Fueled by the development of deep neural networks, breakthroughs have been achieved in plenty of computer vision problems, such as image classification, segmentation, and object detection. These models usually have handers and millions of parameters, which makes them both computational and memory expensive. Motivated by this, this paper proposes a post-training quantization method based on the clipping operation for neural network compression. By quantizing parameters of a model to 8-bit using our proposed methods, its memory consumption is reduced, its computational speed is increased, and its performance is maintained. This method exploits the clipping operation during training so that it saves a large computational cost during quantization. After training, this method quantizes the parameters to 8-bit based on the clipping value. In addition, a fully connected layer compression is conducted using singular value decomposition (SVD), and a novel loss function term is leveraged to further diminish the performance drop caused by quantization. The proposed method is validated on two widely used models, Yolo V3 and Faster R-CNN, for object detection on the PASCAL VOC, COCO, and ImageNet datasets. Performances show it effectively reduces the storage consumption at 18.84% and accelerates the model at 381%, meanwhile avoiding the performance drop (drop < 0.02% in VOC).

An AGV system can be used to transport different-size materials in an assembly line. The hybrid task allocation problem is involved in the assembly line, where both single-AGV tasks and multi-AGV tasks exist. However, there is little research on this problem. The goal of solving this problem is to obtain a task allocation scheme with minimum idle time and maximum system throughput. Since all necessary materials must be delivered to the assembly station before the operation can start, the delivery tasks are not independent of each other in a task group serving the operation. To solve the problem above, a hybrid task allocation method based on a task binding strategy and an improved particle swarm optimization (IPSO) is proposed. Firstly, a mathematical model considering the punctuality of material delivery and the cooperative relationship between tasks is established. Secondly, a task binding strategy and four heuristic rules are devised to improve the quality of randomly- and heuristic-generated individuals in the initial population for model optimization. Thirdly, an IPSO is developed to help the optimization algorithm jump out of local optimums. Finally, a simulation is performed to verify the effectiveness of the proposed methods. The simulation results show that a better scheme can be obtained by our hybrid task allocation method, compared to conventional Genetic Algorithms and PSO algorithms.

Purpose The paper aims to propose general design rules and route plan for automated guided vehicle system (AGVS). The AGVS is applied to automated meter verification areas through a case study of meter verification shop floor to verify the feasibility. Design/methodology/approach The paper gives an appropriate route design for AGVS and proposes an optimized strategy for designed routes and a control system to manage traffic conflict. Findings This case study indicates that the application of AGVS can highly improve the efficiency of manufacturing and production. Besides, a reasonable transportation plan is beneficial in making the system run smoothly and in cutting conveying time. Practical implications The application of AGVS integrates a variety of advanced technologies (i.e. information technology, artificial intelligence, etc.) into the electricity meter verification system, which brings great economic and social benefits via enhancing the verification efficiency and reducing the total labor costs. Originality/value The application proposed in the paper solves the problem that the verification almost relies on workers, labor intensity is high and work efficiency is hard to improve. Furthermore, the general rules and strategies of AGVS transportation can be applied not only to the automated electricity meter verification but also in other industrial areas.

The crankshaft is the core part of an automobile engine, and the accuracy requirements of various shape and position errors are very high. On the basis of a synchronous measurement system, the connecting rod journal is deeply studied, including data processing and roundness evaluation. Firstly, according to the measuring processes of connecting rod journals, the real sampling angle distribution function was established, and the corresponding Gaussian weight function of each sampling angle was calculated. The weight function and the collected data corresponding to the angle were subjected to discrete cyclic convolution operation in the spatial domain to obtain the filtered effective circular contour data. Secondly, the particle swarm optimization algorithm was improved, and its inertia weight was set to decrease nonlinearly to speed up the convergence. A calculation process suitable for the evaluation of journal errors was designed. Then, the improved particle swarm optimization algorithm was used to evaluate the roundness of the corrected rod journal contour data. At last, through multiple measurement experiments, the feasibility and effectiveness of the synchronous measurement scheme and data processing method proposed in this paper are verified.

The crankshaft online measurement system has realized the full inspection function with fast beats, at the same time it requires for high-precision measurement. Considering the effect of ambient temperature and temperature changes on measuring machine, the calibration part, the measured crankshaft and displacement sensor, a temperature compensation method is proposed. Firstly, relationship between calibration part and ambient temperature can be get through the zero calibration. Then use the material properties to obtain compensation values of the calibration part and the measured crankshaft part at different temperatures. Finally, the compensation parameters for displacement sensor can be obtained through the BP algorithm. The improved dragonfly algorithm (DA) is used to optimize the parameters of BP neural network algorithm. Experiments verify the effectiveness of IDA-BP for LVDT in temperature compensation. After temperature compensation, the error range of main journal radius is reduced from 0.0156 mm to 0.0028 mm, the residual error decreased from −0.0282 mm~+0.0018 mm to −0.0058 mm~−0.0008 mm. The influence of temperature changes on the measurement is reduced and measurement accuracy is improved through the temperature compensation method. The effectiveness of the method is proved.

The heterogeneous automatic guided vehicle (AGV) system, composed of several AGVs with different load capability and handling function, has good flexibility and agility to operational requirements. Accurate task travel time prediction (T3P) is vital for the efficient operation of heterogeneous AGV systems. However, T3P remains a challenging problem due to individual task correlations and dynamic changes in model input/output dimensions. To address these challenges, a biomimetics-inspired learning framework based on a radial basis function (RBF) neural network with an improved mayfly algorithm and a selective update strategy (IMA-SURBF) is proposed. Firstly, a T3P model is constructed by using travel-influencing factors as input and task travel time as output of the RBF neural network, where the input/output dimension is determined dynamically. Secondly, the improved mayfly algorithm (IMA), a biomimetic metaheuristic method, is adopted to optimize the initial parameters of the RBF neural network, while a selective update strategy is designed for parameter updates. Finally, simulation experiments on model design, parameter initialization, and comparison with deep learning-based models are conducted in a complex assembly line scenario to validate the accuracy and efficiency of the proposed method.