System for predicting anomalies of machining

1. A system for predicting anomalies of machining such as the anomalies causing a defect on a part machined by a machine tool, comprising:

a signal-processing unit configured to acquire raw data including sensor data and machining parameters from a machine tool and to process the raw data to neuromorphic-circuit input data including features extracted by the signal-processing unit from the raw data;

a neuromorphic circuit connected to the signal-processing unit configured to build a neural network on an integrated circuit, wherein the neural network is trained by training data and the trained neural network is configured to determine anomaly data describing anomalies of the machining in response to the neuromorphic-circuit input data, in particular to the features, wherein each training data set includes a plurality of features and a corresponding anomaly; and

a controller configured to receive the determined anomaly data from the signal-processing unit and to determine at least one action, which can be taken to overcome the determined anomalies, wherein the signal-processing unit is configured to select from the raw data a different combination of data required to determine the anomalies related to different types of machining.

2. The system according to claim 1 , wherein the raw data is obtained during the machining, in particular, without interrupting the machining.

3. The system according to claim 1 , wherein the controller is configured to receive the determined anomaly data and to adjust the machining parameters in response to the determined anomaly data.

4. The system according to claim 1 , wherein the neuromorphic circuit is an integrated circuit and is located in the proximity of the machine tool, preferably the neuromorphic circuit is embedded in the machine tool.

5. The system according to claim 1 , wherein the anomaly data include anomalies indicating a defect of a part machined by a machine tool, and/or defects of the machine tool during operation, in particular the anomaly data comprises the type of the anomalies and the position of the anomalies on the part and/or on the machine tool.

6. The system according to claim 3 , wherein the system further comprises an edge module configured to receive and store one or more of the raw data, anomaly data, and the adjusted machining parameters.

7. The system according to claim 6 , wherein anomalies for one of the following applications can be determined: electro discharge machining, milling, laser machining, and additive manufacturing.

8. The system according to claim 1 , wherein the neuromorphic circuit is configured to store different knowledge spaces for anomaly detection of different machining, wherein the knowledge space defines the correlation between the features provided to the neuro-morphic circuit and the anomaly data.

9. The system according to claim 6 , wherein an analytic unit is included in the edge module and configured to receive part-quality data, which is obtained by inspecting the produced part and to determine optimized training data based on the part-quality data.

10. The system according to claim 9 , wherein the part-quality data is obtained by inspecting a machined part using an optical device.

11. The system according to claim 9 , wherein the optimized training data is used to update the knowledge space stored in the neuromorphic circuit.

12. The system according to claim 6 , wherein the edge module is configured to communicate with a cloud server.

13. A machine tool for machining a part comprising the system according to claim 1 .

14. A method for predicting anomalies of machining, including anomalies causing a defect on a part machined by a machine tool, the method comprising:

acquiring raw data by a signal processing unit, the raw data including sensor data and machining parameters from a machine tool and processing the raw data to neuromorphic-circuit input data including features extracted by the signal processing unit from the raw data;

determining anomaly data related to defects on the part, which is being machined by the machine tool, and/or the defects of the machine tool during operation by a neuro-morphic circuit connected to the signal-processing unit and is configured to build a neural network on an integrated circuit, wherein the neural network is trained by the training data including a plurality of data sets with a plurality of features and a corresponding anomaly, wherein the neural network is configured to determine anomaly data describing anomalies of the machining in response to the neuromorphic-circuit input data, in particular features; and

sending the determined anomaly data by the signal-processing circuit to a controller and determining by the controller at least one action, which can be taken to avoid the determined anomalies, wherein the signal-processing unit is configured to select from the raw data a different combination of data required to determine the anomalies related to different types of machining.

15. The method according to claim 14 , further comprising:

starting machining and performing the anomaly detection according to claim 14 ;

inspecting the part after the machining to generate quality data of the machined part;

comparing the generated quality data with the anomaly data generated during the machining; and

generating optimized training data to train the neuromorphic circuit.