CN110619389A - Load prediction method and system of combined cooling heating and power system based on LSTM-RNN - Google Patents
Load prediction method and system of combined cooling heating and power system based on LSTM-RNN Download PDFInfo
- Publication number
- CN110619389A CN110619389A CN201910898056.6A CN201910898056A CN110619389A CN 110619389 A CN110619389 A CN 110619389A CN 201910898056 A CN201910898056 A CN 201910898056A CN 110619389 A CN110619389 A CN 110619389A
- Authority
- CN
- China
- Prior art keywords
- load
- data
- load prediction
- training
- prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 39
- 238000001816 cooling Methods 0.000 title claims abstract description 38
- 238000010438 heat treatment Methods 0.000 title claims abstract description 34
- 238000012549 training Methods 0.000 claims abstract description 52
- 238000013528 artificial neural network Methods 0.000 claims abstract description 31
- 125000004122 cyclic group Chemical group 0.000 claims abstract description 19
- 230000006403 short-term memory Effects 0.000 claims abstract description 6
- 230000007787 long-term memory Effects 0.000 claims abstract description 5
- 230000015654 memory Effects 0.000 claims description 23
- 238000012360 testing method Methods 0.000 claims description 8
- 238000010219 correlation analysis Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000003860 storage Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 2
- 238000010187 selection method Methods 0.000 claims description 2
- 230000008878 coupling Effects 0.000 abstract description 9
- 238000010168 coupling process Methods 0.000 abstract description 9
- 238000005859 coupling reaction Methods 0.000 abstract description 9
- 238000003062 neural network model Methods 0.000 abstract description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000002474 experimental method Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 230000000052 comparative effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 230000000306 recurrent effect Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000003344 environmental pollutant Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 231100000719 pollutant Toxicity 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 239000002918 waste heat Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Biophysics (AREA)
- Strategic Management (AREA)
- Artificial Intelligence (AREA)
- Human Resources & Organizations (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a load prediction method and a system of a combined cooling, heating and power system based on LSTM-RNN, wherein the method comprises the following steps: receiving heat load, cold load and historical data of electric load, and determining input and output data; training a load prediction network model by taking part of input and output data as training data based on the long-term and short-term memory cyclic neural network; and performing load prediction based on the load prediction network model. According to the invention, the coupling relation among the cold, heat and power loads is excavated by adopting the long and short term memory cyclic neural network model, so that the load prediction precision of the combined cold, heat and power system based on the LSTM-RNN is improved.
Description
Technical Field
The invention belongs to the technical field of load prediction of renewable energy systems, and particularly relates to a Combined Cooling and Heating and Power (CCHP) system load prediction method and system based on a long-short term memory-circulating neural network (LSTM-RNN).
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The combined cooling heating and power system is a comprehensive energy production and utilization system based on energy gradient utilization. In the conventional energy supply system, the power generation efficiency is only about 40%, and the remaining about 60% of energy is wasted. And the combined cooling heating and power system can effectively reduce the emission of pollutants by recovering the waste heat of power generation, so that the energy utilization rate is obviously improved to more than 80 percent. Therefore, the combined cooling, heating and power system has become the most important direction and form of the development of distributed energy.
The accurate prediction of the cooling, heating and power load is a key factor and a basic premise of the optimal design and operation scheduling of the combined cooling, heating and power system, and the prediction precision directly influences the effectiveness of the power supply system and the utilization rate of energy. Therefore, effective prediction of the load of the combined cooling heating and power system has important significance for effective operation of the combined cooling heating and power system and improvement of energy utilization rate.
Over the past few years, a number of related workers have developed rich methods of load prediction. However, the inventors found that most of the literatures predict the cooling, heating and power loads in a univariate manner, and that the coupling relationship between the cooling, heating and power loads is rarely considered. And the load sequence has periodicity, and the partial prediction method is not suitable for long-time periodic load sequence prediction.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a load prediction method and a load prediction system of a combined cooling heating and power system based on LSTM-RNN, wherein the input and output variables are determined based on correlation analysis, and the coupling relation between the cooling, heating and power loads is mined by adopting a long-short term memory cyclic neural network model, so that the load prediction precision is improved.
In order to achieve the above object, one or more embodiments of the present invention provide the following technical solutions:
a load prediction method of a combined cooling heating and power system based on LSTM-RNN comprises the following steps:
receiving heat load, cold load and electrical load historical data;
determining a time period adopted by input data according to the correlation between a certain moment of historical data and a plurality of moments before the certain moment;
determining input and output data according to correlation coefficients between heat load, cold load and historical data of electric load;
inputting the training set into a long-short term memory cyclic neural network, and training a load prediction network model;
and performing load prediction based on the load prediction network model.
Further, determining the time period for the input data comprises:
calculating autocorrelation coefficients and partial correlation coefficients between a certain moment of historical data and a plurality of moments before the moment, and determining continuous time periods when absolute values of the autocorrelation coefficients are all larger than a set threshold value and continuous time periods when absolute values of the partial correlation coefficients are all larger than the set threshold value;
and selecting the time period overlapped with the two time periods as the time period adopted by the input data.
Further, after the input and output data are determined, preprocessing is also performed on the input and output data: data were normalized using zero mean normalization.
Further, the training set selection method comprises the following steps: and defining a time period adopted by input data as a prediction period, forming a group of training data by a plurality of continuous prediction periods, and selecting the training set to be composed of a plurality of groups of continuous training data.
Further, training the load prediction network model comprises:
sequentially inputting each group of training data in the training set to an input layer of the long-short term memory cycle neural network according to a time sequence for iterative training until the network converges;
and after network convergence, error prediction is carried out based on the test set, and parameter optimization is carried out until the error is minimum.
Further, the long-short term memory cycle neural network comprises an input layer, a hidden layer and an output layer, wherein the hidden layer comprises a sigmoid neural network layer and a multiplication structure.
One or more embodiments provide an LSTM-RNN based combined cooling, heating and power system load prediction system, comprising:
the data acquisition module is used for receiving heat load, cold load and historical data of electric load;
the correlation analysis module is used for determining a time period adopted by input data according to the correlation between a certain moment of historical data and a plurality of moments before the certain moment; determining input and output data according to correlation coefficients between heat load, cold load and historical data of electric load;
the model training module is used for inputting a training set into the long-term and short-term memory cyclic neural network and training the load prediction network model;
and the load prediction module is used for predicting the load based on the load prediction network model.
One or more embodiments provide an electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the LSTM-RNN based combined cooling, heating, and power system load prediction method when executing the program.
One or more embodiments provide a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the LSTM-RNN based combined cooling, heating and power system load prediction method.
The above one or more technical solutions have the following beneficial effects:
the time period with obvious influence on load prediction is determined based on autocorrelation and partial correlation analysis, so that training samples are selected, a multivariate prediction method is determined based on the Pearson correlation coefficient, and the accuracy of a prediction result is improved.
The invention adopts a long-short term memory cyclic neural network model, the long-short term memory network is a time cyclic neural network, is suitable for predicting important events with relatively long time sequence middle interval and delay, and can excavate the coupling relation among the cold, heat and power loads, thereby effectively improving the load prediction precision of the combined cooling, heat and power system based on the LSTM-RNN.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description serve to explain the invention and not to limit the invention.
FIG. 1 is a schematic diagram of a conventional recurrent neural network architecture;
FIG. 2 is a flowchart of a method for load prediction of a combined cooling, heating and power system according to one or more embodiments of the invention;
FIG. 3 is a diagram of a model structure of an LSTM prediction algorithm in accordance with one or more embodiments of the present invention;
FIG. 4 is a graph illustrating an analysis of the autocorrelation and partial correlation of thermal load history data in accordance with one or more embodiments of the present invention;
FIG. 5 is a graph illustrating an analysis of auto-correlation and partial correlation of historical data of a cooling load in accordance with one or more embodiments of the present invention;
FIG. 6 is a graph illustrating an autocorrelation and partial correlation analysis of historical data of an electrical load in accordance with one or more embodiments of the present invention;
FIG. 7 is a graph comparing results of heat load multivariate prediction and univariate prediction in one or more embodiments of the invention;
FIG. 8 is a graph comparing results of a multi-variable prediction versus a single-variable prediction of cooling load in one or more embodiments of the invention;
FIG. 9 is a graph comparing results of multi-variable prediction and single-variable prediction of electrical load in accordance with one or more embodiments of the present disclosure;
FIG. 10 is a comparison of predicted results for different prediction methods in one or more embodiments of the invention.
Detailed Description
It is to be understood that the following detailed description is exemplary and is intended to provide further explanation of the invention as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The embodiments and features of the embodiments of the present invention may be combined with each other without conflict.
The noun explains:
a Recurrent Neural Network (RNN), which is an artificial Neural Network in which nodes are directionally connected into a ring, and which is a three-layer Neural Network including an input layer, a hidden layer, and an output layer;
a Long Short-Term Memory network (LSTM) is one kind of cyclic neural network, and can solve the problem of gradient diffusion in the cyclic neural network.
As shown in fig. 1, RNN is a neural network having a feedback structure, and unlike a conventional feedforward neural network, its output is related not only to the current input and network weight, but also to the input of the previous network, and the coupling relationship between the front and the back of the sequence can be mined. The RNN models time by adding a self-join hidden layer at a point in time, in other words, the feedback of the hidden layer is not just output, but also includes the hidden layer at the next moment, which structure may enable the RNN to better utilize the information of the sequence data. RNNs are therefore well suited for prediction of time series.
Example one
The embodiment discloses a load prediction method of a long-short term memory cyclic neural network-based LSTM-RNN (localized surface-to-network neural network) -based combined cooling, heating and power system, as shown in FIG. 2, comprising the following steps:
step 1: and calculating the correlation of the historical data and determining the input and output quantities of the model.
Specifically, in step 1, autocorrelation coefficients, partial correlation coefficients, and pearson correlation coefficients are respectively used to analyze the correlation between historical data, so as to select appropriate data as input and output variables.
The autocorrelation coefficients are used for measuring the degree of correlation of the same event between two different periods, and specifically, with a certain time as a reference, the autocorrelation coefficients of each time of the time and a time period before the time are respectively calculated. The formula can be expressed as:
in the formula, xtLoad data for time t (time in hours in this example), xt+lFor the load data at time t + l, Cov (-) is the data covariance, and Var (-) is the data variance.
The partial correlation coefficient is obtained by independently researching the closeness degree of the correlation between two elements without considering the influence of other elements temporarily, and the calculation method comprises an iteration method, a correlation matrix inversion method and the like. Specifically, with a certain time as a reference, the partial correlation coefficient of each time in a certain time period before the time is calculated.
The method for determining the time period adopted by the input data by comprehensively considering the autocorrelation coefficient and the partial correlation coefficient specifically comprises the following steps:
calculating autocorrelation coefficients and partial correlation coefficients between a certain moment of historical data and a plurality of moments before the moment, and determining continuous time periods when absolute values of the autocorrelation coefficients are all larger than a set threshold value and continuous time periods when absolute values of the partial correlation coefficients are all larger than the set threshold value;
and selecting the time period overlapped with the two time periods as the time period adopted by the input data.
The results of the autocorrelation coefficient and the partial correlation coefficient among the history data of the heat load, the cold load, and the electrical load in the present embodiment are shown in fig. 4, 5, and 6, respectively. As can be seen from analysis, the data at a certain moment has a large correlation not only with the data of the last few hours in the recent past but also with the data of the same moment in the previous day, so the data of the first 1-24 hours at a specific moment is selected as input in the embodiment.
The Pearson correlation coefficient is used for measuring the strength of the correlation relationship among the three loads, and can be calculated as follows:
in the formula, XiAnd YiIs two time sequences, X' is XiY' is YiN is the number of time-series data.
Judging whether the input data is single-factor or multi-factor by calculating the Pearson correlation coefficient between every two of the heat load, the cold load and the historical data of the electric load, and considering that a univariate building prediction model is adopted if the absolute value of the Pearson correlation coefficient between every two is smaller than a set threshold value and the Pearson correlation coefficient between every two is considered to have no correlation; and if the absolute values of the Pearson correlation coefficients between every two pearson correlation coefficients are larger than a set threshold value, the three are considered to have certain coupling, and a multivariate is adopted to establish a prediction model.
The pearson correlation coefficients of the three types of load data acquired in the present embodiment are shown in table 1 below. Analysis shows that the three loads have certain correlation with each other. Multivariate (thermal, cold and electrical) models are therefore used to build the prediction model.
TABLE 1 Pearson correlation coefficient of cooling, heating and power loads
Step 2: and respectively preprocessing the input and output data sets.
Specifically, the data preprocessing in step 2 is to normalize the data by zero-mean normalization, and the data can be calculated as:
where μ is the mean and σ is the standard deviation.
And step 3: the preprocessed data set is divided into a training set, a verification set and a test set.
Defining a time period adopted by the input data determined in the step 1 as a prediction period, forming a group of training data by a plurality of continuous prediction periods, and selecting a training set consisting of a plurality of groups of continuous training data; a prediction period after the training set is a test set; and selecting a part in the training set as a verification set.
Specifically, the prediction period in this embodiment is 24 hours, i.e., one day, and data two months before a certain day is selected as a training set, wherein a selected part in the training set is a verification set, and data on the day of the certain day is used as a test set.
And 4, step 4: and training and updating model parameters by using the processed data, and constructing a proper long-short term memory cyclic neural network prediction model.
And (4) sequentially inputting each group of training data in the training set into the long-short term memory cyclic neural network according to the time sequence, and learning to obtain a prediction model.
In this embodiment, the data of the two months are divided into 7 days to be sequentially input, for example, 1-7 days are used as one group, and 8 days are output; the next 2-8 days as input and the 9 th day as output; then 3-10 days as input, 11 days as output … … and so on until 2 months of data are all input and trained. Specifically, as shown in fig. 3, the LSTM-RNN load prediction model generally has three layers, including an input layer, a hidden layer, and an output layer, where the input layer is data (including cold load, heat load, and electrical load) of 24 hours per day for the first seven days of the current date to be predicted, and a date type, i.e., r is 7 in fig. 3, and the date type is seven days of the week, and is represented by numerals 1 to 7. The hidden layer determines the removal or addition of information in each cell state by means of a "gate" structure, which is a way of selecting information. The method comprises a sigmoid neural network layer and a multiplication structure, wherein the sigmoid neural network layer outputs a numerical value between 0 and 1, which represents how much information of each part can pass through, if the numerical value is 0, no information passes through, and if the numerical value is 1, any quantity is allowed to pass through. The output layer is 24 hours of data (including cold load, heat load, and electrical load) on the test day.
The door structure of the hidden layer comprises three door structures, namely a forgetting door, an input door and an output door.
The forgetting gate is used for determining the information discarded from the cell state, and the gate determines to output a value between 0 and 1 to the cell state Ct-1 by reading the output ht-1 of the previous layer and the current input xt through a sigma function, so as to determine the retention and discard cases of the state, which can be expressed as:
ft＝σ(Wf·[ht-1,xt]+bf)
in the formula, Wf and bf represent the weight matrix and the offset vector of the forgetting gate, respectively.
The input gate is used for determining a new state to be added to the cell, and comprises two parts, wherein the first part uses the sigmoid function to determine a value to be updated, and the other part uses a tanh layer to create a new input vector, which can be represented as:
it＝σ(Wi·[ht-1,xt]+bi)
Ct＝ft*Ct-1+it*tanh(Wc·[ht-1,xt]+bc)
the output gate is used for determining a final output result. Firstly, determining information which needs to be output in a cell state through a sigmoid layer, then processing the cell state information generated by an input gate by using a tanh layer, and multiplying the processed information by the output of the sigmoid layer to obtain a final output result, which can be expressed as:
ot＝σ(Wo·[ht-1,xt]+bo)
ht＝ot*tanh(Ct)
yt＝Wy·ht+by
the sigmoid layer function is as follows:
the tanh layer function is:
and 4, step 4: and iteratively training the LSTM network by using the processed data.
And 5: and judging whether the network is converged, if so, calculating the prediction error of the test set in the next step, and if not, returning to the step 4 to continue iteration.
Step 6: and (4) searching proper parameters until the prediction error of the test set reaches the minimum, and otherwise, returning to the step 3 to reform the LSTM network.
And 7: and carrying out load prediction by using the LSTM network obtained by the training to obtain a result.
Example two
The embodiment aims to provide a combined cooling, heating and power system load prediction system.
In order to achieve the above object, the present embodiment provides a combined cooling, heating and power system load prediction system, including:
the data acquisition module is used for receiving heat load, cold load and historical data of electric load;
the correlation analysis module is used for determining a time period adopted by input data according to the correlation between a certain moment of historical data and a plurality of moments before the certain moment; determining input and output data according to correlation coefficients between heat load, cold load and historical data of electric load;
the model training module is used for inputting a training set into the long-term and short-term memory cyclic neural network and training the load prediction network model;
and the load prediction module is used for predicting the load based on the load prediction network model.
EXAMPLE III
The embodiment aims at providing an electronic device.
An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the program, comprising:
receiving heat load, cold load and electrical load historical data;
determining a time period adopted by input data according to the correlation between a certain moment of historical data and a plurality of moments before the certain moment;
determining input and output data according to correlation coefficients between heat load, cold load and historical data of electric load;
inputting the training set into a long-short term memory cyclic neural network, and training a load prediction network model;
and performing load prediction based on the load prediction network model.
Example four
An object of the present embodiment is to provide a computer-readable storage medium.
A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, performs the steps of:
receiving heat load, cold load and electrical load historical data;
determining a time period adopted by input data according to the correlation between a certain moment of historical data and a plurality of moments before the certain moment;
determining input and output data according to correlation coefficients between heat load, cold load and historical data of electric load;
inputting the training set into a long-short term memory cyclic neural network, and training a load prediction network model;
and performing load prediction based on the load prediction network model.
The steps involved in the above second, third and fourth embodiments correspond to the first embodiment of the method, and the detailed description thereof can be found in the relevant description of the first embodiment. The term "computer-readable storage medium" should be taken to include a single medium or multiple media containing one or more sets of instructions; it should also be understood to include any medium that is capable of storing, encoding or carrying a set of instructions for execution by a processor and that cause the processor to perform any of the methods of the present invention.
Results of the experiment
In order to prove the effectiveness of the method, the Rstudio language is adopted to write programs, two groups of comparison experiments are respectively constructed, one group is that three loads are respectively predicted by adopting a single variable input LSTM prediction model and a multi-variable input LSTM prediction model, and the other group is that the heat loads are respectively predicted by adopting an LSTM prediction model, an LS-SVM prediction model and an RBF neural network model as examples. Training is carried out by using data 24 hours a day in two months before a certain moment, and input and output are carried out by taking 7 days as a period, so as to carry out iterative training. And respectively adopting a mean square error RMSE and a mean absolute percentage error MAPE to evaluate the performance of the prediction result.
In the formula, xiDenotes the actual value, x'iThe predicted value at that time is shown, and n is the number of predicted points. The results of the simulation of the prediction models in the first set of comparative experiments are shown in fig. 7, fig. 8, and fig. 9, respectively. The following table 2 shows evaluation indexes of the three loads by using the respective prediction methods.
TABLE 2 evaluation index of each prediction algorithm
The analysis shows that the mean square error and the mean absolute error of the heat load are 1.4744 and 1.1938 respectively by using a univariate prediction method, and the mean square error and the mean absolute error of the heat load are 1.0314 and 0.7707 respectively by using a multivariate prediction method; the mean square error and the mean absolute error of the cold load are 3.4130 and 0.4076 respectively by using a univariate prediction method, and the mean square error and the mean absolute error of the cold load are 1.9587 and 0.2321 respectively by using a multivariate prediction method; the mean square error and the mean absolute error of the electrical load are 0.1470 and 0.0752 respectively by using a univariate prediction method, and the mean square error and the mean absolute error of the electrical load are 0.0683 and 0.0410 respectively by using a multivariate prediction method. For each type of load, the precision of the multivariable prediction model is higher than that of the univariate prediction model, and the multivariable prediction method can effectively mine the coupling relation between the cooling, heating and power loads, so that the prediction precision of the load is improved.
The results of the simulation of the prediction models in the second set of comparative experiments are shown in fig. 10. The mean square error obtained by predicting the heat load by using the RBF neural network model is 1.0655 and the mean absolute error is 0.7707; the mean square error obtained by predicting the heat load by using the LS-SVM model is 1.7826, and the mean absolute error is 1.3953; the mean square error obtained by predicting the thermal load with the LSTM model is 1.0341, and the mean absolute error is 0.7136. Therefore, compared with the traditional machine learning algorithm, the LSTM algorithm can effectively mine the coupling relation between the current data and the historical data of the time sequence, and the prediction precision of the LSTM algorithm can be effectively improved for loads with strong periodicity.
One or more of the above embodiments have the following technical effects:
the time period with obvious influence on load prediction is determined based on autocorrelation and partial correlation analysis, so that training samples are selected, a multivariate prediction method is determined based on the Pearson correlation coefficient, and the accuracy of a prediction result is improved.
The invention adopts a long-short term memory cyclic neural network model, the long-short term memory network is a time cyclic neural network, is suitable for predicting important events with relatively long time sequence middle interval and delay, and can excavate the coupling relation among the cold, heat and power loads, thereby effectively improving the load prediction precision of the combined cooling, heat and power system based on the LSTM-RNN.
Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using general purpose computer means, or alternatively, they can be implemented using program code that is executable by computing means, such that they are stored in memory means for execution by the computing means, or they are separately fabricated into individual integrated circuit modules, or multiple modules or steps of them are fabricated into a single integrated circuit module. The present invention is not limited to any specific combination of hardware and software.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.
Claims (9)
1. A load prediction method of a combined cooling heating and power system based on LSTM-RNN is characterized by comprising the following steps:
receiving heat load, cold load and electrical load historical data;
determining a time period adopted by input data according to the correlation between a certain moment of historical data and a plurality of moments before the certain moment;
determining input and output data according to correlation coefficients between heat load, cold load and historical data of electric load;
inputting the training set into a long-short term memory cyclic neural network, and training a load prediction network model;
and performing load prediction based on the load prediction network model.
2. The combined cooling, heating and power system load prediction method according to claim 1, wherein determining the time period for the input data comprises:
calculating autocorrelation coefficients and partial correlation coefficients between a certain moment of historical data and a plurality of moments before the moment, and determining continuous time periods when absolute values of the autocorrelation coefficients are all larger than a set threshold value and continuous time periods when absolute values of the partial correlation coefficients are all larger than the set threshold value;
and selecting the time period overlapped with the two time periods as the time period adopted by the input data.
3. The combined cooling heating and power system load prediction method according to claim 1, wherein after the input and output data are determined, preprocessing is further performed on the input and output data: data were normalized using zero mean normalization.
4. The combined cooling heating and power system load prediction method according to claim 2, wherein the training set selection method comprises: and defining a time period adopted by input data as a prediction period, forming a group of training data by a plurality of continuous prediction periods, and selecting the training set to be composed of a plurality of groups of continuous training data.
5. The combined cooling heating and power system load prediction method according to claim 4, wherein training the load prediction network model comprises:
sequentially inputting each group of training data in the training set to an input layer of the long-short term memory cycle neural network according to a time sequence for iterative training until the network converges;
and after network convergence, error prediction is carried out based on the test set, and parameter optimization is carried out until the error is minimum.
6. The combined cooling, heating and power system load prediction method according to claim 5, wherein the long-short term memory cycle neural network comprises an input layer, a hidden layer and an output layer, wherein the hidden layer comprises a sigmoid neural network layer and a multiplication structure.
7. An LSTM-RNN-based combined cooling heating and power system load prediction system is characterized by comprising:
the data acquisition module is used for receiving heat load, cold load and historical data of electric load;
the correlation analysis module is used for determining a time period adopted by input data according to the correlation between a certain moment of historical data and a plurality of moments before the certain moment; determining input and output data according to correlation coefficients between heat load, cold load and historical data of electric load;
the model training module is used for inputting a training set into the long-term and short-term memory cyclic neural network and training the load prediction network model;
and the load prediction module is used for predicting the load based on the load prediction network model.
8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the LSTM-RNN based combined cooling and heating system load prediction method according to any one of claims 1-6 when executing the program.
9. A computer readable storage medium having stored thereon a computer program, which when executed by a processor implements the LSTM-RNN based combined cooling, heating and power system load prediction method according to any of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910898056.6A CN110619389A (en) | 2019-09-23 | 2019-09-23 | Load prediction method and system of combined cooling heating and power system based on LSTM-RNN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910898056.6A CN110619389A (en) | 2019-09-23 | 2019-09-23 | Load prediction method and system of combined cooling heating and power system based on LSTM-RNN |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110619389A true CN110619389A (en) | 2019-12-27 |
Family
ID=68924139
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910898056.6A Pending CN110619389A (en) | 2019-09-23 | 2019-09-23 | Load prediction method and system of combined cooling heating and power system based on LSTM-RNN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110619389A (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932015A (en) * | 2020-08-12 | 2020-11-13 | 武汉中电节能有限公司 | Method and system for predicting cold and heat load of regional cooling and heating |
| CN112116153A (en) * | 2020-09-18 | 2020-12-22 | 上海电力大学 | Park multivariate load joint prediction method for coupling Copula and stacked LSTM network |
| CN112836884A (en) * | 2021-02-07 | 2021-05-25 | 国网福建省电力有限公司经济技术研究院 | Copula-DBiLSTM-based method for accurately predicting multi-element load of comprehensive energy system |
| CN112862179A (en) * | 2021-02-03 | 2021-05-28 | 国网山西省电力公司吕梁供电公司 | Energy consumption behavior prediction method and device and computer equipment |
| CN112862213A (en) * | 2021-03-09 | 2021-05-28 | 中国华能集团清洁能源技术研究院有限公司 | Heat supply demand estimation method, system and equipment based on periodic feedback LSTM |
| CN113159394A (en) * | 2021-03-30 | 2021-07-23 | 国网河北省电力有限公司雄安新区供电公司 | Intelligent power distribution network partition cold and hot load prediction method and device |
| CN113435654A (en) * | 2021-07-05 | 2021-09-24 | 江苏城乡建设职业学院 | ELMAN neural network-based short-term load prediction method |
| CN113469434A (en) * | 2021-06-30 | 2021-10-01 | 国网综合能源服务集团有限公司 | Multi-energy load prediction method, device, prediction equipment and storage medium |
| CN113553356A (en) * | 2021-07-23 | 2021-10-26 | 中海石油(中国)有限公司 | Drilling parameter prediction method and system |
| CN113609762A (en) * | 2021-07-27 | 2021-11-05 | 山东大学 | Electric cooling and heating load combined prediction method and system based on GRU-MTL |
| CN115034518A (en) * | 2022-07-29 | 2022-09-09 | 华北电力大学 | Method and system for predicting multi-element load of cooling, heating and power |
| CN115481788A (en) * | 2022-08-31 | 2022-12-16 | 北京建筑大学 | Load prediction method and system for phase change energy storage system |
| CN116127840A (en) * | 2023-01-05 | 2023-05-16 | 无锡钊源电力电子有限公司 | Data center load prediction method based on data driving |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107239859A (en) * | 2017-06-05 | 2017-10-10 | 国网山东省电力公司电力科学研究院 | The heating load forecasting method of Recognition with Recurrent Neural Network is remembered based on series connection shot and long term |
| CN109949180A (en) * | 2019-03-19 | 2019-06-28 | 山东交通学院 | A kind of the cool and thermal power load forecasting method and system of ship cooling heating and power generation system |
-
2019
- 2019-09-23 CN CN201910898056.6A patent/CN110619389A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107239859A (en) * | 2017-06-05 | 2017-10-10 | 国网山东省电力公司电力科学研究院 | The heating load forecasting method of Recognition with Recurrent Neural Network is remembered based on series connection shot and long term |
| CN109949180A (en) * | 2019-03-19 | 2019-06-28 | 山东交通学院 | A kind of the cool and thermal power load forecasting method and system of ship cooling heating and power generation system |
Non-Patent Citations (3)
| Title |
|---|
| KE LI 等: ""Load forecasting method for CCHP system based on deep learning strategy using LSTM-RNN"", 《IEEE》 * |
| 朱仁崎 等: "《组织行为学原理与实践》", 30 June 2018, 湖南大学出版社 * |
| 金宁德 等: "《非线性信息处理技术》", 31 January 2017, 天津大学出版社 * |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932015A (en) * | 2020-08-12 | 2020-11-13 | 武汉中电节能有限公司 | Method and system for predicting cold and heat load of regional cooling and heating |
| CN112116153B (en) * | 2020-09-18 | 2022-10-04 | 上海电力大学 | Park multivariate load joint prediction method coupling Copula and stacked LSTM network |
| CN112116153A (en) * | 2020-09-18 | 2020-12-22 | 上海电力大学 | Park multivariate load joint prediction method for coupling Copula and stacked LSTM network |
| CN112862179A (en) * | 2021-02-03 | 2021-05-28 | 国网山西省电力公司吕梁供电公司 | Energy consumption behavior prediction method and device and computer equipment |
| CN112836884A (en) * | 2021-02-07 | 2021-05-25 | 国网福建省电力有限公司经济技术研究院 | Copula-DBiLSTM-based method for accurately predicting multi-element load of comprehensive energy system |
| CN112862213A (en) * | 2021-03-09 | 2021-05-28 | 中国华能集团清洁能源技术研究院有限公司 | Heat supply demand estimation method, system and equipment based on periodic feedback LSTM |
| CN112862213B (en) * | 2021-03-09 | 2024-02-20 | 中国华能集团清洁能源技术研究院有限公司 | Heat supply demand prediction method, system and equipment based on periodic feedback LSTM |
| CN113159394A (en) * | 2021-03-30 | 2021-07-23 | 国网河北省电力有限公司雄安新区供电公司 | Intelligent power distribution network partition cold and hot load prediction method and device |
| CN113469434A (en) * | 2021-06-30 | 2021-10-01 | 国网综合能源服务集团有限公司 | Multi-energy load prediction method, device, prediction equipment and storage medium |
| CN113435654A (en) * | 2021-07-05 | 2021-09-24 | 江苏城乡建设职业学院 | ELMAN neural network-based short-term load prediction method |
| CN113435654B (en) * | 2021-07-05 | 2024-02-27 | 江苏城乡建设职业学院 | ELMAN neural network-based short-term load prediction method |
| CN113553356A (en) * | 2021-07-23 | 2021-10-26 | 中海石油(中国)有限公司 | Drilling parameter prediction method and system |
| CN113609762A (en) * | 2021-07-27 | 2021-11-05 | 山东大学 | Electric cooling and heating load combined prediction method and system based on GRU-MTL |
| CN115034518A (en) * | 2022-07-29 | 2022-09-09 | 华北电力大学 | Method and system for predicting multi-element load of cooling, heating and power |
| CN115034518B (en) * | 2022-07-29 | 2024-07-23 | 华北电力大学 | Cold-hot electricity multi-element load prediction method and system |
| CN115481788A (en) * | 2022-08-31 | 2022-12-16 | 北京建筑大学 | Load prediction method and system for phase change energy storage system |
| CN115481788B (en) * | 2022-08-31 | 2023-08-25 | 北京建筑大学 | Phase change energy storage system load prediction method and system |
| CN116127840A (en) * | 2023-01-05 | 2023-05-16 | 无锡钊源电力电子有限公司 | Data center load prediction method based on data driving |
| CN116127840B (en) * | 2023-01-05 | 2023-09-22 | 无锡钊源电力电子有限公司 | Data center load prediction method based on data driving |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110619389A (en) | Load prediction method and system of combined cooling heating and power system based on LSTM-RNN | |
| Mi et al. | Short‐term power load forecasting method based on improved exponential smoothing grey model | |
| CN106022521B (en) | Short-term load prediction method of distributed BP neural network based on Hadoop architecture | |
| CN111027772A (en) | Multi-factor short-term load prediction method based on PCA-DBILSTM | |
| CN111985719B (en) | Power load prediction method based on improved long-term and short-term memory network | |
| CN112884236B (en) | Short-term load prediction method and system based on VDM decomposition and LSTM improvement | |
| Ullah et al. | Deep Learning‐Assisted Short‐Term Power Load Forecasting Using Deep Convolutional LSTM and Stacked GRU | |
| CN115310355A (en) | Multi-energy coupling-considered multi-load prediction method and system for comprehensive energy system | |
| Jiang et al. | Industrial ultra-short-term load forecasting with data completion | |
| Sui et al. | Intelligent grey forecasting model based on periodic aggregation generating operator and its application in forecasting clean energy | |
| Zuo | Integrated forecasting models based on LSTM and TCN for short-term electricity load forecasting | |
| CN116544931B (en) | Power load distribution prediction method based on integrated fragment transformation and time convolution network | |
| CN115528750B (en) | Power grid safety and stability oriented data model hybrid drive unit combination method | |
| Xu et al. | NWP feature selection and GCN-based ultra-short-term wind farm cluster power forecasting method | |
| CN116739130A (en) | Multi-time scale load prediction method of TCN-BiLSTM network | |
| Zhang et al. | Short‐Term Power Load Forecasting Model Design Based on EMD‐PSO‐GRU | |
| CN110659775A (en) | LSTM-based improved electric power short-time load prediction algorithm | |
| CN115619447A (en) | Monthly electricity sales combined prediction method, equipment and medium | |
| CN113792828A (en) | Power grid load prediction method, system, equipment and medium based on deep learning | |
| Lee et al. | Application of temperature prediction model based on LSTNet in telecommunication Room | |
| Ji et al. | An Active Learning based Latency Prediction Approach for Neural Network Architecture | |
| CN118332478B (en) | Photovoltaic power short-term segmentation prediction method and system based on improved EEMD algorithm | |
| Yao et al. | An improved GM (1, 1) model based on iterative weighted least square criterion parameter estimation | |
| CN112952843B (en) | Distributed photovoltaic probability power flow calculation method for power distribution network based on Copula theory | |
| Mi et al. | Research article short-term power load forecasting method based on improved exponential smoothing grey model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |