当前位置：网站首页>LSTM-based distributed energy generation prediction (Matlab code implementation)

LSTM-based distributed energy generation prediction (Matlab code implementation)

2022-08-10 01:25:00 【Electrical Engineering Workshop】

欢迎来到本博客️️️
作者研究：主要研究方向是电力系统和智能算法、机器学习和深度学习.目前熟悉python网页爬虫、机器学习、群智能算法、深度学习的相关内容.希望将计算机和电网有效结合！️️️
博主优势：博客内容尽量做到思维缜密,逻辑清晰,为了方便读者,博主专门做了一个专栏目录,整个专栏只放了一篇文章,足见我对其重视程度：博主专栏目录.做到极度细致,方便大家进行学习！亲民！！！还有我开了一个专栏给女朋友的,很浪漫的喔,代码学累的时候去瞧一瞧,看一看：女朋友的浪漫邂逅.有问题可以私密博主,博主看到会在第一时间回复.
目前更新：电力系统相关知识,期刊论文,算法,机器学习和人工智能学习.
支持：如果觉得博主的文章还不错或者您用得到的话,可以关注一下博主,如果三连收藏支持就更好啦！这就是给予我最大的支持！

欢迎您的到来
个人主页：科研室
所有代码目录：电气工程科研社

【现在公众号名字改为：荔枝科研社】
博主课外兴趣：中西方哲学,送予读者：
做科研,涉及到一个深在的思想系统,需要科研者逻辑缜密,踏实认真,但是不能只是努力,很多时候借力比努力更重要,然后还要有仰望星空的创新点和启发点.当哲学课上老师问你什么是科学,什么是电的时候,不要觉得这些问题搞笑,哲学就是追究终极问题,寻找那些不言自明只有小孩子会问的但是你却回答不出来的问题.在我这个专栏记录我有空时的一些哲学思考和科研笔记：科研和哲思.建议读者按目录次序逐一浏览,免得骤然跌入幽暗的迷宫找不到来时的路,它不足为你揭示全部问题的答案,但若能让人胸中升起一朵朵疑云,也未尝不会酿成晚霞斑斓的别一番景致,万一它居然给你带来了一场精神世界的苦雨,那就借机洗刷一下原来存放在那儿的“真理”上的尘埃吧.
或许,雨过云收,神驰的天地更清朗.......

本文目录如下：️️️

目录
1 长短期记忆神经网络
1.1 网络介绍
1.2 网络训练
2 基于 LSTM Distributed energy generation forecastmatlab仿真结果
3 参考文献
4 Matlab代码
5 写在最后

1 长短期记忆神经网络

1.1 网络介绍

VFAP The heating load data of the system are non-linear data,and has time continuity, It is best to have time to deal with such issues

step recurrent neural network （Recurrent Neural Network,RNN）,But as the amount of collected data increases,RNN The gradient vanishing problem is prone to occur during training, This results in premature data being easily lost during training. LSTM 作为 RNN 的一种变体,可以解决 RNN Gradient vanishing and gradient explosion problems during training, It is mostly used in nonlinear time series forecasting,LSTM 的网络结构如图 5所示.

LSTM 网络结构

网络采用 3 个门控结构, Turn the hidden layer into cells with memory. Its memory function is：

$\left\{\begin{array}{l} f_{t}=\sigma\left(W_{\mathrm{fx}} x_{t}+W_{\mathrm{fh}} h_{t-1}+W_{\mathrm{fc}} h_{t-1}+b_{\mathrm{f}}\right) \\ i_{t}=\sigma\left(W_{\mathrm{ix}} x_{t}+W_{\mathrm{ih}} h_{t-1}+W_{\mathrm{ic}} h_{t-1}+b_{\mathrm{i}}\right) \\ o_{t}=\sigma\left(W_{\mathrm{ox}} x_{t}+W_{\mathrm{oh}} h_{t-1}+W_{\mathrm{oc}} h_{t-1}+b_{\mathrm{o}}\right) \\ \overline{\mathrm{c}}_{t}=\tanh \left(W_{\mathrm{cx}} x_{t}+W_{\mathrm{ch}} h_{t-1}+b_{\mathrm{c}}\right) \\ c_{t}=f_{t} \cdot c_{t-1}+i_{t} \cdot \overline{\mathrm{c}_{t}} \\ h_{t}=o_{t} \tanh \left(c_{t}\right) \end{array}\right.$

式中：ft,it,ot,ct 分别为遗忘门、输入门、Output gates and memory cell state quantities;Wfx,Wix,Wox 为输入层 xt 和隐含层 ht 在 t The associated weight of the moment;Wfh,Wih,Woh for the hidden layer t~t-1 The associated weight of the moment;Wfc,Wic,Woc for cellst~t-1 The associated weight of the moment;Wcx,Wch are the association weights between the cell and the input and the cell and the hidden layer, respectively;bf,bi,bo,bc Offsets for individual gating units and cells;ht-1 is the output of the previous unit cell,ht 为 t The output value of the cell at time;σ 为 sigmoid 激活函数.

1.2 网络训练

The input layer data is specified as VFAP The multi-feature quantities of the heating system are expressed as ：

$\left[\begin{array}{c} \boldsymbol{F}_{1} \\ \boldsymbol{F}_{2} \\ \cdots \\ \boldsymbol{F}_{n} \end{array}\right]=\left[\begin{array}{cccc} f_{1,1} & f_{1,2} & \cdots & f_{1, m} \\ f_{2,1} & f_{2,2} & \cdots & f_{2, m} \\ \cdots & \cdots & \cdots & \cdots \\ f_{n, 1} & f_{n, 2} & \cdots & f_{n, m} \end{array}\right]$

LSTM Sample format for the input layer of the network 为（samples,steps,features）,samples for each training batch,stepsis the feature step size for each slide,features is the input parameter feature quantity. Its dimensions are counted as m,Let the operating data of the heating period system be M,The time step size is t,特征个数为 f,the system is sharedM-t 个样本, 基于 LSTM 神经网络的 VFAP The specific steps of system load prediction are as follows.

2 基于 LSTM Distributed energy generation forecastmatlab仿真结果

3 参考文献

[1]胡洋,程志江,Cui Lan.基于LSTMinverter solar energy-Hourly load forecasting research of air source heat pump system[J].可再生能源,2022,40(07):866-873.DOI:10.13941/j.cnki.21-1469/tk.2022.07.017.

[2]Bi Guihong,赵鑫,李璐,Chen Shilong,Chen Chenpeng.Two-mode decompositionCNN-LSTMAn integrated short-term wind speed prediction model[J/OL].太阳能学报:1-10[2022-08-09].DOI:10.19912/j.0254-0096.tynxb.2021-1307.

[3]赵鑫,Chen Chenpeng,Bi Guihong,Chen Shilong.基于PAM-SSD-LSTM的短期风速预测[J/OL].太阳能学报:1-7[2022-08-09].DOI:10.19912/j.0254-0096.tynxb.2021-0900.