CN113824800B - Big data analysis method and device based on hybrid energy data - Google Patents
Big data analysis method and device based on hybrid energy data Download PDFInfo
- Publication number
- CN113824800B CN113824800B CN202111389440.7A CN202111389440A CN113824800B CN 113824800 B CN113824800 B CN 113824800B CN 202111389440 A CN202111389440 A CN 202111389440A CN 113824800 B CN113824800 B CN 113824800B
- Authority
- CN
- China
- Prior art keywords
- sample
- sensing data
- index
- carbon dioxide
- server
- 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.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2219/00—Indexing scheme relating to application aspects of data processing equipment or methods
- G06F2219/10—Environmental application, e.g. waste reduction, pollution control, compliance with environmental legislation
Abstract
The application discloses a big data analysis method based on hybrid energy data, which comprises the steps of obtaining a first sample server and a second sample server which are bound in advance; acquiring a first sample sensing data set; acquiring a second sample sensing data set; binding and labeling to generate sample data labeled with a first index, calling a deep neural network model, and training the deep neural network model to obtain an index prediction model; respectively acquiring a first sensing data group and a second sensing data group, and acquiring a second index; inputting the data into an index prediction model together for processing to obtain a prediction index value, and calculating the difference value between the prediction index value and a second index; if the difference value is smaller than the preset threshold value, the hybrid energy factory is judged to be normally operated, the purpose of distinguishing fake data is achieved, and the aim of promoting carbon neutralization is really achieved.
Description
Technical Field
The application relates to the field of computers, in particular to a big data analysis method and device based on hybrid energy data.
Background
The goal of carbon neutralization needs to be met in two ways, namely, reducing carbon dioxide emissions and increasing carbon dioxide absorption. Energy plants account for a portion of the carbon dioxide emissions. These energy plants, for example, petrochemical plants (oil refineries, etc.) and electric power plants (thermal power plants, etc.), inevitably emit carbon dioxide during their production. To achieve carbon neutralization, macroscopic regulation is required to control the carbon dioxide emissions of these energy plants as a whole. However, reducing the emission of carbon dioxide increases the cost of the energy plant, and therefore, there are some energy plants that modify the emission data of carbon dioxide to achieve the goal of pseudo-decoration of carbon dioxide emission index. The prior art lacks a technical solution for solving the phenomenon.
Disclosure of Invention
The application provides a big data analysis method based on hybrid energy data, which comprises the following steps:
s1, the big data analysis server acquires a first sample server and a second sample server which are bound in advance; wherein the first sample server corresponds to a first energy plant, the second sample server corresponds to a second energy plant, and the first energy plant is an upstream plant of the second energy plant;
s2, sending a first sample sensing data request message to the first sample server to obtain a first sample sensing data group; wherein the first sample sensing data set is sensed by a plurality of sensors deployed in a first energy plant;
s3, sending a second sample sensing data request message to the second sample server to obtain a second sample sensing data group; wherein the second sample sensing data set is sensed by a plurality of sensors deployed in a second energy plant; the first sample sensing data group and the second sample sensing data group are obtained by sensing at the same time, and the first sample sensing data group and the second sample sensing data group can enable a first energy factory and a second energy factory to jointly meet a first index under macro regulation and control;
s4, binding and labeling the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by using the sample data to obtain an index prediction model;
s5, sending a sensing data request message to a first server and a second server which are bound in advance respectively to obtain a first sensing data group and a second sensing data group respectively, and obtaining a second index under the current macro regulation; wherein the first server corresponds to a third energy plant of the same type as the first energy plant, the second server corresponds to a fourth energy plant of the same type as the second energy plant, the third energy plant is an upstream plant of the fourth energy plant, and the second index is for a community of the third energy plant and the fourth energy plant;
s6, inputting the first sensing data group and the second sensing data group into the index prediction model together for processing to obtain a prediction index value output by the index prediction model, calculating a difference value between the prediction index value and the second index, and judging whether the difference value is smaller than a preset threshold value;
and S7, if the difference is smaller than a preset threshold value, judging that the hybrid energy plant is normally operated.
Further, the types of the numerical values of the first index, the second index and the prediction index are carbon dioxide emission; the first sample sensory data set, the second sample sensory data set, the first sensory data set, and the second sensory data set do not include carbon dioxide emissions.
Further, before sending the sensing data request message to the first server and the second server bound in advance respectively to obtain the first sensing data group and the second sensing data group respectively and obtain the second index under the current macro regulation, the method includes:
s41, acquiring a first sample carbon dioxide emission corresponding to the first sample sensing data group, and acquiring a second sample carbon dioxide emission corresponding to the second sample sensing data group;
s42, performing cross labeling processing to label the second sample carbon dioxide emission amount on the first sample sensing data group, so as to form first sub-training data; simultaneously marking a first sample sensing data group with the first sample carbon dioxide emission so as to form second sub-training data;
s43, calling a preset first neural network model, and training the first neural network model by adopting first sub-training data to obtain a first sub-prediction model;
s44, calling a preset second neural network model, and training the second neural network model by adopting second sub-training data to obtain a second sub-prediction model; wherein the first neural network model is the same as the second neural network model;
the sending of the sensing data request message to the pre-bound first server and the pre-bound second server respectively to obtain the first sensing data set and the second sensing data set respectively, and after obtaining the second index under the current macro regulation, the method includes:
s51, inputting the first sensing data group into the first sub-prediction model for processing to obtain a first predicted carbon dioxide emission output by the first sub-prediction model;
s52, inputting the second sensing data group into the second sub-prediction model for processing to obtain a second predicted carbon dioxide emission output by the second sub-prediction model;
s53, adding the first predicted carbon dioxide emission amount and the second predicted carbon dioxide emission amount to obtain a sum value;
s54, judging whether the difference between the sum and the second index is smaller than a preset threshold value or not;
and S55, if the difference between the sum and the second index is smaller than a preset threshold value, generating an index prediction model processing instruction to instruct to input the first sensing data group and the second sensing data group into the index prediction model together for processing.
Further, the summing processing to sum the first predicted carbon dioxide emissions with the second predicted carbon dioxide emissions to obtain a summed value, comprising:
s521, sending a carbon dioxide emission request message to the first server and the second server simultaneously to obtain the first carbon dioxide emission which is sent by the first server and is not verified, and obtaining the second carbon dioxide emission which is sent by the second server and is not verified;
s522, judging whether the first carbon dioxide emission amount which is not verified is the same as the second predicted carbon dioxide emission amount or not;
s523, if the first carbon dioxide emission amount which is not verified is the same as the second predicted carbon dioxide emission amount, judging whether the second carbon dioxide emission amount which is not verified is the same as the first predicted carbon dioxide emission amount;
and S524, if the unverified second carbon dioxide emission is the same as the first predicted carbon dioxide emission, generating a summation instruction to instruct to perform summation processing.
Further, the binding and labeling the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by using the sample data to obtain the index prediction model includes:
s31, carrying out proportional division processing on the plurality of sample data marked with the first indexes to obtain a plurality of training data and a plurality of verification data;
s32, inputting a plurality of training data into the deep neural network model and training by adopting a gradient descent method to obtain an intermediate index prediction model;
s33, verifying the intermediate index prediction model by adopting a plurality of verification data to obtain a verification result, and judging whether the verification result is passed;
and S34, if the verification result is that the verification is passed, taking the intermediate index prediction model as a final index prediction model.
The application provides a big data analysis device based on hybrid energy data includes:
the system comprises a sample server acquisition unit, a big data analysis server acquisition unit and a big data analysis server acquisition unit, wherein the sample server acquisition unit is used for indicating the big data analysis server to acquire a first sample server and a second sample server which are bound in advance; wherein the first sample server corresponds to a first energy plant, the second sample server corresponds to a second energy plant, and the first energy plant is an upstream plant of the second energy plant;
a first sample sensing data group obtaining unit, configured to instruct to send a first sample sensing data request message to the first sample server, so as to obtain a first sample sensing data group; wherein the first sample sensing data set is sensed by a plurality of sensors deployed in a first energy plant;
a second sample sensing data group obtaining unit, configured to instruct to send a second sample sensing data request message to the second sample server, so as to obtain a second sample sensing data group; wherein the second sample sensing data set is sensed by a plurality of sensors deployed in a second energy plant; the first sample sensing data group and the second sample sensing data group are obtained by sensing at the same time, and the first sample sensing data group and the second sample sensing data group can enable a first energy factory and a second energy factory to jointly meet a first index under macro regulation and control;
the index prediction model acquisition unit is used for indicating that the first sample sensing data group and the second sample sensing data group are bound and labeled so as to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by adopting the sample data so as to obtain an index prediction model;
the sensor data request message sending unit is used for indicating to send sensor data request messages to a first server and a second server which are bound in advance respectively so as to obtain a first sensor data group and a second sensor data group respectively and obtain a second index under current macro regulation and control; wherein the first server corresponds to a third energy plant of the same type as the first energy plant, the second server corresponds to a fourth energy plant of the same type as the second energy plant, the third energy plant is an upstream plant of the fourth energy plant, and the second index is for a community of the third energy plant and the fourth energy plant;
the prediction index value acquisition unit is used for indicating that the first sensing data group and the second sensing data group are input into the index prediction model together for processing so as to obtain a prediction index value output by the index prediction model, calculating a difference value between the prediction index value and the second index, and judging whether the difference value is smaller than a preset threshold value or not;
and the normal operation judging unit is used for indicating that the hybrid energy factory is normally operated if the difference value is smaller than a preset threshold value.
According to the big data analysis method and device based on the hybrid energy data, a first sample server and a second sample server which are bound in advance are obtained; sending a first sample sensing data request message to a first sample server to obtain a first sample sensing data group; sending a second sample sensing data request message to a second sample server to obtain a second sample sensing data group; binding and labeling to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by adopting the sample data to obtain an index prediction model; respectively acquiring a first sensing data group and a second sensing data group, and acquiring a second index under current macro regulation; inputting the first sensing data group and the second sensing data group into the index prediction model together for processing to obtain a prediction index value output by the index prediction model, and calculating a difference value between the prediction index value and the second index; and if the difference is smaller than the preset threshold value, judging that the hybrid energy plant operates normally, achieving the purpose of distinguishing fake data, and facilitating the real realization of the aim of promoting carbon neutralization.
The method utilizes the binding characteristic, and the binding means that two energy factories in upstream and downstream relation are used as a community. In the present application, two energy factories are bound to each other, and correspondingly, two servers are also bound to each other. After binding, the two energy plants can jointly meet the distribution index under macroscopic regulation and control. And the meeting of the indicators between the communities can be performed in any feasible manner, for example, in a fixed ratio agreed in advance, or in a fluctuating ratio (which is reflected in the data of the output and the raw material supply of the upstream and downstream energy plants). The upstream plant of the two energy plants takes all or a part (with a known proportion) of the produced products as raw materials of the downstream plant, and correspondingly, all or a part (with a known proportion) of the raw materials of the downstream plant come from the upstream plant, so that obvious association exists between the upstream plant and the downstream plant, which is the basis for the application to form a community by adopting a binding mode for analysis, but not to analyze the raw materials by a single energy plant.
The binding design is adopted to improve the difficulty of data fake decoration. Specifically, for a single energy plant (in which the index is only for the plant itself), the purpose of finally modifying the carbon dioxide emission can be achieved by modifying a series of sensing data (although the overall modification is difficult to be found compared with the modification of only direct carbon dioxide emission data). However, for the case that two energy plants of the present application are bound and the whole sensing data needs to be modified, it is not feasible to modify the sensing data of only one energy plant, and the sensing data of two energy plants must be modified at the same time, but this difficulty is significantly increased and difficult to implement, so that once the fake data occurs, it is easy to find. On the other hand, in order to realize fine adjustment in macro control, the application aims at a community formed by two energy plants in upstream and downstream relation, so that indexes (such as carbon dioxide emission) of the macro control are aimed at the whole community, fine adjustment possibility can exist between the two energy plants, and implementation flexibility is improved.
Drawings
Fig. 1 is a schematic flowchart of a big data analysis method based on hybrid energy data according to an embodiment of the present application;
fig. 2 is a schematic block diagram illustrating a structure of a big data analysis apparatus based on hybrid energy data according to an embodiment of the present application;
the implementation, functional features and advantages of the objectives of the present application will be further explained with reference to the accompanying drawings.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
Referring to fig. 1, an embodiment of the present application provides a big data analysis method based on hybrid energy data, including the following steps:
s1, the big data analysis server acquires a first sample server and a second sample server which are bound in advance; wherein the first sample server corresponds to a first energy plant, the second sample server corresponds to a second energy plant, and the first energy plant is an upstream plant of the second energy plant;
s2, sending a first sample sensing data request message to the first sample server to obtain a first sample sensing data group; wherein the first sample sensing data set is sensed by a plurality of sensors deployed in a first energy plant;
s3, sending a second sample sensing data request message to the second sample server to obtain a second sample sensing data group; wherein the second sample sensing data set is sensed by a plurality of sensors deployed in a second energy plant; the first sample sensing data group and the second sample sensing data group are obtained by sensing at the same time, and the first sample sensing data group and the second sample sensing data group can enable a first energy factory and a second energy factory to jointly meet a first index under macro regulation and control;
s4, binding and labeling the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by using the sample data to obtain an index prediction model;
s5, sending a sensing data request message to a first server and a second server which are bound in advance respectively to obtain a first sensing data group and a second sensing data group respectively, and obtaining a second index under the current macro regulation; wherein the first server corresponds to a third energy plant of the same type as the first energy plant, the second server corresponds to a fourth energy plant of the same type as the second energy plant, the third energy plant is an upstream plant of the fourth energy plant, and the second index is for a community of the third energy plant and the fourth energy plant;
s6, inputting the first sensing data group and the second sensing data group into the index prediction model together for processing to obtain a prediction index value output by the index prediction model, calculating a difference value between the prediction index value and the second index, and judging whether the difference value is smaller than a preset threshold value;
and S7, if the difference is smaller than a preset threshold value, judging that the hybrid energy plant is normally operated.
Compared with a common big data analysis scheme, the method has the obvious difference that the method is analyzed by taking two pre-bound energy plants as a community, and has the advantages that the difficulty of data fake decoration can be improved, and the possibility of fake decoration on the aspect of the energy plants for sensing data is reduced.
On the other hand, as a macro-control, even though a method of shutting down an energy plant completely and operating another energy plant at full capacity is adopted to reduce the emission of carbon dioxide, it is too extreme, and it is more preferable that both energy plants are normally operated within an allowable range, then binding between two energy plants in an upstream-downstream relationship is performed based on the normal operation, and then indexes are allocated by the two energy plants in the binding relationship in a fixed proportion or flexibly. Since the carbon dioxide emission and the corresponding sensing data are not in a simple linear relationship, it is difficult to implement data falsification by simply modifying the sensing data in proportion, and it is more difficult for two energy plants having a binding relationship.
Hybrid energy in this application means that two energy plants differ from each other in type. An energy plant refers to a plant in the field of energy industry, such as a plant in the fields of coal industry, petroleum industry, and electric power industry, and more specifically, a plant in the field of petrochemical industry (oil refinery, etc.), a plant in the field of electric power (thermal power plant, etc.), which have a large emission amount of carbon dioxide and are necessary for macroscopic regulation.
The binding in the application refers to that two energy plants in the binding are in upstream and downstream relation, wherein the production product of one energy plant is the raw material of the other energy plant, and in the energy industry, the energy plants in the upstream and downstream relation are ubiquitous. For example, an oil refinery is an upstream plant, and downstream plants are petrochemical production plants (e.g., synthetic rubber plants, synthetic fiber plants, etc.).
The index prediction model of the present application can perform prediction processing of an index using a first sensing data group and a second sensing data group as inputs, because:
the first sensing data group is directly related to the emission amount of carbon dioxide of the third energy plant (the index is the emission amount of carbon dioxide at the moment), the second sensing data group is directly related to the emission amount of carbon dioxide of the fourth energy plant, and meanwhile, the two energy plants are in a binding relationship (correspondingly, the first server and the second server are also in a binding relationship, so that the raw material data of the fourth energy plant is related to the first sensing data group corresponding to the third energy plant), and therefore the first sensing data group is also in an indirect relationship with the emission amount of carbon dioxide of the fourth energy plant. The deep neural network model is adopted to discover the complex association relationship, the advantage of the deep neural network model is that the deep neural network model is suitable for discovering the complex relationship which actually exists, and therefore the index prediction model for the hybrid energy plant can be adopted to conduct index prediction.
Therefore, the application relates to macroscopic regulation and control, and can be used as a part of a smart city to be beneficial to the healthy operation of the smart city.
In addition, in the description of the training of the index prediction model, although only the first sample sensing data group and the second sample sensing data group are bound and labeled to generate the sample data labeled with the first index, a preset deep neural network model is called, and the deep neural network model is trained by using the sample data, the number of the sample data is not stated, but those skilled in the art understand that, for the prediction model of machine learning, the prediction model must be obtained by training through a plurality of sample data, and therefore, the number of the sample data is implicitly disclosed herein. The reason why the number of sample data is not written here is to facilitate understanding and to prevent misunderstanding. At this time, one sample data includes a first sample sensing data group and a second sample sensing data group; the process of forming another sample data is also the same.
As described in the above steps S1-S4, the big data analysis server obtains the pre-bound first sample server and second sample server; wherein the first sample server corresponds to a first energy plant, the second sample server corresponds to a second energy plant, and the first energy plant is an upstream plant of the second energy plant; sending a first sample sensing data request message to the first sample server to obtain a first sample sensing data group; wherein the first sample sensing data set is sensed by a plurality of sensors deployed in a first energy plant; sending a second sample sensing data request message to the second sample server to obtain a second sample sensing data group; wherein the second sample sensing data set is sensed by a plurality of sensors deployed in a second energy plant; the first sample sensing data group and the second sample sensing data group are obtained by sensing at the same time, and the first sample sensing data group and the second sample sensing data group can enable a first energy factory and a second energy factory to jointly meet a first index under macro regulation and control; and binding and labeling the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by adopting the sample data to obtain an index prediction model.
The big data analysis server acquires a first sample server and a second sample server which are bound in advance, and the purpose of the big data analysis server is to communicate with the first sample server and the second sample server so as to obtain a corresponding sensing data set. The word of acquisition is a common description in the art, and it can be understood as finding or establishing a communication connection, but those skilled in the art will understand that the description of acquisition is simpler and more convenient and does not affect understanding. The pre-binding of the first and second sample servers effectively means that the first and second energy plants are bound to each other, i.e., the first energy plant is an upstream plant of the second energy plant.
And respectively sending a first sample sensing data request message and a second sample sensing data request message so as to correspondingly obtain a first sample sensing data group and a second sample sensing data group. The method comprises the steps that a server sends a sensing data group to a server, wherein the server is used for sending a sensing data group, and the server is used for sending a sensing data group to the server. The plurality of sensors are different according to the type of the energy plant, for example, for a thermal power plant, the sensing data related to the first index (for example, carbon dioxide emission) should be collected, so the sensors are, for example, a temperature sensor, a pressure sensor, a voltage sensor, a current sensor, a raw material sensor (which can be implemented by using a sensor capable of sensing the weight of the fed material, for example, using a pressure sensor), a gas pressure sensor, etc., the more the sensors are, the more the number of the sensors is, the higher the accuracy of the analysis result of the present application is, because the amount of carbon dioxide to be emitted is necessarily certain after all the influencing factors are sensed by the sensors, which is also the basis for the prediction by the index prediction model of the present application.
In addition, the two energy plants are of different types. Although the sensing data set and the index have a definite corresponding relation, the corresponding relation can be used for forward prediction and cannot be used for backward estimation, namely, if data counterfeiting is needed, the effect of the data to be counterfeited can be known, but the backward data counterfeiting cannot be carried out.
The utility model provides a bind the community of two energy factories, consequently regard as a training data with two sets of sensory data groups jointly, bind and mark the processing to first sample sensory data group and second sample sensory data group promptly to generate mark on the sample data of first index. And calling a preset deep neural network model, and training the deep neural network model by adopting sample data to obtain an index prediction model. The deep neural network model may adopt any feasible model, such as an LSTM neural network model, a convolutional neural network model, a deep residual error network model, and the like. In the training process, a full-supervised learning mode and a semi-supervised learning mode can be adopted, a gradient descent method is adopted for training, and a back propagation algorithm is adopted for updating parameters of each layer of network in the deep neural network model.
The first sample sensing data group and the second sample sensing data group can enable the first energy factory and the second energy factory to jointly meet a first index under macroscopic regulation, and the first index under macroscopic regulation can be met by a community formed by the first energy factory and the second energy factory.
Since the first sample sensing data set directly and uniquely corresponds to the index data (such as carbon dioxide emission) of the corresponding energy plant, and similarly, the second sample sensing data set directly and uniquely corresponds to the index data (such as carbon dioxide emission) of the corresponding energy plant, and the sum of the index data of the two corresponding energy plants is equal to the first index, the index prediction model based on the deep neural network model can be used for forward prediction after full training. It should be noted that the first sample sensory data set is, in fact, capable of affecting the raw material data of the second energy plant, and therefore, the relationship of the first sample sensory data set to the second sample sensory data set is also substantially hidden.
As can be seen from the foregoing, the number of sample data is not explicitly indicated, but the number of sample data for which the index prediction model is trained is inevitably large.
Further, the types of the numerical values of the first index, the second index and the prediction index are carbon dioxide emission; the first sample sensory data set, the second sample sensory data set, the first sensory data set, and the second sensory data set do not include carbon dioxide emissions.
Further, the binding and labeling the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by using the sample data to obtain the index prediction model includes:
s31, carrying out proportional division processing on the plurality of sample data marked with the first indexes to obtain a plurality of training data and a plurality of verification data;
s32, inputting a plurality of training data into the deep neural network model and training by adopting a gradient descent method to obtain an intermediate index prediction model;
s33, verifying the intermediate index prediction model by adopting a plurality of verification data to obtain a verification result, and judging whether the verification result is passed;
and S34, if the verification result is that the verification is passed, taking the intermediate index prediction model as a final index prediction model.
Thereby enabling the final index prediction model to be competent for the forward prediction task for the index. The training data and the verification data are obtained by dividing the same sample data, so that the data in the training process and the data in the verification process are homologous, and the reliability of the model obtained by training can be ensured. Because the labeled sample data is adopted, the training is carried out in a supervised learning (more accurately, semi-supervised learning) mode. In addition, in order to increase the number of sample data and facilitate the training of the model, the method and the device can slightly modify part of parameters, such as a temperature value, a pressure value and the like, when the first sample sensing data group and the second sample sensing data group are acquired, and certainly, the modification must ensure that indexes are basically unchanged (such as carbon dioxide emission is basically unchanged), so that the sensing data group is integrally changed to form new sample data, and the number of the sample data is increased.
As described in the above steps S5-S7, the sensing data request messages are respectively sent to the first server and the second server that are bound in advance, so as to respectively obtain the first sensing data group and the second sensing data group, and obtain the second index under the current macro regulation; wherein the first server corresponds to a third energy plant of the same type as the first energy plant, the second server corresponds to a fourth energy plant of the same type as the second energy plant, the third energy plant is an upstream plant of the fourth energy plant, and the second index is for a community of the third energy plant and the fourth energy plant; inputting the first sensing data group and the second sensing data group into the index prediction model together for processing to obtain a prediction index value output by the index prediction model, calculating a difference value between the prediction index value and the second index, and judging whether the difference value is smaller than a preset threshold value; and if the difference value is smaller than a preset threshold value, judging that the hybrid energy plant operates normally.
Preferably, the first energy plant is the same (not only of the same type) as the third energy plant, while the second energy plant is the same as the fourth energy plant. Because the index can change under the macro regulation, the final purpose of the application is to analyze whether the two energy plants can meet the second index under the current macro regulation, and the fake decoration of data is not carried out in the process. Therefore, the first sensing data set and the second sensing data set also need to be acquired. And all, or a known proportion, of the output of the third energy plant is the feedstock of the fourth energy plant.
Since the relationship between the third energy plant and the fourth energy plant is the same as the relationship between the first energy plant and the second energy plant, the index prediction model can be used to predict the index, thereby obtaining the prediction index value obtained by forward prediction of the sensing data set. And calculating a difference value between the prediction index value and the second index, judging whether the difference value is smaller than a preset threshold value, and if the difference value is smaller than the preset threshold value, indicating that the difference value is in accordance with the expectation, so that the hybrid energy plant is judged to normally operate.
In an embodiment, before sending the sensing data request message to the first server and the second server that are bound in advance respectively to obtain the first sensing data group and the second sensing data group respectively and obtain the second index under the current macro regulation, the method includes:
s41, acquiring a first sample carbon dioxide emission corresponding to the first sample sensing data group, and acquiring a second sample carbon dioxide emission corresponding to the second sample sensing data group;
s42, performing cross labeling processing to label the second sample carbon dioxide emission amount on the first sample sensing data group, so as to form first sub-training data; simultaneously marking a first sample sensing data group with the first sample carbon dioxide emission so as to form second sub-training data;
s43, calling a preset first neural network model, and training the first neural network model by adopting first sub-training data to obtain a first sub-prediction model;
s44, calling a preset second neural network model, and training the second neural network model by adopting second sub-training data to obtain a second sub-prediction model; wherein the first neural network model is the same as the second neural network model;
the sending of the sensing data request message to the pre-bound first server and the pre-bound second server respectively to obtain the first sensing data set and the second sensing data set respectively, and after obtaining the second index under the current macro regulation, the method includes:
s51, inputting the first sensing data group into the first sub-prediction model for processing to obtain a first predicted carbon dioxide emission output by the first sub-prediction model;
s52, inputting the second sensing data group into the second sub-prediction model for processing to obtain a second predicted carbon dioxide emission output by the second sub-prediction model;
s53, adding the first predicted carbon dioxide emission amount and the second predicted carbon dioxide emission amount to obtain a sum value;
s54, judging whether the difference between the sum and the second index is smaller than a preset threshold value or not;
and S55, if the difference between the sum and the second index is smaller than a preset threshold value, generating an index prediction model processing instruction to instruct to input the first sensing data group and the second sensing data group into the index prediction model together for processing.
In order to emphasize the binding relationship between two energy plants and to mention the reliability of the prediction, the present application further employs two sub-models. And when the sub-model is used for marking the training data, a cross marking mode is adopted, namely the first sample sensing data group is marked to actually indicate the second sample carbon dioxide emission amount corresponding to the second energy factory, and the second sample sensing data group is marked to actually indicate the first sample carbon dioxide emission amount corresponding to the first energy factory. Thus, the first sample sensory data set predicts carbon dioxide emissions data for the second energy plant. Since the two energy plants have upstream and downstream relations, namely the corresponding relation between the output and the raw material, the cross prediction also has an exact corresponding relation, so that the prediction can be carried out by adopting a sub prediction model based on a neural network model. The first neural network model may be the same as or different from the deep neural network model. The two sub-prediction models obtained by training can judge whether the predicted value reaches the index under the macro regulation and control and whether the data fake decoration is possible or not again in a mode of adding the values.
Further, the summing processing to sum the first predicted carbon dioxide emissions with the second predicted carbon dioxide emissions to obtain a summed value, comprising:
s521, sending a carbon dioxide emission request message to the first server and the second server simultaneously to obtain the first carbon dioxide emission which is sent by the first server and is not verified, and obtaining the second carbon dioxide emission which is sent by the second server and is not verified;
s522, judging whether the first carbon dioxide emission amount which is not verified is the same as the second predicted carbon dioxide emission amount or not;
s523, if the first carbon dioxide emission amount which is not verified is the same as the second predicted carbon dioxide emission amount, judging whether the second carbon dioxide emission amount which is not verified is the same as the first predicted carbon dioxide emission amount;
and S524, if the unverified second carbon dioxide emission is the same as the first predicted carbon dioxide emission, generating a summation instruction to instruct to perform summation processing.
And judging whether the first unverified carbon dioxide emission is the same as the second predicted carbon dioxide emission or not, wherein the judging is a correct writing method instead of judging whether the first unverified carbon dioxide emission is the same as the first predicted carbon dioxide emission or not, and the judging is a false writing method. Since the first sub-model predicts the carbon dioxide emission corresponding to one energy plant through the sensing data set corresponding to another energy plant, two times of cross judgment are needed here to determine whether the prediction result meets the expectation, and then a summation instruction is generated to instruct summation processing.
According to the big data analysis method based on the hybrid energy data, a first sample server and a second sample server which are bound in advance are obtained; sending a first sample sensing data request message to a first sample server to obtain a first sample sensing data group; sending a second sample sensing data request message to a second sample server to obtain a second sample sensing data group; binding and labeling to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by adopting the sample data to obtain an index prediction model; respectively acquiring a first sensing data group and a second sensing data group, and acquiring a second index under current macro regulation; inputting the first sensing data group and the second sensing data group into the index prediction model together for processing to obtain a prediction index value output by the index prediction model, and calculating a difference value between the prediction index value and the second index; and if the difference is smaller than the preset threshold value, judging that the hybrid energy plant operates normally, achieving the purpose of distinguishing fake data, and facilitating the real realization of the aim of promoting carbon neutralization.
Referring to fig. 2, an embodiment of the present application provides a big data analysis device based on hybrid energy data, including:
a sample server obtaining unit 10, configured to instruct a big data analysis server to obtain a first sample server and a second sample server that are bound in advance; wherein the first sample server corresponds to a first energy plant, the second sample server corresponds to a second energy plant, and the first energy plant is an upstream plant of the second energy plant;
a first sample sensing data group obtaining unit 20, configured to instruct to send a first sample sensing data request message to the first sample server, so as to obtain a first sample sensing data group; wherein the first sample sensing data set is sensed by a plurality of sensors deployed in a first energy plant;
a second sample sensing data group obtaining unit 30, configured to instruct to send a second sample sensing data request message to the second sample server, so as to obtain a second sample sensing data group; wherein the second sample sensing data set is sensed by a plurality of sensors deployed in a second energy plant; the first sample sensing data group and the second sample sensing data group are obtained by sensing at the same time, and the first sample sensing data group and the second sample sensing data group can enable a first energy factory and a second energy factory to jointly meet a first index under macro regulation and control;
the index prediction model obtaining unit 40 is configured to instruct to bind and label the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, call a preset deep neural network model, and train the deep neural network model with the sample data to obtain an index prediction model;
a sensing data request message sending unit 50, configured to instruct to send a sensing data request message to a first server and a second server that are bound in advance, so as to obtain a first sensing data group and a second sensing data group, respectively, and obtain a second index under current macro regulation; wherein the first server corresponds to a third energy plant of the same type as the first energy plant, the second server corresponds to a fourth energy plant of the same type as the second energy plant, the third energy plant is an upstream plant of the fourth energy plant, and the second index is for a community of the third energy plant and the fourth energy plant;
a prediction index value obtaining unit 60, configured to instruct that the first sensing data set and the second sensing data set are input to the index prediction model together for processing, so as to obtain a prediction index value output by the index prediction model, calculate a difference value between the prediction index value and the second index, and determine whether the difference value is smaller than a preset threshold value;
and a normal operation determination unit 70 for indicating that the hybrid energy plant is normally operated if the difference is smaller than a preset threshold.
In one embodiment, the first indicator, the second indicator, and the prediction indicator value are all of the type of carbon dioxide emissions; the first sample sensory data set, the second sample sensory data set, the first sensory data set, and the second sensory data set do not include carbon dioxide emissions.
In an embodiment, before sending the sensing data request message to the first server and the second server that are bound in advance respectively to obtain the first sensing data group and the second sensing data group respectively and obtain the second index under the current macro regulation, the method includes:
acquiring a first sample carbon dioxide emission corresponding to the first sample sensing data set, and acquiring a second sample carbon dioxide emission corresponding to the second sample sensing data set;
performing cross labeling processing to label the first sample sensing data group with the carbon dioxide emission of the second sample, thereby forming first sub-training data; simultaneously marking a first sample sensing data group with the first sample carbon dioxide emission so as to form second sub-training data;
calling a preset first neural network model, and training the first neural network model by adopting first sub-training data to obtain a first sub-prediction model;
calling a preset second neural network model, and training the second neural network model by adopting second sub-training data to obtain a second sub-prediction model; wherein the first neural network model is the same as the second neural network model;
the sending of the sensing data request message to the pre-bound first server and the pre-bound second server respectively to obtain the first sensing data set and the second sensing data set respectively, and after obtaining the second index under the current macro regulation, the method includes:
inputting the first sensing data group into the first sub-prediction model for processing to obtain a first predicted carbon dioxide emission output by the first sub-prediction model;
inputting the second sensing data group into the second sub-prediction model for processing so as to obtain a second predicted carbon dioxide emission output by the second sub-prediction model;
performing a summation process to add the first predicted carbon dioxide emissions to the second predicted carbon dioxide emissions to obtain a summation value;
judging whether the difference between the sum and the second index is smaller than a preset threshold value or not;
and if the difference between the sum value and the second index is smaller than a preset threshold value, generating an index prediction model processing instruction to indicate that the first sensing data group and the second sensing data group are input into the index prediction model together for processing.
In one embodiment, the summing process to sum the first predicted carbon dioxide emissions with the second predicted carbon dioxide emissions to obtain a summed value comprises:
simultaneously sending carbon dioxide emission request messages to a first server and a second server so as to obtain the first carbon dioxide emission which is not verified and sent by the first server and obtain the second carbon dioxide emission which is not verified and sent by the second server;
judging whether the unverified first carbon dioxide emission is the same as the second predicted carbon dioxide emission;
if the first unverified carbon dioxide emission is the same as the second predicted carbon dioxide emission, judging whether the second unverified carbon dioxide emission is the same as the first predicted carbon dioxide emission;
and if the unverified second carbon dioxide emission is the same as the first predicted carbon dioxide emission, generating a summation instruction to instruct summation processing to be carried out.
In an embodiment, the binding and labeling the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model with the sample data to obtain the index prediction model includes:
carrying out proportional division processing on the plurality of sample data marked with the first indexes to obtain a plurality of training data and a plurality of verification data;
inputting a plurality of training data into a deep neural network model and training by adopting a gradient descent method to obtain an intermediate index prediction model;
adopting a plurality of verification data to verify the intermediate index prediction model to obtain a verification result, and judging whether the verification result is passed;
and if the verification result is that the verification is passed, taking the intermediate index prediction model as a final index prediction model.
The operations performed by the units correspond to the steps of the hybrid energy data-based big data analysis method according to the foregoing embodiment one to one, and are not described herein again.
The big data analysis device based on the hybrid energy data obtains a first sample server and a second sample server which are bound in advance; sending a first sample sensing data request message to a first sample server to obtain a first sample sensing data group; sending a second sample sensing data request message to a second sample server to obtain a second sample sensing data group; binding and labeling to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by adopting the sample data to obtain an index prediction model; respectively acquiring a first sensing data group and a second sensing data group, and acquiring a second index under current macro regulation; inputting the first sensing data group and the second sensing data group into the index prediction model together for processing to obtain a prediction index value output by the index prediction model, and calculating a difference value between the prediction index value and the second index; and if the difference is smaller than the preset threshold value, judging that the hybrid energy plant operates normally, achieving the purpose of distinguishing fake data, and facilitating the real realization of the aim of promoting carbon neutralization.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, apparatus, article, or method. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, apparatus, article, or method that includes the element.
The above description is only a preferred embodiment of the present application, and not intended to limit the scope of the present application, and all modifications of equivalent structures and equivalent processes, which are made by the contents of the specification and the drawings of the present application, or which are directly or indirectly applied to other related technical fields, are also included in the scope of the present application.
Claims (10)
1. A big data analysis method based on hybrid energy data is characterized by comprising the following steps:
s1, the big data analysis server acquires a first sample server and a second sample server which are bound in advance; wherein the first sample server corresponds to a first energy plant, the second sample server corresponds to a second energy plant, and the first energy plant is an upstream plant of the second energy plant;
s2, sending a first sample sensing data request message to the first sample server to obtain a first sample sensing data group; wherein the first sample sensing data set is sensed by a plurality of sensors deployed in a first energy plant;
s3, sending a second sample sensing data request message to the second sample server to obtain a second sample sensing data group; wherein the second sample sensing data set is sensed by a plurality of sensors deployed in a second energy plant; the first sample sensing data group and the second sample sensing data group are obtained by sensing at the same time, and the first sample sensing data group and the second sample sensing data group can enable a first energy factory and a second energy factory to jointly meet a first index under macro regulation and control;
s4, binding and labeling the first sample sensing data group and the second sample sensing data group to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by using the sample data to obtain an index prediction model;
s5, sending a sensing data request message to a first server and a second server which are bound in advance respectively to obtain a first sensing data group and a second sensing data group respectively, and obtaining a second index under the current macro regulation; wherein the first server corresponds to a third energy plant of the same type as the first energy plant, the second server corresponds to a fourth energy plant of the same type as the second energy plant, the third energy plant is an upstream plant of the fourth energy plant, and the second index is for a community of the third energy plant and the fourth energy plant;
s6, inputting the first sensing data group and the second sensing data group into the index prediction model together for processing to obtain a prediction index value output by the index prediction model, calculating a difference value between the prediction index value and the second index, and judging whether the difference value is smaller than a preset threshold value;
and S7, if the difference is smaller than a preset threshold value, judging that the hybrid energy plant is normally operated.
2. The big data analysis method based on hybrid energy data according to claim 1, wherein the types of the first index, the second index and the prediction index are carbon dioxide emission; the first sample sensory data set, the second sample sensory data set, the first sensory data set, and the second sensory data set do not include carbon dioxide emissions.
3. The big data analysis method based on hybrid energy data according to claim 2, wherein before sending the sensing data request message to the pre-bound first server and the pre-bound second server to obtain the first sensing data set and the second sensing data set, respectively, and obtain the second index under the current macro regulation, the method comprises:
s41, acquiring a first sample carbon dioxide emission corresponding to the first sample sensing data group, and acquiring a second sample carbon dioxide emission corresponding to the second sample sensing data group;
s42, performing cross labeling processing to label the second sample carbon dioxide emission amount on the first sample sensing data group, so as to form first sub-training data; simultaneously marking a first sample sensing data group with the first sample carbon dioxide emission so as to form second sub-training data;
s43, calling a preset first neural network model, and training the first neural network model by adopting first sub-training data to obtain a first sub-prediction model;
s44, calling a preset second neural network model, and training the second neural network model by adopting second sub-training data to obtain a second sub-prediction model; wherein the first neural network model is the same as the second neural network model;
the sending of the sensing data request message to the pre-bound first server and the pre-bound second server respectively to obtain the first sensing data set and the second sensing data set respectively, and after obtaining the second index under the current macro regulation, the method includes:
s51, inputting the first sensing data group into the first sub-prediction model for processing to obtain a first predicted carbon dioxide emission output by the first sub-prediction model;
s52, inputting the second sensing data group into the second sub-prediction model for processing to obtain a second predicted carbon dioxide emission output by the second sub-prediction model;
s53, adding the first predicted carbon dioxide emission amount and the second predicted carbon dioxide emission amount to obtain a sum value;
s54, judging whether the difference between the sum and the second index is smaller than a preset threshold value or not;
and S55, if the difference between the sum and the second index is smaller than a preset threshold value, generating an index prediction model processing instruction to instruct to input the first sensing data group and the second sensing data group into the index prediction model together for processing.
4. The big data analysis method based on hybrid energy data according to claim 3, wherein the summing the first predicted carbon dioxide emissions plus the second predicted carbon dioxide emissions to obtain a summed value comprises:
s521, sending a carbon dioxide emission request message to the first server and the second server simultaneously to obtain the first carbon dioxide emission which is sent by the first server and is not verified, and obtaining the second carbon dioxide emission which is sent by the second server and is not verified;
s522, judging whether the first carbon dioxide emission amount which is not verified is the same as the second predicted carbon dioxide emission amount or not;
s523, if the first carbon dioxide emission amount which is not verified is the same as the second predicted carbon dioxide emission amount, judging whether the second carbon dioxide emission amount which is not verified is the same as the first predicted carbon dioxide emission amount;
and S524, if the unverified second carbon dioxide emission is the same as the first predicted carbon dioxide emission, generating a summation instruction to instruct to perform summation processing.
5. The big data analysis method based on hybrid energy data according to claim 1, wherein before the binding and labeling of the first sample sensing data set and the second sample sensing data set to generate sample data labeled with the first index, and invoking a preset deep neural network model, training the deep neural network model with the sample data to obtain the index prediction model, the method comprises:
s31, carrying out proportional division processing on the plurality of sample data marked with the first indexes to obtain a plurality of training data and a plurality of verification data;
s32, inputting a plurality of training data into the deep neural network model and training by adopting a gradient descent method to obtain an intermediate index prediction model;
s33, verifying the intermediate index prediction model by adopting a plurality of verification data to obtain a verification result, and judging whether the verification result is passed;
and S34, if the verification result is that the verification is passed, taking the intermediate index prediction model as a final index prediction model.
6. A big data analysis device based on hybrid energy data is characterized by comprising:
the system comprises a sample server acquisition unit, a big data analysis server acquisition unit and a big data analysis server acquisition unit, wherein the sample server acquisition unit is used for indicating the big data analysis server to acquire a first sample server and a second sample server which are bound in advance; wherein the first sample server corresponds to a first energy plant, the second sample server corresponds to a second energy plant, and the first energy plant is an upstream plant of the second energy plant;
a first sample sensing data group obtaining unit, configured to instruct to send a first sample sensing data request message to the first sample server, so as to obtain a first sample sensing data group; wherein the first sample sensing data set is sensed by a plurality of sensors deployed in a first energy plant;
a second sample sensing data group obtaining unit, configured to instruct to send a second sample sensing data request message to the second sample server, so as to obtain a second sample sensing data group; wherein the second sample sensing data set is sensed by a plurality of sensors deployed in a second energy plant; the first sample sensing data group and the second sample sensing data group are obtained by sensing at the same time, and the first sample sensing data group and the second sample sensing data group can enable a first energy factory and a second energy factory to jointly meet a first index under macro regulation and control;
the index prediction model acquisition unit is used for indicating that the first sample sensing data group and the second sample sensing data group are bound and labeled so as to generate sample data labeled with the first index, calling a preset deep neural network model, and training the deep neural network model by adopting the sample data so as to obtain an index prediction model;
the sensor data request message sending unit is used for indicating to send sensor data request messages to a first server and a second server which are bound in advance respectively so as to obtain a first sensor data group and a second sensor data group respectively and obtain a second index under current macro regulation and control; wherein the first server corresponds to a third energy plant of the same type as the first energy plant, the second server corresponds to a fourth energy plant of the same type as the second energy plant, the third energy plant is an upstream plant of the fourth energy plant, and the second index is for a community of the third energy plant and the fourth energy plant;
the prediction index value acquisition unit is used for indicating that the first sensing data group and the second sensing data group are input into the index prediction model together for processing so as to obtain a prediction index value output by the index prediction model, calculating a difference value between the prediction index value and the second index, and judging whether the difference value is smaller than a preset threshold value or not;
and the normal operation judging unit is used for indicating that the hybrid energy factory is normally operated if the difference value is smaller than a preset threshold value.
7. The big data analysis device based on hybrid energy data according to claim 6, wherein the types of the first index, the second index and the prediction index are carbon dioxide emission; the first sample sensory data set, the second sample sensory data set, the first sensory data set, and the second sensory data set do not include carbon dioxide emissions.
8. The big data analysis device based on hybrid energy data according to claim 7, wherein before sending the sensing data request message to the pre-bound first server and the pre-bound second server to obtain the first sensing data set and the second sensing data set, respectively, and obtain the second index under the current macro regulation, the big data analysis device comprises:
acquiring a first sample carbon dioxide emission corresponding to the first sample sensing data set, and acquiring a second sample carbon dioxide emission corresponding to the second sample sensing data set;
performing cross labeling processing to label the first sample sensing data group with the carbon dioxide emission of the second sample, thereby forming first sub-training data; simultaneously marking a first sample sensing data group with the first sample carbon dioxide emission so as to form second sub-training data;
calling a preset first neural network model, and training the first neural network model by adopting first sub-training data to obtain a first sub-prediction model;
calling a preset second neural network model, and training the second neural network model by adopting second sub-training data to obtain a second sub-prediction model; wherein the first neural network model is the same as the second neural network model;
the sending of the sensing data request message to the pre-bound first server and the pre-bound second server respectively to obtain the first sensing data set and the second sensing data set respectively, and after obtaining the second index under the current macro regulation, the method includes:
inputting the first sensing data group into the first sub-prediction model for processing to obtain a first predicted carbon dioxide emission output by the first sub-prediction model;
inputting the second sensing data group into the second sub-prediction model for processing so as to obtain a second predicted carbon dioxide emission output by the second sub-prediction model;
performing a summation process to add the first predicted carbon dioxide emissions to the second predicted carbon dioxide emissions to obtain a summation value;
judging whether the difference between the sum and the second index is smaller than a preset threshold value or not;
and if the difference between the sum value and the second index is smaller than a preset threshold value, generating an index prediction model processing instruction to indicate that the first sensing data group and the second sensing data group are input into the index prediction model together for processing.
9. The big data analysis device according to claim 8, wherein the summing process for summing the first predicted carbon dioxide emission and the second predicted carbon dioxide emission to obtain a sum value comprises:
simultaneously sending carbon dioxide emission request messages to a first server and a second server so as to obtain the first carbon dioxide emission which is not verified and sent by the first server and obtain the second carbon dioxide emission which is not verified and sent by the second server;
judging whether the unverified first carbon dioxide emission is the same as the second predicted carbon dioxide emission;
if the first unverified carbon dioxide emission is the same as the second predicted carbon dioxide emission, judging whether the second unverified carbon dioxide emission is the same as the first predicted carbon dioxide emission;
and if the unverified second carbon dioxide emission is the same as the first predicted carbon dioxide emission, generating a summation instruction to instruct summation processing to be carried out.
10. The big data analysis device based on hybrid energy data according to claim 6, wherein before the binding and labeling processing of the first sample sensing data set and the second sample sensing data set to generate sample data labeled with the first index, and invoking a preset deep neural network model, training the deep neural network model with the sample data to obtain the index prediction model, the method comprises:
carrying out proportional division processing on the plurality of sample data marked with the first indexes to obtain a plurality of training data and a plurality of verification data;
inputting a plurality of training data into a deep neural network model and training by adopting a gradient descent method to obtain an intermediate index prediction model;
adopting a plurality of verification data to verify the intermediate index prediction model to obtain a verification result, and judging whether the verification result is passed;
and if the verification result is that the verification is passed, taking the intermediate index prediction model as a final index prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111389440.7A CN113824800B (en) | 2021-11-23 | 2021-11-23 | Big data analysis method and device based on hybrid energy data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111389440.7A CN113824800B (en) | 2021-11-23 | 2021-11-23 | Big data analysis method and device based on hybrid energy data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113824800A CN113824800A (en) | 2021-12-21 |
| CN113824800B true CN113824800B (en) | 2022-02-11 |
Family
ID=78918066
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111389440.7A Active CN113824800B (en) | 2021-11-23 | 2021-11-23 | Big data analysis method and device based on hybrid energy data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113824800B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105975799A (en) * | 2016-06-01 | 2016-09-28 | 广东电网有限责任公司电力科学研究院 | Method and system for calculating carbon emissions |
| CN107798467A (en) * | 2017-10-11 | 2018-03-13 | 杭州市环境保护科学研究院 | Water pollution burst accident based on deep learning quickly meet an urgent need assess and decision-making technique |
| CN108793289A (en) * | 2018-07-12 | 2018-11-13 | 刘守凯 | A kind of intelligentized industrial wastewater treatment system |
| CN111127849A (en) * | 2019-12-19 | 2020-05-08 | 浙江天禄环境科技有限公司 | Accident early warning method comprehensively considering meteorological and chemical plant peripheral area |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| BR112019017301A2 (en) * | 2017-10-09 | 2020-04-22 | Bl Technologies, Inc. | intelligent methods and systems for health diagnosis of a water treatment plant, anomaly detection and control |
-
2021
- 2021-11-23 CN CN202111389440.7A patent/CN113824800B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105975799A (en) * | 2016-06-01 | 2016-09-28 | 广东电网有限责任公司电力科学研究院 | Method and system for calculating carbon emissions |
| CN107798467A (en) * | 2017-10-11 | 2018-03-13 | 杭州市环境保护科学研究院 | Water pollution burst accident based on deep learning quickly meet an urgent need assess and decision-making technique |
| CN108793289A (en) * | 2018-07-12 | 2018-11-13 | 刘守凯 | A kind of intelligentized industrial wastewater treatment system |
| CN111127849A (en) * | 2019-12-19 | 2020-05-08 | 浙江天禄环境科技有限公司 | Accident early warning method comprehensively considering meteorological and chemical plant peripheral area |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113824800A (en) | 2021-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pan et al. | Data-driven estimation of building energy consumption with multi-source heterogeneous data | |
| Tang et al. | Perturbation analysis of evidential reasoning rule | |
| CN104657792A (en) | Early warning method based on rule engine and intelligent prediction | |
| Yang et al. | A hybrid deep learning and mechanistic kinetics model for the prediction of fluid catalytic cracking performance | |
| CN110991866B (en) | Machine learning-based platform area data missing value completion method and device | |
| KR20150018681A (en) | Apparatus and method for predicting product condition by using datamining in manufacturing process | |
| CN104881715B (en) | A kind of method that paper plant predicts pulp property according to waste paper proportioning | |
| WO2012029154A1 (en) | Internal abnormality diagnosis method, internal abnormality diagnosis system, and decision tree generating method for internal abnormality diagnosis of oil-filled electric apparatus utilizing gas concentration in oil | |
| CN102013148A (en) | Multi-information fusion fire hazard detection method | |
| CN108052092A (en) | A kind of subway electromechanical equipment abnormal state detection method based on big data analysis | |
| Pohjoranta et al. | Validation of neural network-based fault diagnosis for multi-stack fuel cell systems: Stack voltage deviation detection | |
| CN113824800B (en) | Big data analysis method and device based on hybrid energy data | |
| WO2011162592A1 (en) | A method and system for validating energy measurement in a high pressure gas distribution network | |
| Kadiyala et al. | Vector time series models for prediction of air quality inside a public transportation bus using available software | |
| CN114740159B (en) | Natural gas energy metering component acquisition method and Internet of things system | |
| Lepore et al. | Neural network based control charting for multiple stream processes with an application to HVAC systems in passenger railway vehicles | |
| Yi et al. | A fault diagnosis method of oil-immersed transformer based on improved harris hawks optimized random forest | |
| CN110390454A (en) | A kind of analysis of petrochemical equipment risk of leakage and management system | |
| Shaban et al. | A double neural network approach for the identification and parameter estimation of control chart patterns | |
| Liu et al. | Evidential Ensemble Preference-Guided Learning Approach for Real-Time Multimode Fault Diagnosis | |
| Sun et al. | Traceability analysis for low-voltage distribution network abnormal line loss using a data-driven power flow model | |
| Anderson et al. | A Bayesian reliability analysis exploring the effect of scheduled maintenance on wind turbine time to failure | |
| Medina et al. | Comparative analysis of artificial neural networks and statistical models applied to demand forecasting | |
| CN113923058B (en) | Big data early warning method and device for non-instant energy data analysis | |
| Yu et al. | Machine learning for data verification in emissions trading system |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder |
Address after: 430000 Office and production buildings (Building 5-6, Dongfang Industrial Park) in 2MA plot of Wuhan Economic and Technological Development Zone, Hubei Province (Ji HCY-5 # 311) Patentee after: Wuhan Chaoyun Technology Co.,Ltd. Address before: 430000 No. 8, floor 1-3, building 5, phase II, Tainan 2nd Road North, Gaoqiao 4th Road West, Gaoqiao 3rd Road East tiandihui logistics business headquarters construction project, Changqing street, Dongxihu District, Wuhan City, Hubei Province Patentee before: Wuhan Chaoyun Technology Co.,Ltd. |
|
| CP02 | Change in the address of a patent holder |