智能制造背景下，切削加工逐步走向高精度、高效能。数字孪生作为新兴概念，可以解决加工中心的切削信息和虚拟模型数据融合困难的问题，加速加工智能化和自动化过程。铣削过程的信号监测的难度较大，有效的温度和振动信号的实时监测在改善加工质量、降低安全风险等方面发挥着重要作用。目前，刀具嵌入热电偶法是主流的铣削温度测量方法，但难以实现存在稳定信号传输和长期供电。对于振动的监测以在机床、工件等位置安装附加外置传感器为主要手段，获取的信号存在一致性差、准确性低、衰减严重等问题，难以满足要求。本文设计了嵌入热电偶的立铣刀和集成数据处理功能的智能刀柄进行切削温度在线感知。在铣刀后刀面近切削刃附近加工凹槽以布置极细热电偶传感器，最大限度接近最大切削温度区；在刀柄本体设计机械结构，集成包括信号采集和ZigBee无线传输等后端处理装置；基于电磁感应，利用主轴的旋转动能设计自供电方案，实现旋转加工过程信号监测的无源化。基于温度感知智能刀柄，进行了不同加工参数的切削实验，验证了测温方案的可行性。通过工件预埋热电偶的方法测量瞬时的铣削最高温度真值，修正有限元仿真模型，得到本方案的测量误差。智能刀柄的分辨率为0.25 ℃，量程0~700 ℃，更新速度为10 Hz，平均误差为8.53%。设计了集成加速度传感器的智能刀柄进行高质量测振。基于BT40-ER32刀柄，选择合理的内腔结构安装传感器，接近振动源而减小失真；优化信号处理装置，设计高度集成的前置调理、高速采集和WiFi传输方案，并实现电路板与刀柄的可靠装配。基于设计的智能测振刀柄，进行了不同加工参数的切削实验探究切削振动的一般规律，并与传统的有线式主轴箱测振和工件测振进行对比实验，用多个指标论证了测振刀柄的优异性能。测振刀柄的分辨率为4.88 mg，量程-50~50 g，更新速度超过10 kHz。针对刀具磨损状态识别进行了实验研究。从时域、频域和小波域进行特征信号的提取，基于费舍尔判别和BP神经网络构建了刀具磨损判别模型，准确率为90.0%。本文分别通过嵌入传感器的方法，设计了无源无线的铣削温度感知刀柄和集成在线处理和无线传输功能的智能测振刀柄，可实现原位、在线的铣削过程的信号实时监测。基于测振刀柄，提出一种刀具磨损状态判别的分类模型并进行了优化。
In the context of intelligent manufacturing, machining is gradually moving towards high-precision, high-efficiency. As a new concept, ‘digital twin’ can help solve the problem of cutting data fusion between machine center in physical world and virtual models, accelerating intelligent and automated procedure. Signal monitoring of milling process is especially hard and effective real-time monitoring of temperature and vibration signals can help improve machining accuracy and reduce machining safety risks. Currently, embedding thermocouple into tool chip is the most common way to measure milling temperature, facing problems of unsteady signal transfer and long-term power supply. The main method for monitoring vibration is to install additional external sensors at machining system such as machine tools and workpieces. The acquired signals cannot meet the usage requirements because of various problems, such as poor consistency, low accuracy, and severe signal attenuation.In this paper, to carry out online temperature sensing, an end milling cutter with built-in thermocouple and an intelligent tool holder integrated with data processing devices are designed. To arrange thin thermocouple sensors, groove is processed on major flank of the milling tool as close to cutting edge as possible to get greatest proximity to the maximum cutting temperature zone; The mechanical assembly structure of handle is designed to integrate back-end data processing devices, including signal acquisition and ZigBee wireless transmission schemes; Based on electromagnetic induction, a self-powered plan is proposed using the rotational kinetic energy of the spindle to achieve passive signal monitoring of milling process. Based on temperature sensing intelligent handle, cutting experiments with different machining parameters were conducted to verify the feasibility. In this paper, ‘workpiece with pre-embedded thermocouples’ method is used to get the instantaneous maximum milling temperature as true value, then to modify the finite element analysis simulation model, and ultimately obtain the measurement error of the smart handle. The resolution of smart handle is 0.25 ℃, the range is 0~700 ℃, the update speed is 10Hz, and the average measurement error is 8.53%.An intelligent handle with embedded acceleration sensor is designed to get vibration signals with high quality. An internal cavity of BT40 handle is designed to install ICP vibration sensor, close to vibration source to reduce signal distortion; Signal processing device is optimized with highly integrated signal preconditioned scheme, high-speed acquisition and WiFi wireless transfer solutions, and reliable mechanical assembly of circuit boards and handle is achieved. Based on the designed intelligent vibration measuring handle, cutting experiments with different machining parameters were carried out to explore the disciplines of vibration signals, and a comparative experiment was conducted with traditional wired methods of headstock and workpiece vibration measurement. The excellent performance of the intelligent vibration measuring tool handle was demonstrated from multiple indicators. The resolution of the vibration measuring handle is 4.88 mg, with a range of -50 g~50 g, and an update speed exceeding 10 kHz.Research on tool wear state recognition is finished. Feature extraction from time domain, frequency domain and wavelet domain is explored and recognition model based om Fisher Discriminant and neural network is proposed, reaching an accuracy of 90.0%.In summary, a wireless passive tool holder with milling temperature sensing function and a smart vibration measuring handle with online processing and wireless transmission function are designed by different built-in sensors, which can be used to achieve in-situ online signal monitoring of milling process. Based on vibration sensing handle, a tool wear state recognition model is designed and optimized.
