A hydraulic gear pump assembly detection system

Abstract

The application discloses a hydraulic gear pump assembly detection system and relates to the technical field of product quality inspection. The hydraulic gear pump assembly detection system comprises a prediction model establishing module, an actual transmission comparison module and a multi-working condition extension analysis module. Through early fusion of multi-modal data and self-adaptive feature extraction, a precise prediction model is constructed, potential abnormalities can be effectively identified in advance, prediction data sets or abnormal reports can be output in advance, and detection efficiency is improved. The drive gear pump is operated at low speed, the predicted and actual gap data are further compared, the deviation is calculated, the abnormal angle is marked, and the error source is analyzed, so that problem positioning is more accurate. Finally, the influence of speed and temperature is considered, a digital twin model is constructed to simulate different working conditions, error sources are positioned in combination with multi-dimensional data, and optimization suggestions are output, so that the assembly efficiency is improved. Dynamic detection is carried out in combination with multi-parameter analysis, and abnormal analysis is more accurate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of product quality inspection, in particular to a hydraulic gear pump assembly detection system. BACKGROUND

[0002] Publication No. CN118735139B discloses a hydraulic gear pump assembly detection system, relating to the field of hydraulic gear pump assembly detection. The hydraulic gear pump assembly detection system comprises: a data acquisition subsystem for acquiring the angle data, distance data and surface data of the components of the assembled hydraulic gear pump; a data analysis subsystem for analyzing the angle data, distance data and surface data of the components of the assembled hydraulic gear pump respectively to obtain the angle detection index, linear size detection index and surface detection index of the assembled hydraulic gear pump. The invention integrates multiple detection modules and technical means to comprehensively evaluate the assembly quality of the hydraulic gear pump, not limited to a single quality index, but covering multiple aspects such as angle, size and surface, thereby significantly improving the accuracy and systematicness of detection and ensuring that each assembly link meets high quality standards.

[0003] However, as shown in the prior art, the current hydraulic gear pump is detected after assembly, which generally directly detects the size data of each part in a static state to determine part error and assembly error. However, in a static state, a single angle cannot comprehensively detect the error data of each running angle or even at different speeds and temperatures, which relates to whether it can run normally in actual operation. Although the running state of the finished product is checked in the later stage, it cannot be checked for each device, which may result in missed detection, and it is difficult to find the cause after the problem occurs, which greatly reduces the efficiency. If comprehensive detection is performed during the assembly stage, the efficiency and quality can be effectively improved. Therefore, the existing technology still has deficiencies in quality inspection of the assembled hydraulic gear pump. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides a hydraulic gear pump assembly detection system to solve the above problems.

[0005] To achieve the above purpose, the present application realizes the following technical scheme: a hydraulic gear pump assembly detection system, comprising:

[0006] A prediction model establishment module: collects image, vibration and pressure multi-modal data and performs preprocessing; adopts an early fusion strategy to fuse image, vibration signal and pressure data at the feature layer; adaptively extracts gear tooth shape and gap width features, and introduces an attention mechanism to automatically focus on the area most sensitive to gap changes; constructs a mathematical model to predict gap changes, and performs model training and optimization; calculates the change rate of the predicted gap, identifies potential abnormalities, and outputs a predicted gap data set or an abnormality report;

[0007] Actual transmission comparison module: drive gear pump low-speed operation, synchronous recording image and time stamp, angle information, using three-dimensional reconstruction technology to construct three-dimensional model of gear pump; compare the predicted data with the gap data in the three-dimensional model; calculate the mean and standard deviation of the gear gap, determine the fluctuation value, identify the periodic characteristics; compare the actual gap data with the predicted value, calculate the deviation, mark the abnormal angle and analyze the error source;

[0008] Multi-condition extended analysis module: collect data at different speeds, analyze the influence of speed on gap fluctuation and jitter frequency; monitor the working temperature of the gear pump, analyze the influence of thermal expansion on the gap, and correct the prediction model parameters; build a digital twin model of the gear pump to simulate the running state under different working conditions; introduce reinforcement learning algorithm to optimize assembly process parameters; combine multi-dimensional data to locate the main error source and output assembly optimization suggestions.

[0009] Preferably, the content of the multi-modal data acquisition and fusion performed by the prediction model establishment module includes:

[0010] Data acquisition:

[0011] Use a high-resolution industrial camera to collect images of the assembled gear pump, covering 0°~360° full angle, and define the angle step B, a total of images are collected;

[0012] Integrate vibration sensors and pressure sensors to synchronously collect vibration signals V(t) and pressure data P(t) of the gear pump during operation;

[0013] Data preprocessing:

[0014] Preprocessing steps of denoising, graying, and edge enhancement are performed on the images to extract gear profile and gap features;

[0015] Filter and normalize the vibration signals and pressure data to remove noise and outliers;

[0016] Multi-modal fusion:

[0017] An early fusion strategy is adopted, and a multi-modal fusion algorithm is used to fuse the preprocessed image data I, vibration signals V, and pressure data P at the feature layer. The fused feature vector F is represented as:

[0018] F=Fusion(I,V,P)=W I ·f I (I)+W V ·f V (V)+W P ·f P (P);

[0019] where f I , f V , f P are feature extraction functions for images, vibration signals and pressure data respectively, W I , W V , W P are weight coefficients.

[0020] Preferably, the content of the prediction model establishment module performing adaptive feature extraction includes:

[0021] Feature extraction: extract key parameters such as gear tooth profile and gap width based on deep learning or traditional CV algorithm;

[0022] Introduce attention mechanism, automatically focus on the area in the image that is most sensitive to gap change, and the attention weight A is calculated as:

[0023] A = Softmax(W a ·h);

[0024] where h is the feature vector, W a is the weight matrix of the attention mechanism, and the Softmax function is used to calculate the attention weight A, ensuring that the sum of all weights is 1, thereby representing the relative probability of the sensitivity of different areas to gap change;

[0025] Adaptive adjustment: automatically adjust the parameters and strategies of feature extraction according to different working states of the gear pump and image characteristics, use an adaptive learning rate adjustment algorithm, and dynamically adjust the learning rate η according to the gradient size:

[0026] ;

[0027] where η0 is the initial learning rate, β1 and β2 are momentum parameters, m t , v t are the first and second momentums respectively, is a small constant to prevent division by zero.

[0028] Preferably, the content of the prediction model establishment module performing mathematical model construction and potential anomaly identification includes:

[0029] Model selection: construct a dynamic transmission prediction model, the input is the angle sequence θ = [θ1, θ2, …, θ N ] and the fused multi-modal feature F, and the output is the gap prediction value corresponding to the angle ;

[0030] Model training: use historical data to train the prediction model and optimize the model parameters; the loss function L can be defined as the mean square error:

[0031] ;

[0032] wherein g i is the actual gap value;

[0033] Rate of change calculation: calculate the rate of change r i of the predicted gap with respect to the angle

[0034] ;

[0035] Set a rate of change mutation threshold T r , if the rate of change mutation |r i |>T r , it is determined that there is a potential anomaly;

[0036] Abnormal pre-judgment: if there is no anomaly, output the predicted gap dataset; if there is an anomaly, mark the abnormal angle and output an abnormal report.

[0037] Preferably, the actual transmission comparison module includes the following contents for acquiring images and constructing a three-dimensional model at low speed:

[0038] Drive and collect: drive the gear pump to run at low speed, set the number of collected frames, and record the timestamp t i and angle information θ i ;

[0039] Image grouping: group by angle, extract image set I θ at the same angle;

[0040] Three-dimensional reconstruction: use three-dimensional reconstruction technology to reconstruct the collected two-dimensional images into a three-dimensional model M of the gear pump.

[0041] Preferably, the actual transmission comparison module includes the following contents for comparing and analyzing based on the three-dimensional model and actual data:

[0042] Comparison model analysis: compare the gap data g M in the three-dimensional model and the predicted data , calculate the deviation e i :

[0043] ;

[0044] Through a visualization tool, intuitively display the assembly error and gap distribution;

[0045] Gap fluctuation calculation: for the image set I θ at the same angle, calculate the mean μ g and standard deviation σ g of the gear gap:

[0046] ;

[0047] ;

[0048] Combining mean μ g , standard deviation σ g , engineering experience and design requirements to determine the fluctuation value;

[0049] Periodic feature recognition:

[0050] Through Fourier transform or wavelet analysis, the periodic feature f p of the gap fluctuation is identified, and the gear jitter condition is associated;

[0051] Angle-by-angle deviation calculation: compare the actual gap data g with the predicted value angle-by-angle, calculate the deviation e i ;

[0052] Abnormal marking and analysis: set a deviation threshold, if the deviation exceeds the threshold, mark it as an abnormal angle, and analyze the error source in combination with the jitter feature.

[0053] Preferably, the multi-working-condition extended analysis module includes analysis of speed sensitivity, which includes:

[0054] Data acquisition: repeat the acquisition and analysis process at low speed, and continue to acquire data at medium and high speeds;

[0055] Influence analysis: analyze the influence of speed v on the standard deviation σ g of the gap fluctuation, the jitter frequency f d , and establish a speed-error correlation model:

[0056] σ g =a·v+b;

[0057] f d =c·v d ;

[0058] Wherein, a, b, c, d are model parameters.

[0059] Preferably, the multi-working-condition extended analysis module includes analysis of temperature sensitivity, which includes:

[0060] Temperature monitoring: integrate a temperature sensor to monitor the working temperature T of the gear pump;

[0061] Data comparison: repeat the image acquisition and comparison at different temperatures to analyze the influence of thermal expansion on the gap g; the thermal expansion coefficient α can be expressed as:

[0062] ;

[0063] Wherein, Delta g is the gap change, Delta T is the temperature change, g0 is the initial gap;

[0064] Model correction: according to the temperature influence, correct the prediction model parameters, improve the accuracy of the model at different temperatures.

[0065] Preferably, the content of the multi-condition extended analysis module for constructing the digital twin model includes:

[0066] Model construction: construct the digital twin model M' of the gear pump, and map the actually collected data to the digital twin model in real time;

[0067] Simulation and verification: by simulating the running state of the gear pump under different conditions, the performance and reliability are predicted; and compared with the actual detection results, the assembly process is optimized.

[0068] Preferably, the content of the multi-condition extended analysis module based on reinforcement learning for assembly optimization includes:

[0069] Agent definition: introduce a reinforcement learning algorithm, take the assembly process parameters p as the action of the agent, and take the detection result r as the reward signal;

[0070] Optimization process: constantly try different assembly parameter combinations, let the agent learn the assembly process; set the reward function R, the detection result r includes the deviation and jitter frequency of the gap, then the R calculation formula is:

[0071] ;

[0072] Wherein, lambda is the weight coefficient, balancing the influence of the gap deviation and the jitter frequency;

[0073] Step-by-step optimization of assembly parameters: use the policy gradient algorithm to update the assembly parameters, improve the quality and reliability of the gear pump;

[0074] Data analysis: combine the three-dimensional data of angle, speed and temperature, and locate the main error source through PCA or decision tree algorithm;

[0075] Optimization suggestion: output the assembly optimization suggestion combined with historical experience.

[0076] The present application provides a kind of hydraulic gear pump assembly detection system. Compared with prior art, it has the following beneficial effects:

[0077] 1、The hydraulic gear pump assembly detection system, through early fusion of multi-modal data and adaptive feature extraction, combined with attention mechanism focusing on sensitive areas, constructs a precise prediction model, which can effectively identify potential abnormalities in the first step, output prediction data set or abnormal report in advance, and improve detection efficiency; while the actual transmission comparison module drives the gear pump to run at low speed, uses three-dimensional reconstruction technology to build a model, further compares the predicted and actual gap data, calculates the deviation, marks the abnormal angle and analyzes the error source, so that the problem positioning is more accurate; finally, the multi-working-condition extended analysis module considers the influence of speed and temperature, corrects the model parameters, constructs a digital twin model to simulate different working conditions, introduces a reinforcement learning algorithm to optimize the assembly process parameters, locates the error source combined with multi-dimensional data and outputs optimization suggestions, and improves the assembly efficiency; and dynamic detection combined with multi-parameter analysis, the abnormal analysis is more accurate.

[0078] 2、The hydraulic gear pump assembly detection system, high-resolution industrial cameras collect images at full angle, combined with vibration and pressure sensors to collect data synchronously, providing rich and accurate information sources for subsequent analysis; the images and signals are preprocessed for targeted, effectively removing noise and outliers, enhancing the accuracy of feature extraction and improving data quality; the early fusion strategy combines advanced algorithms to fuse multi-modal data at the feature level, fully exploiting the correlation between data and enriching feature vector information; introduce attention mechanism, automatically focus on sensitive areas, combined with adaptive adjustment, improve the relevance and adaptability of feature extraction, enhance the robustness of the model to different working conditions; build a dynamic prediction model, calculate the change rate and set the threshold, accurately identify potential abnormalities, and output reports in time to ensure the assembly quality of the gear pump.

[0079] 3、The hydraulic gear pump assembly detection system collects images at low speed and records information synchronously, uses advanced three-dimensional reconstruction technology after grouping by angle, constructs a precise three-dimensional model of the gear pump, provides a reliable basis for subsequent comparison and analysis; compare the three-dimensional model gap data with the predicted data, calculate the deviation and visualize the display, intuitively present the assembly error and gap distribution, facilitate quick problem positioning; calculate the mean and standard deviation of the gear gap to determine the fluctuation value, identify periodic characteristics, and help understand the gap change of the gear pump during operation; calculate the deviation by angle, set the threshold to mark the abnormal angle, and analyze the error source combined with the jitter characteristics, which can accurately find out the root cause of the assembly problem, providing a strong basis for improving assembly quality.

[0080] 4、The hydraulic gear pump assembly detection system collects data at different speeds and temperatures, analyzes the influence of gap fluctuation, jitter frequency and thermal expansion, establishes a correlation model, makes the understanding of the performance of the gear pump more comprehensive, corrects the prediction model parameters according to the influence of temperature, improves the accuracy of the model under different working conditions, enhances the adaptability and reliability of the model, constructs a digital twin model, maps real-time data, simulates different working conditions, predicts performance and reliability, and compares with actual data to verify, and provides strong support for optimizing assembly process; introduce reinforcement learning algorithm, intelligently adjust assembly parameters, combine multi-dimensional data to locate error sources, output optimization suggestions, and effectively improve the quality and reliability of the gear pump. BRIEF DESCRIPTION OF DRAWINGS

[0081] Figure 1 The system module block diagram of the present application is shown in the following table:

[0082] Figure 2 The working process diagram of the prediction model establishment module of the present application is shown in the following table:

[0083] Figure 3 The working process diagram of the actual transmission comparison module of the present application is shown in the following table:

[0084] Figure 4 The working process diagram of the multi-condition extension analysis module of the present application is shown in the following table. DETAILED DESCRIPTION

[0085] The technical solutions in the embodiments of the present application will be described in detail below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0086] The present application discloses a kind of hydraulic gear pump assembly detection system, and provides the following four technical solutions:

[0087] Figure 1 The first embodiment is shown: it include:

[0088] Prediction model establishment module: collect image, vibration and pressure multi-modal data and pre-process;Adopt early fusion strategy, fuse image, vibration signal and pressure data in feature layer;Self-adaptive extraction gear tooth profile, gap width feature, and introduce attention mechanism to automatically focus on the area most sensitive to gap change;Build mathematical model to predict gap change, and carry out model training and optimization;Calculate the rate of change of predicted gap, identify potential anomalies, and output predicted gap dataset or anomaly report;

[0089] Actual transmission comparison module: drive the gear pump to run at low speed, synchronize the image and timestamp, angle information, and construct a three-dimensional model of the gear pump using three-dimensional reconstruction technology; compare the predicted data with the gap data in the three-dimensional model; calculate the mean and standard deviation of the gear gap, determine the fluctuation value, and identify the periodic characteristics; compare the actual gap data with the predicted value, calculate the deviation, mark the abnormal angle, and analyze the error source;

[0090] Multi-condition extended analysis module: collect data at different speeds, analyze the influence of speed on gap fluctuation and jitter frequency; monitor the working temperature of the gear pump, analyze the influence of thermal expansion on the gap, and correct the prediction model parameters; construct a digital twin model of the gear pump to simulate the running state under different working conditions; introduce a reinforcement learning algorithm to optimize the assembly process parameters; combine multi-dimensional data to locate the main error source and output assembly optimization suggestions.

[0091] Through early fusion of multi-modal data and adaptive feature extraction, combined with attention mechanism focusing on sensitive areas, a precise prediction model is constructed, which can effectively identify potential abnormalities in advance, output prediction data set or abnormal report in advance, and improve detection efficiency; the actual transmission comparison module drives the gear pump to run at low speed, constructs a model using three-dimensional reconstruction technology, further compares the predicted and actual gap data, calculates the deviation, marks the abnormal angle, and analyzes the error source, making the problem positioning more accurate; finally, the multi-condition extended analysis module considers the influence of speed and temperature, corrects the model parameters, constructs a digital twin model to simulate different working conditions, introduces a reinforcement learning algorithm to optimize the assembly process parameters, combines multi-dimensional data to locate the error source and outputs optimization suggestions, improving the assembly efficiency; and dynamic detection is performed combined with multi-parameter analysis, and the abnormal analysis is more accurate.

[0092] Figure 2 The second embodiment is shown, and the main difference from the first embodiment is that the work content of the prediction model establishment module specifically includes:

[0093] Multi-modal data acquisition and fusion:

[0094] Data acquisition:

[0095] A high-resolution industrial camera (resolution such as 2048x1536) is used to collect images of the assembled gear pump, covering 0°~360° full angle, and defining an angle step B, a total of images are collected;

[0096] A vibration sensor (sampling frequency set to 1000Hz) and a pressure sensor (accuracy ±0.1%) are integrated to synchronously collect vibration signals V(t) and pressure data P(t) of the gear pump during operation;

[0097] High-resolution industrial cameras capture images at full angle, combined with vibration and pressure sensors to collect data synchronously, providing rich and accurate information sources for subsequent analysis.

[0098] Data preprocessing:

[0099] Image preprocessing steps: denoising (using Gaussian filter algorithm, standard deviation σ=1), grayscale, edge enhancement (using Canny edge detection, threshold T1=50, T2=150) to extract gear profile and gap features;

[0100] Filtering (using low-pass filter, cutoff frequency f c =50Hz) and normalization processing of vibration signals and pressure data to remove noise and outliers;

[0101] Targeted preprocessing of images and signals effectively removes noise and outliers, enhances the accuracy of feature extraction, and improves data quality.

[0102] Multi-modal fusion:

[0103] Early fusion strategy using multi-modal fusion algorithm (such as the combination of convolutional neural network and recurrent neural network) to fuse preprocessed image data I, vibration signal V and pressure data P at the feature layer. The fused feature vector F is represented as:

[0104] F=Fusion(I,V,P)=W I ·f I (I)+W V ·f V (V)+W P ·f P (P);

[0105] Where f I , f V , f P are feature extraction functions for image, vibration signal and pressure data, respectively, and W I , W V , W P are weight coefficients.

[0106] Early fusion strategy combined with advanced algorithms to fuse multi-modal data at the feature layer, fully exploiting the correlation between data and enriching feature vector information.

[0107] Adaptive feature extraction:

[0108] Feature extraction: extract key parameters such as gear tooth shape and gap width based on deep learning (such as YOLO, Mask R-CNN) or traditional CV algorithms (such as Canny edge detection + Hough transform);

[0109] Introduce attention mechanism, automatically focus on the area in the image that is most sensitive to gap changes, attention weight A is calculated as:

[0110] A = Softmax(W a ·h);

[0111] Where h is the feature vector, W a is the weight matrix of the attention mechanism, and the Softmax function is used to calculate the attention weight A, ensuring that the sum of all weights is 1, thus representing the relative probability of different areas being sensitive to gap changes.

[0112] Adaptive adjustment: according to the different working conditions of the gear pump and the characteristics of the image, automatically adjust the parameters and strategies of feature extraction, use adaptive learning rate adjustment algorithm (such as Adam optimizer), learning rate η is dynamically adjusted according to the gradient size:

[0113] ;

[0114] Where η0 is the initial learning rate, β1 and β2 are momentum parameters, m t and v t are the first and second order momentum, is a small constant to prevent division by zero.

[0115] Introduce attention mechanism, automatically focus on sensitive areas, combined with adaptive adjustment, improve the pertinence and adaptability of feature extraction, enhance the robustness of the model to different working conditions.

[0116] Construct mathematical model and identify potential anomalies:

[0117] Model selection: build a dynamic transmission prediction model (such as LSTM or Transformer), input is the angle sequence θ = [θ1, θ2, …, θ N ] and the fused multi-modal feature F, output is the gap prediction value corresponding to the angle ;

[0118] Model training: use historical data to train the prediction model, optimize model parameters; the loss function L can be defined as mean square error:

[0119] ;

[0120] Where g i is the actual gap value;

[0121] Change rate calculation: calculate the change rate r i of the predicted gap with respect to the angle:

[0122] ;

[0123] Set the rate of change of the threshold T r , if the rate of change of ∣r i ∣>T r , it is determined that there is a potential anomaly;

[0124] Abnormal prediction: if there is no anomaly, output the predicted gap data set; if there is an anomaly, mark the abnormal angle and output the abnormal report.

[0125] Build a dynamic prediction model, accurately identify potential anomalies through rate of change calculation and threshold setting, and output reports in a timely manner to ensure the quality of gear pump assembly.

[0126] Figure 3 The third embodiment is shown, and the main difference from the first embodiment is that the actual transmission comparison module includes:

[0127] Collect images at low speed and build a three-dimensional model:

[0128] Drive and collect: drive the gear pump to run at low speed (such as 10 rpm), set the collection frame number (such as 10 frames per second), and record the timestamp t i and angle information θ i ;

[0129] Image grouping: group by angle, extract image set I θ of the same angle;

[0130] Three-dimensional reconstruction: use three-dimensional reconstruction technology (such as structured light scanning or stereo vision) to reconstruct the collected two-dimensional images into a three-dimensional model M of the gear pump.

[0131] Collect images at low speed and record information synchronously, use advanced three-dimensional reconstruction technology after grouping by angle, and build a precise three-dimensional model of the gear pump, providing a reliable basis for subsequent comparative analysis.

[0132] Based on the three-dimensional model and the actual data for comparative analysis:

[0133] Compare model analysis: compare the gap data g M in the three-dimensional model and the predicted data , calculate the deviation e i :

[0134] ;

[0135] Through the visualization tool, the assembly error and gap distribution are intuitively displayed;

[0136] Compare the three-dimensional model gap data with the predicted data, calculate the deviation and visually display it, and intuitively present the assembly error and gap distribution, which facilitates quick problem positioning.

[0137] Gap fluctuation calculation: Calculate the mean value μ of the gear gap for the same angle of the image set I θ g And the standard deviation σ g :

[0138] ;

[0139] ;

[0140] Determine the fluctuation value (such as ±0.05mm) by combining the mean value μ g , the standard deviation σ g , engineering experience and design requirements;

[0141] Periodic feature identification:

[0142] Identify the periodic feature f of the gap fluctuation through Fourier transform or wavelet analysis p , and correlate the gear rattle situation;

[0143] Calculate the mean value and standard deviation of the gear gap to determine the fluctuation value and identify the periodic feature, which helps to better understand the gap change of the gear pump during operation.

[0144] Angular deviation calculation: Compare the actual gap data g with the predicted value angle by angle, and calculate the deviation e i (For example, absolute error ≤0.03mm is qualified);

[0145] Abnormal marking and analysis: Set the deviation threshold, if the deviation exceeds the threshold, mark it as an abnormal angle, and analyze the error source (such as assembly eccentricity, tooth profile wear) combined with the rattle feature.

[0146] Calculate the deviation angle by angle, set the threshold to mark the abnormal angle, and analyze the error source combined with the rattle feature, which can accurately find out the root cause of the assembly problem and provide a strong basis for improving the assembly quality.

[0147] Figure 4 The fourth embodiment is shown, and the main difference from the first embodiment is that the working content of the multi-working condition expansion analysis module specifically includes:

[0148] Analysis of speed sensitivity:

[0149] Data collection: Repeat the collection and analysis process at low speed, continue to collect data at medium speed (50rpm) and high speed (100rpm);

[0150] Influence analysis: Analyze the influence of speed v on the gap fluctuation standard deviation σ g , the rattle frequency f d , and establish a speed-error correlation model:​

[0151] σ g =a·v+b;

[0152] f d =c·v d ;

[0153] wherein a, b, c, d are model parameters.

[0154] By collecting data at different speeds and temperatures, the influence of gap fluctuation, dithering frequency and thermal expansion is analyzed, and a correlation model is established, so that the understanding of the performance of the gear pump is more comprehensive.

[0155] Analysis of temperature sensitivity:

[0156] wherein a, b, c, d are model parameters.

[0157] By collecting data at different speeds and temperatures, the influence of gap fluctuation, dithering frequency and thermal expansion is analyzed, and a correlation model is established, so that the understanding of the performance of the gear pump is more comprehensive.

[0158] Analysis of temperature sensitivity:

[0159] Data comparison: Repeat image collection and comparison at different temperatures to analyze the influence of thermal expansion on the gap g; the thermal expansion coefficient α can be expressed as:

[0160] ;

[0161] wherein Δg is the gap change, ΔT is the temperature change, and g0 is the initial gap.

[0162] Model correction: According to the influence of temperature, correct the prediction model parameters to improve the accuracy of the model at different temperatures.

[0163] According to the influence of temperature, correct the prediction model parameters to improve the accuracy of the model at different working conditions, enhance the adaptability and reliability of the model.

[0164] Construct a digital twin model:

[0165] Model construction: Construct the digital twin model M' of the gear pump, and map the actual collected data in real time to the digital twin model;

[0166] Simulation and verification: Through the simulation of the running state of the gear pump under different working conditions, the performance and reliability are predicted; and compared with the actual detection results, the assembly process is optimized.

[0167] Construct a digital twin model, map real-time actual data, simulate different working conditions, predict performance and reliability, and compare with actual results to provide strong support for optimizing assembly process.

[0168] Assembly optimization based on reinforcement learning:

[0169] Agent definition: Introduce reinforcement learning algorithm, take assembly process parameters p (such as bearing preload, gear shape machining accuracy) as the action of agent, and take detection result r as reward signal;

[0170] Optimization process: Continuously try different assembly parameter combinations to let the agent learn the assembly process; Set reward function R, and detection result r includes gap deviation and dither frequency, then R calculation formula is:

[0171] ;

[0172] Where λ is the weight coefficient, balancing the influence of gap deviation and dither frequency;

[0173] Step-by-step optimization of assembly parameters: Use policy gradient algorithm (such as REINFORCE) to update assembly parameters to improve the quality and reliability of gear pump;

[0174] Data analysis: Combine angle, speed, and temperature three-dimensional data to locate the main error source through PCA or decision tree algorithm;

[0175] Optimization suggestions: Output assembly optimization suggestions (such as adjusting bearing preload, improving gear shape machining accuracy) combined with historical experience.

[0176] Introduce reinforcement learning algorithm, intelligently adjust assembly parameters, locate error sources combined with multi-dimensional data, output optimization suggestions, effectively improve the quality and reliability of gear pump.

[0177] A factory originally used traditional manual detection method to detect the gap of hydraulic gear pump and evaluate the assembly quality, which had problems such as low detection efficiency, large error, and difficulty in covering all working conditions. In order to improve the detection efficiency and accuracy, the factory decided to use this scheme to test the hydraulic gear pump. The comparison of test results with the original process test results is shown in Table 1 below:

[0178] Table 1 Comparison of test results

[0179]

[0180] In summary, the factory uses this scheme to test the hydraulic gear pump, which has significantly improved detection efficiency, smaller detection error, more comprehensive working condition coverage, higher abnormal identification accuracy, more effective assembly optimization suggestions, and lower detection cost compared to the original method.

[0181] Meanwhile, the contents not described in detail in this specification all belong to the existing technology known to those skilled in the art.

[0182] It is to be understood that the terminology used herein such as first and second, and the like, is only used to distinguish one entity or action from another entity or action, and does not necessarily require or imply any such actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0183] While embodiments of the present application have been shown and described with reference to particular embodiments thereof, it will be understood by those skilled in the art that various changes in form and details can be made therein without departing from the spirit and scope of the application. The scope of the application is defined by the appended claims and their equivalents.

Claims

1. A hydraulic gerotor pump assembly inspection system, characterized by, The application relates to a gear pump fault prediction system and method. The prediction model establishment module: collects image, vibration and pressure multi-modal data and performs pretreatment; an early fusion strategy is adopted to fuse the image, vibration signal and pressure data at a feature layer; Adaptive extraction of gear tooth profile and gap width features, and introduction of an attention mechanism to automatically focus on the area most sensitive to gap changes; a mathematical model is constructed to predict gap changes, and model training and optimization are performed; the rate of change of the predicted gap is calculated, potential abnormalities are identified, and a predicted gap dataset or an abnormality report is output; Actual transmission comparison module: drive the gear pump to run at low speed, synchronously record the image and time stamp, and angle information, and construct a three-dimensional model of the gear pump by using a three-dimensional reconstruction technology; compare the predicted data with the gap data in the three-dimensional model; calculate the mean and standard deviation of the gear gap, determine the fluctuation value, and identify the periodic characteristics; Compare the actual gap data with the predicted value, calculate the deviation, mark the abnormal angle and analyze the error source; Multi-working condition extended analysis module: collect data at different speeds, analyze the influence of speed on gap fluctuation and dithering frequency, monitor the working temperature of the gear pump, analyze the influence of thermal expansion on the gap, and correct the parameters of the prediction model; a digital twin model of the gear pump is constructed to simulate the running state under different working conditions; a reinforcement learning algorithm is introduced to optimize the assembly process parameters; combined with multi-dimensional data, the main error source is located, and assembly optimization suggestions are output.

2. A hydraulic gerotor assembly inspection system as claimed in claim 1, wherein: The prediction model establishment module includes the following contents of multi-modal data collection and fusion: Data collection: The images of the assembled gear pump are collected using a high-resolution industrial camera, covering the full angle of 0°~360°, and an angle step B is defined, and a total of Zhang images; Integrate vibration sensors and pressure sensors to synchronously collect vibration signals V(t) and pressure data P(t) during the running of the gear pump; Data preprocessing: Perform denoising, grayscale, and edge enhancement preprocessing steps on the image to extract gear profile and gap features; Filter and normalize the vibration signal and pressure data to remove noise and outliers; Multi-modal fusion: An early fusion strategy is adopted, and a multi-modal fusion algorithm is used to fuse the preprocessed image data I, vibration signal V and pressure data P at the feature layer. The fused feature vector F is represented as: F = Fusion (I, V, P) = W I • f I (I) + W V • f V (V) + W P • f P (P); where f I , f V , and f P are feature extraction functions of the image, vibration signal, and pressure data, respectively, and W I , W V , and W P are weight coefficients.

3. A hydraulic gerotor assembly inspection system as claimed in claim 2, wherein: The prediction model establishment module includes the following contents of adaptive feature extraction: Feature extraction: extract the gear tooth profile and gap width key parameters based on deep learning or traditional CV algorithms; Introduce an attention mechanism to automatically focus on the area in the image that is most sensitive to gap changes. The attention weight A is calculated as: A = Softmax(W a • h); where h is the feature vector, W a is the weight matrix of the attention mechanism, and the Softmax function is used to calculate the attention weight A, ensuring that the sum of all weights is 1, thereby representing the relative probability of the sensitivity of different regions to gap changes; Adaptive adjustment: automatically adjust the parameters and strategies of feature extraction according to different working states of the gear pump and image characteristics. Use an adaptive learning rate adjustment algorithm. The learning rate η is dynamically adjusted according to the gradient size: ； where η0is the initial learning rate, β1, β2are momentum parameters, m t , v t are the first and second order momentum, respectively, is a small constant to prevent division by zero.

4. The hydraulic gear pump assembly detection system of claim 1, wherein: The prediction model establishment module includes the following contents of constructing a mathematical model and identifying potential abnormalities: Model selection: a dynamic transmission prediction model is constructed, the input is an angle sequence θ=[θ1, θ2, …, θ N ] and the fused multi-modal feature F, and the output is the gap prediction value corresponding to the angle ; Model training: use historical data to train the prediction model and optimize the model parameters. The loss function L can be defined as the mean square error: ； where g i is the actual gap value; Rate of change calculation: Calculate rate of change of predicted gap with angle r i : ； Setting a rate of change mutation threshold T r , if the rate of change mutation |r i | > T r , then determine that there is a potential anomaly; Abnormality prediction: if there is no abnormality, output the predicted gap dataset; if there is an abnormality, mark the abnormal angle and output an abnormality report.

5. A hydraulic gear pump assembly detection system as set forth in claim 1, characterized in that: The actual transmission comparison module includes the following contents of collecting images and constructing a three-dimensional model at low speed: Drive and collect: drive gear pump to run at low speed, set the number of collection frames, synchronous record timestamp t i With angle information θ i ; Image grouping: Grouping by angle, extracting image set I of same angle θ ; Three-dimensional reconstruction: adopt a three-dimensional reconstruction technology to reconstruct the collected two-dimensional images into a three-dimensional model M of the gear pump.

6. A hydraulic gerotor assembly inspection system as claimed in claim 5, wherein: The actual transmission ratio comparison module includes the following content based on the comparison analysis of the three-dimensional model and the actual data: Comparative model analysis: comparing gap data g in comparative three-dimensional model M and predicted data , calculating deviation e i : ； Through the visualization tool, the assembly error and gap distribution are intuitively displayed; Gap fluctuation calculation: for the set of images I of the same angle θ , the mean value μ of the gear gap is calculated g and the standard deviation σ g : ； ； Combining the mean μ g , the standard deviation σ g , engineering experience and design requirements to determine the fluctuation value; Periodic feature recognition: By Fourier transform or wavelet analysis, the periodicity of the gap fluctuations f is identified p , the gear rattle situation is correlated; Angle-wise deviation calculation: Calculate the deviation e of the actual gap data g from the predicted value Angle-wise comparison, calculate deviation e i ; Abnormal marking and analysis: set the deviation threshold, if the deviation exceeds the threshold, mark it as an abnormal angle, and analyze the error source combined with the jitter feature.

7. A hydraulic gear pump assembly detection system as set forth in claim 1, characterized in that: The multi-working condition extended analysis module includes the following content for the analysis of speed sensitivity: Data acquisition: repeat the acquisition and analysis process under low speed transmission, and continue to collect data under medium speed and high speed; Influence analysis: analysis of the influence of the speed v on the standard deviation σ of the gap fluctuations g , the dithering frequency f d , establishment of a speed-error correlation model: σ g = a · v + b; f d = c · v d ; Wherein, a, b, c, d are model parameters.

8. A hydraulic gerotor assembly inspection system as claimed in claim 7, wherein: The multi-working condition extended analysis module includes the following content for the analysis of temperature sensitivity: Temperature monitoring: integrate temperature sensors to monitor the working temperature T of the gear pump; Data comparison: repeat the image acquisition and comparison under different temperatures, analyze the influence of thermal expansion on the gap g; the thermal expansion coefficient α can be expressed as: ； Where Δg is the gap change, ΔT is the temperature change, and g0 is the initial gap; Model correction: according to the temperature influence, correct the prediction model parameters, and improve the accuracy of the model under different temperatures.

9. A hydraulic gerotor assembly inspection system as claimed in claim 8, wherein: The multi-working condition extended analysis module includes the following content for constructing a digital twin model: Model construction: construct the digital twin model M' of the gear pump, and map the actual collected data into the digital twin model in real time; Simulation and verification: simulate the running state of the gear pump under different working conditions, predict the performance and reliability, and compare and verify with the actual detection results to optimize the assembly process.

10. A hydraulic gerotor assembly inspection system as claimed in claim 9, wherein: The multi-working condition extended analysis module includes the following content for assembly optimization based on reinforcement learning: Agent definition: introduce reinforcement learning algorithm, take the assembly process parameters p as the action of the agent, and take the detection result r as the reward signal; Optimization process: constantly try different assembly parameter combinations to make the agent learn the assembly process; set the reward function R, the detection result r includes the deviation of the gap and the jitter frequency, then the R calculation formula is: ； Where λ is the weight coefficient, balancing the influence of the gap deviation and the jitter frequency; Step-by-step optimization of assembly parameters: use the policy gradient algorithm to update the assembly parameters, improve the quality and reliability of the gear pump; Data analysis: combine the three-dimensional data of angle, speed and temperature, and locate the main error source through PCA or decision tree algorithm; Optimization suggestion: output the assembly optimization suggestion combined with historical experience.

Images