Showing papers by "Mitsubishi Electric published in 2020"

Highly porous nature of a primitive asteroid revealed by thermal imaging

Abstract: Additional co-authors: Tsuneo Matsunaga, Takeshi Imamura, Takehiko Wada, Sunao Hasegawa, Jorn Helbert, Thomas G. Muller, Jens Biele, Matthias Grott, Maximilian Hamm, Marco Delbo, Naru Hirata, Naoyuki Hirata, Yukio Yamamoto, Seiji Sugita, Noriyuki Namiki, Kohei Kitazato, Masahiko Arakawa, Shogo Tachibana, Hitoshi Ikeda, Masateru Ishiguro, Koji Wada, Chikatoshi Honda, Rie Honda, Yoshiaki Ishihara, Koji Matsumoto, Moe Matsuoka, Tatsuhiro Michikami, Akira Miura, Tomokatsu Morota, Hirotomo Noda, Rina Noguchi, Kazunori Ogawa, Kei Shirai, Eri Tatsumi, Hikaru Yabuta, Yasuhiro Yokota, Manabu Yamada, Masanao Abe, Masahiko Hayakawa, Takahiro Iwata, Masanobu Ozaki, Hajime Yano, Satoshi Hosoda, Osamu Mori, Hirotaka Sawada, Takanobu Shimada, Hiroshi Takeuchi, Ryudo Tsukizaki, Atsushi Fujii, Chikako Hirose, Shota Kikuchi, Yuya Mimasu, Naoko Ogawa, Go Ono, Tadateru Takahashi, Yuto Takei, Tomohiro Yamaguchi, Kent Yoshikawa, Fuyuto Terui, Takanao Saiki, Satoru Nakazawa, Makoto Yoshikawa, Seiichiro Watanabe & Yuichi Tsuda Output Status: Forthcoming/Available Online

Trajectory tracking for autonomous vehicles on varying road surfaces by friction-adaptive nonlinear model predictive control

Abstract: We propose an adaptive nonlinear model predictive control (NMPC) for vehicle tracking control. The controller learns in real time a tyre force model to adapt to a varying road surface that is only ...

Anomaly Detection in Smart Home Operation From User Behaviors and Home Conditions

Abstract: As several home appliances, such as air conditioners, heaters, and refrigerators, were connecting to the Internet, they became targets of cyberattacks, which cause serious problems such as compromising safety and even harming users. We have proposed a method to detect such attacks based on user behavior. This method models user behavior as sequences of user events including operation of home IoT (Internet of Things) devices and other monitored activities. Considering users behave depending on the condition of the home such as time and temperature, our method learns event sequences for each condition. To mitigate the impact of events of other users in the home included in the monitored sequence, our method generates multiple event sequences by removing some events and learning the frequently observed sequences. For evaluation, we constructed an experimental network of home IoT devices and recorded time data for four users entering/leaving a room and operating devices. We obtained detection ratios exceeding 90% for anomalous operations with less than 10% of misdetections when our method observed event sequences related to the operation. In this article, we also discuss the effectiveness of our method by comparing with a method learning users’ behavior by Hidden Markov Models.

Quantum key distribution with correlated sources

Abstract: In theory, quantum key distribution (QKD) offers information-theoretic security. In practice, however, it does not due to the discrepancies between the assumptions used in the security proofs and the behavior of the real apparatuses. Recent years have witnessed a tremendous effort to fill the gap, but the treatment of correlations among pulses has remained a major elusive problem. Here, we close this gap by introducing a simple yet general method to prove the security of QKD with arbitrarily long-range pulse correlations. Our method is compatible with those security proofs that accommodate all the other typical device imperfections, thus paving the way toward achieving implementation security in QKD with arbitrary flawed devices. Moreover, we introduce a new framework for security proofs, which we call the reference technique. This framework includes existing security proofs as special cases, and it can be widely applied to a number of QKD protocols.

Graphene Plasmonics in Sensor Applications: A Review.

Abstract: Surface plasmon polaritons (SPPs) can be generated in graphene at frequencies in the mid-infrared to terahertz range, which is not possible using conventional plasmonic materials such as noble metals. Moreover, the lifetime and confinement volume of such SPPs are much longer and smaller, respectively, than those in metals. For these reasons, graphene plasmonics has potential applications in novel plasmonic sensors and various concepts have been proposed. This review paper examines the potential of such graphene plasmonics with regard to the development of novel high-performance sensors. The theoretical background is summarized and the intrinsic nature of graphene plasmons, interactions between graphene and SPPs induced by metallic nanostructures and the electrical control of SPPs by adjusting the Fermi level of graphene are discussed. Subsequently, the development of optical sensors, biological sensors and important components such as absorbers/emitters and reconfigurable optical mirrors for use in new sensor systems are reviewed. Finally, future challenges related to the fabrication of graphene-based devices as well as various advanced optical devices incorporating other two-dimensional materials are examined. This review is intended to assist researchers in both industry and academia in the design and development of novel sensors based on graphene plasmonics.

High-performance graphene/InSb heterojunction photodetectors for high-resolution mid-infrared image sensors

Abstract: Graphene/semiconductor heterojunction photodetectors have the potential to outperform conventional infrared (IR) sensors. A high-performance graphene/indium antimonide (InSb) heterojunction photodetector for high-resolution mid-IR image sensors was developed using an undoped InSb substrate with low carrier density. The width of the depletion layer at the graphene/InSb interface was increased because of the low carrier density of undoped InSb, which resulted in a low dark current for the proposed device. In addition, the low carrier density resulted in an increased capacitance change due to photocarrier generation for the depletion layer upon light irradiation. As a result, the carrier density modulation effect of graphene was also amplified owing to the photogating effect. Consequently, low dark currents on the order of nanoamperes and high responsivities of over 2 A/W were achieved over a wide voltage range of −0.05 to −0.5 V using our proposed photodetector. The best performance achieved for the developed photodetector corresponded to a specific detectivity (D*) of 2.28 × 1010 cm Hz1/2/W and a noise-equivalent-power of 0.13 pW/Hz1/2, which makes these detectors suitable for use in mid-IR image sensors, realizing a low dark current of just −7.5 nA over the wide wavelength range of 3–5 μm. These results demonstrate that an IR image sensor with detection performance better than those developed using existing graphene-based photodetectors can be obtained by adopting a simple graphene/InSb heterojunction with undoped InSb, owing to the photogating effect.

Dynamic bandwidth allocation scheme for network-slicing-based TDM-PON toward the beyond-5G era

Abstract: In the coming 5G mobile communications era, a huge number of wireless access points will be densely deployed for networks such as mobile fronthaul and Internet-of-Things networks. A network-slicing-based time-division-multiplexing passive optical network (TDM-PON) is one solution for cost-effectively accommodating these access points. It enables us to converge these multiple networks that have different requirements by virtually isolating each sub-network on a single TDM-PON. To achieve this TDM-PON convergence, we propose a dynamic bandwidth allocation scheme that performs uplink bandwidth allocations in a different manner at every sub-network in order to simultaneously satisfy each sub-network requirement. Experiments show that the scheme can simultaneously provide low-latency transmission, bandwidth guarantee, and an auto-discovery process.

MERLIN: Parameter-Free Discovery of Arbitrary Length Anomalies in Massive Time Series Archives

Abstract: Time series anomaly detection remains a perennially important research topic. If anything, it is a task that has become increasingly important in the burgeoning age of IoT. While there are hundreds of anomaly detection methods in the literature, one definition, time series discords, has emerged as a competitive and popular choice for practitioners. Time series discords are subsequences of a time series that are maximally far away from their nearest neighbors. Perhaps the most attractive feature of discords is their simplicity. Unlike many parameter laden methods, discords require only a single parameter to be set by the user: the subsequence length. In this work we argue that the utility of discords is reduced by sensitivity to this single user choice. The obvious solution to this problem, computing discords of all lengths then selecting the best anomalies (under some measure), seems to be computationally untenable. However, in this work we introduce MERLIN, an algorithm that can efficiently and exactly find discords of all lengths in massive time series archives.

Efficient Power Control for Satellite-Borne Batteries Using Q-Learning in Low-Earth-Orbit Satellite Constellations

Abstract: Recently, changes in the relationship between satellite communication networks and ground communication networks have led to an increase in the demands imposed on satellite communication networks. In this category of networks, the low earth orbit (LEO) satellite constellations, which cover the entire surface of Earth through cooperation among many small satellites in low orbit, are currently attracting attention. LEO satellites consume power when communicating with terrestrial terminals. However, in the absence of solar light, these satellites must operate using battery energy only. The power consumption during these periods places a heavy load on the satellite battery and can shorten their lifetimes. This entails a significant cost for satellite communication networks. In this letter, we apply Q-learning to the power allocation problem in satellite-to-ground communication using LEO satellites. Using this method, we can extend the lifetime of the LEO satellite battery by sharing the workload of overworked satellites with adjacent satellites with a lower load. The effects of the proposed method on the battery lifetime of the LEO satellites are verified.

Model-Based Analysis and Quantification of Bearing Faults in Induction Machines

Abstract: The detection of rolling-element bearing fault can be accomplished by monitoring and interpreting a variety of signals, including the vibration, the acoustic noise, and the stator current. The existence of a bearing fault as well as its specific fault type can be readily determined by performing frequency spectral analysis on the monitored signals with various signal processing techniques. However, this traditional approach, despite being simple and intuitive, is not able to identify the severity of a bearing fault in a quantitative manner. Moreover, it is often times tedious and time-consuming to apply this approach to electric machines with different power ratings, as the bearing fault threshold values need to be manually calibrated for each motor running at every possible speed and carrying any possible load. This article, thus, proposes a quantitative approach to estimate the bearing fault severity based on the airgap displacement profile, which is reconstructed from the mutual inductance variation profile estimated from a quantitative electrical model that takes the stator current as input. In addition, the accuracy of the developed electrical model and the estimated bearing fault severity are validated by the simulation and experimental results, and the explicit airgap variation profile is reconstructed with the superposition of multiple Fourier series terms estimated from the stator current via the proposed scheme. The proposed method offers a quantitative and universal bearing fault indicator for induction machines with any power ratings and operating under any speed and load conditions.

Unsupervised Sound Separation Using Mixture Invariant Training

Abstract: In recent years, rapid progress has been made on the problem of single-channel sound separation using supervised training of deep neural networks. In such supervised approaches, a model is trained to predict the component sources from synthetic mixtures created by adding up isolated ground-truth sources. Reliance on this synthetic training data is problematic because good performance depends upon the degree of match between the training data and real-world audio, especially in terms of the acoustic conditions and distribution of sources. The acoustic properties can be challenging to accurately simulate, and the distribution of sound types may be hard to replicate. In this paper, we propose a completely unsupervised method, mixture invariant training (MixIT), that requires only single-channel acoustic mixtures. In MixIT, training examples are constructed by mixing together existing mixtures, and the model separates them into a variable number of latent sources, such that the separated sources can be remixed to approximate the original mixtures. We show that MixIT can achieve competitive performance compared to supervised methods on speech separation. Using MixIT in a semi-supervised learning setting enables unsupervised domain adaptation and learning from large amounts of real world data without ground-truth source waveforms. In particular, we significantly improve reverberant speech separation performance by incorporating reverberant mixtures, train a speech enhancement system from noisy mixtures, and improve universal sound separation by incorporating a large amount of in-the-wild data.

Satellite Constellation Pattern Optimization for Complex Regional Coverage

Abstract: The use of regional-coverage satellite constellations is on the rise, urging the need for an optimal constellation design method for complex regional coverage. Traditional constellations are often ...

Transformer-Based Long-Context End-to-End Speech Recognition.

Abstract: This paper presents an approach to long-context end-to-end automatic speech recognition (ASR) using Transformers, aiming at improving ASR accuracy for long audio recordings such as lecture and conversational speeches. Most end-to-end ASR systems are basically designed to recognize independent utterances, but contextual information (e.g., speaker or topic) over multiple utterances is known to be useful for ASR. There are some prior studies on RNN-based models that utilize such contextual information, but very few on Transformers, which are becoming more popular in end-to-end ASR. In this paper, we propose a Transformer-based architecture that accepts multiple consecutive utterances at the same time and predicts an output sequence for the last utterance. This is repeated in a slidingwindow fashion with one-utterance shifts to recognize the entire recording. Based on this framework, we also investigate how to design the context window and train the model effectively in monologue (one speaker) and dialogue (two speakers) scenarios. We demonstrate the effectiveness of our approach using monologue benchmarks on CSJ and TED-LIUM3 and dialogue benchmarks on SWITCHBOARD and HKUST, showing significant error reduction from single-utterance ASR baselines with or without speaker i-vectors.

Considerations on the Use of Digital Signal Processing in Future Optical Access Networks

Abstract: In this invited article, we address the impacts of the rise of signal processing for increased capacity per wavelength and better receiver sensitivities in next generation optical access networks. We start by recalling the main channel limitations of currently deployed intensity modulated, directly detected, passive optical networks. Then, with the intention of providing a benchmarking of signal processing approaches used in communication systems other than optical access, we provide a historic perspective on how digital signal processing emerged in copper access systems, and we evaluate powerful techniques envisaged for future mobile generation in both air interface and radio access networks. We also assess signal processing in light of multi-vendor interoperability by providing insights on burst mode operation and the needed protocol and monitoring procedures. Interoperability with regard to optical transceivers is also considered. Last but not least, we evaluate power consumption of digital signal processing and forward error correction solutions used in disruptive, coherent transmission approaches.

Experimental Field Trials on MU-MIMO Transmissions for High SHF Wide-Band Massive MIMO in 5G

Abstract: The field trial results for massive multi-user multiple-input multiple-output (MU-MIMO) transmission at 28 GHz with 500 MHz bandwidth are reported in this paper. In this experiment, hybrid beamforming with block diagonal precoding is employed, in which channel state information is acquired by uplink sounding assuming channel reciprocity in a time division duplex system. 8-user MU-MIMO scenarios with various user distributions are considered, and the downlink transmission performance and propagation channel characteristics are evaluated. The trial results show that digital precoding improves the system performance and that over 20 Gbps total throughput can be achieved for a variety of user distributions. The performance of the precoding is evaluated in terms of eliminating inter-user interference.

Efficient Exploration in Constrained Environments with Goal-Oriented Reference Path

Abstract: In this paper, we consider the problem of building learning agents that can efficiently learn to navigate in constrained environments. The main goal is to design agents that can efficiently learn to understand and generalize to different environments using high-dimensional inputs (a 2D map), while following feasible paths that avoid obstacles in obstacle-cluttered environment. To achieve this, we make use of traditional path planning algorithms, supervised learning, and reinforcement learning algorithms in a synergistic way. The key idea is to decouple the navigation problem into planning and control, the former of which is achieved by supervised learning whereas the latter is done by reinforcement learning. Specifically, we train a deep convolutional network that can predict collision-free paths based on a map of the environment– this is then used by an reinforcement learning algorithm to learn to closely follow the path. This allows the trained agent to achieve good generalization while learning faster. We test our proposed method in the recently proposed Safety Gym suite that allows testing of safety-constraints during training of learning agents. We compare our proposed method with existing work and show that our method consistently improves the sample efficiency and generalization capability to novel environments.

Condition Monitoring of Ball-Screw Drives Based on Frequency Shift

Abstract: Ball-screw drives are one of the common wear parts in machine tools. The wear will affect the positioning accuracy of motion axes and can cause a high rejection rate in production. Therefore, it is interesting to develop a method, which can detect the wear condition of ball screws at an early stage. Stiffness loss and backlash in ball-screw drives are often associated with wear. This article proposes a novel method to detect stiffness loss and backlash for ball-screw drives without preload based only on machine internal data. To simulate the wear of a ball screw, a spacer sleeve has been designed, which effectively reduces the axial stiffness of the feed axes. This design allows a quick modification of the wear status of the machine axes and can be utilized to test if the wear status is detectable by analyzing the machine internal data. A multimass model is developed and validated by experimental data, which is used to simulate the dynamic behavior of the machine axis dependent on wear.

Photogating for small high-responsivity graphene middle-wavelength infrared photodetectors

Abstract: We demonstrated a middle-wavelength infrared (MWIR) graphene photodetector using the photogating effect. This effect was induced by photosensitizers situated around a graphene channel that coupled incident light and generated a large electrical charge. The graphene-based MWIR photodetector consisted of a top graphene channel, source–drain electrodes, an insulator layer, and a photosensitizer, and its photoresponse characteristics were determined by current measurements. Irradiation of the graphene channel of the vacuum cooled device by an MWIR laser generated a clear photoresponse, as evidenced by modulation of the output current during irradiation. The MWIR photoresponse with the photogating effect was 100 times greater than that obtained from conventional graphene photodetectors without the photogating effect. The device maintained its MWIR photoresponse at temperatures up to 150 K. The effect of the graphene channel size on the responsivity was evaluated to assess the feasibility of reducing the photodetector area, and decreasing the channel area from 100 to 25 μm2 improved the responsivity from 61.7 to 321.0 AW − 1. The results obtained in our study will contribute to the development of high-performance graphene-based IR imaging sensors.

Flexibility-Enhanced HTS System for Disaster Management: Responding to Communication Demand Explosion in a Disaster

Abstract: Natural disaster interrupts essential services due to the facility damage and lack of power supply. Especially, significant communication outages occurs in a wide area for hierarchical network such as cellular communication system in case core nodes are damaged or congested. In order to provide alternative communication ability, high throughput satellite (HTS) is one of the ideal candidates for disaster management because it provides operative communication for a wide area regardless of the availability of regular terrestrial infrastructures. However, conventional HTS relays data with predetermined beam bandwidth and connection, it is inefficient when the communication demand explodes in a disaster area. Therefore, this paper proposes novel frequency allocation technique with flexibility-enhanced HTS system for disaster management to respond to communication demand explosion. While related research works consider user-link resource allocation, this paper focuses on how to control feeder-link and user-link bandwidths in case that the both links can be assigned at continuous frequency such as Ka-band. Furthermore, our proposal addresses resilient satellite network by selecting optimum gateway based on the traffic demand at neighboring user-link beams. The effectiveness of our proposal is verified through simulation results.

Lightweight Authenticated Encryption Mode Suitable for Threshold Implementation.

Abstract: This paper proposes tweakable block cipher (TBC) based modes \(\mathsf {PFB\_Plus}\) and \(\mathsf {PFB}\omega \) that are efficient in threshold implementations (TI). Let t be an algebraic degree of a target function, e.g. \(t=1\) (resp. \(t>1\)) for linear (resp. non-linear) function. The d-th order TI encodes the internal state into \(d t + 1\) shares. Hence, the area size increases proportionally to the number of shares. This implies that TBC based modes can be smaller than block cipher (BC) based modes in TI because TBC requires s-bit block to ensure s-bit security, e.g. PFB and Romulus, while BC requires 2s-bit block. However, even with those TBC based modes, the minimum we can reach is 3 shares of s-bit state with \(t=2\) and the first-order TI (\(d=1\)).

Effects of temperature on electrical treeing and partial discharges in epoxy/silica nanocomposites

Abstract: This article reports the effect of temperature on tree propagation and phase resolved partial discharge (PRPD) pattern in epoxy/silica nanocomposite. The results show that tree shapes in 0-wt% and 5-wt% samples change from branch-like to bush-like as temperature rises. The tensile stress tends to decrease when the temperature is high, and a tree can select numerous paths to propagate; however, tree shapes in 10-wt% samples keep bush-like regardless of the temperature. It is suggested that nano-particle effect of inhibiting trees is more likely to occur at high temperatures rather than at room temperature. When a tree starts to propagate, the PRPD pattern tends to form wing-like one regardless of the amount of nano-particle and the temperature.

CSTBT™ based Split-Gate RC-IGBT with Low Loss and EMI Noise

Abstract: CSTBT™ faces a complex tradeoff among E on/off -V CEsat -SOA and EMI noise. We propose a split-gate CSTBT™ to break up the complex tradeoff and introduce this novel structure to the RC-IGBT for the first time. The Miller capacitance was drastically reduced by using the proposed structure while maintaining the Blocking Voltage, which contributes to approximately a 15% reduction in total loss without any sacrifices of EMI noise and/or SOA.

Air-conditioning apparatus

Abstract: A vehicle air-conditioning apparatus includes: an indoor heat exchanger that evaporates a refrigerant; a cooperative device group that forms, with the indoor heat exchanger, a refrigeration cycle; a waste heat recovery part that recovers waste heat from a waste heat source and a heat-release part that releases, to an indoor heat exchanger, heat transmitted to the heat-release part, and that is switchable between (a) a heat-releasing state in which the waste heat recovered by the waste heat recovery part is transmitted to the heat-release part, and (b) a heat-release-stopped state in which the heat recovered by the waste heat recovery part is less transmittable than when in the heat-releasing state; and a controller to switch between (i) a heat-release-stopped state when the refrigerant is being exchanged between the indoor heat exchanger and the cooperative device group, and (ii) a heat-releasing state when the exchange of the refrigerant is stopped.

Carrier density modulation and photocarrier transportation of graphene/InSb heterojunction middle-wavelength infrared photodetectors

Abstract: The photoresponse mechanism of graphene/InSb heterojunction middle-wavelength infrared (MWIR) photodetectors was investigated. The devices comprised a graphene/InSb heterojunction as a carrier-injection region and an insulator region of graphene on tetraethyl orthosilicate (TEOS) for photogating. The MWIR photoresponse was significantly amplified with an increase in the graphene/TEOS cross-sectional area by covering the entire detector with graphene. The graphene-channel dependence of the MWIR photoresponse indicated that the graphene carrier density was modulated by photocarrier accumulation at the TEOS/InSb boundary, resulting in photogating. The dark current of the devices was suppressed by a decrease in the graphene/InSb carrier-injection region and the formation of the heterojunction using an n-type InSb substrate. The results indicate that photocarrier transportation was dominated by the formation of a Schottky barrier at the interface of the graphene/InSb heterojunction and a Fermi-level shift under bias application. The high-responsivity and low-dark-current photoresponse mechanism was attributed to the graphene/InSb heterojunction diode behavior and the photogating effect. The devices combining the aforementioned features had a noise equivalent power of 0.43 pW / Hz1/2. The results obtained in our study will contribute to the development of high-performance graphene-based IR image sensors.

Hayabusa2 Landing Site Selection: Surface Topography of Ryugu and Touchdown Safety

Abstract: One of the primary goals of Hayabusa2 is to land on the asteroid Ryugu to collect its surface materials. The key for a successful touchdown is to find a promising landing site that meets both scientific and engineering requirements. Due to the limited availability of pre-arrival information about Ryugu, the landing site selection (LSS) must be conducted based on proximity observations over a limited length of time. In addition, Ryugu was discovered to possess an unexpectedly high abundance of boulders with an absence of wide and flat areas, further complicating the LSS. To resolve these problems, we developed a systematic and stepwise LSS process with a focus on the surface topography of Ryugu and the associated touchdown safety. The proposed LSS scheme consists of two phases: Phase-I LSS, a comprehensive survey of potential landing areas at the 100-m scale based on the global mapping of Ryugu, and Phase-II LSS, a narrowing-down process of the candidate landing sites at the 10-m scale using high-resolution images and a local terrain model. To verify the feasibility of a precision landing at the target site, we also investigated the landing dispersion via a Monte Carlo simulation, which incorporates the effect of the irregular surface gravity field. One of the major characteristics of the Hayabusa2 LSS developed in this study is the iterative feedback between LSS analyses on the ground and actual spacecraft operations near the target asteroid. Using the newly developed method, we chose a landing site with a radius of 3 m, and Hayabusa2 successfully conducted its first touchdown on February 21, 2019. This paper reports the methodology and results of the stepwise iterative LSS for the first Hayabusa2 touchdown. The touchdown operation results reconstructed from flight data are also provided, demonstrating the validity of the adopted LSS strategy.

Optical in-process height measurement system for process control of laser metal-wire deposition

Abstract: We propose an in-process height measurement system for a weld bead and feedback control system for wire-feeding speed for high-quality laser deposition. Metal additive manufacturing, especially laser metal-wire deposition, is effective for complex shape fabrication and repair processing. However, we must control the gap between a weld bead and a feed wire in an optimal range for high-quality deposition. Conventionally, the Z-stage pitch for multi-layer deposition must be precisely adjusted by each deposition shape. In this paper, we design an in-process height measurement system that is integrated in a laser processing head, which measures the weld bead height by a line section method. We decreased the influence of the intense thermal radiation generated from a melt pool by inserting the band-pass filter of the line beam's wavelength in the imaging system and optimizing its line laser power. Consequently, our system can measure the weld bead height near the melt pool, which is 4 mm in front of it. Next we show that our proposed system can measure the weld bead height during wire-laser metal deposition with 50-μm accuracy by comparing its value to the true value. Finally, we achieved a cylinder shape deposition of 50-mm height, regardless of the Z-stage pitch and the cylinder diameter of the multi-layer deposition, by controlling the wire-feeding speed based on the measured weld bead height.