Abstract: Flexible manufacturing system is to automate material flow and information flow based on processing automation. In the metal cutting process, due to the failure of tool wear and damage can not be found in time, will lead to interruption of the cutting process, resulting in scrapped parts or machine tool damage, or even the entire flexible manufacturing system to stop running, causing great economic losses, and therefore should An on-line detection and monitoring device is provided in a flexible manufacturing system. To this end, the performance and composition of the tool management system in the flexible manufacturing system are discussed, and a method for detecting the tool in a flexible manufacturing system using a neural network is introduced.
Keywords: computer application; flexible manufacturing system; theoretical research; tool management system; wear and damage of tool; neural network
CLC number: TP108 Document code: B Article ID: 1003─188X(2005)03─0119─02
In a flexible manufacturing system, there are many types of parts to be machined, the parts processing process is more complicated, and the processes are highly concentrated, and the types, sizes, and quantities of tools required are many. As the parts to be machined change and the tool wears and breaks, it is necessary to force the tool change and random tool change at regular intervals. During the operation of the system, the tools are frequently exchanged between the machine tools, the machine tool and the tool magazine, and the transportation, management and monitoring of the tool flow are very complicated. Therefore, there is a need for an advanced, practical, and fully functional tool management system to realize tasks such as scheduling, storage, and information management of tools in a flexible manufacturing system.
1 The tool management system should have the performance
The tool management system should have the following features: First, to manage a large number of tools, usually hundreds to thousands of tools; Second, a high degree of automation of tool delivery, high-performance, intelligent robot as a tool for the flow of tools; The third is to collect tool information automatically and accurately; the fourth is to use a large database to achieve the optimization of tool scheduling and dynamic and static management; Fifth, automatic on-line detection of tool life and tool wear, damage management, and can achieve online tool change function. For this purpose, a tool management system in a flexible component processing system for a key component will be described as an example.
2 The composition of the tool management system
The system uses VB6.0 programming system software as a development platform and consists of 6 program blocks and 5 external files.
2.1 Program blocks
2.1.1 The user login module mainly implements protection measures for the system, prevents illegal users from entering the system, and ensures the security of system data and system operation.
2.1.2 System Operation The control module mainly implements system initialization to complete start-stop control, task queuing, and interface management for various modules of online tool management, and confirms the information of the machining center tool magazine and the central tool magazine.
2.1.3 Tool offline management module mainly realizes the management of tool magazine, performs tool demand analysis, tool assembly planning, tool grinding and presetting, tool code generation, paste and information input, tool component management, and tool purchase List, provide the necessary tools to the line in time.
2.1.4 Tool online management module mainly realizes the management of tool activities in the flexible manufacturing system, ensures that the machining center gets the correct tool at the right time, completes the demand and supply of the machining center tool, and uses appropriate strategies to achieve a reasonable choice of tool , scheduling, and the release of tool shipping instructions and calculation of tool life remaining.
2.1.5 The system information management module mainly completes the missing knife inspection, tool storage management (including new tool storage and tool return from the flexible automatic line), tool delivery management, wear and tear tool management and database management.
2.1.6 The system status monitoring module mainly implements the monitoring of tool and equipment faults in the system and provides necessary resource status information for real-time management of the tool management system. The six program blocks are the main components of the system. They are called in accordance with a certain logical sequence, so that the system has a complete and reasonable tool management function. At the same time, it is a mutually independent subroutine module. Data is passed between subroutines. interconnected.
2.2 External Database Files
2.2.1 Tool files, main parameters for all tools, 1 parameter per 1 tool.
2.2.2 Machine Tool File The main parameters of the machine are stored. One type of machine parameter is used as a record.
2.2.3 The process data file stores a variety of process data related to processing, and the data of one processing method is used as a record.
2.2.4 The cutting condition coefficient file stores various factors that determine the cutting conditions.
2.2.5 Abrasion and Damaged Tool Files The main parameters of wear and broken tools are stored. The parameters of each tool are recorded as 1 record.
3 Online inspection of tool wear and breakage
There are many methods for on-line detection of tool wear and breakage, including power detection, acoustic emission detection, learning mode, and force detection. This paper introduces a method of using neural networks to detect tools in flexible manufacturing systems.
3.1 Establishment of tool load model
The load on the tool during cutting is related to many factors. According to the requirements of on-line testing, only a few large influencing factors are considered, namely spindle speed, feed speed, cutting depth, and cutting performance of the machining material. The tool load model is
F = f (s, v, h, m)
In the formula:
F - load vector
h — depth of cut;
s - spindle speed;
m - the cutting performance of the material;
v — Feed amount.
Obviously, the above formula can only show that the load is related to each influencing factor. The corresponding relational formula can be established by the mathematical method of differential geometry or experimental methods, but it is not ideal for on-line detection. Here, the neural network technology is applied to process the load model of the tool.
3.2 Neural Network Technology
The neural network system of this tool load adaptive control adopts a 3-layer BP structure. According to the above analysis, obviously the input layer has 4 nodes and the output layer has 3 nodes, ie the size of the load in the XYZ direction. Taking into account the characteristics of the load adaptive control system, it can be considered that the load is a continuous function of the feedrate. According to the Kolmogorov theorem, the number of nodes in the intermediate hidden layer should be 2 times the number of input points plus 1. Therefore, the neural network structure consists of 4 nodes in the input layer, 9 nodes in the middle layer, and 3 nodes in the output layer. According to the above analysis, each node is given four values, and their different combinations are used as sample input data, so that 256 samples can be obtained. The specific approach is: divide the input into roughly 4 equal parts within the possible range of change, and use the experimental method to measure the load value under each input condition. After obtaining 256 samples, offline learning is performed to obtain the weights of each node. Thus, the learned neural network establishes a corresponding tool load model to provide conditions for on-line inspection of the tool.
3.3 Tool on-line detection principle The principle of on-line tool detection is shown in Figure 1.
First, the cutting depth and feed amount of the tool are measured, and the neural network controller is input together with the spindle speed and the type of machining material. The load calculation is performed by the neural network controller and the resulting load is input to the detector. The result of the detector output is compared with the input signal. If the load exceeds the crack propagation load under the fatigue condition of the tool, the feed rate of the tool is reduced and the feed rate reduction is fed back to the CNC controller. The CNC controller makes corresponding controls so that the size of the load changes to a safe level.
4 Conclusion
In a flexible manufacturing system, a good tool management system and online inspection technology can not only improve processing productivity and reduce labor costs, but also play a key role in optimizing product mix and reducing failure rate.
references:
[1] Zhang Genbao. Automatic Manufacturing System [M]. Beijing: Mechanical Industry Press, 2002.
[2] Zhang Peizhong. Flexible Manufacturing System [M]. Beijing: Mechanical Industry Press, 1998.
[3] Ouyang Puren. FMS Tool Management System Architecture Research [J]. Journal of Nanjing University of Science and Technology, 1996, (6): 32-34.