A real-time fault diagnosis and location method for thermal measuring points of a bucket wheel machine

Abstract

The present application relates to the technical field of fault diagnosis and positioning of bucket wheel machines, and specifically discloses a method for real-time fault diagnosis and positioning of thermal measurement points of a bucket wheel machine, to solve the problems of poor data quality, low fault diagnosis accuracy and inaccurate positioning of traditional thermal measurement points. The present application collects data of thermal measurement points of a bucket wheel machine and environmental interference factor data; removes outliers in the thermal data using the Rahiada criterion, and performs noise reduction processing based on MREMD-VMD dual mode; diagnoses the fault type and confidence of the measurement points through a double-input CNN-LSTM model, and triggers a fault positioning signal for the thermal measurement points; establishes a three-dimensional coordinate library for the bucket wheel machine, and obtains real-time coordinates of the fault thermal measurement points by combining UWB positioning and pose compensation; and sends the fault type and positioning coordinates of the thermal measurement points to an operation and maintenance terminal for display. The present application improves the signal-to-noise ratio of thermal data and the accuracy of fault diagnosis, realizes high-precision positioning under dynamic working conditions, reduces equipment downtime, and ensures operational reliability.

Description

Technical Field

[0001] This invention relates to the field of bucket wheel excavator fault diagnosis and location technology, specifically a method for fault diagnosis and real-time location of thermal measurement points in bucket wheel excavators. Background Technology

[0002] Bucket wheel excavators, also known as bucket wheel stacker-reclaimers, are core equipment in bulk material conveying systems, used in material yard operations in industries such as thermal power, coal, and metallurgy. They are responsible for material storage and retrieval, and their operational reliability directly impacts the continuity of the production process. Thermal monitoring points, acting as the "sensory nerves" for monitoring the operating status of key components of the bucket wheel excavator, collect thermal data such as temperature, pressure, and flow rate, which are crucial for determining whether components are malfunctioning. Therefore, building an integrated operation and maintenance system around thermal monitoring points for accurate fault diagnosis and real-time location tracking has become a key technological direction for ensuring the reliable operation of bucket wheel excavators. However, current fault diagnosis and location technologies for bucket wheel excavator thermal monitoring points still face many technical bottlenecks, making it difficult to meet the modern industrial demands for "high precision, high real-time performance, and high reliability" in equipment operation and maintenance. Specific problems include:

[0003] The highly interfering operating environment of bucket wheel excavators leads to severe noise pollution and baseline drift in the collected thermal data. Current technologies often employ single filtering algorithms (such as moving average filtering or Kalman filtering) or traditional empirical mode decomposition (EMD) for data preprocessing. The former can only filter out high-frequency random noise and cannot eliminate nonlinear interference from dust and vibration; the latter, while capable of decomposing nonlinear and non-stationary signals, is easily affected by baseline drift, and the resulting intrinsic mode functions (IMFs) still contain a large amount of invalid noise components, making it difficult to extract effective information reflecting fault characteristics. Low-quality thermal data directly leads to distortion in the input of subsequent fault diagnosis models, creating potential for misdiagnosis and missed diagnosis.

[0004] Existing methods for diagnosing faults at thermal measurement points of bucket wheel excavators largely rely on traditional machine learning models, such as support vector machines, random forests, or single deep learning models, such as using only CNNs to extract local features or only LSTMs to capture time-series trends. These methods have two major limitations: First, they fail to fully integrate the dual-dimensional information of thermal data and environmental interference factors. Interference factors such as dust concentration and vibration frequency are strongly coupled with thermal data. Ignoring this information will cause the model to be unable to distinguish between real fault signals and interference pseudo-signals, resulting in a significant decrease in diagnostic robustness under complex operating conditions. Second, they lack feature extraction capabilities. Traditional models struggle to uncover the nonlinear fault features hidden in thermal data. Single CNN or LSTM models cannot simultaneously take into account the collaborative extraction of local fault features and time-series change trends, leading to low diagnostic accuracy and poor reliability of the output fault confidence, thus failing to provide a clear basis for operation and maintenance decisions.

[0005] Accurately locating thermal measurement points is crucial for timely maintenance and troubleshooting during the operation of bucket wheel excavators. However, traditional positioning methods, such as those based on manual marking or simple distance measurement, suffer from low accuracy. Manual marking is susceptible to environmental and human factors, leading to inaccurate marking positions; simple distance measurement methods also struggle to achieve high-precision positioning due to the accumulation of measurement errors. On large equipment like bucket wheel excavators, even minor positioning errors can prevent maintenance personnel from quickly and accurately locating the faulty measurement point, thus prolonging maintenance time and increasing costs. For example, if a critical thermal measurement point on the bucket wheel excavator malfunctions, inaccurate positioning may require maintenance personnel to spend a significant amount of time searching for the fault location on the equipment, impacting maintenance efficiency and potentially causing secondary damage.

[0006] To address the aforementioned shortcomings, a technical solution is provided. Summary of the Invention

[0007] The purpose of this invention is to provide a method for diagnosing faults and locating real-time positions of thermal measuring points in bucket wheel excavators, so as to solve the problems mentioned in the background.

[0008] The objective of this invention can be achieved through the following technical solution: a method for fault diagnosis and real-time location positioning of thermal measuring points in a bucket wheel excavator, comprising the following steps:

[0009] S1. Thermal Measurement Point Data Acquisition: Thermal data and environmental interference factor data of each thermal measurement point of the bucket wheel excavator are collected through a sensor array installed on the excavator. The thermal data of each measurement point includes: the temperature T measured by the temperature sensor at each measurement point. t i The pressure P measured by the pressure sensors at each thermal measurement point t i Flow rate measured by flow sensors at each thermal measurement point Where t is the timestamp, i.e., the time when the thermal data was collected; where i is the number of the thermal measurement point, i = 1, 2, ..., n, and n is the total number of thermal measurement point numbers; environmental interference factor data includes: dust concentration C measured by the dust concentration sensor. t The vibration frequency f measured by the vibration sensor t ;

[0010] The collected thermal data is converted into digital signals using an analog-to-digital converter, and step S2 is executed.

[0011] S2. Preprocessing of thermal measurement point data: Based on the Laida criterion, outlier detection is performed on the thermal data of each thermal measurement point. Outliers are identified and removed to obtain the thermal data of each thermal measurement point after outlier processing. At the same time, based on the MREMD-VMD dual-modal approach, noise reduction processing is performed on the thermal data of each thermal measurement point after outlier processing to obtain the noise-reduced thermal data of each thermal measurement point.

[0012] S3. Fault Diagnosis: Input the denoised thermal data and environmental interference factor data of each thermal measuring point into the fault diagnosis model of the bucket wheel excavator thermal measuring point to obtain the fault type and confidence level of each thermal measuring point; wherein, the fault diagnosis model of the bucket wheel excavator thermal measuring point is constructed by a dual-input CNN-LSTM model; if the confidence level corresponding to the fault type of a certain thermal measuring point is greater than or equal to the preset confidence level threshold, it is determined to be a valid fault, triggering the fault location signal of the thermal measuring point, and executing step S4;

[0013] S4. Measurement point positioning: Establish a coordinate library of thermal measurement points for bucket wheel excavators. When a fault positioning signal for a thermal measurement point is received, analyze the real-time coordinates of the thermal measurement point.

[0014] S5. Display: Based on the fault type and the location coordinates of the thermal measurement points of the bucket wheel excavator, send the information to the operation and maintenance terminal of the bucket wheel excavator management personnel to perform the corresponding display operation.

[0015] Furthermore, the process of identifying and removing outliers includes the following steps:

[0016] Thermal data at each thermal measurement point The thermal data mentioned include temperature T t i Pressure P t i and traffic The mean value of thermal data at each thermal measurement point was obtained by mean calculation. The standard deviation of thermal data at each thermal measurement point is obtained by calculating the standard deviation. Based on the Laida criterion, the judgment interval for normal thermal data is determined as follows: If the thermal data of a certain thermal measuring point is not within the normal thermal data judgment range, then the thermal data of that thermal measuring point is judged as an outlier and is removed accordingly. Based on this, the thermal data of each thermal measuring point after outlier removal is obtained.

[0017] Furthermore, the process of acquiring the thermal data of each thermal measurement point after noise reduction is as follows:

[0018] S201, EMD Initial Decomposition: For the thermal data at each thermal measurement point after outlier processing, perform EMD decomposition to obtain several intrinsic mode function (IMF) components and one residual term, satisfying: Where j is the index of the intrinsic mode function (IMF) component, j = 1, 2, ..., o, o is the total number of IMF component indices, and r(i,t) is the residual term;

[0019] S202, Sliding median filter fitting trend term: for each component (N is the sequence length), a sliding median filter is used to fit the trend. Let the sliding window length be L, then the median of the h-th window is the data within the window. The value at the middle position after sorting in ascending order is denoted as med. h After traversing all windows, the trend sequence is obtained. j (i,t)={med1,med2,...,med N-L+1};

[0020] S203. Eliminating Baseline Drift: Subtract the trend term from the IMF components to obtain the detrended IMF, the formula is: IMF′ j (i,t)=IMF j (i,t)-trend j (i,t), where IMF′ j (i,t) represents the detrended IMF;

[0021] S204, Phase Space Reconstruction: For each detrended IMF component Phase space reconstruction is performed using the delayed coordinate method. Let the time delay be τ and the phase space dimension be m. Then the reconstructed phase space vector is: Where f represents the starting point index of the phase space vector, f = 1, 2, ..., N-(m-1)τ, ensuring that the last element does not exceed the original sequence length N;

[0022] S205. Arrangement Patterns and Entropy Calculation: Arrange the reconstructed phase space vectors in ascending order to generate arrangement patterns; for an m-dimensional phase space, there are m! arrangement patterns. Calculate the probability P of each pattern. s (s=1,2,...,m!), through the formula The permutation entropy H was calculated. p ;

[0023] S206. Filtering valid IMFs: Setting permutation entropy threshold If a detrended IMF component IMF′ j The permutation entropy H of (i,t) p Less than the set permutation entropy threshold Then it is determined to be a noisy IMF; if a certain detrended IMF component IMF′ j The permutation entropy H of (i,t) p Greater than or equal to the set permutation entropy threshold This is then determined to be a valid IMF; the valid IMFs obtained through this screening are denoted as IMF′. eff (i,t);

[0024] S207. Variational Problem Construction of VMD: Based on VMD, a constrained variational problem is constructed, decomposing the signal into K band-limited IMFs. For the noise IMFu(i,t), the objective function is to minimize the bandwidth-weighted sum of each IMF, with the constraint that the sum of each IMF equals the original signal. Its variational form is:

[0025]

[0026] Constraints:

[0027] Where δ(i,t) represents the Dirac function. The kernel is represented by , and * denotes the convolution operation. For time derivative, Represented as a frequency shift factor, u k The center frequency of (i,t) shifts to ω k ω k Let the center frequency of the k-th IMF be denoted as . Represented as L 2 Norm;

[0028] S208. Solving the variational problem: Introducing the Lagrange multiplier λ(i,t) and the quadratic penalty factor ε, we construct the augmented Lagrange function to solve this variational problem. The formula is as follows:

[0029] The optimal decomposition is solved iteratively using the alternating direction multiplier method (ADMM).

[0030] S209. Effective Information Extraction and Signal Reconstruction: The various modes u obtained by VMD decomposition of the noisy IMF. k (i,t) is subjected to a Fast Fourier Transform to obtain the spectrum. Effective modal components matching the characteristic frequency range of thermal faults were selected based on this range. Finally, these effective modal components are reconstructed with all the effective IMFs selected in step S206 to obtain the denoised thermal data of each thermal measurement point. The formula is:

[0031] Furthermore, the fault diagnosis model for the thermal measurement points of the bucket wheel excavator is constructed using a dual-input CNN-LSTM model, including:

[0032] Acquire thermal data, environmental interference factor data and their corresponding fault type labels for each thermal measurement point after MREMD-VMD dual-modal noise reduction.

[0033] The thermal data, environmental interference factor data and their corresponding fault type labels of several historical thermal measurement points after MREMD-VMD dual-modal noise reduction are divided into training set, test set and validation set, with a ratio of 7:2:1 between training set, test set and validation set.

[0034] The dual-input CNN-LSTM model is selected as the base model. The dual-input CNN-LSTM model includes a first input layer and a second input layer. Each input layer is followed by a CNN feature extraction layer and an LSTM temporal analysis layer. The features are then fused through a fully connected layer, and finally the output layer outputs the fault type and confidence level.

[0035] The base model is trained on the training set, the cross-entropy loss function is used, and the Adam optimizer is used on the validation set to adjust the learning rate, batch size and other hyperparameters to obtain the pre-trained model.

[0036] By validating the pre-trained model on the test set, the final result is a fault diagnosis model for the bucket wheel turbine thermal measuring points, with the input being the denoised thermal data and environmental interference factor data of each thermal measuring point, and the output being the fault type and corresponding confidence level of each thermal measuring point.

[0037] Furthermore, the establishment of the bucket wheel excavator thermal measurement point coordinate library includes: constructing a three-dimensional model of the bucket wheel excavator, taking the intersection of the center line of the bucket wheel excavator's travel track and the rotation center of the slewing platform as the coordinate origin, defining a three-dimensional rectangular coordinate system, wherein the X-axis is parallel to the travel track, and the direction of the bucket wheel excavator's forward material stacking is the positive direction; the Y-axis is perpendicular to the travel track, and the direction away from the material stacking is the positive direction; and the Z-axis is perpendicular to the ground, and the vertical upward direction is the positive direction.

[0038] A high-precision laser rangefinder was used to measure the three-dimensional coordinates of each thermal measurement point of the bucket wheel machine and to calibrate each point. A set number of UWB positioning base stations were deployed in the working area of ​​the bucket wheel machine, and a UWB positioning tag was configured for each thermal measurement point. The UWB positioning tag was bound to the thermal measurement point number.

[0039] A mapping table is established in the server database, consisting of thermal measurement point number, UWB positioning tag ID, and initial three-dimensional coordinates, forming a dedicated coordinate library for the thermal measurement points of the bucket wheel excavator.

[0040] Furthermore, the analysis process of the real-time coordinates of the thermal measurement point is as follows: After receiving the fault location signal of the thermal measurement point, the number of the fault thermal measurement point is obtained and the UWB positioning tag configured for the fault thermal measurement point is matched. A wireless signal is sent according to the UWB positioning tag configured for the fault thermal measurement point. The UWB base station receives the wireless signal sent by the UWB positioning tag configured for the fault thermal measurement point. The distances d1, d2, d3, and d4 between the UWB positioning tag configured for the fault thermal measurement point and each base station are calculated using the TOF algorithm.

[0041] Obtain the coordinates (X1,Y1,Z1), (X2,Y2,Z2), (X3,Y3,Z3), and (X4,Y4,Z4) of four UWB positioning base stations, and combine them with the distances d1, d2, d3, and d4, using the polygonal positioning principle algorithm formula. The real-time three-dimensional coordinates (X, Y, F, Z) of the UWB positioning tag configured at the fault thermal measurement point were calculated. real ,Y real Z real This refers to the real-time three-dimensional coordinates of the fault thermal measurement point;

[0042] The real-time three-dimensional coordinates (X) of the fault thermal measurement point real ,Y real Z real The coordinates of the faulty thermal measuring point are compared with the initial coordinates (X0, Y0, Z0) in the thermal measuring point coordinate library, and then the result is determined according to the formula. The coordinate deviations ΔX, ΔY, and ΔZ are calculated. A coordinate deviation threshold is set. If ΔX, ΔY, and ΔZ are all less than the set deviation threshold, the real-time three-dimensional coordinates (X, Y, and Z) are directly output. real ,Y real Z real ) represents the final positioning coordinates of the fault thermal measurement point. If any deviation among ΔX, ΔY, and ΔZ is greater than the set deviation threshold, the real-time position and attitude parameters of the bucket wheel excavator are obtained. The position and attitude transformation is compensated by the coordinate system transformation matrix, and the total transformation matrix T is obtained through analysis.

[0043] Represent the real-time 3D coordinates as homogeneous coordinate vectors. Multiplying by the total transformation matrix T yields the homogeneous vector of the corrected coordinates, as shown in the formula. Take the first three dimensions, that is, the final location coordinates of the fault thermal measurement point are the corrected measurement point coordinates (X). corrected ,Y corrected Z corrected ).

[0044] The beneficial effects of this invention are:

[0045] This invention uses the Laida criterion to detect and remove outliers from thermal data at various thermal measurement points. This efficiently eliminates extreme values ​​caused by instantaneous sensor malfunctions, such as jumps caused by dust obstructing the sensor or instantaneous peak values ​​caused by vibration and impact, ensuring the validity and accuracy of the data foundation and providing reliable data support for subsequent fault diagnosis and location. Furthermore, the MREMD-VMD dual-modal approach is used to denoise the outlier-processed thermal data. First, MREMD is used to eliminate baseline drift. Then, a sliding median filter is added to the EMD decomposition to fit the trend term, effectively eliminating noise caused by slow changes in ambient temperature and long-term equipment aging. The signal shift caused by factors such as bearing wear is analyzed to identify local thermal changes caused by faults, such as sudden temperature changes due to bearing wear. Then, VMD (Vibration Mode Decomposition) is used to extract effective information from the noise. The selected noise IMF components are band-limited and combined with the characteristic frequencies of thermal faults to screen effective modes, avoiding the problem of traditional single-mode filtering that mistakenly deletes effective signals. After dual-mode processing, the signal-to-noise ratio of the thermal data is effectively improved, avoiding problems such as outlier interference, baseline drift, and noise contamination in the thermal data collected by the bucket wheel excavator under harsh conditions of high dust and strong vibration. This provides high-purity input for subsequent fault diagnosis, avoiding misdiagnosis or missed diagnosis caused by data distortion from the source.

[0046] This invention employs a dual-input CNN-LSTM model to construct a fault diagnosis model for thermal measurement points of bucket wheel excavators. It fully considers the characteristics of thermal data and environmental interference factor data, and uses several historical thermal data points, environmental interference factor data, and their corresponding fault type labels after MREMD-VMD dual-modal denoising to train and validate the model. This ensures the quality and reliability of the model training data and provides the model with generalization ability and diagnostic accuracy.

[0047] This invention constructs a three-dimensional model of a bucket wheel excavator, defines a three-dimensional rectangular coordinate system with a specific point as the origin, and uses a high-precision laser rangefinder to measure the three-dimensional coordinates of each thermal measurement point and perform point-by-point calibration. A mapping table of thermal measurement point number, UWB tag ID, and initial three-dimensional coordinates is established in the server database, forming a dedicated coordinate library that provides a precise reference benchmark for the positioning of thermal measurement points. When a thermal measurement point positioning signal is triggered, the wireless signal sent by the UWB positioning tag bound to the faulty measurement point is received through a UWB base station. The distance between the tag and each base station is calculated using the Time-of-Flight (TOF) algorithm, and combined with the coordinates of four UWB positioning base stations, the real-time three-dimensional coordinates of the measurement point are calculated using a polygonal positioning principle algorithm, achieving real-time positioning of the thermal measurement point. Furthermore, considering that the walking, turning, and pitching movements of the bucket wheel excavator during operation will cause changes in the equipment's posture, the real-time coordinates calculated by UWB are corrected. By obtaining the real-time posture parameters of the bucket wheel excavator and analyzing the total transformation matrix, the real-time coordinates are corrected, solving the impact of equipment posture changes on the spatial position of the measurement point and ensuring the real-time performance and accuracy of positioning during dynamic operation.

[0048] This invention sends the fault type and the location coordinates of the measuring points of the bucket wheel excavator to the operation and maintenance terminal of the bucket wheel excavator management personnel to perform corresponding display operations. This enables the management personnel to understand the fault status and location of the equipment in a timely manner, facilitates rapid maintenance measures, reduces equipment downtime, and improves the operating efficiency and reliability of the equipment. Attached Figure Description

[0049] The invention will now be further described with reference to the accompanying drawings.

[0050] Figure 1 This is a flowchart of the method steps of the present invention.

[0051] Figure 2 This is a logical diagram illustrating the noise reduction process of thermal data from various thermal measurement points based on the MREMD-VMD dual-modal approach for outlier processing according to the present invention.

[0052] Figure 3 This is a schematic diagram illustrating the analysis logic of the real-time coordinates of the thermal measurement points in this invention. Detailed Implementation

[0053] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0054] Please see Figure 1 As shown, this invention provides a method for fault diagnosis and real-time location positioning of thermal measuring points in a bucket wheel excavator, comprising the following steps:

[0055] S1. Thermal Measurement Point Data Acquisition: Thermal data and environmental interference factor data of each thermal measurement point of the bucket wheel excavator are collected through a sensor array installed on the excavator. The thermal data of each thermal measurement point includes:

[0056] Temperature T measured by temperature sensors at each thermal measurement point t i ;

[0057] The pressure P measured by the pressure sensor at each thermal measuring point t i ;

[0058] Flow rate measured by flow sensors at each thermal measurement point Where t is the timestamp, i.e. the time when the thermal data was collected; where i is the number of the thermal measurement point, i = 1, 2, ..., n, and n is the total number of thermal measurement point numbers;

[0059] Environmental interference factor data includes: dust concentration C measured by a dust concentration sensor. t The vibration frequency f measured by the vibration sensor t ;

[0060] The collected thermal data is converted into digital signals using an analog-to-digital converter, and step S2 is then executed.

[0061] It should be noted that the thermal data from the thermal measurement points directly reflect the thermal state of key components of the bucket wheel excavator, such as bearings, hydraulic systems, and conveying systems. The basis for selecting dust concentration and vibration frequency as environmental interference factors is that the high dust concentration in the bucket wheel excavator's operating environment easily adheres to the sensors, affecting their accuracy, and the strong vibration couples to the thermal signals, producing pseudo-fluctuations. These two factors are the most critical environmental factors interfering with the thermal measurement points.

[0062] S2. Preprocessing of thermal measurement point data: Based on the Laida criterion, outlier detection is performed on the thermal data of each thermal measurement point. Outliers are identified and removed to obtain the thermal data of each thermal measurement point after outlier processing. At the same time, based on the MREMD-VMD dual-modal approach, noise reduction processing is performed on the thermal data of each thermal measurement point after outlier processing to obtain the noise-reduced thermal data of each thermal measurement point.

[0063] Specifically, the process of identifying and removing outliers includes the following steps:

[0064] Thermal data at each thermal measurement point The thermal data mentioned include temperature T t i Pressure P t i and traffic The mean value of thermal data at each thermal measurement point was obtained by mean calculation. The standard deviation of thermal data at each thermal measurement point is obtained by calculating the standard deviation. Based on the Laida criterion, the judgment interval for normal thermal data is determined as follows: If the thermal data of a certain thermal measuring point is not within the normal thermal data judgment range, then the thermal data of that thermal measuring point is judged as an outlier and is removed accordingly. Based on this, the thermal data of each thermal measuring point after outlier removal is obtained.

[0065] It should be noted that anomaly removal is performed to eliminate extreme values ​​caused by momentary sensor malfunctions, ensuring data validity.

[0066] Specifically, please refer to Figure 2 As shown, the process of acquiring the thermal data of each thermal measurement point after noise reduction is as follows:

[0067] S201, EMD Initial Decomposition: For the thermal data at each thermal measurement point after outlier processing, perform EMD decomposition to obtain several intrinsic mode function (IMF) components and one residual term, satisfying: Where j is the index of the intrinsic mode function (IMF) component, j = 1, 2, ..., o, o is the total number of IMF component indices, and r(i,t) is the residual term;

[0068] S202, Sliding median filter fitting trend term: for each component (N is the sequence length), a sliding median filter is used to fit the trend. Let the sliding window length be L, then the median of the h-th window is the data within the window. The value at the middle position after sorting in ascending order is denoted as med. h After traversing all windows, the trend sequence is obtained. j (i,t)={med1,med2,...,med N-L+1};

[0069] S203. Eliminating Baseline Drift: Subtract the trend term from the IMF components to obtain the detrended IMF, the formula is: IMF′ j (i,t)=IMF j (i,t)-trend j (i,t), where IMF′ j (i,t) represents the detrended IMF;

[0070] It should be noted that Empirical Mode Decomposition (EMD) can decompose nonlinear and non-stationary thermal data into several intrinsic mode functions (IMFs). However, direct decomposition is susceptible to baseline drift interference, such as the overall trend shift of the signal caused by environmental full-scale. MREMD improves the trend by combining EMD decomposition and median regression fitting, effectively eliminating baseline drift caused by slowly changing factors such as ambient temperature, and focusing on the thermal changes of the measurement itself, such as temperature abrupt changes caused by faults.

[0071] S204, Phase Space Reconstruction: For each detrended IMF component Phase space reconstruction is performed using the delayed coordinate method. Let the time delay be τ and the phase space dimension be m. Then the reconstructed phase space vector is: Where f represents the starting point index of the phase space vector, f = 1, 2, ..., N-(m-1)τ, ensuring that the last element does not exceed the original sequence length N;

[0072] It should be noted that, by using the delayed coordinate method, the one-dimensional thermal data sequence is transformed into a set of vectors in the higher phase space. The trajectories of these vectors can fully reflect the dynamic characteristics of the thermal system, providing a basis for subsequent permutation entropy calculation.

[0073] S205. Arrangement Patterns and Entropy Calculation: Arrange the reconstructed phase space vectors in ascending order to generate arrangement patterns; for an m-dimensional phase space, there are m! arrangement patterns. Calculate the probability P of each pattern. s (s=1,2,...,m!), through the formula The permutation entropy H was calculated. p ;

[0074] S206. Filtering valid IMFs: Setting permutation entropy threshold If a detrended IMF component IMF′ j The permutation entropy H of (i,t) p Less than the set permutation entropy threshold Then it is determined to be a noisy IMF; if a certain detrended IMF component IMF′ j The permutation entropy H of (i,t) p Greater than or equal to the set permutation entropy threshold This is then determined to be a valid IMF; the valid IMFs obtained through this screening are denoted as IMF′. eff (i,t);

[0075] It should be noted that permutation entropy quantifies the complexity and randomness of time series. The smaller the entropy value, the higher the noise ratio, and the larger the entropy value, the higher the effective signal ratio.

[0076] S207. Variational Problem Construction of VMD: Based on VMD, a constrained variational problem is constructed, decomposing the signal into K band-limited IMFs. For the noise IMFu(i,t), the objective function is to minimize the bandwidth-weighted sum of each IMF, with the constraint that the sum of each IMF equals the original signal. Its variational form is:

[0077]

[0078] Constraints:

[0079] Where δ(i,t) represents the Dirac function. The kernel is represented by , and * denotes the convolution operation. For time derivative, Represented as a frequency shift factor, u k The center frequency of (i,t) shifts to ω k ω k Let the center frequency of the k-th IMF be denoted as . Represented as L 2 Norm;

[0080] S208. Solving the variational problem: Introducing the Lagrange multiplier λ(i,t) and the quadratic penalty factor ε, we construct an augmented Lagrange function to solve the variational problem. The formula is as follows:

[0081] The optimal decomposition is solved iteratively using the alternating direction multiplier method (ADMM).

[0082] It should be noted that the optimal decomposition is obtained by iteratively updating the modal components u using the alternating direction multiplier method (ADMM). k Center frequency ω k By using the Lagrange multiplier λ, the augmented Lagrange function is gradually minimized until a convergence condition is met, where the convergence condition is that the relative error is less than a preset tolerance, i.e. in This is represented by the frequency domain representation of the k-th modal component at the (n+1)-th iteration, derived from the Fourier transform of the time-domain signal u. k (i,t) is derived from this. Let ||k|| represent the frequency domain representation of the k-th modal component in the (n+1)-th iteration, ||·||² represent the Euclidean norm used to quantify the energy scale of the frequency domain signal, and tolerance represent the noise tolerance, indicating that the relative change of the modal component is small enough to be acceptable. For example, tolerance = 10. -6 .

[0083] S209. Effective Information Extraction and Signal Reconstruction: The various modes u obtained by VMD decomposition of the noise IMF. k (i,t) is subjected to a Fast Fourier Transform to obtain the spectrum. Effective modal components matching the characteristic frequency range of thermal faults were selected based on this range.

[0084] It should be noted that the characteristic frequency range of thermal failures is determined based on the thermal properties of the bucket wheel excavator's thermal system and known failure modes. For example, in the diagnosis of bucket wheel excavator bearing failures, different failure types, such as inner ring failures, outer ring failures, and rolling element failures, will correspond to specific failure characteristic frequencies. These frequencies can be determined through theoretical calculations or experience.

[0085] It should be further explained that the specific steps for screening out the effective modal components that match the characteristic frequency range of thermal faults are as follows: the spectrum of each mode is compared and analyzed with the characteristic frequency range of thermal faults. If the spectrum of a certain mode has a significant amplitude peak in the characteristic frequency range of thermal faults, it is determined that the mode contains information related to thermal faults and is marked as an effective modal component; otherwise, if the spectrum of a certain mode does not contain any components that match the characteristic frequency of thermal faults, it is determined that the mode is noise or other irrelevant information and is discarded; thus, the effective modal components are obtained statistically.

[0086] Finally, these effective modal components are reconstructed with all the effective IMFs selected in step S206 to obtain the denoised thermal data of each thermal measurement point. The formula is:

[0087] It should be noted that Variational Mode Decomposition (VMD) is an adaptive, non-recursive mode decomposition method that can extract masked effective information from noisy IMFs. After MREMD-VMD dual-modal denoising, the signal-to-noise ratio of the thermal data is improved, and a high signal-to-noise ratio input is provided for the subsequent CNN-LSTM fault diagnosis model, ensuring diagnostic accuracy.

[0088] S3. Fault Diagnosis: The denoised thermal data and environmental interference factor data of each thermal measurement point are input into the fault diagnosis model of the bucket wheel excavator thermal measurement point to obtain the fault type and confidence level of each thermal measurement point; the fault diagnosis model of the bucket wheel excavator thermal measurement point is constructed by a dual-input CNN-LSTM model, including:

[0089] Acquire thermal data, environmental interference factor data and their corresponding fault type labels for each thermal measurement point after MREMD-VMD dual-modal noise reduction.

[0090] It should be noted that the thermal data of various thermal measurement points after MREMD-VMD dual-modal noise reduction includes the thermal data of various thermal measurement points after MREMD-VMD dual-modal noise reduction corresponding to various types of fault type labels.

[0091] The thermal data, environmental interference factor data and their corresponding fault type labels of several historical thermal measurement points after MREMD-VMD dual-modal noise reduction are divided into training set, test set and validation set, with a ratio of 7:2:1 between training set, test set and validation set.

[0092] The dual-input CNN-LSTM model is selected as the base model. The dual-input CNN-LSTM model includes a first input layer and a second input layer. Each input layer is followed by a CNN feature extraction layer and an LSTM temporal analysis layer. The features are then fused through a fully connected layer, and finally the output layer outputs the fault type and confidence level.

[0093] It should be noted that the CNN feature extraction layer is used to extract local features from thermal data and interference factor data. It includes two convolutional layers and one pooling layer. The number of filters in the first and second convolutional layers are 32 and 64, respectively, and the kernel sizes of the first and second convolutional layers are 3×1 and 3×1, respectively. The pooling layer uses max pooling with a kernel size of 2×1. The LSTM temporal analysis layer is used to capture the changing trend of thermal data over time. It includes two hidden layers. The number of neurons in the first and second hidden layers are 128 and 64, respectively. The fully connected layer is used to fuse the output features of CNN and LSTM.

[0094] The base model is trained on the training set, the cross-entropy loss function is used, and the Adam optimizer is used on the validation set to adjust the learning rate, batch size and other hyperparameters to obtain the pre-trained model.

[0095] By validating the pre-trained model on the test set, the final result is a fault diagnosis model for the thermal measurement points of the bucket wheel engine, with the input being the denoised thermal data and environmental interference factor data of each thermal measurement point, and the output being the fault type and corresponding confidence level of each thermal measurement point.

[0096] If the confidence level corresponding to the fault type of a certain thermal measuring point is greater than or equal to the preset confidence level threshold, it is determined to be a valid fault, triggering the thermal measuring point fault location signal and executing step S4.

[0097] For example, the confidence threshold is 95%.

[0098] S4. Measuring point positioning: Establish a coordinate library of thermal measuring points for the bucket wheel excavator. When a fault positioning signal for a thermal measuring point is received, analyze the real-time coordinates of the thermal measuring point.

[0099] Specifically, a coordinate library of thermal measurement points for the bucket wheel excavator is established, including: constructing a three-dimensional model of the bucket wheel excavator, taking the intersection of the center line of the bucket wheel excavator's travel track and the rotation center of the slewing platform as the origin of the coordinate system, defining a three-dimensional rectangular coordinate system, where the X-axis is parallel to the travel track and the direction of the bucket wheel excavator's forward stacking is the positive direction, the Y-axis is perpendicular to the travel track and the direction away from the stacking is the positive direction, and the Z-axis is perpendicular to the ground and the vertical upward direction is the positive direction;

[0100] A high-precision laser rangefinder was used to measure the three-dimensional coordinates of each thermal measurement point of the bucket wheel machine and to calibrate each point. A set number of UWB positioning base stations were deployed in the working area of ​​the bucket wheel machine, and a UWB positioning tag was configured for each thermal measurement point. The UWB positioning tag was bound to the thermal measurement point number.

[0101] A mapping table is established in the server database, consisting of thermal measurement point number, UWB positioning tag ID, and initial three-dimensional coordinates, forming a dedicated coordinate library for the thermal measurement points of the bucket wheel excavator.

[0102] In one specific embodiment, the number of positioning base stations is set to four, and the deployment of the four positioning base stations ensures that there is no signal obstruction between the UWB positioning tag of the thermal measurement point and the positioning base stations.

[0103] Specifically, please refer to Figure 3 As shown, the analysis process of the real-time coordinates of the thermal measurement point is as follows: After receiving the fault location signal of the thermal measurement point, the number of the fault thermal measurement point is obtained and the UWB positioning tag configured for the fault thermal measurement point is matched. The wireless signal is sent according to the UWB positioning tag configured for the fault thermal measurement point. The UWB base station receives the wireless signal sent by the UWB positioning tag configured for the fault thermal measurement point. The distances d1, d2, d3, and d4 between the UWB positioning tag configured for the fault thermal measurement point and each base station are calculated using the TOF algorithm.

[0104] It should be noted that the specific calculation process for obtaining the distance between the UWB positioning tag configured at the fault thermal measurement point and each base station is as follows:

[0105] Obtain the time t1 when the UWB positioning tag configured at the faulty thermal measurement point sends a wireless pulse signal to the first UWB positioning base station and the time t2 when the first base station receives the signal; obtain the time t3 when the first UWB positioning base station sends a response pulse signal to the UWB positioning tag configured at the faulty thermal measurement point and the time t4 when the UWB positioning tag configured at the faulty thermal measurement point receives the response signal; according to the formula The distance d1 between the UWB positioning tag configured at the faulty thermal measurement point and the first base station is calculated, where c represents the speed of light; similarly, the distances d2, d3, and d4 between the UWB positioning tag configured at the faulty thermal measurement point and the second, third, and fourth UWB positioning base stations are calculated.

[0106] Obtain the coordinates (X1,Y1,Z1), (X2,Y2,Z2), (X3,Y3,Z3), and (X4,Y4,Z4) of four UWB positioning base stations, and combine them with the distances d1, d2, d3, and d4, using the polygonal positioning principle algorithm formula. The real-time three-dimensional coordinates (X, Y, F, Z) of the UWB positioning tag configured at the fault thermal measurement point were calculated. real ,Y real Zreal This refers to the real-time three-dimensional coordinates of the fault thermal measurement point;

[0107] The real-time three-dimensional coordinates (X) of the fault thermal measurement point real ,Y real Z real The coordinates of the faulty thermal measuring point are compared with the initial coordinates (X0, Y0, Z0) in the thermal measuring point coordinate library, and then the result is determined according to the formula. The coordinate deviations ΔX, ΔY, and ΔZ are calculated. A coordinate deviation threshold is set. If ΔX, ΔY, and ΔZ are all less than the set deviation threshold, the real-time three-dimensional coordinates (X, Y, and Z) are directly output. real ,Y real Z real ) represents the final positioning coordinates of the fault thermal measurement point. If any deviation among ΔX, ΔY, and ΔZ is greater than the set deviation threshold, the real-time position and attitude parameters of the bucket wheel excavator are obtained. The position and attitude transformation is compensated by the coordinate system transformation matrix, and the total transformation matrix T is obtained through analysis.

[0108] It should be noted that the walking, turning, and pitching movements of the bucket wheel excavator during operation will cause changes in the equipment's posture, requiring correction of the real-time coordinates calculated by UWB.

[0109] It should be noted that the specific analysis process of the total transformation matrix T is as follows:

[0110] Real-time pose parameters are obtained from the bucket wheel excavator's PLC system or sensors. These pose parameters include travel displacement s, rotation angle θ, and pitch angle α. It should be noted that travel displacement s refers to the distance traveled along the X-axis (travel track direction), rotation angle θ refers to the rotation angle of the slewing platform around the Z-axis, and pitch angle α refers to the pitch angle of the cantilever around the horizontal axis.

[0111] In a three-dimensional coordinate system, if a point P = (x, y, z) is translated along the X-axis, the new coordinates are P' = (x + s, y, z). To uniformly represent translation and rotation using matrix multiplication, homogeneous coordinates are introduced, extending the three-dimensional point (x, y, z) into a four-dimensional vector. homogeneous coordinates of any point Right now

[0112] In a three-dimensional coordinate system, when point P = (x, y, z) rotates about the Z-axis by θ, the Z-coordinate remains unchanged, while the changes in the X and Y coordinates follow the rotation law of planar polar coordinates, i.e. homogeneous coordinates of any point Right now

[0113] In a three-dimensional coordinate system, when point P = (x, y, z) rotates about the Y-axis by α, the Y-coordinate remains unchanged, while the changes in the X and Z coordinates follow the rotation law of planar polar coordinates, i.e. homogeneous coordinates of any point Right now

[0114] According to T trans R rot R pitch According to the formula T = T trans ·R rot ·R pitch The total transformation matrix T is calculated.

[0115] Represent the real-time 3D coordinates as homogeneous coordinate vectors. Multiplying by the total transformation matrix T yields the homogeneous vector of the corrected coordinates, as shown in the formula. Take the first three dimensions, that is, the final location coordinates of the fault thermal measurement point are the corrected measurement point coordinates (X). corrected ,Y corrected Z corrected ).

[0116] It should be noted that after correcting the real-time coordinates, the impact of the bucket wheel excavator's walking, rotating, and pitching movements on the spatial position of the measuring points was resolved, ensuring the real-time performance and accuracy of positioning during dynamic operation.

[0117] S5. Display: Based on the fault type and the location coordinates of the thermal measurement points of the bucket wheel excavator, send the information to the operation and maintenance terminal of the bucket wheel excavator management personnel to perform the corresponding display operation.

[0118] In one specific embodiment, the present invention performs outlier detection and removal on thermal data from various thermal measurement points based on the Laida criterion. This efficiently removes extreme values ​​caused by instantaneous sensor malfunctions, such as jump data caused by dust obstructing the sensor or instantaneous peak values ​​caused by vibration and impact, ensuring the validity and accuracy of the data foundation and providing reliable data support for subsequent fault diagnosis and location. Furthermore, the MREMD-VMD dual-modal approach is used to denoise the outlier-processed thermal data. First, MREMD is used to eliminate baseline drift, and then a sliding median filter is added to fit the trend term based on EMD decomposition, effectively eliminating noise caused by slow changes in ambient temperature and long-term equipment malfunctions. The signal shift caused by aging and other factors during long-term operation is analyzed, focusing on local thermal changes caused by faults, such as sudden temperature changes due to bearing wear. Then, effective information in the noise is mined through VMD, and the selected noise IMF components are band-limited decomposed. Effective modes are selected in combination with the characteristic frequencies of thermal faults, avoiding the problem of traditional single filtering that may accidentally delete effective signals. After dual-mode processing, the signal-to-noise ratio of thermal data is effectively improved, avoiding problems such as outlier interference, baseline drift and noise mixing in the thermal data collected by the bucket wheel excavator under harsh conditions of high dust and strong vibration. This provides high-purity input for subsequent fault diagnosis, avoiding misdiagnosis or missed diagnosis caused by data distortion from the source.

[0119] The fault diagnosis model for the thermal measurement points of the bucket wheel excavator, constructed by using a dual-input CNN-LSTM model, fully considers the characteristics of thermal data and environmental interference factor data. It uses several historical thermal data and environmental interference factor data of each thermal measurement point after MREMD-VMD dual-modal denoising, as well as their corresponding fault type labels, to train and validate the model, ensuring the quality and reliability of the model training data and providing the model's generalization ability and diagnostic accuracy.

[0120] A 3D model of the bucket wheel excavator was constructed, defining a 3D Cartesian coordinate system with a specific point as the origin. A high-precision laser rangefinder was used to measure the 3D coordinates of each thermal measurement point and perform point-by-point calibration. A mapping table of thermal measurement point number, UWB tag ID, and initial 3D coordinates was established in the server database, forming a dedicated coordinate library that provides a precise reference benchmark for the positioning of thermal measurement points. When the positioning signal of a thermal measurement point is triggered, the wireless signal sent by the UWB positioning tag bound to the faulty measurement point is received through the UWB base station. The distance between the tag and each base station is calculated using the TOF algorithm, and combined with the coordinates of the four UWB positioning base stations, the real-time 3D coordinates of the measurement point are calculated using the polygonal positioning principle algorithm, achieving real-time positioning of the thermal measurement point. Furthermore, considering the changes in equipment posture caused by the bucket wheel excavator's walking, rotating, and pitching movements, the real-time coordinates calculated by UWB are corrected. By obtaining the real-time posture parameters of the bucket wheel excavator and analyzing the total transformation matrix, the real-time coordinates are corrected, solving the impact of equipment posture changes on the spatial position of the measurement point and ensuring the real-time performance and accuracy of positioning during dynamic operation.

[0121] Based on the fault type and location coordinates of the thermal measurement points of the bucket wheel excavator, the data is sent to the operation and maintenance terminal of the bucket wheel excavator management personnel to perform corresponding display operations. This enables management personnel to understand the fault status and location of the equipment in a timely manner, facilitates rapid maintenance measures, reduces equipment downtime, and improves the operating efficiency and reliability of the equipment.

[0122] Some of the data in the above formula are calculated by removing dimensions and taking their numerical values. The formula is the closest to the real situation obtained by software simulation of a large amount of collected data. The preset parameters and preset thresholds in the formula are set by those skilled in the art according to the actual situation or obtained through simulation of a large amount of data.

[0123] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

Claims

1. A method for fault diagnosis and real-time location positioning of thermal measuring points of a bucket wheel excavator, characterized in that, include: S1. Thermal Measurement Point Data Acquisition: Collect thermal data and environmental interference factor data from various thermal measurement points of the bucket wheel excavator. The thermal data for each thermal measurement point includes: temperature measured by the temperature sensor; pressure measured by the pressure sensor; and flow rate measured by the flow sensor. The environmental interference factor data includes: dust concentration measured by the dust concentration sensor and vibration frequency measured by the vibration sensor. The collected thermal data is converted into digital signals using an analog-to-digital converter, and step S2 is executed. S2. Preprocessing of Thermal Measurement Data: Based on the Laida criterion, outlier detection is performed on the thermal data of each thermal measurement point. Outliers are identified and removed to obtain the outlier-processed thermal data of each thermal measurement point. Simultaneously, noise reduction processing is performed on the outlier-processed thermal data of each thermal measurement point based on the MREMD-VMD dual-modality method to obtain the noise-reduced thermal data of each thermal measurement point. The process of obtaining the noise-reduced thermal data of each thermal measurement point is as follows: EMD Initial Decomposition: EMD decomposition is performed on the thermal data of each thermal measurement point after outlier processing, yielding several intrinsic mode function (IMF) components and one residual term, satisfying the following: Where j is the index of the intrinsic mode function (IMF) component, j = 1, 2, ..., o, and o is the total number of IMF component indices. For residual terms; Sliding median filter fits the trend term: for each component Let N be the sequence length. A sliding median filter is used to fit the trend. Let the sliding window length be L. Then the median of the h-th window is the data within that window. The value at the middle position after sorting in ascending order is denoted as After traversing all windows, the trend sequence is obtained. ; Eliminating baseline drift: Subtract the trend term from the IMF components to obtain the detrended IMF, as shown in the formula: ,in Represented as the detrended IMF; S3. Fault Diagnosis: Input the denoised thermal data and environmental interference factor data of each thermal measuring point into the fault diagnosis model of the bucket wheel excavator thermal measuring point to obtain the fault type and confidence level of each thermal measuring point; wherein, the fault diagnosis model of the bucket wheel excavator thermal measuring point is constructed by a dual-input CNN-LSTM model; if the confidence level corresponding to the fault type of a certain thermal measuring point is greater than or equal to the preset confidence level threshold, it is determined to be a valid fault, triggering the fault location signal of the thermal measuring point, and executing step S4; S4. Measurement point positioning: Establish a coordinate library of thermal measurement points for bucket wheel excavators. When a fault positioning signal for a thermal measurement point is received, analyze the real-time coordinates of the thermal measurement point. S5. Display: Send the fault type and the location coordinates of the thermal measurement points of the bucket wheel excavator to the operation and maintenance terminal of the bucket wheel excavator management personnel to perform the corresponding display operation.

2. The method for fault diagnosis and real-time location positioning of thermal measuring points of a bucket wheel excavator according to claim 1, characterized in that, The process of identifying and removing outliers includes the following steps: For the thermal data of each thermal measuring point, the mean value of the thermal data of each thermal measuring point is obtained by calculating the mean value, and the standard deviation of the thermal data of each thermal measuring point is obtained by calculating the standard deviation. The judgment interval of normal thermal data is determined based on the Laida criterion. If the thermal data of a certain thermal measuring point is not within the judgment interval of normal thermal data, the thermal data of that thermal measuring point is judged as an outlier and is removed accordingly. Based on this, the thermal data of each thermal measuring point after outlier processing is obtained.

3. The method for fault diagnosis and real-time location positioning of thermal measuring points of a bucket wheel excavator according to claim 1, characterized in that, The process of acquiring the thermal data of each thermal measurement point after noise reduction also includes: Phase space reconstruction: For each detrended IMF component, phase space reconstruction is performed using the delayed coordinate method. The time delay and phase space dimension are set to reconstruct the phase space vector. Arrangement Pattern and Entropy Calculation: Arrange the reconstructed phase space vectors in ascending order to generate an arrangement pattern; for an m-dimensional phase space, there are a total of Given a number of permutation patterns, the probability of each pattern occurring is calculated, and the permutation entropy is obtained by using a formula. Filtering valid IMFs: Set a permutation entropy threshold. If the permutation entropy of a detrended IMF component is less than the set permutation entropy threshold, it is determined to be a noisy IMF; if the permutation entropy of a detrended IMF component is greater than or equal to the set permutation entropy threshold, it is determined to be a valid IMF; thus, valid IMFs are obtained through this filtering process. VMD variational problem construction: Based on VMD, a constrained variational problem is constructed, which decomposes the signal into K band-limited IMFs. For noise, the objective function is to minimize the bandwidth weighted sum of each IMF, and the constraint condition is that the sum of each IMF is equal to the original signal. Solving the variational problem: By introducing Lagrange multipliers and a quadratic penalty factor, an augmented Lagrange function is constructed to solve the variational problem; and the Alternating Direction Multiplier Method (ADMM) is used iteratively to find the optimal decomposition. Effective information extraction and signal reconstruction: The modes obtained by VMD decomposition of the noise IMF are subjected to Fast Fourier Transform to obtain the spectrum. Effective mode components that match the characteristic frequency range of thermal faults are selected based on the characteristic frequency range of thermal faults. Finally, these effective mode components are reconstructed with all the selected effective IMFs to obtain the thermal data of each thermal measurement point after noise reduction.

4. The method for fault diagnosis and real-time location positioning of thermal measuring points of a bucket wheel excavator according to claim 1, characterized in that, The fault diagnosis model for the thermal measurement points of the bucket wheel excavator is constructed using a dual-input CNN-LSTM model, including: Acquire thermal data, environmental interference factor data and their corresponding fault type labels for each thermal measurement point after MREMD-VMD dual-modal noise reduction. The thermal data, environmental interference factor data, and their corresponding fault type labels of several historical thermal measurement points after MREMD-VMD dual-modal noise reduction are divided into training set, test set, and validation set. The dual-input CNN-LSTM model is selected as the base model. The dual-input CNN-LSTM model includes a first input layer and a second input layer. Each input layer is followed by a CNN feature extraction layer and an LSTM temporal analysis layer. The features are then fused through a fully connected layer, and finally the output layer outputs the fault type and confidence level. The base model is trained on the training set, the cross-entropy loss function is used, and the Adam optimizer is used on the validation set to adjust the learning rate, batch size and other hyperparameters to obtain the pre-trained model. By validating the pre-trained model on the test set, the final result is a fault diagnosis model for the bucket wheel turbine thermal measuring points, with the input being the denoised thermal data and environmental interference factor data of each thermal measuring point, and the output being the fault type and corresponding confidence level of each thermal measuring point.

5. The method for fault diagnosis and real-time location positioning of thermal measuring points of a bucket wheel excavator according to claim 1, characterized in that, The establishment of the bucket wheel excavator thermal measurement point coordinate library includes: constructing a three-dimensional model of the bucket wheel excavator, taking the intersection of the center line of the bucket wheel excavator's travel track and the rotation center of the slewing platform as the coordinate origin, defining a three-dimensional rectangular coordinate system, wherein the X-axis is parallel to the travel track and the direction of the bucket wheel excavator's forward stacking is the positive direction, the Y-axis is perpendicular to the travel track and the direction away from the stacking is the positive direction, and the Z-axis is perpendicular to the ground and the vertical upward direction is the positive direction; A high-precision laser rangefinder was used to measure the three-dimensional coordinates of each thermal measurement point of the bucket wheel excavator and to perform point-by-point calibration. A mapping table of measurement point number, UWB tag ID, and initial three-dimensional coordinates is established in the server database to form a dedicated coordinate library for the thermal measurement points of the bucket wheel excavator. A set number of UWB positioning base stations are deployed in the operating area of ​​the bucket wheel excavator, and a UWB positioning tag is configured for each thermal measurement point. The measurement point number and UWB tag ID are then established in the coordinate system.

6. The method for fault diagnosis and real-time location positioning of thermal measuring points of a bucket wheel excavator according to claim 1, characterized in that, The analysis process of the real-time coordinates of the thermal measuring points is as follows: Upon receiving a fault location signal from a thermal measurement point, the faulty thermal measurement point number is obtained and matched with the UWB positioning tag configured for the faulty thermal measurement point. A wireless signal is sent according to the UWB positioning tag configured for the faulty thermal measurement point. The UWB base station receives the wireless signal sent by the UWB positioning tag configured for the faulty thermal measurement point and calculates the distance between the UWB positioning tag configured for the faulty thermal measurement point and each base station using the TOF algorithm. The coordinates of four UWB positioning base stations are obtained. Combined with the distance, the real-time three-dimensional coordinates of the UWB positioning tag configured at the fault thermal measurement point are calculated using the polygonal positioning principle algorithm. That is, the real-time three-dimensional coordinates of the fault thermal measurement point. The real-time three-dimensional coordinates of the faulty thermal measuring point are compared with the initial coordinates of the faulty thermal measuring point in the thermal measuring point coordinate library. The coordinate deviation is calculated according to the formula, and a coordinate deviation threshold is set. If all coordinate deviations are less than the set deviation threshold, the real-time coordinates are directly output. If any of the coordinate deviations is greater than the set deviation threshold, the real-time position and posture parameters of the bucket wheel excavator are obtained. The position and posture transformation is compensated by the coordinate system transformation matrix, and the total transformation matrix is ​​obtained through analysis. The real-time 3D coordinates are represented as homogeneous coordinate vectors, and multiplied by the total transformation matrix to obtain the homogeneous vector of the corrected coordinates; the first three dimensions, i.e., the real-time coordinates, are taken as the corrected coordinates of the measured points.

7. A method for fault diagnosis and real-time location positioning of thermal measuring points of a bucket wheel excavator according to claim 6, characterized in that, The specific calculation process for obtaining the distance between the UWB positioning tag configured at the fault thermal measurement point and each base station is as follows: Obtain the time when the UWB positioning tag configured at the faulty thermal measurement point sends a wireless pulse signal to the first UWB positioning base station and the time when the first base station receives the signal; obtain the time when the first UWB positioning base station sends a response pulse signal to the UWB positioning tag configured at the faulty thermal measurement point and the time when the UWB positioning tag configured at the faulty thermal measurement point receives the response pulse signal; calculate the distance between the UWB positioning tag configured at the faulty thermal measurement point and the first base station according to the formula; similarly, analyze the distances between the UWB positioning tag configured at the faulty thermal measurement point and the second, third, and fourth UWB positioning base stations.

Images