基于SSAE-IARO-BiLSTM的工业过程故障诊断研究

针对工业过程故障诊断精度低的问题,提出了一种基于栈式稀疏自编码网络(stacked sparse auto-encoder network, SSAE)和改进人工兔算法优化双向长短时记忆神经网络(improved artificial rabbit algorithm optimized bidirectional long short-term memory neural network, IARO-BiLSTM)的故障诊断方法。首先,利用SSAE网络强大的特征提取能力,实现对原始数据进行降维处理;其次,引入Circle混沌映射以达到丰富种群数量的目的,提出权重系数和Levy飞行机制改进人工兔算法的位置更新公式,提高人工兔算法的寻优能力,进而对BiLSTM网络的参数进行优化。最后,利用优化后的BiLSTM网络实现对故障的识别和分类。通过选取多组数据集进行验证,结果表明,基于SSAE-IARO-BiLSTM故障诊断方法能够准确地对故障进行识别和分类,且诊断准确率可达98%以上。

Stochastic Programming for Hub Energy Management Considering Uncertainty Using Two-Point Estimate Method and Optimization Algorithm

To maximize energy profit with the participation of electricity,natural gas,and district heating networks in the day-ahead market,stochastic scheduling of energy hubs taking into account the uncertainty of photovoltaic and wind resources,has been carried out.This has been done using a new meta-heuristic algorithm,improved artificial rabbits optimization(IARO).In this study,the uncertainty of solar and wind energy sources is modeled using Hang’s two-point estimating method(TPEM).The IARO algorithm is applied to calculate the best capacity of hub energy equipment,such as solar and wind renewable energy sources,combined heat and power(CHP)systems,steamboilers,energy storage,and electric cars in the day-aheadmarket.The standard ARO algorithmis developed to mimic the foraging behavior of rabbits,and in this work,the algorithm’s effectiveness in avoiding premature convergence is improved by using the dystudynamic inertia weight technique.The proposed IARO-based scheduling framework’s performance is evaluated against that of traditional ARO,particle swarm optimization(PSO),and salp swarm algorithm(SSA).The findings show that,in comparison to previous approaches,the suggested meta-heuristic scheduling framework based on the IARO has increased energy profit in day-ahead electricity,gas,and heating markets by satisfying the operational and energy hub limitations.Additionally,the results show that TPEM approach dependability consideration decreased hub energy’s profit by 8.995%as compared to deterministic planning.

Simultaneous allocation of renewable energy sources and custom power quality devices in electrical distribution networks using artificial rabbits optimization

This study suggests an optimal renewable energy source(RES)allocation and distribution-static synchronous compensator(D-STATCOM)and passive power filters(PPFs)for an electrical distribution network(EDN)to improve its performance and power quality(PQ).First,the latest metaheuristic artificial rabbits optimization(ARO)is used to locate and size solar photovoltaic(PV),wind turbine(WT)and D-STATCOM units.In the second stage,ratings of single-tuned PPFs and D-STATCOMs at the RESs are determined,considering non-linear loads in the network.The multi-objective function reduces power loss,improves the voltage stability index(VSI)and limits total harmonic distortion.Simulations using the IEEE 33-bus EDN compared the ARO results with those of previous studies.In the first scenario,ideally integrated D-STATCOMs,PVs and WTs reduced losses by 34.79%,64.74%and 94.15%,respectively.VSI increases from 0.6965 to 0.7749,0.8804 and 0.967.The optimal WT integration of the first scenario outperformed the PVs and D-STATCOMs.The second step optimizes the WTs and PQ devices for non-linear loads.WTs and D-STATCOMs reduce the maximum total harmonic distortion of the voltage waveform by 5.21%with non-linear loads to 3.23%,while WTs and PPFs reduce it to 4.39%.These scenarios demonstrate how WTs and D-STATCOMs can improve network performance and PQ.The computational efficiency of ARO is compared to that of the pathfinder algorithm,future search algorithm,butterfly optimization algorithm and coyote optimization algorithm.ARO speeds up convergence and improves solution quality and comprehension.