjust the current estimation state by combining the correction coefficient and the estimation deviation value, and calculate the corrected estimation state. The calculation expression is:

[0129] ;

[0130] In the formula, represents the estimated status before correction, represents the revised estimated status, represents the correction factor, Represents the estimated deviation value.

[0131] It should be noted that: this technical solution uses evaluation results and error feedback information to make real-time corrections to the estimation results in the filtering process to improve the estimation accuracy. First, by comparing the current estimated value with the actual measured value, the estimated deviation value is obtained. Then, the particle filter algorithm is used to calculate the correction coefficient based on the deviation value. The correction coefficient combines the covariance matrix and weight information of the particle state to express the degree of influence of the current deviation on the correction of the estimated state. Finally, the estimated state at the current moment is adjusted in combination with the correction coefficient and the deviation value to obtain the corrected estimated state. Through such a dynamic correction mechanism, it is ensured that the difference between the estimated result and the actual measured value during the filtering process can be adjusted in real time, thereby improving the accuracy and robustness of the system.

[0132] In S5, based on the real-time correction results, the overall accuracy of the dredger rake tube spatial angle measurement system is evaluated, the causes of inaccurate recognition results are analyzed, and the spatial angle measurement is re-performed, including:

[0133] According to the real-time correction result, the estimated deviation value and correction coefficient in the process of measuring the spatial angle of the dredger rake pipe are used to construct a combined comprehensive feature vector as the input of the machine learning model. The machine learning model is trained in combination with historical data. According to the training result, the accuracy evaluation result of the whole system is output. The machine learning model is a deep learning model. The deep learning model includes an input layer: for receiving the combined comprehensive feature vector, a hidden layer: composed of a multi-layer fully connected neural network, for extracting nonlinear features, an output layer: outputting the accuracy evaluation result of the whole system, and using historical data to perform supervised learning training on the deep learning model. The accuracy label in the historical data is the accuracy score of the whole system.

[0134] Obtain the accuracy score of the dredger rake tube spatial angle measurement system, and determine whether the accuracy score is greater than or equal to a preset threshold. If so, it is classified as an accurate recognition result; if not, it is classified as an inaccurate recognition result.

[0135] By comparing the similarity between the inaccurate recognition result and the historical measurement data, if the similarity is lower than the preset threshold, it is determined that the current measurement value is inconsistent with the historical data characteristics; according to the cause of the inaccurate recognition, the measurement parameters and processes are dynamically adjusted, including: recalibrating the sensor and adjusting the parameters of the sampling frequency; re-measuring the rake tube space angle to obtain a new measurement value, and correcting the measurement value based on the correction coefficient, calculating a new corrected estimate, reconstructing the comprehensive feature vector, combining the corrected estimate and measurement value with the historical data, and updating the overall accuracy evaluation result of the system.

[0136] The working principle of the present invention is to optimize the estimation of the rake tube posture by collecting the motion data of the hull and the rake tube in real time and using a variety of data processing techniques, such as median filtering, feature extraction, fuzzy reasoning and particle filtering. The method gradually improves the measurement accuracy of the system through dynamic feedback and error correction. In particular, the accuracy of the system is evaluated through a machine learning model, and automatic adjustments are made based on the evaluation results. For inaccurate recognition results, the system can analyze and locate the cause of the error, re-measure the spatial angle, ensure the accuracy and stability of the measurement results, and adapt to complex working environments and dynamic changes.

[0137] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0138] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave) mode. The computer-readable storage medium can be any available medium that a computer can access or a server or data center data storage device containing one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0139] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0140] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0141] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A method for measuring the spatial angle of a dredger rake pipe, characterized in that: The following steps are involved: S1: Real-time collection of motion data of the hull and rake tube, including acceleration, angular velocity and attitude information; S2: Separate nonlinear effects and optimize attitude estimation based on hull and rake motion data; S3: Based on the optimized posture estimation, the difference between the estimation result and the actual measurement value is evaluated through real-time error feedback calculation; S4: Based on the evaluation results, the estimation results in the filtering process are corrected in real time using the error feedback information; S5: Based on the real-time correction results, the overall accuracy of the dredger rake tube spatial angle measurement system is evaluated, and for inaccurate identification results, the causes of inaccurate identification are analyzed, and the spatial angle measurement is re-performed.

2. The method for measuring the spatial angle of a dredger rake pipe according to claim 1, characterized in that: Get the motion data of the hull and rake tube, and use median filtering to remove high-frequency noise; The acceleration, angular velocity and attitude information of the hull and rake tube are extracted and a comprehensive feature vector is constructed.

3. The method for measuring the spatial angle of a dredger rake pipe according to claim 2, characterized in that: The construction process of the comprehensive feature vector is: Obtain the motion data of the hull and rake tube, and extract time domain features from the acceleration, angular velocity and attitude data, including mean, standard deviation, maximum, minimum, peak value and variance; Use fast Fourier transform to perform frequency domain analysis on acceleration and angular velocity data and extract spectrum features such as frequency distribution, spectrum density and frequency components; Combined with time-frequency analysis technology, time-frequency domain features are extracted to capture the time-varying characteristics in the data; The extracted time domain, frequency domain and time-frequency domain features are combined to form a preliminary feature vector; The preliminary feature vector is trained using the gradient boosted tree algorithm to automatically select the features most relevant to the target prediction; During the gradient boosting tree training process, new feature combinations and cross-features are automatically generated to improve feature expression capabilities; the influence of each feature on the final goal is evaluated, irrelevant features are removed, the most qualified features are retained, and the optimized features are combined to generate the final comprehensive feature vector.

4. The method for measuring the spatial angle of a dredger rake pipe according to claim 1, characterized in that: The separation of nonlinear effects and optimization of attitude estimation specifically include: Obtain comprehensive feature vectors, and use singular spectrum decomposition to convert the comprehensive feature vectors of the hull and the rake tube into input features for fuzzy inference system processing; formulate fuzzy rules to describe the relationship between the hull sway and the rake tube posture; use adaptive methods to adjust the membership function and rule weight of fuzzy rules based on real-time data to dynamically optimize the performance of the inference system; The collected motion data is input into the fuzzy inference system, and the nonlinear effects between the hull sway and the rake tube attitude are separated based on the output of the fuzzy inference system; Through the membership value of fuzzy reasoning, the influence of hull sway and the estimated value of rake tube posture are calculated to achieve dynamic separation of nonlinear influences; According to the real-time feedback data, the fuzzy rules and membership functions in the fuzzy reasoning system are dynamically updated. Through the output of the fuzzy reasoning system, combined with the real-time data and the optimized model, accurate posture estimation results of the hull and the rake tube are provided.

5. The method for measuring the spatial angle of a dredger rake pipe according to claim 1, characterized in that: The differences between the evaluation results and the actual measured values ​​include: According to the optimized posture estimation results, the actual measured value is compared with the estimated value, and the estimated deviation value is calculated. The calculation process is: The difference between the estimated value and the actual measured value is calculated to obtain the estimated deviation. The estimated deviation is combined with the prior distribution and the likelihood function to obtain the posterior distribution, and the deviation estimate is updated. The calculation expression is: ； In the formula, represents the actual measured value, represents the estimation bias, Indicates the actual measured value given Then, estimate the deviation Update, Indicates the given estimated deviation In the case of The probability of represents the probability distribution of the estimation deviation, represents the normalization constant, ensuring that the total probability of the posterior distribution is 1; Estimation bias It follows a normal distribution, and the prior distribution is: ； In the formula, represents the prior variance of the estimated bias, represents the covariance matrix of the estimates, represents the transpose operation, Represents the dimension of the data; Substituting the prior distribution and likelihood function into the Bayesian formula, we get the posterior distribution, and the calculation expression is: ； In the formula, Indicates an estimated value; By maximizing the posterior distribution, the maximum posterior estimate of the estimated deviation is obtained, and the calculation expression is: ； In the formula, represents the maximum a posteriori estimate, represents the maximum value function; By taking the derivative of the posterior distribution and setting it to 0, we get the estimated deviation value. The calculation expression is: ; Estimated deviation value In the formula, represents the estimated deviation value, represents the precision matrix of the actual measurement data, A precision matrix representing the a priori estimates.

6. The method for measuring the spatial angle of a dredger rake pipe according to claim 1, characterized in that: The method of using error feedback information to correct the estimation result in the filtering process in real time specifically includes: Obtain the estimated deviation value, calculate the correction coefficient based on the estimated deviation value in combination with the particle filter algorithm, adjust the estimation result at the current moment according to the correction coefficient, and obtain the corrected estimation state.

7. The method for measuring the spatial angle of a dredger rake pipe according to claim 6, characterized in that: The process of obtaining the correction coefficient is as follows: Get the estimated deviation value and calculate the correction coefficient based on the estimated deviation value. The calculation expression is: ； In the formula, represents the correction factor, represents the estimated deviation value, represents the transpose operation, Indicates the first Particles, represents the total number of particles, represents the time step, represents the covariance matrix of the state estimate, Particle weight.

8. The method for measuring the spatial angle of a dredger rake pipe according to claim 1, characterized in that: The overall accuracy of the dredger rake tube spatial angle measurement system is evaluated, specifically including: According to the real-time correction result, the estimated deviation value and correction coefficient in the process of measuring the spatial angle of the dredger rake pipe are used to construct a combined comprehensive feature vector as the input of the machine learning model. The machine learning model is trained in combination with historical data. According to the training result, the accuracy evaluation result of the whole system is output. The machine learning model is a deep learning model. The deep learning model includes an input layer: for receiving the combined comprehensive feature vector, a hidden layer: composed of a multi-layer fully connected neural network, for extracting nonlinear features, an output layer: outputting the accuracy evaluation result of the whole system, and using historical data to perform supervised learning training on the deep learning model. The accuracy label in the historical data is the accuracy score of the whole system.

9. The method for measuring the spatial angle of a dredger rake pipe according to claim 8, characterized in that: According to the overall accuracy evaluation results of the system, the quality recognition results of the system are divided into accurate recognition results and inaccurate recognition results, including: Obtain the accuracy score of the dredger rake tube spatial angle measurement system, and determine whether the accuracy score is greater than or equal to a preset threshold. If so, it is classified as an accurate recognition result; if not, it is classified as an inaccurate recognition result.

10. The method for measuring the spatial angle of a dredger rake pipe according to claim 1, characterized in that: The inaccurate recognition result, analyzing the cause of the inaccurate recognition, and re-measuring the spatial angle specifically includes: By comparing the similarity between the inaccurate recognition result and the historical measurement data, if the similarity is lower than the preset threshold, it is determined that the current measurement value is inconsistent with the historical data characteristics; according to the cause of the inaccurate recognition, the measurement parameters and processes are dynamically adjusted, including: recalibrating the sensor and adjusting the parameters of the sampling frequency; re-measuring the rake tube space angle to obtain a new measurement value, and correcting the measurement value based on the correction coefficient, calculating a new corrected estimate, reconstructing the comprehensive feature vector, combining the corrected estimate and measurement value with the historical data, and updating the overall accuracy evaluation result of the system.

Images