fish-kong/VMD-SSA-LSSVM-for-power-forecast

Stars
1
Language
Created about 4 years ago
Updated about 4 years ago

fish-kong/VMD-SSA-LSSVM-for-power-forecast

fish-kong

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

In this paper, LSSVM is used for short-term power load forecasting, and a short-term power load forecasting model based on LSSVM is proposed. At the same time, a Sparrow Algorithm (SSA) model is established to optimize the parameters of LLSVM to improve the forecasting accuracy. However, studies have shown that if a time series forecast model is built directly on the original series, the forecast data will lag the actual data. Such a model is meaningless. This is mainly due to the autocorrelation in the time series data, so I use VMD decomposition The method decomposes the original sequence, then models each sequence separately, and finally adds the results of each sequence test set as the final result. The comparative analysis results show that the prediction accuracy of this model is better than that of many other prediction models, and this model shows better performance in short-term load forecasting.

CSDN-code-link

由于CSDN博客里面不能直接上代码链接，涉嫌营销推广，因此建一个github仓库用于整理这些代码链接

Yolov5-instance-seg-tensorrt

fish-kong/Yolov5-Instance-Seg-Tensorrt-CPP

Yolov8-instance-seg-tensorrt

based on the yolov8，provide pt-onnx-tensorrt transcode and infer code by c++

Yolov5-obb-Tensorrt-Infer

face-recognize-by-comera

1、结合opencv，利用特征提取方法（LDA LBP PCA）进行特征提取建立模型库；2、利用电脑摄像头进行拍照，每隔3秒提取一个正面照进行特征提取，然后与模型库中的样本进行余弦距离相似度计算，实现人脸匹配识别

-rnn-each-types-of-rnn-for-regression

利用各种循环网络进行回归拟合，包括rnn，lstm，nlstm，bilstm

texture-classification-based-on-BPNN-and-dictionary

代码主要包括：1。特征提取首先对文本信息进行分词处理，采用基于字符串匹配的方法：假如一段叫：李二狗就是一个傻逼基于匹配的方法就是依次截取一到多个词，并与字典库进行匹配。如二狗，如果匹配到字典中有这个词，则将其分为一个词；当取到“狗就”，发现字典中没有与之匹配的，则说明这个不是一个词语，进行顺序操作，最优将这段话分为：李二狗就是一个傻逼。 2. 得到分词后的文本之后，就是转换成数字编码，因此电脑没办法识别汉字。这一部分叫特征表示，即用数字的方式表示中文文本，采用的方法是基于词带模型的特征表示：词带就是字典--程序中那个dictionary.mat。我们将分词处理之后的文本中的每一个词语，分别与字典中的词进行匹配，只要出现过就为1，否则为0。如字典中的词含有：李周吴郑王他妈的就是大傻逼一个三炮也是瓜娃子，一共13词（当然正常的词典都是上万个词），将1中得到的词语与之匹配，则李二狗就是一个傻逼对应的数字编码就应该是 1 0 0 0 0 0 1 0 1 1 0 0 0 3，通过2我们将文本表示成了数字，但是这样的表示通常都是稀疏的（因为一般字典都含有上万个词，所以得到的数字表示大部分都是0），为此我们利用降维方法，消除掉这些冗余特征。这里我们采用的PCA（主成分分析）进行降维，并降至15维。 4. 文本分类，采用的就是bp网络代码修改的地方不多，主要就是超参数的选择，（1）如pca的降维数，维数过高，包含冗余数据，过低又会删除掉重要信息。（2）bp网络结构的调整，如隐含层节点数，学习率，等

PSO-IPSO-LSTM-regression

this project use pso and ipso to optim lstm's hyperparams, include learning rate,hidden-nodes and training epoch number. and finally use ipso-lstm for power load forcast.

objective-detection

利用LBP进行特征提取，SVM进行2分类器建模；利用滑动窗口实现目标检测

Fusion-CNN-multi-CNN-for-feature-extraction-and-SVM-for-fault-diagnosis-

西储大学滚动轴承故障诊断

CNN_GRU-Regression

This project use CNN+GRU in tensorflow1.x/python to implement regression about time_series.The main content is to predict the wind power at the current time based on the wind speed and wind power data at the historical time。

Two-stream-CNN-for-rolling-bear-fault-diagnosis

Based on the dual-flow CNN, a new bearing fault diagnosis model is proposed. The model is composed of 2D-CNN and 1D-CNN. Among them, 2D-CNN takes wavelet time-frequency map as input, and 1D-CNN takes original vibration signal as input. After the feature extraction is implemented by the convolutional layer and the pooling layer, the output of the pooling layer of the two is spliced using a fully connected layer, and then the fault classification is achieved through the fully connected layer

rolling-bear-fault-diagnosis-based-wavelet-time-frequency-map-and-CNN

The wavelet time-frequency map is obtained from the original vibration signal, and then input into CNN to realize fault diagnosis, the test set has the highest diagnostic accuracy

dbn-tool-box

dbn tool box deep belief network tool box

CNN-PSO-SVM-rolling-bear-fault-diagnosis

In python, by using CNN(LENET-5) as a feature extractor to supervised learning from orignal data, then extract the output data of fc1 then put into a pso-svm for classification

supervised-LPP

supervised LPP, This algorithm used kernel algorithm and label date when calculate the Euclidean distance D , ,by contract, another supervised LPP used label-data when calculate the weights matrix

Wind-power-prediction-based-on-EEMD-CNN-LSTM

Wind power prediction based on EEMD-CNN-LSTM

Recurring-PCA-LPP-papers

2 days later, found a interesting paper. this paper combined PCA with LPP, and formed a comprehensive algorithm for data dimension deduction. in CNKI

SSA-LSTM-power-forecast

Python/tensorflow1.x，采用麻雀搜索算法对LSTM的超参数进行优化，需要的加我qq2919218574，有偿出售的

Wavelet-LSTM

by using tf2.0 construct a single-layer WNN (wavelet neural network)、Multi-layer WNN and a wavelet LSTM。采用tf2.0搭建小波神经网络、多隐层小波神经网络于小波长短时记忆网络，详情可以看我的csdn博客：

BBAVectors-tensortt-deploy

Whale-optimization-algorithm

python，Whale optimization algorithm used for function minimum optimization

SSA-DELM-regression

Using the sparrow search algorithm to optimize the deep extreme learning machine to achieve regression tasks, add me qq2919218574

2D-DCAE-RESNET-for-image-de-noising

提出基于误差学习的二卷卷积降噪自动编码器，采用tensorflow1.x 框架，详情移步我的csdn博客https://blog.csdn.net/qq_41043389/article/details/105037283

Network-intrusion-detection-based-on-feature-selection-PSO-CTree-

Aiming at the KDD data set using onehot encoding for feature value conversion, there is a problem of redundant features. This paper proposes to use particle swarm optimized algorithm combined with decision tree to achieve feature selection and detection classification.

IPSO_GRU-Regression

Using tensorflow1.x/python to implement a IPSO_GRU for regression，IPSO's here, it was mainly uesd to optimize hyper-parameters includess learning rate ,hidden_layer's number.

EEMD-IWOA-LSSVM-for-power-prediction

（MATLAB CODE） Establishing a time series forecasting model for PV power prediction based on LSSVM，Due to the autocorrelation of the original power data sequence, the predicted value and the actual value lag, so EEMD(EMD/CEEMD) is used to decompose the original sequence, and then the decomposed components are modeled in turn. To further mention accuracy, the improved whale-optimization-algorithm （IWOA）is used to optimize the lssvm。 (MATLAB代码)采用最小二乘支持向量机LSSVM建立光伏功率预测的时间序列预测模型，由于数据本身的自相关性，导致得到的预测值与实际值存在滞后。针对这个问题，首先对光伏功率序列进行EEMD得到imf分量，然后对各分量进行LSSVM建模。最后为进一步提高精度，采用改进的鲸鱼优化算法优化的lsvvm的核参数与惩罚参数，需要的可以加我qq2919218574 ，程序是matlab代码

Imporved-weed-algorithm-to-optimize-BP-network

Adopt improved weed algorithm to optimize the weight of BP network and achieve better regression fitting

CS-LSTM-for-powerload-forecast

python3.3 tensorflow1.x，Using LSTM to construct a time series forecast model for short-term power forecast, and using Cuckoo Search(CS) algorthim to optimize the iteration,learning rate, hidden-layer nodes.

Rotation-box-dimension-tool-and-dimension-file-adjustment

Rotation box dimension tool and dimension file adjustment

Time-frequency-graph-transformation-toolbox

Time-frequency graph transformation toolbox

EMD-SA-DBN-for-Wind-speed-forecast

Establishing a time series forecasting model for wind speed prediction based on DBN，Due to the autocorrelation of the wind speed sequence, the predicted value and the actual value lag, so EMD is used to decompose the wind speed sequence, and then the decomposed components are modeled in turn. To further mention accuracy, the simulated annealing algorithm is used to optimize the DBN。

SSA-DBN-Regression

麻雀优化DBN用于回归，最近写了好多麻雀优化的，我就不写博客了，需要买代码直接加我qq2919218574。Sparrow optimized DBN for regression. Recently I wrote a lot of sparrow optimized ones. I won’t write a blog anymore. If you need to buy the code and please add me qq2919218574 directly.

BiLSTM-attention-for-power-forecast

python3，tensorflow1.x，利用双向长短时记忆网络加注意力机制构建时间序列预测模型，然后用功率数据集进行验证，需要的加我qq2919218574，还可以用各自优化算法进行该模型的超参数寻优，收费的哟

sparrow-search-algorithm-for-kelm-regression

The sparrow search algorithm is a new heuristic search algorithm proposed in 2020. I use python to write and use it to optimize the extreme learning machine and the nuclear extreme learning machine to complete the regression task.

GA-DBN-for-multi-output-regression

Use DBN to complete the multi-input multi-output regression task, and at the same time, use the genetic algorithm to optimize the hidden layer parameter selection

Exchange-rate-forecast-based-on-EMD-and-BP

BP is used for time series modeling of exchange rate prediction. For the correlation problems in the sequence, EMD is used to decompose the original sequence, then each subsequence is decomposed to establish a BP model, and finally the prediction results of each subsequence are added as the final the result of

GA-DBN-Classification

Using MATLAB and DeepBeliefNetworksToolbox to implement a GA-DBN for classification task, Genetic algorithm（GA） is used to optimize the neuron's number of each hidden-layer

WOA-LSTM-time-series-forecast

python3.3 tensorflow1.x，Using LSTM to construct a time series forecast model for short-term power forecast, and using Whale optimization algorithmto （woa）optimize the iteration,learning rate, hidden-layer nodes. 利用tensorflow1.x，使用lstm进行短期电力负荷预测建模，并采用鲸鱼算法对lstm的迭代次数，学习率隐含层节点数进行寻优，详情看我博客： https://blog.csdn.net/qq_41043389（这个是布谷鸟的，我改了优化算法，所以就没写博客）需要的代码的可以加我qq2919218574,出售的，不是免费，不是免费，不是免费

GA-DA-for-feature-select

In the Python environment, the genetic algorithm and the dragonfly algorithm are first combined to form a modified dragonfly genetic algorithm, and the effectiveness of the improved method is verified by using the test function. Secondly, the improved algorithm and decision tree are applied to feature selection (classification task) to remove redundant features from the original features

sparrow-search-algorithm-for-cnn-s-hyper-parameters

在matlab2018a中，采用麻雀算法SSA对CNN的超参数进行寻优，包括卷积核大小，数量，全连接层神经元数，迭代次数，学习率等，有回归有分类，需要的加我qq2919218574，代码收费的哟

SSA-DBN-classification

Combining the advantages of deep belief network (DBN) in extracting features and processing high-dimensional and non-linear data, a classification method based on deep belief network is proposed. This method uses the Fourier spectrum (FFT) of the original time domain signal to train a deep confidence network through deep learning. Its advantage is that the method does not need to set parameters when performing FFT on the signal, and directly uses all spectral components for modeling, so there is no need for complexity The feature selection method has strong versatility and adaptability. Finally, in order to further enhance the classification accuracy of DBN, the Sparrow Search Algorithm (SSA) is used to optimize the weight parameters of DBN. The experimental results show that the method proposed in this paper can effectively improve the classification and recognition accuracy.

EMD-GA-DBN-Regression

（MATLAB CODE） Establishing a time series forecasting model for wind speed prediction based on DBN，Due to the autocorrelation of the wind speed sequence, the predicted value and the actual value lag, so EMD is used to decompose the wind speed sequence, and then the decomposed components are modeled in turn. To further mention accuracy, the simulated annealing algorithm is used to optimize the DBN。 (MATLAB代码)采用深度置信网络DBN建立风速预测的时间序列预测模型，由于数据本身的自相关性，导致得到的预测值与实际值存在滞后。针对这个问题，首先对风速序列进行EMD得到imf分量，然后对各分量进行DBN建模。最后为进一步提高精度，采用模拟退火算法的DBN各隐含层节点进行优化，需要的可以加我qq2919218574 ，效果可以看csdn博客https://blog.csdn.net/qq_41043389/article/details/104517495 程序是matlab代码

SWT-ISSA-LSTM-time-series-forecast

The long and short-term memory network LSTM has attracted attention in recent years for the short-term time series prediction problem. However, because this method is a deep learning method, it usually faces the influence of many hyperparameters. As we all know, the setting of deep learning hyperparameters is not There is no clear guideline, and most of them use empirical methods, such as learning rate 1e-3, 1e-4, etc. The number of iterations is set according to the change of loss curve, etc. This method is simple to try and find the effect is better A group of people, time-consuming and labor-intensive. To this end, this paper will use the improved sparrow search algorithm ISSA to optimize the parameters of LSTM, while using synchronous squeeze wavelet SWT to filter the original data to reduce noise, and use the data of the noise reduction meeting to model, and finally use the example verification to show that, The prediction effect of the SWT-ISSA-LSTM model is better. 长短时记忆网络LSTM在针对短时时间序列预测问题上近来年受到大家的关注，但由于该方法为深度学习方法，通常面临着众多超参数的影响，而众所周知，关于深度学习超参数的设置并没有一直明确的指导方针，大多采用经验方法，比如学习率1e-3,1e-4啥的，迭代次数根据loss曲线的变化等进行设置，这种方法说简单的就是无限尝试，找到效果比较好的一组，耗时耗力。为此，本文将采用改进的麻雀搜索算法ISSA对 LSTM进行参数优化 , 同时采用同步挤压小波SWT对原始数据进行滤波降噪，并采用降噪会的数据进行建模，最后用实例验证表明 ,SWT-ISSA-LSTM 模型的预测效果更佳。