开关磁阻电动机转矩脉动的智能抑制方法

提出了一种基于瞬时转矩控制的抑制开关磁阻电动机转矩脉动的方法。先定义转矩分配函数 ,用以对各相转矩进行分配 ,保证各相瞬时转矩之和为一恒定值 ;其次通过矩角特性反演出各相电流指令 ,然后 ,通过滞环电流闭环控制电机 ;依负载要求而给定的转矩给定值 ,则是通过模糊神经网络控制器的输出而给定的。文中介绍了该智能控制器的设计方法。最后 ,通过数字仿真结果证实该方法不仅有效地抑制了转矩的脉动 ,而且具有期望的瞬态响应特性     

Glioblastoma Extracellular Vesicle-Specific Peptides Inhibit EV-Induced Neuronal Cytotoxicity

WHO Grade 4 IDH-wild type astrocytoma (GBM) is the deadliest brain tumor with a poor prognosis. Meningioma (MMA) is a more common “benign” central nervous system tumor but with significant recurrence rates. There is an urgent need for brain tumor biomarkers for early diagnosis and effective treatment options. Extracellular vesicles (EVs) are tiny membrane-enclosed vesicles that play essential functions in cell-to-cell communications among tumor cells. We aimed to identify epitopes of brain tumor EVs by phage peptide libraries. EVs from GBM plasma, MMA plasma, or brain tumor cell lines were used to screen phage-displayed random peptide libraries to identify high-affinity peptides. We purified EVs from three GBM plasma pools (23 patients), one MMA pool (10 patients), and four brain tumor cell lines. We identified a total of 21 high-affinity phage peptides (12 unique) specific to brain tumor EVs. The peptides shared high sequence homologies among those selected by the same EVs. Dose–response ELISA demonstrated that phage peptides were specific to brain tumor EVs compared to controls. Peptide affinity purification identified unique brain tumor EV subpopulations. Significantly, GBM EV peptides inhibit brain tumor EV-induced complement-dependent cytotoxicity (necrosis) in neurons. We conclude that phage display technology could identify specific peptides to isolate and characterize tumor EVs.     

自动离合器位置跟踪神经元自适应PID控制

针对自动离合器位置难以准确快速跟踪控制的问题,在采用蜗卷弹簧作为补偿装置改善驱动机构性能基础上,设计了一种具有前馈作用的增益可调神经元自适应PID控制器,利用前馈控制加快控制器的响应速度,利用神经元的自学习功能实现权值的自动调整,采用PSD算法实现增益的自动调整.实验结果表明,与常规控制器相比,该控制器用于离合器位置跟踪控制动、静态误差小,能够适应离合器负荷和目标接合速度的非线性变化,具有较好的鲁棒性,能够满足离合器位置跟踪控制要求.     

Engineered Neutral Phosphorous Dendrimers Protect Mouse Cortical Neurons and Brain Organoids from Excitotoxic Death

Nanoparticles are playing an increasing role in biomedical applications. Excitotoxicity plays a significant role in the pathophysiology of neurodegenerative diseases, such as Alzheimer’s or Parkinson’s disease. Glutamate ionotropic receptors, mainly those activated by N-methyl-D-aspartate (NMDA), play a key role in excitotoxic death by increasing intraneuronal calcium levels; triggering mitochondrial potential collapse; increasing free radicals; activating caspases 3, 9, and 12; and inducing endoplasmic reticulum stress. Neutral phosphorous dendrimers, acting intracellularly, have neuroprotective actions by interfering with NMDA-mediated excitotoxic mechanisms in rat cortical neurons. In addition, phosphorous dendrimers can access neurons inside human brain organoids, complex tridimensional structures that replicate a significant number of properties of the human brain, to interfere with NMDA-induced mechanisms of neuronal death. Phosphorous dendrimers are one of the few nanoparticles able to gain access to the inside of neurons, both in primary cultures and in brain organoids, and to exert pharmacological actions by themselves.     

CX3CL1 Action on Microglia Protects from Diet-Induced Obesity by Restoring POMC Neuronal Excitability and Melanocortin System Activity Impaired by High-Fat Diet Feeding

Both hypothalamic microglial inflammation and melanocortin pathway dysfunction contribute to diet-induced obesity (DIO) pathogenesis. Previous studies involving models of altered microglial signaling demonstrate altered DIO susceptibility with corresponding POMC neuron cytological changes, suggesting a link between microglia and the melanocortin system. We addressed this hypothesis using the specific microglial silencing molecule, CX3CL1 (fractalkine), to determine whether reducing hypothalamic microglial activation can restore POMC/melanocortin signaling to protect against DIO. We performed metabolic analyses in high fat diet (HFD)-fed mice with targeted viral overexpression of CX3CL1 in the hypothalamus. Electrophysiologic recording in hypothalamic slices from POMC-MAPT-GFP mice was used to determine the effects of HFD feeding and microglial silencing via minocycline or CX3CL1 on GFP-labeled POMC neurons. Finally, mice with hypothalamic overexpression of CX3CL1 received central treatment with the melanocortin receptor antagonist SHU9119 to determine whether melanocortin signaling is required for the metabolic benefits of CX3CL1. Hypothalamic overexpression of CX3CL1 increased leptin sensitivity and POMC gene expression, while reducing weight gain in animals fed an HFD. In electrophysiological recordings from hypothalamic slice preparations, HFD feeding was associated with reduced POMC neuron excitability and increased amplitude of inhibitory postsynaptic currents. Microglial silencing using minocycline or CX3CL1 treatment reversed these HFD-induced changes in POMC neuron electrophysiologic properties. Correspondingly, blockade of melanocortin receptor signaling in vivo prevented both the acute and chronic reduction in food intake and body weight mediated by CX3CL1. Our results show that suppressing microglial activation during HFD feeding reduces DIO susceptibility via a mechanism involving increased POMC neuron excitability and melanocortin signaling.     

The Functional Characterization of GCaMP3.0 Variants Specifically Targeted to Subcellular Domains

Calcium (Ca²⁺) ions play a pivotal role in physiology and cellular signaling. The intracellular Ca²⁺ concentration ([Ca²⁺]_i) is about three orders of magnitude lower than the extracellular concentration, resulting in a steep transmembrane concentration gradient. Thus, the spatial and the temporal dynamics of [Ca²⁺]_i are ideally suited to modulate Ca²⁺-mediated cellular responses to external signals. A variety of highly sophisticated methods have been developed to gain insight into cellular Ca²⁺ dynamics. In addition to electrophysiological measurements and the application of synthetic dyes that change their fluorescent properties upon interaction with Ca²⁺, the introduction and the ongoing development of genetically encoded Ca²⁺ indicators (GECI) opened a new era to study Ca²⁺-driven processes in living cells and organisms. Here, we have focused on one well-established GECI, i.e., GCaMP3.0. We have systematically modified the protein with sequence motifs, allowing localization of the sensor in the nucleus, in the mitochondrial matrix, at the mitochondrial outer membrane, and at the plasma membrane. The individual variants and a cytosolic version of GCaMP3.0 were overexpressed and purified from E. coli cells to study their biophysical properties in solution. All versions were examined to monitor Ca²⁺ signaling in stably transfected cell lines and in primary cortical neurons transduced with recombinant Adeno-associated viruses (rAAV). In this comparative study, we provide evidence for a robust approach to reliably trace Ca²⁺ signals at the (sub)-cellular level with pronounced temporal resolution.     

An Artificial Sensory Neuron with Tactile Perceptual Learning

Sensory neurons within skin form an interface between the external physical reality and the inner tactile perception. This interface enables sensory information to be organized identified, and interpreted through perceptual learning—the process whereby the sensing abilities improve through experience. Here, an artificial sensory neuron that can integrate and differentiate the spatiotemporal features of touched patterns for recognition is shown. The system comprises sensing, transmitting, and processing components that are parallel to those found in a sensory neuron. A resistive pressure sensor converts pressure stimuli into electric signals, which are transmitted to a synaptic transistor through interfacial ionic/electronic coupling via a soft ionic conductor. Furthermore, the recognition error rate can be dramatically decreased from 44% to 0.4% by integrating with the machine learning method. This work represents a step toward the design and use of neuromorphic electronic skin with artificial intelligence for robotics and prosthetics.     

Information-Theoretic Competitive Learning with Inverse Euclidean Distance Output Units

In this paper, we propose a new information theoretic competitive learning method. We first construct a learning method in single-layered networks, and then we extend it to supervised multi-layered networks. Competitive unit outputs are computed by the inverse of Euclidean distance between input patterns and connection weights. As distance is smaller, competitive unit outputs are stronger. In realizing competition, neither the winner-take-all algorithm nor the lateral inhibition is used. Instead, the new method is based upon mutual information maximization between input patterns and competitive units. In maximizing mutual information, the entropy of competitive units is increased as much as possible. This means that all competitive units must equally be used in our framework. Thus, no under-utilized neurons or dead neurons are generated. When using multi-layered networks, we can improve noise-tolerance performance by unifying information maximization and minimization. We applied our method with single-layered networks to a simple artificial data problem and an actual road classification problem. In both cases, experimental results confirmed that the new method can produce the final solutions almost independently of initial conditions, and classification performance is significantly improved. Then, we used multi-layered networks, and applied them to a character recognition problem and a political data analysis. In these problem, we could show that noise-tolerance performance was improved by decreasing information content on input patterns to certain points.     

Neural network models

The remarkable processing capabilities of the nervous system must derive at least in part from the large numbers of neurons participating (roughly 10¹⁰), since the timescales involved are of the order of milliseconds, rather than the nanoseconds of modern computers. We summarise common features of the neural network models which attempt to capture this behaviour and describe the many levels of parallelism which they exhibit. A range of models has been implemented on the SIMD (ICL Distributed Array Processor) and MIMD (Meiko Computing Surface) hardware at Edinburgh. Examples include: (i) training algorithms in the context of the Hopfield net, with specific application to the storage of words and text with content-addressable memory; (ii) the back-propagation training algorithm for the multi-layer perception; (iii) image restoration with Hopfield and Tank analogue neurons, and (iv) the Durbin and Willshaw elastic net, as applied to the travelling salesman problem.