Balancing machine automatic test system for dynamic balance of rotating body

Abstract

The invention belongs to the technical field of dynamic balance test, and discloses a balancing machine automatic test system for dynamic balance of a rotating body, which comprises a trial operation setting module for setting the rotating speed of a driving shaft in a trial operation stage of a balancing machine and marking the rotating speed as a trial operation rotating speed; the first acquisition module is used for collecting environment data of the balancing machine, and the preprocessing module is used for carrying out feature extraction on the environment data to obtain environment feature data; the model building module is used for collecting point cloud data of the to-be-detected rotating body in the operation stage and building a three-dimensional model of the to-be-detected rotating body according to the point cloud data; the second acquisition module is used for acquiring the centrifugal force of the to-be-detected rotating body when the rotating speed of the driving shaft reaches the pilot running rotating speed; the high-precision balance test is realized by automatically adjusting the calibration parameters of the balancing machine, incomplete or wrong manual input data is avoided, human intervention is effectively reduced, and the calibration efficiency and precision are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of dynamic balancing test, and more specifically, to an automatic testing system of a balancing machine for dynamic balancing of a rotating body. Background Art

[0002] Rotating bodies (such as impellers, fans, motor rotors, etc.) often generate large vibrations and noises during operation due to their uneven distribution of rotating mass, which affects the operating performance and life of the equipment. In order to solve this problem, dynamic balancing tests are usually required during the manufacturing or installation phase of the rotating body to ensure that the rotating body remains balanced when rotating at high speeds. However, traditional dynamic balancing test systems still face many problems during the detection process.

[0003] Environmental factors interfere with test accuracy, the inspection process of complex workpieces is cumbersome, and the three-dimensional shape and quality are difficult to automatically identify. As a result, the parameter calibration of traditional balancing machines before testing relies on manual input of relevant information. This increases the difficulty of data collection and is easily affected by human factors, such as incomplete or erroneous data entry, which leads to inaccurate calibration of balancing machine parameters, which in turn leads to large deviations in balancing test results, affecting efficiency and accuracy.

[0004] In view of this, the present invention proposes an automatic testing system of a balancing machine for dynamic balancing of a rotating body to solve the above-mentioned problem. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned purpose, the present invention provides the following technical solution: an automatic testing system of a balancing machine for dynamic balancing of a rotating body, comprising:

[0006] The trial operation setting module is used to set the drive shaft speed during the trial operation phase of the balancing machine, which is marked as the trial operation speed;

[0007] The first acquisition module is used to collect the environmental data of the balancing machine.

[0008] A preprocessing module is used to extract features from environmental data to obtain environmental feature data;

[0009] A model building module is used to collect point cloud data of the rotating body to be detected during the operation phase, and to build a three-dimensional model of the rotating body to be detected based on the point cloud data;

[0010] The second acquisition module acquires the centrifugal force of the rotating body to be detected when the driving shaft speed reaches the test running speed;

[0011] The first analysis module is used to determine the radius of the rotating body to be detected according to the three-dimensional model, and calculate the mass of the rotating body to be detected according to the radius, the test running speed and the centrifugal force;

[0012] The second analysis module is used to input the three-dimensional model into the rotating body recognition model to obtain the name and components of the rotating body to be detected;

[0013] The third analysis module obtains the volume of the rotating body to be detected according to the three-dimensional model, and calculates the density of the rotating body to be detected according to the volume and mass;

[0014] A segmentation analysis module, which inputs the three-dimensional model into a pre-trained component segmentation model corresponding to the name, obtains an independent model of each component, and determines the size data of each component according to the independent model;

[0015] The adaptive calibration module selects a parameter calibration model corresponding to the name, inputs environmental characteristic data, radius, mass, and dimension data of each component into the parameter calibration model, obtains balancing machine parameter calibration data, and calibrates the balancing machine.

[0016] Furthermore, the training method of the rotating body recognition model includes:

[0017] Collecting b groups of three-dimensional models in advance, setting corresponding names and components for the b groups of three-dimensional models, where b is an integer greater than 1, and converting the three-dimensional models, the corresponding names and components into a corresponding set of feature vectors;

[0018] Each group of feature vectors is used as the input of the rotating body recognition model. The rotating body recognition model takes a group of predicted names and components corresponding to each group of three-dimensional models as output, and takes the actual names and components corresponding to each group of three-dimensional models as prediction targets; minimizing the sum of prediction errors of all three-dimensional models is used as the training target; the rotating body recognition model is trained until the sum of prediction errors reaches convergence and the training is stopped; the rotating body recognition model is a deep neural network model.

[0019] Furthermore, the method for calculating the mass of the rotating body to be detected according to the radius and the centrifugal force includes:

[0020] ω1=2πn;

[0021] In the formula, m is mass, F is centrifugal force, r is radius, ω1 is test run angular velocity, π is pi, and n is test run speed.

[0022] Furthermore, the training method of the component segmentation model includes:

[0023] Collect H groups of three-dimensional models with component area annotations in advance, and save the annotated three-dimensional models in a format that can be understood by computers; divide the H groups of three-dimensional models with component area annotations into a training set and a test set, use the three-dimensional models in the training set as the input of a component segmentation model, use the component area in the training set as the output of the component segmentation model, use the test set to test the component segmentation model, and output a component segmentation model that meets the prediction accuracy, wherein the component segmentation model is a convolutional neural network model or a graph neural network model.

[0024] Furthermore, the environmental data includes temperature, humidity and noise; the environmental characteristic data includes temperature average value, temperature standard deviation, humidity average value, humidity standard deviation and noise average value and noise standard deviation.

[0025] Furthermore, the method for obtaining the environmental characteristic data includes:

[0026] The temperatures collected during the trial run phase are used to establish a temperature analysis set, and the temperature average and temperature standard deviation of the temperature analysis set are calculated; the humidity collected during the trial run phase is used to establish a humidity analysis set, and the humidity average and humidity standard deviation of the humidity analysis set are calculated; the noise collected during the trial run phase is used to establish a noise analysis set, and the noise average and noise standard deviation of the noise analysis set are calculated.

[0027] Furthermore, the parameter calibration data includes sensor sensitivity setting data, test rotation speed, and imbalance tolerance setting data.

[0028] Furthermore, it also includes an early warning module, which is used to determine whether to generate an early warning instruction based on the predicted test speed and the test centrifugal force generated when the rotating body to be detected reaches the predicted test speed; if an early warning instruction is generated, the detection of the rotating body to be detected is stopped.

[0029] Furthermore, the method for determining whether to generate an early warning instruction includes:

[0030] If the predicted test speed exceeds the rated speed of the balancing machine, an early warning instruction is generated, otherwise no early warning instruction is generated;

[0031] If the rotating body to be tested reaches the predicted test speed and the test centrifugal force generated exceeds the centrifugal force borne by the balancing machine, an early warning instruction is generated, otherwise no early warning instruction is generated.

[0032] Furthermore, the test centrifugal force calculation method includes:

[0033] ω=2πz;

[0034] To test the centrifugal force, m is the mass of the rotating body to be tested, r is the radius of the rotating body to be tested, ω2 2 is the test angular velocity, π is the pi, and z is the test rotation speed.

[0035] The technical effects and advantages of the automatic testing system of a balancing machine for dynamic balancing of a rotating body of the present invention are as follows:

[0036] Before the formal test, a trial run phase is set up. During the trial run phase, the environmental data of the balancing machine is collected, a three-dimensional model of the rotating body to be tested is constructed, the radius of the rotating body to be tested is obtained, and the mass of the rotating body to be tested is calculated based on the radius, trial run speed and centrifugal force. There is no need to weigh it through other weighing equipment, which simplifies the test process, improves the test efficiency, and reduces the error caused by manual participation. The three-dimensional model is input into the rotating body recognition model to obtain the name and components of the rotating body to be tested, so that the computer system can autonomously identify the rotating body to be tested and its structural information, calculate the density of the rotating body to be tested based on the three-dimensional model and the mass, identify the material of the rotating body to be tested by density, and input the three-dimensional model into the pre-trained component segmentation model corresponding to the name to obtain an independent model of each component to achieve automatic segmentation. For complex rotating bodies to be tested, before the dynamic imbalance detection, the dimensions of each component need to be recorded, and the balancing machine needs to set the test speed according to the height difference of each part of the component. The setting of the test speed must ensure that the dynamic imbalance force at different height positions can be accurately captured when the rotating body to be tested reaches the running state.

[0037] Then, by analyzing the data obtained above, the calibration parameters of the balancing machine can be automatically adjusted to achieve high-precision balancing tests, avoiding incomplete or erroneous data entry, effectively reducing human intervention, improving calibration efficiency and accuracy, and making the dynamic detection balancing machine more intelligent and easy to use. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a schematic diagram of an automatic testing system for a balancing machine for dynamic balancing of a rotating body according to Embodiment 1 of the present invention;

[0039] Figure 2 This is a schematic diagram of an automatic testing system for a balancing machine for dynamic balancing of a rotating body according to Example 2 of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] Example 1

[0042] See also Figure 1 As shown, the automatic testing system of a balancing machine for dynamic balancing of a rotating body described in this embodiment includes a trial operation setting module, a first acquisition module, a preprocessing module, a model building module, a first analysis module, a second acquisition module, a second analysis module, a third analysis module, a segmentation analysis module and an adaptive calibration module, and each module is connected by wire and / or wirelessly.

[0043] The trial operation setting module is used to set the drive shaft speed during the trial operation phase of the balancing machine, which is marked as the trial operation speed. The trial operation phase is the phase before the balancing machine formally performs a dynamic balancing test on the rotating body to be tested. The trial operation phase is used to obtain relevant data of the rotating body to be tested. The drive shaft speed during the trial operation phase is a preset value.

[0044] The first acquisition module is used to collect the environmental data of the balancing machine, including temperature, humidity and noise. Changes in temperature, humidity and noise will affect the accuracy of the balancing machine sensor, especially the vibration sensor. High temperature may slow down the sensor's response speed, and humidity will cause unstable sensor data. If environmental calibration is not performed, the measured data may deviate greatly, resulting in sensor inaccuracy and data errors, affecting the test results.

[0045] The preprocessing module is used to extract features from environmental data to obtain environmental feature data, which includes temperature average, temperature standard deviation, humidity average, humidity standard deviation, noise average and noise standard deviation.

[0046] Methods for obtaining environmental characteristic data include:

[0047] The temperatures collected during the trial run phase are used to establish a temperature analysis set, and the temperature average and temperature standard deviation of the temperature analysis set are calculated; the humidity collected during the trial run phase is used to establish a humidity analysis set, and the humidity average and humidity standard deviation of the humidity analysis set are calculated; the noise collected during the trial run phase is used to establish a noise analysis set, and the noise average and noise standard deviation of the noise analysis set are calculated.

[0048] The model building module is used to collect point cloud data of the rotating body to be detected during the operation stage, and to build a three-dimensional model of the rotating body to be detected based on the point cloud data; the point cloud data is scanned by a high-frequency laser beam of a laser scanner or an optical sensor on the surface of the rotating body to be detected, and the reflection time and angle are recorded to form point cloud data. The point cloud data can represent the three-dimensional coordinates of multiple points on the surface of the rotating body to be detected.

[0049] The specific construction methods of the 3D model include:

[0050] Filter out noise data; align point cloud data collected from multiple angles to ensure that data from different perspectives can be correctly integrated.

[0051] Use the Delaunay triangulation algorithm or other mesh generation algorithms to convert point cloud data into a triangular mesh surface to form the three-dimensional geometric shape of the object; use the fitting algorithm to fit the discrete points on the point cloud surface into a continuous smooth surface to generate a smooth three-dimensional model.

[0052] The first analysis module is used to determine the radius of the rotating body to be detected based on the three-dimensional model (the three-dimensional model can reflect the radius of the rotating body to be detected), and calculate the mass of the rotating body to be detected based on the radius, trial running speed and centrifugal force; the mass of the rotating body to be detected will affect its behavior during rotation. For example, a large fan impeller and a small motor rotor have completely different mass distributions. If targeted calibration is not performed, the balancing machine cannot identify the mass difference between different objects, resulting in inaccurate measurement results; for example, objects with larger mass require more sensitive sensor settings during calibration, while objects with smaller mass require different measurement ranges.

[0053] The second acquisition module is used to collect the centrifugal force of the rotating body to be detected when the driving shaft speed reaches the trial operation speed during the operation stage.

[0054] Methods for calculating the mass of the rotating body to be detected based on the radius and centrifugal force include:

[0055] ω1=2πn;

[0056] In the formula, m is mass, F is centrifugal force, r is radius, ω1 is test run angular velocity, π is pi, and n is test run speed. Calculating the mass of the rotating body to be tested by radius and centrifugal force can facilitate objects that cannot be accurately measured by static methods, especially large objects or objects that are inconvenient to install. Secondly, for the rotating body to be tested that needs to be weighed before the test, there is no need to weigh it with other weighing equipment, which simplifies the test process, improves test efficiency, and reduces errors caused by manual participation.

[0057] The second analysis module is used to input the three-dimensional model into the rotating body recognition model to obtain the name and components of the rotating body to be detected.

[0058] The training method of the rotating body recognition model includes:

[0059] Collect b groups of three-dimensional models in advance, set corresponding names and components for the b groups of three-dimensional models, b is an integer greater than 1, and convert the three-dimensional models and the corresponding names and components into a corresponding set of feature vectors.

[0060] Each group of feature vectors is used as the input of a rotating body recognition model. The rotating body recognition model takes a group of predicted names and components corresponding to each group of three-dimensional models as output, and takes the actual names and components corresponding to each group of three-dimensional models as prediction targets, where the actual names and components are pre-set names and components corresponding to the three-dimensional models; minimizing the sum of prediction errors of all three-dimensional models is used as a training target; the rotating body recognition model is trained until the sum of prediction errors reaches convergence and the training is stopped.

[0061] The above-mentioned rotating body recognition model is specifically a deep neural network model; it includes an input layer, a hidden layer and an output layer; each hidden layer includes multiple neurons, each neuron is connected to the neurons in the next layer, and the connection contains weights, which determine the importance and influence of data transmission in the neural network; an activation function is applied to each neuron between the hidden layer and the output layer, and the activation function introduces nonlinearity, allowing the network to learn more complex patterns and features.

[0062] The third analysis module obtains the volume of the rotating body to be detected based on the three-dimensional model, calculates the density of the rotating body to be detected based on the volume and mass, and identifies the material of the rotating body to be detected through the density. The vibration characteristics of different materials are different. For example, metal materials are usually more rigid and have smaller vibration amplitudes, while composite materials may have larger vibration amplitudes. If the balancing machine is not calibrated according to the material properties, it will not be able to accurately identify the vibration characteristics of the composite materials, resulting in large errors in the measurement results. During correction, the rotating body to be detected will not work properly due to the errors, which will ultimately affect the balance and performance of the rotating body to be detected.

[0063] The segmentation analysis module inputs the 3D model into the pre-trained component segmentation model corresponding to the name, obtains an independent model of each component, and realizes automated segmentation; determines the dimensional data of each component according to the independent model of each component; the dimensional data includes height and length. For complex rotating bodies to be tested, it is necessary to record the dimensions of each component before dynamic imbalance detection. The balancing machine needs to set the test speed according to the height difference of each part of the component. The setting of the test speed must ensure that the dynamic imbalance force at different height positions can be accurately captured when the rotating body to be tested reaches the running state, thereby improving the detection accuracy.

[0064] The purpose of analyzing and segmenting the components through the component segmentation model is that the three-dimensional model is constructed by collecting point cloud data to construct the rotating body to be detected. Therefore, as a computer system, it is impossible to independently identify the area of ​​each component in the three-dimensional model. Therefore, it is necessary to identify it through the component segmentation model before the computer system can independently complete the segmentation, so as to record the size of each component next.

[0065] The training method of the component segmentation model includes:

[0066] Collect H groups of 3D models with component area annotations in advance, such as the 3D model of the impeller, and annotate the shaft and blade areas of the 3D model. For example, use 3D modeling software (such as Blender, MeshLab) to mark the areas of the 3D model; save the annotated 3D model in a format that can be understood by a computer (such as labeled point cloud data), in which each vertex or triangle facet of the 3D model has a corresponding label (shaft or blade).

[0067] The H groups of 3D models with component annotations are divided into training sets and test sets. The 3D models in the training set are used as the input of the component segmentation model, and the component regions in the training set are used as the output of the component segmentation model. The component segmentation model is tested using the test set, and the component segmentation model that meets the prediction accuracy is output. The component segmentation model uses a convolutional neural network model (CNN) graph neural network model.

[0068] The adaptive calibration module selects the parameter calibration model corresponding to the name, inputs the environmental characteristic data, radius, mass, and the dimension data of each component into the parameter calibration model, and obtains the balancing machine parameter calibration data corresponding to the name, and the parameter calibration data includes the sensor sensitivity setting data, the test speed, and the imbalance tolerance setting data. The system automatically adjusts the sensor sensitivity and the speed imbalance tolerance to calibrate the balancing machine parameters according to the predicted balancing machine parameter calibration data, and then starts the formal measurement.

[0069] The test speed refers to the speed of the balancing machine when the dynamic balance test is performed on the rotating body to be tested. The rotating bodies to be tested with different characteristics need to be measured at different speeds. Larger objects (the size can be known from the volume and radius of the rotating body to be tested) require a lower speed to ensure safety, while smaller objects can run at a higher speed. During calibration, the balancing machine will adjust the speed according to the characteristics of the rotating body to be tested to avoid the balancing machine measuring at the wrong speed, resulting in inaccurate results.

[0070] Unbalance tolerance refers to the vibration generated by an object when it rotates. Larger or more massive objects require a larger amount of imbalance, and vice versa. During calibration, the balancing machine needs to know the imbalance tolerance of the object to be tested in order to provide appropriate correction solutions. If it is not calibrated according to the characteristics of the object, the balancing machine may incorrectly judge the severity of the imbalance.

[0071] This embodiment sets a trial operation stage before the formal test. During the trial operation stage, the environmental data of the balancing machine is collected, a three-dimensional model of the rotating body to be detected is constructed, the radius of the rotating body to be detected is obtained, and the mass of the rotating body to be detected is calculated according to the radius, the trial operation speed and the centrifugal force. There is no need to weigh it through other weighing equipment, which simplifies the test process, improves the test efficiency, and reduces the error caused by manual participation. The three-dimensional model is input into the rotating body recognition model to obtain the name and components of the rotating body to be detected, so that the computer system can autonomously identify the rotating body to be detected and its structural information, calculate the density of the rotating body to be detected according to the three-dimensional model and the mass, identify the material of the rotating body to be detected by the density, and input the three-dimensional model into the pre-trained component segmentation model corresponding to the name to obtain an independent model of each component, thereby realizing automatic segmentation. For a complex rotating body to be detected, before the dynamic imbalance detection, the size of each component needs to be recorded, and the balancing machine needs to set the test speed according to the height difference of each part of the component. The setting of the test speed must ensure that the dynamic unbalanced force at different height positions can be accurately captured when the rotating body to be detected reaches the running state.

[0072] Then, by analyzing the data obtained above, the calibration parameters of the balancing machine can be automatically adjusted to achieve high-precision balancing tests, avoiding incomplete or erroneous data entry, effectively reducing human intervention, improving calibration efficiency and accuracy, and making the dynamic detection balancing machine more intelligent and easy to use.

[0073] Example 2

[0074] See also Figure 2 As shown, the present application also provides an automatic testing system for a balancing machine for dynamic balancing of a rotating body, which is used to solve the problem of early warning before officially entering the test phase when the predicted test speed exceeds the rated speed of the balancing machine, or when the rotating body to be tested reaches the predicted test speed and the test centrifugal force generated exceeds the centrifugal force that the balancing machine can withstand, so as to avoid the test causing substantial damage to the balancing machine. The automatic testing system for the balancing machine also includes an early warning module.

[0075] The early warning module is used to determine whether to generate an early warning instruction based on the predicted test speed and the test centrifugal force generated when the rotating body to be detected reaches the predicted test speed; if the early warning instruction is generated, the detection of the rotating body to be detected is stopped to avoid damage to the balancing machine after entering the test phase.

[0076] Methods for determining whether to generate an early warning instruction include:

[0077] If the predicted test speed exceeds the rated speed of the balancing machine, an advance warning instruction is generated, otherwise no early warning instruction is generated.

[0078] If the rotating body to be tested reaches the predicted test speed and the test centrifugal force generated exceeds the centrifugal force borne by the balancing machine, an early warning instruction is generated, otherwise no early warning instruction is generated.

[0079] ω=2πz;

[0080] To test the centrifugal force, m is the mass of the rotating body to be tested, r is the radius of the rotating body to be tested, ω2 2 is the test angular velocity, π is the pi, and z is the test rotation speed.

[0081] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0082] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An automatic testing system for a balancing machine for dynamic balancing of a rotating body, characterized in that: include: The trial operation setting module is used to set the drive shaft speed during the trial operation phase of the balancing machine, which is marked as the trial operation speed; The first acquisition module is used to collect the environmental data of the balancing machine. A preprocessing module is used to extract features from environmental data to obtain environmental feature data; A model building module is used to collect point cloud data of the rotating body to be detected during the operation phase, and to build a three-dimensional model of the rotating body to be detected based on the point cloud data; The second acquisition module acquires the centrifugal force of the rotating body to be detected when the driving shaft speed reaches the test running speed; The first analysis module is used to determine the radius of the rotating body to be detected according to the three-dimensional model, and calculate the mass of the rotating body to be detected according to the radius, the test running speed and the centrifugal force; The second analysis module is used to input the three-dimensional model into the rotating body recognition model to obtain the name and components of the rotating body to be detected; The third analysis module obtains the volume of the rotating body to be detected according to the three-dimensional model, and calculates the density of the rotating body to be detected according to the volume and mass; A segmentation analysis module, which inputs the three-dimensional model into a pre-trained component segmentation model corresponding to the name, obtains an independent model of each component, and determines the size data of each component according to the independent model; The adaptive calibration module selects a parameter calibration model corresponding to the name, inputs environmental characteristic data, radius, mass, and dimension data of each component into the parameter calibration model, obtains balancing machine parameter calibration data, and calibrates the balancing machine.

2. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 1, characterized in that: The training method of the rotating body recognition model includes: Collecting b groups of three-dimensional models in advance, setting corresponding names and components for the b groups of three-dimensional models, where b is an integer greater than 1, and converting the three-dimensional models, the corresponding names and components into a corresponding set of feature vectors; Each group of feature vectors is used as the input of the rotating body recognition model. The rotating body recognition model takes a group of predicted names and components corresponding to each group of three-dimensional models as output, and takes the actual names and components corresponding to each group of three-dimensional models as prediction targets; minimizing the sum of prediction errors of all three-dimensional models is used as the training target; the rotating body recognition model is trained until the sum of prediction errors reaches convergence and the training is stopped; the rotating body recognition model is a deep neural network model.

3. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 1, characterized in that: Methods for calculating the mass of the rotating body to be detected based on the radius and centrifugal force include: In the formula, m is mass, F is centrifugal force, r is radius, ω1 is test run angular velocity, π is pi, and n is test run speed.

4. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 1, characterized in that: The training method of the component segmentation model includes: Collect H groups of three-dimensional models with component area annotations in advance, and save the annotated three-dimensional models in a format that can be understood by computers; divide the H groups of three-dimensional models with component area annotations into a training set and a test set, use the three-dimensional models in the training set as the input of a component segmentation model, use the component area in the training set as the output of the component segmentation model, use the test set to test the component segmentation model, and output a component segmentation model that meets the prediction accuracy, wherein the component segmentation model is a convolutional neural network model or a graph neural network model.

5. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 1, characterized in that: The environmental data include temperature, humidity and noise; the environmental characteristic data include temperature average, temperature standard deviation, humidity average, humidity standard deviation and noise average and noise standard deviation.

6. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 1, characterized in that: The method for obtaining the environmental characteristic data includes: The temperatures collected during the trial run phase are used to establish a temperature analysis set, and the temperature average and temperature standard deviation of the temperature analysis set are calculated; the humidity collected during the trial run phase is used to establish a humidity analysis set, and the humidity average and humidity standard deviation of the humidity analysis set are calculated; the noise collected during the trial run phase is used to establish a noise analysis set, and the noise average and noise standard deviation of the noise analysis set are calculated.

7. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 1, characterized in that: The parameter calibration data includes sensor sensitivity setting data, test rotation speed, and imbalance tolerance setting data.

8. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 1, characterized in that: It also includes an early warning module, which is used to determine whether to generate an early warning instruction according to the predicted test speed and the test centrifugal force generated when the rotating body to be detected reaches the predicted test speed; If an early warning instruction is generated, the detection of the rotating body to be detected is stopped.

9. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 8, characterized in that: Methods for determining whether to generate an early warning instruction include: If the predicted test speed exceeds the rated speed of the balancing machine, an early warning instruction is generated, otherwise no early warning instruction is generated; If the rotating body to be tested reaches the predicted test speed and the test centrifugal force generated exceeds the centrifugal force borne by the balancing machine, an early warning instruction is generated, otherwise no early warning instruction is generated.

10. The automatic testing system for a balancing machine for dynamic balancing of a rotating body according to claim 9, characterized in that: The test centrifugal force calculation method comprises: To test the centrifugal force, m is the mass of the rotating body to be tested, r is the radius of the rotating body to be tested, ω2 2 is the test angular velocity, π is the pi, and z is the test rotation speed.

Images