Aluminum alloy radiator laser powder bed melting additive manufacturing online acoustic monitoring system and device
By using an online acoustic monitoring system to identify defects in the laser powder bed melting process of aluminum alloy heat sinks in real time, the problem of defect detection in parts in powder-laying laser additive manufacturing has been solved, improving production efficiency and yield, and meeting airworthiness requirements.
Patent Information
- Application Number
- CN202410813210.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-24
- Publication Date
- 2025-12-26
AI Technical Summary
In existing technologies, powder-spread laser additive manufacturing parts are prone to defects such as holes, lack of fusion, and cracks. Furthermore, offline inspection methods result in long delivery cycles and high costs, and cannot achieve 100% non-destructive testing.
An online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks is adopted. It includes acoustic sensors, data acquisition and processing modules, defect identification modules, early warning feedback modules, and three-dimensional model storage modules. It identifies defects and issues early warnings by monitoring acoustic signals in real time, and optimizes production parameters in combination with laser and powder supply systems.
It enables real-time detection of anomalies in the production process, improves detection accuracy, reduces material waste, lowers scrap rates and production costs, and meets airworthiness requirements.
Smart Images
Figure CN121199136A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of additive manufacturing, more particularly, the present application relates to an aluminum alloy radiator laser powder bed fusion additive manufacturing online acoustic monitoring system and device. BACKGROUND
[0002] In recent years, laser additive manufacturing technology has developed rapidly. Compared with traditional forming technology, laser additive manufacturing technology is more suitable for manufacturing complex metal parts, can shorten the production cycle of parts, and reduce production cost. At present, powder laying laser additive manufacturing technology has been widely used and influenced in the aerospace industry. Many aerospace manufacturers have included additive manufacturing technology in their manufacturing business to produce complex aircraft components that are difficult to manufacture by traditional manufacturing processes. However, due to the influence of powder characteristics, process parameters, external environment and other factors, powder laying laser additive manufacturing parts are prone to defects such as holes, incomplete fusion, cracks, etc., thereby affecting the product yield and process stability of the parts. According to the airworthiness regulation, the civil aircraft additive manufacturing complex structure parts must be 100% non-destructive tested.
[0003] At present, the commonly used non-destructive testing methods are mainly offline testing, and the offline testing methods include ultrasonic, X-ray, eddy current, CT testing, etc. However, offline testing leads to long part delivery cycle and high cost, and due to the shielding of complex structure, not all offline testing methods can achieve 100% testing of the parts. Therefore, in order to improve the accuracy of non-destructive testing and manufacture high-quality aircraft structural parts, online monitoring of the powder laying laser additive manufacturing process is very important.
[0004] Online monitoring can timely find defects. When the defects are located on the outer surface of the semi-finished workpiece, the local defects in the manufacturing process can be eliminated by re-melting the defect position, thereby improving the acceptance rate of the parts and shortening the manufacturing cycle. At the same time, it can meet the requirements of airworthiness regulations and realize airworthiness installation. However, only testing the outer surface of the workpiece, the cooperation between each layer during production, such as the generation of connection gap, etc. cannot be determined, which will also cause defects in the completed product.
[0005] Therefore, the present application provides an aluminum alloy radiator laser powder bed fusion additive manufacturing online acoustic monitoring system and device. SUMMARY
[0006] In order to overcome the above-mentioned defects of the prior art, the present application provides an aluminum alloy radiator laser powder bed fusion additive manufacturing online acoustic monitoring system and device to solve the problems raised in the above background.
[0007] In order to achieve the above object, the present application provides the following technical scheme: an online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy radiators, comprising an acoustic sensor, a data acquisition and processing module, a defect identification module, an early warning feedback module, a three-dimensional model storage module and a comparison module, the acoustic sensor is connected with the data acquisition and processing module, the data acquisition and processing module is connected with the defect identification module, the defect identification module is connected with the early warning feedback module, the three-dimensional model storage module is connected with the comparison module, and the comparison module is connected with the defect identification module;
[0008] The acoustic sensor is used for capturing acoustic signals generated in the laser powder bed fusion process.
[0009] The data acquisition and processing module is used for digitizing the signals collected by the acoustic sensor.
[0010] The defect identification module is used for establishing an identification model to perform pattern recognition and defect classification on the processed acoustic signals.
[0011] The early warning feedback module is used for early warning of the identified defects and feedback of specific defect information.
[0012] The three-dimensional model storage module is used for storing the slice data converted from the three-dimensional model of the aluminum alloy radiator.
[0013] The comparison module is used for comparing the processed acoustic signal data with the slice data converted from the three-dimensional model of the actual aluminum alloy radiator identified by the identification model.
[0014] Preferably, the data acquisition and processing module running process comprises:
[0015] The received acoustic signals are amplified, filtered and denoised:
[0016] Then the feature vectors are extracted from the processed acoustic signals for input into the identification model to identify the defects.
[0017] Preferably, the identification model comprises an input layer, a hidden layer and an output layer, the input layer is connected with the hidden layer, and the hidden layer is connected with the output layer.
[0018] The input layer is used for receiving the feature vectors extracted in the data acquisition and processing module, the features including the frequency, amplitude and duration of the acoustic signals, and the number of input layer neurons matches the dimension of the feature vectors.
[0019] The hidden layer is used for extracting useful features from the input data and transmitting them to the output layer, and the number of hidden layers is the same as the number of extracted useful features.
[0020] The output layer outputs the prediction result of the defect type according to the features transmitted by the hidden layer, by first calculating the original output of each feature vector and then calculating the probability of each category.
[0021] Preferably, the specific training process of the identification model is:
[0022] First, collect acoustic signal samples containing various defect types, perform feature extraction and labeling, and divide the data set into a training set, a validation set, and a test set.
[0023] During the training process, the neural network maps the input data to the output data through forward propagation. In each neuron, the input data is multiplied by the weight and added to the bias, and then it is nonlinearly transformed by the activation function.
[0024] The loss function is calculated according to the output of the neural network and the true label, and the loss function is used to measure the difference between the neural network prediction result and the true result.
[0025] The gradient of the loss function with respect to the neural network parameters is calculated by the backpropagation algorithm, and the weights and biases are updated using the optimization algorithm to minimize the loss function.
[0026] Finally, repeat the above training process until the performance of the neural network on the validation set reaches the optimal or reaches the preset number of iterations.
[0027] Preferably, the output formula of each feature vector of the hidden layer is:
[0028]
[0029] where zj is the output of the jth feature vector in the hidden layer; wij is the weight from the ith feature vector in the input layer to the jth feature vector in the hidden layer; xi is the value of the ith feature vector in the input layer; bj is the bias of the jth node in the hidden layer; and n is the number of feature vectors in the input layer. i j
[0030] Preferably, the activation value formula of the hidden layer is:
[0031] aj = f(zj)
[0032] where f(·) is an activation function, and the specific formula of the activation function is:
[0033]
[0034] Preferably, the specific prediction formula of the output layer is: wherein the output formula of each feature vector of the hidden layer is:
[0035]
[0036] where y is the predicted value, wij is the weight from the jth feature vector of the hidden layer to the output layer; a j is the activation value of the jth feature vector of the hidden layer; b k is the bias of the output layer; m is the feature vector of the hidden layer.
[0037] Preferably, the formula for calculating the original output of each feature vector in the output layer is:
[0038]
[0039] Preferably, the formula for calculating the probability of each category in the output layer is:
[0040]
[0041] where P(y=Ck) is the probability of predicting the kth category; zk is the original output of the kth feature vector in the output layer; C is the total number of categories.
[0042] An aluminum alloy radiator laser powder bed fusion additive manufacturing online acoustic monitoring device, characterized by comprising a laser, an online acoustic monitoring system, a powder supply system, a forming bin and a workbench, the forming bin is arranged on the workbench, one side of the forming bin is connected with the powder supply system, the top of the forming bin is provided with the laser, the laser is connected with the online acoustic monitoring system, and the online acoustic monitoring system is connected with the powder supply system.
[0043] The technical effects and advantages of the present application are:
[0044] 1. By monitoring the process of laser melting metal powder in real time, the extracted acoustic wave features can be analyzed according to the extracted acoustic wave data, the error between the predicted value and the actual value is reduced, the detection accuracy is improved, so that abnormal conditions in the production process can be found in time, and the connection between each layer can be analyzed, so that unqualified products can be avoided, and material waste can be reduced;
[0045] 2. By connecting and cooperating the laser, the online acoustic monitoring system and the powder supply system, the production parameters can be monitored and adjusted in real time, the production process can be optimized, the production efficiency can be improved, the waste rate and the production cost can be reduced, and the overall economic benefit can be improved. BRIEF DESCRIPTION OF DRAWINGS
[0046] Figure 1 It is a schematic diagram of the overall system structure of the present application.
[0047] Figure 2 It is a schematic diagram of the system working step flow of the present application. DETAILED DESCRIPTION
[0048] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.
[0049] As shown in the accompanying drawings Figure 1 and Figure 2 The aluminum alloy radiator laser powder bed fusion additive manufacturing online acoustic monitoring system comprises an acoustic sensor, a data acquisition and processing module, a defect identification module, an early warning feedback module, a three-dimensional model storage module and a comparison module, the acoustic sensor is connected with the data acquisition and processing module, the data acquisition and processing module is connected with the defect identification module, the defect identification module is connected with the early warning feedback module, the three-dimensional model storage module is connected with the comparison module, and the comparison module is connected with the defect identification module.
[0050] The acoustic sensor is used for capturing acoustic signals generated in the laser powder bed fusion process.
[0051] The data acquisition and processing module is used for digitizing the signals collected by the acoustic sensor.
[0052] The defect identification module is used for establishing an identification model to perform pattern recognition and defect classification on the processed acoustic signals.
[0053] The early warning feedback module is used for early warning of the identified defects and feedback of specific defect information.
[0054] The three-dimensional model storage module is used for storing slice data converted from the three-dimensional model of the aluminum alloy radiator.
[0055] The comparison module is used for comparing the processed acoustic signal data with the slice data converted from the three-dimensional model of the actual aluminum alloy radiator identified by the identification model.
[0056] In specific implementation, first, a three-dimensional model of a qualified product storage aluminum alloy heat sink is designed, and is converted into slice data by computer-aided software and stored in a three-dimensional model storage module. In operation, an acoustic sensor is used to capture acoustic signals generated in the laser powder bed melting process, a data acquisition and processing module is used to digitize the signals collected by the acoustic sensor, a defect recognition module is used to perform pattern recognition and defect classification on the processed acoustic signals, actual model data of the aluminum alloy heat sink in production is recognized, and a comparison module is used to compare the actual model data of the aluminum alloy heat sink with slice data converted from the three-dimensional model of the actual aluminum alloy heat sink recognized by the recognition model, to reflect differences and determine difference points, and a warning feedback module is used to warn the recognized defects and feed back specific defect information.
[0057] In one embodiment, the data acquisition and processing module running process includes:
[0058] The received acoustic signals are amplified, filtered and denoised:
[0059] Then, feature vectors are extracted from the processed acoustic signals for input into the recognition model for defect recognition.
[0060] In one embodiment, the recognition model includes an input layer, a hidden layer and an output layer, the input layer is connected to the hidden layer, and the hidden layer is connected to the output layer.
[0061] The input layer is used to receive the feature vectors extracted by the data acquisition and processing module, the features include frequency, amplitude, duration, etc. of the acoustic signals, and the number of input layer neurons matches the dimension of the feature vectors.
[0062] The hidden layer is used to extract useful features from the input data and pass them to the output layer, and the number of hidden layers is the same as the number of useful features extracted.
[0063] The output layer outputs the prediction result of the defect type according to the features passed by the hidden layer, by first calculating the original output of each feature vector, and then calculating the probability of each category.
[0064] In one embodiment, the specific training process of the recognition model is as follows:
[0065] First, acoustic signal samples containing various defect types are collected, and feature extraction and labeling are performed, and the data set is divided into a training set, a validation set and a test set.
[0066] During training, the neural network maps input data to output data through forward propagation, where at each neuron, the input data is multiplied by weights and added to a bias, then passed through an activation function for non-linear transformation.
[0067] The loss function is calculated based on the output of the neural network and the true labels, which measures the difference between the neural network's prediction and the true result.
[0068] The gradient of the loss function with respect to the neural network parameters is calculated through the backpropagation algorithm, and the weights and biases are updated using optimization algorithms to minimize the loss function.
[0069] Finally, the above training process is repeated until the performance of the neural network on the validation set is optimal or reaches the preset number of iterations.
[0070] In one embodiment, the output formula of each feature vector in the hidden layer is:
[0071]
[0072] where zj is the output of the jth feature vector in the hidden layer; wij is the weight from the ith feature vector in the input layer to the jth feature vector in the hidden layer; x i is the value of the ith feature vector in the input layer; b j is the bias of the jth node in the hidden layer; n is the number of feature vectors in the input layer.
[0073] In one embodiment, the activation value formula of the hidden layer is:
[0074] aj=f(zj)
[0075] where f(·) is the activation function, and the specific formula of the activation function is:
[0076]
[0077] In one embodiment, the specific prediction formula of the output layer is: in which the output formula of each feature vector in the hidden layer is:
[0078]
[0079] where y is the predicted value, wij is the weight from the jth feature vector in the hidden layer to the output layer; a j is the activation value of the jth feature vector in the hidden layer; b k is the bias of the output layer; m is the number of feature vectors in the hidden layer.
[0080] In one embodiment, the original output formula of each feature vector in the output layer is calculated as:
[0081]
[0082] In one embodiment, the formula for calculating the probability of each class in the output layer is:
[0083]
[0084] where P(y=Ck) is the probability of being predicted as the kth class; zk is the original output of the kth feature vector in the output layer; C is the total number of classes.
[0085] An online acoustic monitoring device for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks, characterized in that it comprises a laser, an online acoustic monitoring system, a powder supply system, a forming bin and a workbench, the forming bin is arranged on the workbench, one side of the forming bin is connected with the powder supply system, the top end of the forming bin is provided with the laser, the laser is connected with the online acoustic monitoring system, and the online acoustic monitoring system is connected with the powder supply system.
[0086] In specific implementation, in the additive manufacturing process of aluminum alloy heat sinks and other metal parts, the laser powder bed fusion additive manufacturing device scans a cross section of the part according to a predetermined trajectory, melts the metal powder and makes it solidify rapidly. At the same time, the acoustic monitoring system monitors the acoustic changes in the manufacturing process in real time and transmits the data to the data processing and analysis system. The system identifies potential defects by analyzing the changes in acoustic signals and feeds back the results to the operator or the automatic control system. The operator can make corresponding processing according to the prompt of the system, such as adjusting the manufacturing parameters, suspending the manufacturing process, etc., to ensure the manufacturing quality of aluminum alloy heat sinks and other metal parts.
[0087] Finally, it should be noted that: the above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.
Claims
1. An online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks, characterized in that, It includes an acoustic sensor, a data acquisition and processing module, a defect identification module, an early warning feedback module, a three-dimensional model storage module, and a comparison module. The acoustic sensor is connected to the data acquisition and processing module, the data acquisition and processing module is connected to the defect identification module, the defect identification module is connected to the early warning feedback module, the three-dimensional model storage module is connected to the comparison module, and the comparison module is connected to the defect identification module. The acoustic sensor is used to capture acoustic signals generated during the laser powder bed melting process; The data acquisition and processing module is used to digitize the signals acquired by the acoustic sensor; The defect identification module is used to establish an identification model to perform pattern recognition and defect classification on the processed acoustic signal; The early warning feedback module is used to issue early warnings for identified defects and provide feedback on specific defect information; The three-dimensional model storage module is used to store slice data converted from the three-dimensional model of the aluminum alloy heat sink; The comparison module is used to compare the processed acoustic signal data with the slice data converted from the three-dimensional model of the actual aluminum alloy heat sink identified by the recognition model.
2. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 1, characterized in that, The data acquisition and processing module operation flow includes: The received acoustic signal is amplified and then filtered and denoised. Then, feature vectors are extracted from the processed acoustic signals and used as inputs into the recognition model to identify defects.
3. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 2, characterized in that, The recognition model includes an input layer, a hidden layer, and an output layer, wherein the input layer is connected to the hidden layer, and the hidden layer is connected to the output layer; The input layer is used to receive feature vectors extracted by the data acquisition and processing module. The features include the frequency, amplitude, and duration of the sound wave signal. The number of neurons in the input layer matches the dimension of the feature vector. The hidden layers are used to extract useful features from the input data and pass them to the output layer. The number of hidden layers is the same as the number of useful features extracted. The output layer outputs the prediction result of the defect type based on the features passed from the hidden layer. This is done by first calculating the original output of each feature vector and then calculating the probability of each category.
4. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 3, characterized in that, The specific training process of the recognition model is as follows: First, we collect acoustic signal samples containing various defect types, extract and label their features, and divide the dataset into training, validation and test sets. During training, the neural network maps input data to output data through forward propagation. In each neuron, the input data is multiplied by weights and a bias is added, and then a nonlinear transformation is performed through an activation function. The loss function is calculated based on the output of the neural network and the true label. The loss function measures the difference between the neural network's prediction and the true result. The gradient of the loss function with respect to the neural network parameters is calculated using the backpropagation algorithm, and the weights and biases are updated using an optimization algorithm to minimize the loss function. Finally, repeat the above training process until the neural network achieves optimal performance on the validation set or reaches the preset number of iterations.
5. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 4, characterized in that, The output formula for each feature vector of the hidden layer is: Where zj is the output of the j-th feature vector in the hidden layer; wij is the weight from the i-th feature vector in the input layer to the j-th feature vector in the hidden layer; x i It is the value of the i-th feature vector in the input layer; b j is the bias of the j-th node in the hidden layer; n is the number of feature vectors in the input layer.
6. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 5, characterized in that, The activation value formula for the hidden layer is: aj = f(zj) Where f(·) is the activation function, and the specific formula of the activation function is:
7. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 6, characterized in that, The specific prediction formula for the output layer is as follows: The output formula for each feature vector of the hidden layer is as follows: Where y is the predicted value, wij is the weight from the j-th feature vector in the hidden layer to the output layer; a j b is the activation value of the j-th feature vector in the hidden layer; k is the bias of the output layer; m is the feature vector of the hidden layer.
8. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 7, characterized in that, The original output formula for calculating each feature vector in the output layer is as follows:
9. The online acoustic monitoring system for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks according to claim 8, characterized in that, The formula for calculating the probability of each category in the output layer is as follows: Where P(y=Ck) is the probability of being predicted as the kth class; zk is the original output of the kth feature vector in the output layer; and C is the total number of classes.
10. An online acoustic monitoring device for laser powder bed fusion additive manufacturing of aluminum alloy heat sinks, characterized in that, The device includes a laser, an online acoustic monitoring system, a powder supply system, a molding chamber, and a worktable. The worktable is equipped with a molding chamber, one side of which is connected to the powder supply system. The top of the molding chamber is equipped with a laser, which is connected to the online acoustic monitoring system. The online acoustic monitoring system is connected to the powder supply system.