CN113962425A - Heating data generation method, device, equipment and computer storage medium - Google Patents
Heating data generation method, device, equipment and computer storage medium Download PDFInfo
- Publication number
- CN113962425A CN113962425A CN202110880007.7A CN202110880007A CN113962425A CN 113962425 A CN113962425 A CN 113962425A CN 202110880007 A CN202110880007 A CN 202110880007A CN 113962425 A CN113962425 A CN 113962425A
- Authority
- CN
- China
- Prior art keywords
- data
- heat supply
- training
- supply data
- generator
- 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 58
- 238000003860 storage Methods 0.000 title claims abstract description 16
- 238000010438 heat treatment Methods 0.000 title claims description 65
- 238000012549 training Methods 0.000 claims abstract description 161
- 230000002776 aggregation Effects 0.000 claims description 14
- 238000004220 aggregation Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 11
- 238000007781 pre-processing Methods 0.000 claims description 9
- 230000007613 environmental effect Effects 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 230000004931 aggregating effect Effects 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 7
- 238000013135 deep learning Methods 0.000 abstract description 6
- 238000013473 artificial intelligence Methods 0.000 abstract description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 23
- 230000006870 function Effects 0.000 description 19
- 238000004891 communication Methods 0.000 description 9
- 238000009826 distribution Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000004913 activation Effects 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000033228 biological regulation Effects 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 235000019580 granularity Nutrition 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000001556 precipitation Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000013256 Gubra-Amylin NASH model Methods 0.000 description 1
- 238000003324 Six Sigma (6σ) Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000004883 computer application Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- 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
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic 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
- G06N3/048—Activation functions
-
- 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
- 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)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Economics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Linguistics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Probability & Statistics with Applications (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The invention discloses a heat supply data generation method, a heat supply data generation device, heat supply equipment and a computer storage medium, and relates to the technologies of Internet of things, big data, deep learning and the like under the artificial intelligence technology. The specific implementation scheme is as follows: obtaining a training sample by using historical heat supply data; training a confrontation model comprising a generator and a discriminator by using the training sample; wherein the generator generates heat supply data of the simulated training sample by using the input random signal; the discriminator discriminates the training sample and the heat supply data which is generated by the generator and simulates the training sample; the training target of the generator is to minimize the difference between the generated heat supply data and the training sample, and the training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent; and generating heat supply data by using the generator obtained after training. The problem that a large amount of heat supply data are difficult to obtain and high in cost can be effectively solved.
Description
Technical Field
The utility model relates to a computer application technology field especially relates to technologies such as thing networking, big data, deep learning under the artificial intelligence technical field.
Background
The central heating system mainly comprises a heat source, a heat exchange station and a user. The supply and demand of heat is a dynamically balanced process. The insufficient heat supply amount can cause the heat supply quality of the system to be reduced, and the overhigh heat supply amount can cause the heat supply cost to be increased, so that the prediction of the proper heat supply amount is very important. Traditional heat load prediction relies on experts to estimate the heat demand of the current year from historical data of the past year. With the development of artificial intelligence technology, the heat demand can be effectively predicted in real time by mining the historical data rule and establishing a heat load prediction model based on a machine learning method, and a direction is provided for realizing the purpose of 'heat supply on demand'. However, the effectiveness and generalization performance of machine learning is limited by the data volume and quality of the training data. However, the collection of a large amount of heat supply data is limited by the time length of a heat supply season on one hand; on the other hand, huge economic cost is needed for high-frequency acquisition of primary heating regulation data.
Disclosure of Invention
In view of the above, the present disclosure provides a heating data generation method, apparatus, device and computer storage medium, so as to solve the problems of difficulty and high cost in acquiring a large amount of heating data.
According to a first aspect of the present disclosure, there is provided a heating data generation method, including:
obtaining a training sample by using historical heat supply data;
training a confrontation model comprising a generator and a discriminator by using the training sample; wherein the generator generates heat supply data of the simulated training sample by using the input random signal; the discriminator discriminates the training sample and the heat supply data which is generated by the generator and simulates the training sample; the training target of the generator is to minimize the difference between the generated heat supply data and the training sample, and the training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent;
and generating heat supply data by using the generator obtained after training.
According to a second aspect of the present disclosure, there is provided a heating data generation apparatus including:
the sample acquisition unit is used for acquiring a training sample by using historical heat supply data;
a model training unit for training a confrontation model including a generator and a discriminator using the training samples; wherein the generator generates heat supply data of the simulated training sample by using the input random signal; the discriminator discriminates the training sample and the heat supply data which is generated by the generator and simulates the training sample; the training target of the generator is to minimize the difference between the generated heat supply data and the training sample, and the training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent;
and the data generation unit is used for generating heat supply data by using the generator obtained after training is finished.
According to a third aspect of the present disclosure, there is provided an electronic device comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method as described above.
According to a fourth aspect of the present disclosure, there is provided a non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method as described above.
According to a fifth aspect of the disclosure, a computer program product comprising a computer program which, when executed by a processor, implements the method as described above.
It should be understood that the statements in this section do not necessarily identify key or critical features of the embodiments of the present disclosure, nor do they limit the scope of the present disclosure. Other features of the present disclosure will become apparent from the following description.
Drawings
The drawings are included to provide a better understanding of the present solution and are not to be construed as limiting the present disclosure. Wherein:
fig. 1 is a flowchart of a heating data generation method provided in an embodiment of the present disclosure;
FIG. 2 is a schematic diagram of collecting training samples provided by an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of an countermeasure model provided by an embodiment of the disclosure;
fig. 4 is a structural diagram of a heating data generation apparatus according to an embodiment of the present disclosure;
fig. 5 is a block diagram of an electronic device for implementing a heating data generation method according to an embodiment of the present disclosure.
Detailed Description
Exemplary embodiments of the present disclosure are described below with reference to the accompanying drawings, in which various details of the embodiments of the disclosure are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the present disclosure. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Fig. 1 is a flowchart of a heating data generation method provided in an embodiment of the present disclosure, where an execution subject of the method may be a heating data generation device, and the device may be an application located in a computer terminal, or may also be a functional unit such as a Software Development Kit (SDK) or a plug-in located in the application located in the computer terminal, or may also be located at a server side, which is not particularly limited in this embodiment of the present disclosure. As shown in fig. 1, the method may include the steps of:
in 101, training samples are obtained using historical heating data.
Training a confrontation model comprising a generator and a discriminator using the training samples at 102; the generator generates heat supply data of the simulated training sample by using the input random signal; the discriminator discriminates the training sample and the heat supply data which is generated by the generator and simulates the training sample; the training target of the generator is to minimize the difference between the generated heat supply data and the training sample, and the training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent.
In 103, heating data is generated using the generator obtained after the training is completed.
It can be seen that the generator can well simulate the data distribution of the historical heating data through the pair learning of the generator and the discriminator in the countermeasure model. The trained generator can generate a large amount of heating data which accord with the distribution of the heating data. The method can effectively solve the problems of difficulty in obtaining a large amount of heat supply data and high cost.
The above steps are described in detail with reference to the following examples. First, step 101, i.e., "training samples are obtained using historical heating data" will be described in detail.
When historical heating data is obtained first, the following data can be obtained but not limited to:
first data: environmental data. Such as air temperature, precipitation, humidity, wind, sensible temperature, weather type (e.g., cloudy, sunny, rain, snow, haze, etc.). The environmental data may be collected by sensors such as temperature sensors, humidity sensors, wind sensors, etc. The system can also be acquired in real time at a preset frequency through an API (application program interface) provided by official platforms such as the China aeronautical network, and the like, and the mode can directly acquire more abundant environmental data, so that the system is low in cost and high in data accuracy. The collection is typically done at a city as a granularity, other granularities such as district, county, etc. may also be employed.
Second data: heat source device parameters. Such as a net water supply pressure, a net water return pressure, a net heat supply flow, a net water supply temperature, a net water return temperature, accumulated electricity, circulating pump water consumption, heat, and the like. The heat source device parameters can be acquired in real time at preset frequency by such as a pressure sensor, a temperature sensor, a flow meter, a water meter, an electric meter, a heat meter and the like.
Third data: and (4) equipment parameters of the heat exchange station. Such as the pressure of the two-network water supply, the pressure of the two-network water return, the heat supply flow of the two-network, the temperature of the two-network water supply, the temperature of the two-network water return, the accumulated electric quantity, the water consumption of the circulating pump, the heat quantity and the like. The heat exchange station equipment parameters can be acquired in real time at preset frequency by such as a pressure sensor, a temperature sensor, a flow meter, a water meter, an electric meter, a heat meter and the like.
The above-mentioned "one net" and "two nets" will be briefly described here. The first network refers to a pipeline system for high-temperature water from a heat source to enter a district heat exchange station for heat exchange, and the second network refers to a pipeline system for supplying heat from the district heat exchange station to users. Wherein the temperature and pressure of one web are relatively high and the temperature and pressure of the two webs are relatively low.
Fourth data: and (4) user side thermal parameters. Such as indoor temperature, humidity, etc. of the user terminal. The real-time acquisition can be achieved by a temperature sensor, a humidity sensor and the like at a preset frequency.
It can be seen that the second to fourth data are collected by the field sensor and then can be transmitted to the intelligent gateway through a PLC (Power Line Communication) protocol. The intelligent gateway connects to an internet of things Core suite (IoT Core), and then writes data into a TSDB (Time series database) in real Time.
In addition, it should be particularly noted that, when historical heat supply data is acquired in the present disclosure, a sliding window with a preset duration may be used to acquire the historical heat supply data, and the historical heat supply data of the time sequence in each sliding window is respectively used as a training sample. That is, each training sample in this disclosure is time series data. For example, as shown in fig. 2, the size of the sliding window is 252 days, and then the historical heating data collected from day 1 to day 252 is taken as training sample 1; the sliding window slides backwards for one day, and historical heating data collected from day 2 to day 253 is used as a training sample 2; the sliding window slides backward one day, with historical heating data collected from day 3 to day 254 as training sample 3, and so on. The mode of collecting the training samples through the sliding window can effectively improve the acquisition quantity of the training samples.
Still further, the collected historical heating data may be pre-processed in view of the possible anomalies that may exist in the collected historical heating data. The pretreatment that is performed may include, but is not limited to, the following:
the first pretreatment: the outlier is deleted.
Occasional failure of sensors, human observation errors, etc. may cause anomalies in the collected historical heating data (i.e., time series data). Aiming at each dimension of historical heat supply data, determining a reasonable interval according to one of the mean value or the median value of the dimension and the standard deviation; and deleting the data which do not meet the reasonable interval.
The historical heating data referred to in the present disclosure may be time series data, and the heating data for each point in time may be multidimensional, with each dimension representing a field. For example, the indoor temperature may be one dimension and the indoor humidity may be one dimension. In the embodiment of the disclosure, the acquired abnormal data can be deleted by adopting a six-Sigma (6Sigma) method.
For each dimension of the historical heat supply data, when a reasonable interval is determined, the following formula (1) can be adopted to determine the upper limit UCL of the reasonable interval, and the formula (2) is adopted to determine the lower limit LCL of the reasonable interval:
UCL＝CL+k*σ (1)
LCL＝CL-k*σ (2)
where CL may be the mean or median of the heating data for the dimension. k takes a preset empirical or experimental value, for example 3.σ is the standard deviation of the heating data for this dimension.
And determining a reasonable interval as [ LCL, UCL ], and deleting abnormal data which do not meet the reasonable interval in the dimensional heat supply data. The goal of this preprocessing is to have the data evenly distributed within a finite interval.
And (2) second pretreatment: the missing value is filled.
Due to sensor differences and the influence of external disturbance, a small amount of missing values may exist in the finally acquired heating data. And therefore, missing value filling can be carried out on each dimension of the historical heating data by a linear interpolation method.
And (3) third pretreatment: and (6) data aggregation.
Since each training sample contains multidimensional heating data. The heat supply data are different due to the fact that data sources are different, and the acquisition frequencies corresponding to the heat supply data with different dimensions are possibly different. In the embodiment of the disclosure, data aggregation may be performed on the historical heat supply data at the lowest acquisition frequency adopted in acquiring the historical heat supply data, and an average value is taken for each dimension of data obtained by aggregation respectively as a sample value of the dimension of data.
For example, the maximum frequency of data of the chinese air grid is 5 minutes, and the collection frequency of the heat source device parameters, the heat exchange station device parameters and the user side heat parameters can reach 30 seconds. The heat supply data collected in every 5 minutes can be aggregated by adopting a frequency of 5 minutes, and the average value of each dimension of aggregated data is used as a sample value of the dimension data at a time point. By the method, the training samples can be fused with different data sources, and the data of the different data sources are aligned in the training samples, so that the data are guaranteed to be more diverse.
And (4) fourth pretreatment: and (5) discretizing the features.
Most of equipment parameters such as heat source equipment parameters, heat exchange station equipment parameters and the like belong to continuous data. On the one hand, the search space dimension of the continuous data is high, and the generation model in the countermeasure model is difficult to search for the nash equilibrium point. On the other hand, the data distribution of different dimensions is different, and the data scale is also different. In view of this, discretization processing can be performed on each dimension of continuous data of the historical heating data in an equal frequency binning mode, so that robustness of the heating data on abnormal data is enhanced. The equal frequency binning is to bin the continuous data in a way of dividing the continuous data by a frequency percentage, so that each interval has the same number of samples. Since the equal frequency binning mode is a mature technology at present, it will not be described in detail here.
And performing one-hot (one-hot) encoding on the data obtained after equal frequency binning, and directly performing one-hot encoding on the originally discrete data. Discretization can handle categorical data as opposed to continuous data.
The above step 102, i.e., "training the confrontation model including the generator and the discriminator using the training sample", is described in detail with reference to the embodiments below.
The traditional countermeasure model is widely applied to picture generation, and the countermeasure model is applied to the heat supply industry to generate heat supply data in a large scale.
As shown in fig. 3, the confrontation model includes a generator and an arbiter. In the disclosed embodiment, the input to the generator is a random signal, which may be, for example, random noise. The output is the heating data of the generated simulated training samples. The aim is to minimize the difference between the generated heat supply data and the training samples, and make the discrimination result of the discriminator wrong as much as possible, that is, the generator learns the data distribution characteristics of the training samples as much as possible, aiming at making the discriminator unable to discriminate the heat supply data and the training samples generated by the generator.
The input of the discriminator is the training sample and the heat supply data which is generated by the generator and simulates the training sample, and the output of the discriminator is the discrimination result of the training sample and the heat supply data generated by the generator. Namely, the training sample is used as real heat supply data, the heat supply data generated by the generator is used as fake heat supply data, and the discriminator judges the true and false heat supply data. The training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent, namely to discriminate the true heat supply data from the false heat supply data as much as possible. And obtaining a loss function according to the judgment result of the discriminator, and alternately training the generator and the discriminator by utilizing the loss function so as to update the model parameters of the generator and the discriminator.
The countermeasure model used in the present disclosure may employ DCGAN (Deep Convolutional adaptive adaptation network). Compared with conventional GAN (Generative adaptive Networks, Generative countermeasure network), the DCGAN employed in the present disclosure has the following improvements:
the first point is as follows: both the generator and the arbiter discard the pooling layer in CNN (Convolutional Neural Network). The discriminator reserves the whole structure of CNN, and the generator replaces the volume base layer with the transposed volume layer to recover the original dimensionality of the data.
And a second point: and a BN (Batch Normalization) layer is adopted behind each layer in the discriminator and the generator, so that the training problem caused by poor initialization is favorably processed, the model training is accelerated, and the training stability is improved.
And a third point: all fully connected layers were replaced with 1 x 1 convolutional layers.
A fourth point: in the generator, all layers use a reli (Rectified Linear Unit) activation function, except for the output layer which uses a Sigmoid activation function.
And fifth, the method comprises the following steps: LeakyReLU activation functions are used at all layers of the discriminator to prevent gradient sparsity.
The countermeasure model used in this disclosure may also employ WGAN (Wasserstein generated adaptive Networks, bulldozer-generated countermeasure network). In other words, based on DCGAN, for the problems of asynchronous training of generators and discriminators, non-convergence of training, pattern collapse, and the like which may exist in the countermeasure model, the learkyrelu activation function output by the discriminators is removed, and the W Distance (bulldozer Distance) is used as the output.
The penalty function for the arbiter J (D) is as follows:
J(D)＝Ez～Pz[fw(G(z))]-Ex～Pr[fw(x)] (3)
the loss function of generator J (G) is as follows:
j(G)＝-Ez～Pz[fw(G(z))] (4)
pr is the training sample set, i.e. the true heating data distribution, Pz is the random signal set, e.g. the standard normal distribution, z is the training sample, x is the random signal, fw() Is a Lipschitz function based on W distance, Ez～PzRepresenting the expected value of the corresponding random signal distribution.
Compared with the traditional GAN model, JS divergence (Jensen-Shannon divergence) is replaced by W distance. The W distance has Lipschitz continuity mathematically, and the problems of unstable gradient, mode collapse and the like of the GAN can be solved.
Two approaches are introduced in WGAN: one is to cut the weight of the discriminator to limit the weight of the discriminator in a certain range during training. The weight clipping means that not all weights are updated by the same gradient amplitude in the process of training the arbiter, and a weight range can be set, and the weight exceeding the range in the arbiter is set to be 0, i.e. the connection is clipped.
The other is to blend a gradient penalty into the discriminator, that is, add a gradient penalty term to the loss function of the discriminator, for example, add a constraint of L2-norm.
By the two methods, the discriminator can meet the Lipschitz continuity, and the training speed is improved.
In each round of training, the model parameters of the generator can be fixed first, and n can be performed on the discriminatorcriticA sub-training in which ncriticIs a positive integer. M real samples (namely training samples obtained by collecting historical heat supply data) are sampled in each training, and m is a positive integer. And m counterfeit samples (i.e., generated heating data) are generated by the generator by outputting random signal data. And (3) regarding the forged sample as a negative sample, regarding the real sample as a positive sample, calculating a loss function of the discriminator, updating model parameters of the discriminator through gradient descent, and limiting the size of the model parameters. To reach ncriticAfter the secondary training, fixing the model parameters of the discriminator, training the generator, generating m forged samples by the generator by outputting random signal data, calculating a loss function of the generator, updating the model parameters of the generator through gradient descent, and limiting the size of the model parameters.
And performing iteration round by round according to the above mode, and finishing the training when an iteration stop condition is reached. Where the iteration stop condition may be such as the generator model parameters converging or the number of iterations reaching a preset threshold number of iterations, etc.
As a preferred embodiment, a method such as RMSProp (learning rate adaptive) may be used when the gradient decreases. The RMSProp is a deep learning network optimization algorithm, and after the learning rate is set, the learning rate of each parameter is different through dividing the global learning rate parameter by the square root of the sum of squares of historical gradients controlled by an attenuation coefficient. The RMSProp can make greater progress in a more gradual direction of the parameter space, thereby speeding up the training. It is more suitable for use in WGAN.
In addition, since the training samples in the present disclosure are Time-series heating data and are Time-series data, in order to more appropriately measure the similarity between the training samples and the heating data generated by the generator, a DTW (Dynamic Time Warping) algorithm may be employed. The idea of DTW is to calculate a distance matrix between points of two sequences, and find a path from the top left corner to the top right corner of the matrix, so that the path sum is minimum, and then this minimum path sum can be used as the similarity between the two sequences.
After the collected historical heating data is preprocessed as mentioned in the above embodiment, a part of the collected historical heating data can be used as a training sample set, and another part of the collected historical heating data can be used as a test set. After the training is finished, the confrontation model obtained by the training can be tested by using the test set, whether the confrontation model has an overfitting phenomenon or not is judged, and model parameters are adjusted and optimized.
The above step 103, i.e., "generating heating data by using the generator obtained after the training" will be described in detail with reference to the embodiment.
After the training of the confrontation model is completed, only the generator in the confrontation model needs to be utilized in generating the heating data. When random signals are input into the generator, the generator can generate heat supply data similar to historical heat supply data in distribution, and the heat supply data are time sequence data with preset time length. A large amount of heat supply data can be generated conveniently, quickly and at low cost through the generator.
The heat supply industry generally adopts the heat supply data of the past year to predict the heat load of the current year or the future, and a deep learning method represented by a neural network can be adopted during prediction. However, the application of deep learning in the heating field is limited by the lack of available heating data. The above-mentioned embodiments of the present disclosure may use the heat supply data generated by the trained generator as the training data of the heat load prediction model, and may ensure the quality of the trained heat load prediction model because a large amount of high-quality heat supply data can be generated. The embodiment of the disclosure applies the countermeasure learning technology to heat supply data generation in the heat supply industry, greatly reduces the cost and period of heat supply data collection, and provides data guarantee for the training of the heat load prediction model. The threshold of deep learning applied to the heat supply industry is reduced, and transformation and digital development of the heat supply industry are facilitated.
The above is a detailed description of the method provided by the present disclosure, and the following is a detailed description of the apparatus provided by the present disclosure with reference to the embodiments.
Fig. 4 is a heating data generation apparatus provided in an embodiment of the present disclosure, and as shown in fig. 4, the apparatus may include: the sample acquisition unit 401, the model training unit 402, and the data generation unit 403 further include a preprocessing unit 404. The main functions of each component unit are as follows:
and a sample obtaining unit 401, configured to obtain a training sample by using the historical heat supply data.
A model training unit 402 for training a confrontation model including a generator and a discriminator using training samples; the generator generates heat supply data of the simulated training sample by using the input random signal; the discriminator discriminates the training sample and the heat supply data which is generated by the generator and simulates the training sample; the training target of the generator is to minimize the difference between the generated heat supply data and the training sample, and the training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent.
A data generating unit 403, configured to generate heating data by using the generator obtained after the training is completed.
As a preferred embodiment, the sample obtaining unit 401 is specifically configured to collect historical heat supply data by using sliding windows with preset durations, and use the historical heat supply data of time sequences in the sliding windows as training samples respectively.
Wherein, historical heating data includes: the system comprises environmental data, heat source equipment parameters, heat exchange station equipment parameters and user side heat parameters.
The environmental data may be collected by sensors such as temperature sensors, humidity sensors, wind sensors, etc. And the data can be acquired in real time at a preset frequency through an API (application programming interface) provided by official platforms such as the China air network. May include air temperature, precipitation, humidity, wind, sensible temperature, weather type (e.g., cloudy, sunny, rain, snow, haze, etc.).
The heat source device parameters can be acquired in real time at preset frequency by such as a pressure sensor, a temperature sensor, a flow meter, a water meter, an electric meter, a heat meter and the like. The system can comprise one-network water supply pressure, one-network water return pressure, one-network heat supply flow, one-network water supply temperature, one-network water return temperature, accumulated electric quantity, circulating pump water consumption, heat and the like.
The heat exchange station equipment parameters can be acquired in real time at preset frequency by such as a pressure sensor, a temperature sensor, a flow meter, a water meter, an electric meter, a heat meter and the like. The method can comprise two-network water supply pressure, two-network backwater pressure, two-network heat supply flow, two-network water supply temperature, two-network backwater temperature, accumulated electric quantity, circulating pump water consumption, heat and the like.
Preferably, the preprocessing unit 404 is configured to preprocess the collected historical heat supply data for the sample obtaining unit to use; wherein the preprocessing comprises at least one of deleting outliers, filling missing values, data aggregation, and feature discretization.
Deleting outliers includes: aiming at each dimension of historical heat supply data, determining a reasonable interval according to one of the mean value or the median value of the dimension and the standard deviation; and deleting the data which do not meet the reasonable interval.
Fill missing values include: and filling missing values in each dimension of the historical heat supply data by adopting a linear interpolation method.
The data aggregation comprises the following steps: and performing data aggregation on the historical heat supply data at the lowest acquisition frequency adopted in the acquisition of the historical heat supply data, and respectively taking the average value of each dimension data obtained by aggregation as the sample value of the dimension data.
The feature discretization comprises the following steps: and discretizing each dimension of continuous data of the historical heat supply data by adopting an equal-frequency box dividing mode.
As one implementation, the above-mentioned countermeasure model may employ DCGAN. As another implementation, the above-described countermeasure model may employ WGAN.
If the confrontation model adopts WGAN, the model training unit 402 adopts RMSProp descent method during the training process.
As a preferred embodiment, two methods are introduced in WGAN: one is to cut the weight of the discriminator to limit the weight of the discriminator in a certain range during training. The other is to blend a gradient penalty into the discriminator, that is, add a gradient penalty term to the loss function of the discriminator, for example, add a constraint of L2-norm. By the two methods, the discriminator can meet the Lipschitz continuity, and the training speed is improved.
The model training unit 402 may first fix the model parameters of the generator and perform n on the discriminators in each round of trainingcriticA sub-training in which ncriticIs a positive integer. M real samples (namely training samples obtained by collecting historical heat supply data) are sampled in each training, and m is a positive integer. And m counterfeit samples (i.e., generated heating data) are generated by the generator by outputting random signal data. And (3) regarding the forged sample as a negative sample, regarding the real sample as a positive sample, calculating a loss function of the discriminator, updating model parameters of the discriminator through gradient descent, and limiting the size of the model parameters. To reach ncriticAfter the secondary training, fixing the model parameters of the discriminator, training the generator, generating m forged samples by the generator by outputting random signal data, calculating a loss function of the generator, updating the model parameters of the generator through gradient descent, and limiting the size of the model parameters.
And performing iteration round by round according to the above mode, and finishing the training when an iteration stop condition is reached. Where the iteration stop condition may be such as the generator model parameters converging or the number of iterations reaching a preset threshold number of iterations, etc.
The data generating unit 403 is specifically configured to input a random signal into a generator obtained after training is completed, to obtain time sequence heat supply data with a preset time length, and to use the time sequence heat supply data as training data of a heat load prediction model.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, as for the apparatus embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
In the technical scheme of the disclosure, the acquisition, storage, application and the like of the personal information of the related user all accord with the regulations of related laws and regulations, and do not violate the good customs of the public order.
The present disclosure also provides an electronic device, a readable storage medium, and a computer program product according to embodiments of the present disclosure.
As shown in fig. 5, is a block diagram of an electronic device of a heating data generation method according to an embodiment of the present disclosure. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the disclosure described and/or claimed herein.
As shown in fig. 5, the apparatus 500 comprises a computing unit 501 which may perform various appropriate actions and processes in accordance with a computer program stored in a Read Only Memory (ROM)502 or a computer program loaded from a storage unit 508 into a Random Access Memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored. The calculation unit 501, the ROM 502, and the RAM 503 are connected to each other by a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.
A number of components in the device 500 are connected to the I/O interface 505, including: an input unit 506 such as a keyboard, a mouse, or the like; an output unit 507 such as various types of displays, speakers, and the like; a storage unit 508, such as a magnetic disk, optical disk, or the like; and a communication unit 509 such as a network card, modem, wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange information/data with other devices through a computer network such as the internet and/or various telecommunication networks.
The computing unit 501 may be a variety of general-purpose and/or special-purpose processing components having processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various dedicated Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, and so forth. The calculation unit 501 executes the respective methods and processes described above, such as the square heating data generation method. For example, in some embodiments, the heating data generation method may be implemented as a computer software program tangibly embodied in a machine-readable medium, such as storage unit 508.
In some embodiments, part or all of the computer program may be loaded and/or installed onto the device 500 via the ROM 802 and/or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the heating data generation method described above may be performed. Alternatively, in other embodiments, the computing unit 501 may be configured to perform the heating data generation method by any other suitable means (e.g., by means of firmware).
Various implementations of the systems and techniques described here can be realized in digital electronic circuitry, integrated circuitry, Field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for implementing the methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller 30, cause the functions/operations specified in the flowchart and/or block diagram to be performed. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic, speech, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), Wide Area Networks (WANs), and the Internet.
The computer system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The Server may be a cloud Server, which is also called a cloud computing Server or a cloud host, and is a host product in a cloud computing service system, so as to solve the defects of high management difficulty and weak service expansibility existing in the traditional physical host and virtual Private Server (VPs) service. The server may also be a server of a distributed system, or a server incorporating a blockchain.
It should be understood that various forms of the flows shown above may be used, with steps reordered, added, or deleted. For example, the steps described in the present application may be executed in parallel, sequentially, or in different orders, and are not limited herein as long as the desired results of the technical solutions disclosed in the present disclosure can be achieved.
The above detailed description should not be construed as limiting the scope of the disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions may be made in accordance with design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present disclosure should be included in the scope of protection of the present disclosure.
Claims (17)
1. A heating data generation method, comprising:
obtaining a training sample by using historical heat supply data;
training a confrontation model comprising a generator and a discriminator by using the training sample; wherein the generator generates heat supply data of the simulated training sample by using the input random signal; the discriminator discriminates the training sample and the heat supply data which is generated by the generator and simulates the training sample; the training target of the generator is to minimize the difference between the generated heat supply data and the training sample, and the training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent;
and generating heat supply data by using the generator obtained after training.
2. The method of claim 1, wherein the using historical heating data to derive training samples comprises:
and acquiring the historical heat supply data by adopting a sliding window with preset time length, and respectively taking the historical heat supply data of the time sequence in each sliding window as training samples.
3. A method according to claim 1 or 2, wherein the historical heating data comprises: the system comprises environmental data, heat source equipment parameters, heat exchange station equipment parameters and user side heat parameters.
4. The method of claim 1, wherein the using historical heating data to derive training samples comprises:
preprocessing the collected historical heat supply data, wherein the preprocessing comprises at least one of deleting abnormal values, filling missing values, aggregating data and discretizing characteristics;
the deleting outliers comprises: aiming at each dimension of historical heat supply data, determining a reasonable interval according to one of the mean value or the median value of the dimension and the standard deviation; deleting data which do not meet the reasonable interval;
the fill missing values include: filling missing values of each dimension of the historical heat supply data by adopting a linear interpolation method;
the data aggregation comprises: performing data aggregation on historical heat supply data at the lowest acquisition frequency adopted in the acquisition of the historical heat supply data, and respectively taking the average value of each dimension data obtained by aggregation as a sample value of the dimension data;
the feature discretization comprises: and discretizing each dimension of continuous data of the historical heat supply data by adopting an equal-frequency box dividing mode.
5. The method of claim 1, wherein the countermeasure model employs a deep convolutional countermeasure network (DCGAN) or a bulldozer countermeasure network (WGAN).
6. The method of claim 5, wherein if the countermeasure model employs WGAN, a RMSProp descent method is employed during training.
7. The method of claim 2, wherein the generating heating data with the generator obtained after training comprises:
and inputting a random signal into the generator obtained after the training is finished to obtain time sequence heat supply data with preset time length, wherein the time sequence heat supply data are used as training data of a heat load prediction model.
8. A heating data generation apparatus comprising:
the sample acquisition unit is used for acquiring a training sample by using historical heat supply data;
a model training unit for training a confrontation model including a generator and a discriminator using the training samples; wherein the generator generates heat supply data of the simulated training sample by using the input random signal; the discriminator discriminates the training sample and the heat supply data which is generated by the generator and simulates the training sample; the training target of the generator is to minimize the difference between the generated heat supply data and the training sample, and the training target of the discriminator is to discriminate the training sample and the heat supply data generated by the generator to the maximum extent;
and the data generation unit is used for generating heat supply data by using the generator obtained after training is finished.
9. The apparatus according to claim 8, wherein the sample obtaining unit is specifically configured to collect the historical heat supply data by using sliding windows with preset durations, and use the historical heat supply data of the time sequence in each sliding window as training samples respectively.
10. The apparatus of claim 8 or 9, wherein the historical heating data comprises: the system comprises environmental data, heat source equipment parameters, heat exchange station equipment parameters and user side heat parameters.
11. The apparatus of claim 8, further comprising a pre-processing unit for pre-processing the collected historical heating data for utilization by the sample acquisition unit; wherein the preprocessing comprises at least one of deleting outliers, filling missing values, data aggregation, and feature discretization;
the deleting outliers comprises: aiming at each dimension of historical heat supply data, determining a reasonable interval according to one of the mean value or the median value of the dimension and the standard deviation; deleting data which do not meet the reasonable interval;
the fill missing values include: filling missing values of each dimension of the historical heat supply data by adopting a linear interpolation method;
the data aggregation comprises: performing data aggregation on historical heat supply data at the lowest acquisition frequency adopted in the acquisition of the historical heat supply data, and respectively taking the average value of each dimension data obtained by aggregation as a sample value of the dimension data;
the feature discretization comprises: and discretizing each dimension of continuous data of the historical heat supply data by adopting an equal-frequency box dividing mode.
12. The apparatus of claim 8, wherein the countermeasure model employs a deep convolutional countermeasure network (DCGAN) or a bulldozer countermeasure network (WGAN).
13. The apparatus of claim 12, wherein if the countermeasure model employs WGAN, the model training unit employs RMSProp descent method during training.
14. The apparatus according to claim 9, wherein the data generating unit is specifically configured to input a random signal into the generator obtained after the training is completed, and obtain time-series heating data with a preset duration, which is used as training data of a thermal load prediction model.
15. An electronic device, comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.
16. A non-transitory computer readable storage medium having stored thereon computer instructions for causing the computer to perform the method of any one of claims 1-7.
17. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110880007.7A CN113962425A (en) | 2021-08-02 | 2021-08-02 | Heating data generation method, device, equipment and computer storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110880007.7A CN113962425A (en) | 2021-08-02 | 2021-08-02 | Heating data generation method, device, equipment and computer storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113962425A true CN113962425A (en) | 2022-01-21 |
Family
ID=79460606
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110880007.7A Pending CN113962425A (en) | 2021-08-02 | 2021-08-02 | Heating data generation method, device, equipment and computer storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113962425A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115622797A (en) * | 2022-11-16 | 2023-01-17 | 北京国电通网络技术有限公司 | Power consumption information generation suite based on privacy protection and power consumption information generation method |
| CN117235655A (en) * | 2023-11-15 | 2023-12-15 | 北明天时能源科技(北京)有限公司 | Intelligent heat supply abnormal condition identification method and system based on federal learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110598843A (en) * | 2019-07-23 | 2019-12-20 | 中国人民解放军63880部队 | Generation countermeasure network organization structure based on discriminator sharing and training method thereof |
| CN111126700A (en) * | 2019-12-25 | 2020-05-08 | 远景智能国际私人投资有限公司 | Energy consumption prediction method, device, equipment and storage medium |
| CN111178626A (en) * | 2019-12-30 | 2020-05-19 | 苏州科技大学 | Building energy consumption prediction method and monitoring prediction system based on WGAN algorithm |
| CN111191835A (en) * | 2019-12-27 | 2020-05-22 | 国网辽宁省电力有限公司阜新供电公司 | IES incomplete data load prediction method and system based on C-GAN transfer learning |
| CN112560981A (en) * | 2020-12-24 | 2021-03-26 | 北京百度网讯科技有限公司 | Training method, apparatus, device, program and storage medium for generating countermeasure model |
| CN112561197A (en) * | 2020-12-23 | 2021-03-26 | 国网江苏省电力有限公司南京供电分公司 | Power data prefetching and caching method with active defense influence range |
| CN113111589A (en) * | 2021-04-25 | 2021-07-13 | 北京百度网讯科技有限公司 | Training method of prediction model, method, device and equipment for predicting heat supply temperature |
-
2021
- 2021-08-02 CN CN202110880007.7A patent/CN113962425A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110598843A (en) * | 2019-07-23 | 2019-12-20 | 中国人民解放军63880部队 | Generation countermeasure network organization structure based on discriminator sharing and training method thereof |
| CN111126700A (en) * | 2019-12-25 | 2020-05-08 | 远景智能国际私人投资有限公司 | Energy consumption prediction method, device, equipment and storage medium |
| CN111191835A (en) * | 2019-12-27 | 2020-05-22 | 国网辽宁省电力有限公司阜新供电公司 | IES incomplete data load prediction method and system based on C-GAN transfer learning |
| CN111178626A (en) * | 2019-12-30 | 2020-05-19 | 苏州科技大学 | Building energy consumption prediction method and monitoring prediction system based on WGAN algorithm |
| CN112561197A (en) * | 2020-12-23 | 2021-03-26 | 国网江苏省电力有限公司南京供电分公司 | Power data prefetching and caching method with active defense influence range |
| CN112560981A (en) * | 2020-12-24 | 2021-03-26 | 北京百度网讯科技有限公司 | Training method, apparatus, device, program and storage medium for generating countermeasure model |
| CN113111589A (en) * | 2021-04-25 | 2021-07-13 | 北京百度网讯科技有限公司 | Training method of prediction model, method, device and equipment for predicting heat supply temperature |
Non-Patent Citations (4)
| Title |
|---|
| 尚成国 等: "《山西科技文献资源整合与数据挖掘技术研究》", vol. 1, 31 December 2016, 科学技术文献出版社, pages: 205 - 209 * |
| 徐宗本 等: "《数据只能研究前沿》", vol. 1, 31 May 2021, 上海交通大学出版社, pages: 50 - 53 * |
| 杨博雄: "《深度学习理论与实践》", vol. 1, 30 September 2020, 北京邮电大学出版社, pages: 197 - 198 * |
| 梁俊杰 等: ""生成对抗网络GAN综述"", 《计算机科学与探索》, vol. 14, no. 1, 28 November 2019 (2019-11-28), pages 1 - 17 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115622797A (en) * | 2022-11-16 | 2023-01-17 | 北京国电通网络技术有限公司 | Power consumption information generation suite based on privacy protection and power consumption information generation method |
| CN117235655A (en) * | 2023-11-15 | 2023-12-15 | 北明天时能源科技(北京)有限公司 | Intelligent heat supply abnormal condition identification method and system based on federal learning |
| CN117235655B (en) * | 2023-11-15 | 2024-02-02 | 北明天时能源科技(北京)有限公司 | Intelligent heat supply abnormal condition identification method and system based on federal learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113962425A (en) | Heating data generation method, device, equipment and computer storage medium | |
| CN105225000B (en) | A kind of wind power probability model nonparametric probability method based on Fuzzy Ordered optimization | |
| CN113408808B (en) | Training method, data generation device, electronic equipment and storage medium | |
| CN117078048A (en) | Digital twinning-based intelligent city resource management method and system | |
| CN108205713A (en) | A kind of region wind power prediction error distribution determination method and device | |
| CN114357670A (en) | Power distribution network power consumption data abnormity early warning method based on BLS and self-encoder | |
| CN116523140A (en) | Method and device for detecting electricity theft, electronic equipment and storage medium | |
| WO2024109487A1 (en) | Load shedding testing method and apparatus for pumped storage unit, device, and medium | |
| CN117332898A (en) | New energy small time scale power time sequence rolling prediction method based on machine learning | |
| CN117971487A (en) | High-performance operator generation method, device, equipment and storage medium | |
| CN117332896A (en) | New energy small time scale power prediction method and system for multilayer integrated learning | |
| CN115169731A (en) | Smart campus energy consumption prediction method, device, equipment and medium | |
| CN112488805A (en) | Long-renting market early warning method based on multiple regression time series analysis | |
| CN117131315B (en) | Out-of-tolerance electric energy meter determining method and medium based on solving multi-element quadratic function extremum | |
| CN114896826B (en) | Planet boundary layer parameterization method based on physics and residual error attention network | |
| CN116151034B (en) | Insulator core rod crisping prediction method, device, equipment and medium | |
| CN118100151A (en) | Power grid load prediction method, device, equipment and storage medium | |
| CN115081703A (en) | Gas quantity prediction method, device, equipment and storage medium | |
| CN117251809A (en) | Power grid time sequence data anomaly detection method, device, equipment and storage medium | |
| CN118014018A (en) | Building energy consumption prediction method, device, equipment and storage medium | |
| CN114742153A (en) | Power utilization behavior analysis method based on one graph of power distribution network | |
| CN118548041A (en) | Characterization method and device for residual gas volume of water-flooding gas reservoir, electronic equipment and medium | |
| CN116151854A (en) | User type determining method, device, equipment and storage medium | |
| CN118627418A (en) | Method and device for predicting water inflow of underground water seal oil depot based on ensemble learning | |
| CN114991748A (en) | Offshore oilfield extract optimization method and device, electronic equipment and storage medium |
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 |