CN113780405B - Air conditioner parameter regression optimization method based on deep neural network - Google Patents
Air conditioner parameter regression optimization method based on deep neural network Download PDFInfo
- Publication number
- CN113780405B CN113780405B CN202111044715.3A CN202111044715A CN113780405B CN 113780405 B CN113780405 B CN 113780405B CN 202111044715 A CN202111044715 A CN 202111044715A CN 113780405 B CN113780405 B CN 113780405B
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- air conditioner
- regression
- matrix
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Air Conditioning Control Device (AREA)
Abstract
The invention discloses an air conditioner parameter regression optimization method based on a deep neural network, and relates to the technical field of air conditioner production. The invention comprises data acquisition: acquiring various sample data by acquiring a platinum resistor, a temperature sensor and a main control board; data preprocessing: preprocessing the acquired data to acquire training sample data; feature selection: expert knowledge and practical experience are introduced to extract monitoring data to be used as input of the neural network; model training: selecting ResNet-152 as a backbone network, and carrying out regression prediction on the characteristics extracted by the backbone network by using a binary decision model; real-time data analysis: inputting the established model to obtain a group of high-dimensional abstract features, and taking the abstract features as the input of a regression tree, and analyzing the output data of the regression tree. The invention quickens the testing speed of the air conditioner, adopts the self-learning neural network to carry out real-time parameter adjustment on the air conditioner data, and can quicken the research and development speed of the machine.
Description
Technical Field
The invention belongs to the technical field of air conditioner production, and particularly relates to an air conditioner parameter regression optimization method based on a deep neural network.
Background
The air conditioner parameter adjustment is a significant research topic in the heating and ventilation field, and parameters which can be adjusted by a just-manufactured air conditioner prototype include refrigerant type, refrigerant filling amount, main valve opening, auxiliary valve opening and main frequency. An appropriate set of parameters may allow the machine to achieve or exceed design performance and design energy consumption. The conventional method requires setting initialization air conditioning parameters according to recorded data, and then adjusting parameters except for refrigerant types and refrigerant filling amounts. When the performance and cop of the machine reach the limit under the current parameters, the refrigerant filling amount is increased, and the process is repeated until the end point of debugging is reached when the refrigerant filling amount is increased so as not to influence the performance and cop of the machine. Returning to the above-mentioned parameter selection process, it requires a lot of manpower to put into experiments, and has a certain requirement on human experience, and when there are too many machines designed in the project, a lengthy test flow will have a serious influence on the project progress.
The neural network is also called artificial neural network, which is a learning algorithm model for simulating the behavior characteristics of animal neural network and carrying out distributed parallel information processing. It is a complex network formed by interconnecting a large number of simple neurons (processing units) in a hierarchical manner. Through a complex connection mode, the method can fit any complex function for processing various data analysis problems. Neural networks are generally divided into an input layer, a hidden layer and an output layer. The input layer comprises a large number of neurons for receiving linear or nonlinear data, the hidden layer comprises one or more layers of neurons, and abstract features in the data are extracted through connection of the hidden layer with neurons of other layers, so that the purpose of simulating various models is achieved. The features extracted by the hidden layer get the data we want after entering the output layer. The neural network may be classified into a single-layer neural network and a multi-layer neural network according to the difference of hidden layers. Now due to the rapid development of neural networks, we have many effective network architectures and pre-trained neural networks that have been validated. Modern neural network architectures are divided into backbone networks and recurrent networks. The backbone network is used for extracting the characteristics, and the regression network is used for receiving the characteristics and predicting.
Disclosure of Invention
The invention aims to provide an air conditioner parameter regression optimization method based on a deep neural network, which is characterized in that monitoring data is stored in a distributed mode and calculated in parallel through a cloud platform, so that parameter adjustment can be performed on a plurality of experimental machines to be tested at the same time, and the problems of difficult test, high test cost and low test efficiency of the existing air conditioner are solved.
In order to solve the technical problems, the invention is realized by the following technical scheme:
the invention relates to an air conditioner parameter regression optimization method based on a deep neural network, which comprises the following steps:
step S1, data acquisition: acquiring various detection amounts by acquiring a platinum resistance temperature sensor and a main control board to obtain sample data of a system in operation;
step S2, data preprocessing: sequentially carrying out abnormal data removal, denoising, normalization, reduction and data vectorization on the obtained data to obtain preprocessed training sample data;
step S3, feature selection: introducing expert knowledge and practical experience to extract monitoring data influencing the performance cop of the air conditioner from the preprocessed training sample data as the input of the neural network;
step S4, model training: selecting a feature extraction layer of ResNet-152 as a backbone network for feature extraction, and carrying out regression prediction on features extracted by the backbone network by using a binary decision model;
step S5, real-time data analysis: after preprocessing and feature selection, inputting the real-time collected data into a built model to obtain a group of high-dimensional abstract features, and taking the group of abstract features as the input of a regression tree, wherein the output of the regression tree is the data analysis result;
step S6, self-learning: the monitoring data generated each time in real time is used as a new training set to help the neural network to adapt and perfect itself.
In the step S1, the system acquires various detection amounts by acquiring the platinum resistance temperature sensor and the main control board in real time, and uses the detection amounts as sample data when the system operates, and also can acquire historical detection amounts of different machines as sample data through a database; the sample data is used for training of the neural network.
As a preferable solution, in the step S2, when one data in the set of sample data is lost, the obtained sample data needs to be processed according to different cases of the data volume:
when the lost data belongs to the category attribute, the data is subjected to one-time thermal coding, namely a real category is added to the data;
when the lost data belongs to the numerical value type and the data in the same category is less than three pieces, replacing the lost data by using the average value or the mode of similar data;
when the lost data belongs to the numerical type and the data in the same category is more than three pieces, the group of data is deleted.
As a preferred embodiment, in the step S2, the normalization process is used to map the values of the numerical value sequence in the dataset to a common scale; the normalization processing method comprises the following steps:
(1) Z normalization treatment:
(2) Maximum and minimum normalization processing:
(3) Unit vector normalization processing:
where X represents a matrix of data sets, where m represents m pieces of data when the matrix is [ m, n ] in size, and n represents a dimension of the data.
As a preferable technical solution, in the step S2, the main component is used to reduce the dimension, and the steps are as follows:
provided with data X of (mxn), then:
step S21: forming the original data into a matrix X of n rows and m columns according to columns;
step S22: zero-equalizing each row of X, namely subtracting the average value of the row;
step S23: solving covariance matrix
Step S24: obtaining eigenvalues and corresponding eigenvectors of the covariance matrix;
step S25: arranging the eigenvectors into a matrix according to the corresponding eigenvalue from top to bottom, and taking the first k rows to form a matrix P;
step S26: y=px, which is the data after dimension reduction to k dimension.
As a preferred technical solution, in the step S4, a binary decision model is constructed by using a least squares regression tree generation algorithm, and in an input space where the training data set is located, each region is recursively divided into two sub-regions and an output value of each sub-region is determined, and the step of constructing the binary decision tree is as follows:
step S41, selecting an optimal component variable j and a segmentation point S, and solving a formula:
wherein the traversal variable j scans the segmentation point s for the fixed segmentation variable j, and selects the pair (j, s) that minimizes the above equation, R 1 And R is 2 The method comprises the steps of carrying out a first treatment on the surface of the Representing a partition of the regression tree corresponding to the input space; c 1 And c 2 Is the output of each division, y i Is a component of y in the original data;
step S42: dividing the regions by the selected pairs (j, s) and determining the corresponding output values:
R1(j,s)＝{x|x(j)≤s},R2(j,s)＝{x|x(j)＞s}；
wherein s represents a j-th variable; r is R 1 And R is 2 Is the region divided by j and s;r represents m All input instances x above i Corresponding output y i Is the average value of (2);
step S43: continuing to call the step S41 and the step S42 for the two sub-areas until a stopping condition is met;
step S44: dividing the input space into M sub-regions R 1 、R 2 ，R 3 ，...，R M Generating a decision tree f (x):
i is an indication function.
The invention has the following beneficial effects:
1. the invention accelerates the testing speed of the air conditioner, adopts the self-learning neural network to carry out real-time parameter adjustment on the air conditioner data, and can accelerate the research and development speed of the machine;
2. according to the invention, the monitoring data is stored in a distributed manner and calculated in parallel through the cloud platform, so that parameters of a plurality of experimental machines to be tested can be adjusted at the same time, and the testing difficulty in the peak period can be solved more gracefully;
3. on the basis of the method, the learning capacity of the algorithm is added, so that the fault recognition capacity can be continuously improved, new characteristics which are not summarized manually can be obtained through continuous self-learning, the data monitored in real time can fluctuate along with seasons, temperatures and evaporator types, compressor manufacturers design different conditions, and the change rule of the data can be well adapted through self-learning.
Of course, it is not necessary for any one product to practice the invention to achieve all of the advantages set forth above at the same time.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed for the description of the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a step diagram of an air conditioner parameter regression optimization method based on a deep neural network;
FIG. 2 is a block diagram of ResNet-152;
FIG. 3 is a diagram of a basic neural network architecture;
fig. 4 is a single neuron model.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, the invention discloses a deep neural network-based air conditioner parameter regression optimization method, which comprises the following steps:
step S1, data acquisition: acquiring various detection amounts by acquiring a platinum resistance temperature sensor and a main control board to obtain sample data of a system in operation;
in the step S1, the system acquires various detection amounts through acquiring a platinum resistance temperature sensor and a main control board in real time, and the detection amounts are used as sample data when the system operates, and also can acquire historical detection amounts of different machines through a database to be used as the sample data; the sample data is used for training the neural network, so that the neural network can monitor the data in real time and tune the parameters.
Step S2, data preprocessing: sequentially carrying out abnormal data removal, denoising, normalization, reduction and data vectorization on the obtained data, and obtaining preprocessed training sample data, wherein the preprocessed training sample data is used for preprocessing the acquired data and converting the preprocessed data into a data format suitable for data mining;
in step S2, when data is collected, for various reasons, when one of the data in the group is lost, it is necessary to process this case according to different cases of the data amount:
when the lost data belongs to the category attribute, the data is subjected to one-time thermal coding, namely a real category is added to the data, if the data represents four seasons, four categories are theoretically needed, but because of the existence of the lost data, a fifth category can be added, namely no record exists, and the corresponding one-time thermal coding is 00001;
when the lost data belongs to the numerical value type and the data in the same category is less than three pieces, replacing the lost data by using the average value or the mode of similar data;
when the lost data belongs to the numerical type and the data in the same category is more than three pieces, the group of data is deleted.
In step S2, the normalization process is used to map the values of the numerical columns in the dataset to a common scale, so as to avoid the problem of numerical distortion caused by overlarge differences of the data, and for the neural network, normalization is not required for each column of data, and only when the features have different ranges; in this document, the normalization processing method used is as follows:
(1) Z normalization treatment:
(2) Maximum and minimum normalization processing:
(3) Unit vector normalization processing:
where X represents a matrix of data sets, where m represents m pieces of data when the matrix is [ m, n ] in size, and n represents a dimension of the data.
Data dimension reduction: in general, we prefer to add more features, but when the feature types exceed a certain number, the performance of the neural network decreases, and the feature space dimension increases and becomes thinner due to the continuous addition of features. This sparsity can lead to neural network models that are prone to overfitting, requiring projection of data from high latitude space into low dimensional space to create new features, which can yield information rich non-redundant features.
In step S2, the main component reduces the dimension when in use, and the steps are as follows:
provided with data X of (mxn), then:
step S21: forming the original data into a matrix X of n rows and m columns according to columns;
step S22: zero-equalizing each row of X, namely subtracting the average value of the row;
step S23: solving covariance matrix
Step S24: obtaining eigenvalues and corresponding eigenvectors of the covariance matrix;
step S25: arranging the eigenvectors into a matrix according to the corresponding eigenvalue from top to bottom, and taking the first k rows to form a matrix P;
step S26: y=px, which is the data after dimension reduction to k dimension.
Data vectorization: the data is converted into a structure that is easy to handle by the neural network.
Step S3, feature selection: the method comprises the steps of acquiring and preprocessing monitoring data to obtain a large amount of training sample data, wherein factors influencing the performance cop of the machine are far less than the acquired data, so that expert knowledge and practical experience are introduced to extract the monitoring data influencing the performance cop of the air conditioner from the preprocessed training sample data to serve as input of a neural network;
referring to fig. 2-4, in step S4, model training: the feature extraction layer of ResNet-152 is selected as a backbone network for feature extraction, regression prediction is performed on features extracted by the backbone network by using a binary decision model, and the parameters of the whole network are huge (about 1.3 hundred million), so that the whole network architecture cannot be well trained by using the existing data, and the direct consequence is that the whole network can show over fitting or under fitting to input data, so that generalization performance is poor. Therefore, we select the ResNet-152 model which is trained on the ImageNet as our initialization network, and the data volume required by network fitting is greatly reduced by using the transfer learning technology;
in step S4, a binary decision model is constructed by using a least squares regression tree generation algorithm, and in the input space where the training data set is located, each region is recursively divided into two sub-regions and the output value of each sub-region is determined, and the step of constructing a binary decision tree is as follows:
the decision tree generation model is as follows:
input: an output data set D of the backbone network;
and (3) outputting: regression tree f (x).
In the input space where the training data set is located, recursively dividing each region into two sub-regions and determining the output value of each sub-region to construct a binary decision tree;
step S41, selecting an optimal component variable j and a segmentation point S, and solving a formula:
wherein the traversal variable j scans the segmentation point s for the fixed segmentation variable j, and selects the pair (j, s) that minimizes the above equation, R 1 And R is 2 The method comprises the steps of carrying out a first treatment on the surface of the Representing a partition of the regression tree corresponding to the input space; c 1 And c 2 Is the output of each division, y i Is a component of y in the original data;
step S42: dividing the regions by the selected pairs (j, s) and determining the corresponding output values:
R1(j,s)＝{x|x(j)≤s},R2(j,s)＝{x|x(j)＞s}；
wherein s represents a j-th variable; r is R 1 And R is 2 Is the region divided by j and s;r represents m All input instances x above i Corresponding output y i Is the average value of (2);
step S43: continuing to call the step S41 and the step S42 for the two sub-areas until a stopping condition is met;
step S44: dividing the input space into M sub-regions R 1 、R 2 ，R 3 ，...，R M Generating a decision tree f (x):
i is an indication function;
the output f (x) is the required parameter, and the number of regression trees to be trained can be set according to the required output number.
Step S5, real-time data analysis: after preprocessing and feature selection, the real-time collected data is input into an established model, a group of high-dimensional abstract features are obtained after connection and conversion of each layer of neurons of a neural network, the abstract features are used as the input of a regression tree, the output of the regression tree is the data analysis result, and in the debugging process, a machine can be debugged according to the output of the regression tree;
step S6, self-learning: the neural network is not invariable after being trained by the training set, and monitoring data generated in real time each time can be used as a new training set to help the neural network to adapt and perfect, for example, the experience of compressors of the same nameplate but different brands is different, and the performance improvement and cop change after frequency adding are also different, so that the model is required to be continuously corrected and perfected.
It should be noted that, in the above system embodiment, each unit included is only divided according to the functional logic, but not limited to the above division, so long as the corresponding function can be implemented; in addition, the specific names of the functional units are also only for distinguishing from each other, and are not used to limit the protection scope of the present invention.
In addition, those skilled in the art will appreciate that all or part of the steps in implementing the methods of the embodiments described above may be implemented by a program to instruct related hardware, and the corresponding program may be stored in a computer readable storage medium.
The preferred embodiments of the invention disclosed above are intended only to assist in the explanation of the invention. The preferred embodiments are not exhaustive or to limit the invention to the precise form disclosed. Obviously, many modifications and variations are possible in light of the above teaching. The embodiments were chosen and described in order to best explain the principles of the invention and the practical application, to thereby enable others skilled in the art to best understand and utilize the invention. The invention is limited only by the claims and the full scope and equivalents thereof.
Claims (5)
1. The air conditioner parameter regression optimization method based on the deep neural network is characterized by comprising the following steps of:
step S1, data acquisition: acquiring various detection amounts by acquiring a platinum resistance temperature sensor and a main control board to obtain sample data of a system in operation;
step S2, data preprocessing: sequentially carrying out abnormal data removal, denoising, normalization, reduction and data vectorization on the obtained data to obtain preprocessed training sample data;
step S3, feature selection: introducing expert knowledge and practical experience to extract monitoring data influencing the performance cop of the air conditioner from the preprocessed training sample data as the input of the neural network;
step S4, model training: selecting a feature extraction layer of ResNet-152 as a backbone network for feature extraction, and carrying out regression prediction on features extracted by the backbone network by using a binary decision model;
step S5, real-time data analysis: after preprocessing and feature selection, inputting the real-time collected data into a built model to obtain a group of high-dimensional abstract features, and taking the group of abstract features as the input of a regression tree, wherein the output of the regression tree is the data analysis result;
step S6, self-learning: the monitoring data generated in real time each time is used as a new training set to help the neural network to adapt and perfect itself;
in the step S2, when one of the obtained sample data is lost, the processing is required according to different cases of the data amount:
when the lost data belongs to the category attribute, the data is subjected to one-time thermal coding, namely a real category is added to the data;
when the lost data belongs to the numerical value type and the data in the same category is less than three pieces, replacing the lost data by using the average value or the mode of similar data;
when the lost data belongs to the numerical type and the data in the same category is more than three pieces, the group of data is deleted.
2. The air conditioner parameter regression optimization method based on the deep neural network according to claim 1, wherein in the step S1, the system acquires various detection amounts by acquiring the platinum resistance temperature sensor and the main control board in real time, and uses the detection amounts as sample data when the system operates, and also can acquire the historical detection amounts of different machines as sample data through a database; the sample data is used for training of the neural network.
3. The method according to claim 1, wherein in the step S2, the normalization process is used to map the values of the numerical value sequence in the dataset to a common scale; the normalization processing method comprises the following steps:
(1) Z normalization treatment:
(2) Maximum and minimum normalization processing:
(3) Unit vector normalization processing:
where X represents a matrix of data sets, where m represents m pieces of data when the matrix is [ m, n ] in size, and n represents a dimension of the data.
4. The air conditioner parameter regression optimization method based on the deep neural network according to claim 1, wherein in the step S2, the dimension of the adopted main component is reduced, and the steps are as follows:
provided with data X of (mxn), then:
step S21: forming the original data into a matrix X of n rows and m columns according to columns;
step S22: zero-equalizing each row of X, namely subtracting the average value of the row;
step S23: solving covariance matrix
Step S24: obtaining eigenvalues and corresponding eigenvectors of the covariance matrix;
step S25: arranging the eigenvectors into a matrix according to the corresponding eigenvalue from top to bottom, and taking the first k rows to form a matrix P;
step S26: y=px, which is the data after dimension reduction to k dimension.
5. The method according to claim 1, wherein in the step S4, a binary decision model is constructed by using a least squares regression tree generation algorithm, and in the input space where the training data set is located, each region is recursively divided into two sub-regions and the output value of each sub-region is determined, and the step of constructing the binary decision tree is as follows:
step S41, selecting an optimal component variable j and a segmentation point S, and solving a formula:
wherein the traversal variable j scans the segmentation point s for the fixed segmentation variable j, and selects the pair (j, s) that minimizes the above equation, R 1 And R is 2 Representing the correspondence of regression trees to inputsA division into spaces; c 1 And c 2 Is the output of each division, y i Is a component of y in the original data;
step S42: dividing the regions by the selected pairs (j, s) and determining the corresponding output values:
wherein s represents a j-th variable; r is R 1 And R is 2 Is the region divided by j and s;r represents m All input instances x above i Corresponding output y i Is the average value of (2);
step S43: continuing to call the step S41 and the step S42 for the two sub-areas until a stopping condition is met;
step S44: dividing the input space into M sub-regions R 1 、R 2 ，R 3 ，...，R M Generating a decision tree f (x):
i is an indication function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111044715.3A CN113780405B (en) | 2021-09-07 | 2021-09-07 | Air conditioner parameter regression optimization method based on deep neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111044715.3A CN113780405B (en) | 2021-09-07 | 2021-09-07 | Air conditioner parameter regression optimization method based on deep neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113780405A CN113780405A (en) | 2021-12-10 |
| CN113780405B true CN113780405B (en) | 2023-07-25 |
Family
ID=78841500
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111044715.3A Active CN113780405B (en) | 2021-09-07 | 2021-09-07 | Air conditioner parameter regression optimization method based on deep neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113780405B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114543403A (en) * | 2022-03-07 | 2022-05-27 | 温格智能制造(安徽)科技有限公司 | Joint control method based on Internet of things filling equipment |
Family Cites Families (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106765932A (en) * | 2016-12-14 | 2017-05-31 | 深圳达实智能股份有限公司 | The Energy Efficiency Ratio Forecasting Methodology and device of central air conditioner system refrigeration host computer |
| CN106874581B (en) * | 2016-12-30 | 2021-03-30 | 浙江大学 | Building air conditioner energy consumption prediction method based on BP neural network model |
| CN108507117A (en) * | 2017-10-13 | 2018-09-07 | 上海智容睿盛智能科技有限公司 | A kind of Air-conditioning system sensor method for diagnosing faults based on wavelet neural network |
| JP6955450B2 (en) * | 2018-01-17 | 2021-10-27 | 日立グローバルライフソリューションズ株式会社 | Air conditioning control system and air conditioning control method |
| US11416739B2 (en) * | 2018-01-29 | 2022-08-16 | Lawrence Livermore National Security, Llc | Optimization control technology for building energy conservation |
| CN110968341B (en) * | 2018-09-30 | 2024-01-19 | 上海汽车集团股份有限公司 | Method and device for setting air conditioner parameters |
| CN109631238A (en) * | 2019-01-28 | 2019-04-16 | 宁波溪棠信息科技有限公司 | A kind of control system and control method improving air-conditioning system operational energy efficiency |
| CN109959123A (en) * | 2019-03-11 | 2019-07-02 | 浙江工业大学 | A kind of energy-saving method for air conditioner based on genetic algorithm and shot and long term memory Recognition with Recurrent Neural Network |
| CN109961186A (en) * | 2019-03-22 | 2019-07-02 | 大唐环境产业集团股份有限公司 | Desulphurization system operating parameter prediction technique based on decision tree and BP neural network |
| CN110542178B (en) * | 2019-09-02 | 2020-06-02 | 上海德昂科技有限公司 | Air conditioner and refrigerator room control method and system with self-learning capability |
| CN110848892B (en) * | 2019-10-11 | 2021-08-31 | 广东寰球智能科技有限公司 | Energy-saving operation method, device and equipment of ice storage system and storage medium |
| CN111059737A (en) * | 2019-12-16 | 2020-04-24 | 珠海格力电器股份有限公司 | Air conditioner indoor unit control device and method and air conditioner indoor unit |
| CN111256294B (en) * | 2020-01-17 | 2021-01-05 | 深圳市得益节能科技股份有限公司 | Model prediction-based optimization control method for combined operation of water chilling unit |
| CN114186469A (en) * | 2020-09-14 | 2022-03-15 | 青岛海信日立空调系统有限公司 | Air conditioning system energy efficiency prediction method and air conditioning system |
| CN113177594B (en) * | 2021-04-29 | 2022-06-17 | 浙江大学 | Air conditioner fault diagnosis method based on Bayesian optimization PCA-extreme random tree |
| CN113268913B (en) * | 2021-06-24 | 2022-02-11 | 广州鼎泰智慧能源科技有限公司 | Intelligent building air conditioner cooling machine system operation optimization method based on PSO-ELM algorithm |
-
2021
- 2021-09-07 CN CN202111044715.3A patent/CN113780405B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN113780405A (en) | 2021-12-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111337768B (en) | Deep parallel fault diagnosis method and system for dissolved gas in transformer oil | |
| CN109408389B (en) | Code defect detection method and device based on deep learning | |
| CN111460728B (en) | Method and device for predicting residual life of industrial equipment, storage medium and equipment | |
| CN110609524B (en) | Industrial equipment residual life prediction model and construction method and application thereof | |
| CN106991355B (en) | Face recognition method of analytic dictionary learning model based on topology maintenance | |
| CN112699605B (en) | Charging pile fault element prediction method and system | |
| CN105469376A (en) | Method and device for determining picture similarity | |
| CN111046961B (en) | Fault classification method based on bidirectional long-time and short-time memory unit and capsule network | |
| CN112147432A (en) | BiLSTM module based on attention mechanism, transformer state diagnosis method and system | |
| CN117290800A (en) | Timing sequence anomaly detection method and system based on hypergraph attention network | |
| CN113780405B (en) | Air conditioner parameter regression optimization method based on deep neural network | |
| CN116400168A (en) | Power grid fault diagnosis method and system based on depth feature clustering | |
| CN117076936A (en) | Time sequence data anomaly detection method based on multi-head attention model | |
| CN112418175A (en) | Rolling bearing fault diagnosis method and system based on domain migration and storage medium | |
| CN114169091A (en) | Method for establishing prediction model of residual life of engineering mechanical part and prediction method | |
| CN116595465A (en) | High-dimensional sparse data outlier detection method and system based on self-encoder and data enhancement | |
| CN116245019A (en) | Load prediction method, system, device and storage medium based on Bagging sampling and improved random forest algorithm | |
| CN117235565A (en) | Transformer fault diagnosis model construction method and device | |
| Sharma et al. | A semi-supervised generalized vae framework for abnormality detection using one-class classification | |
| CN113469013B (en) | Motor fault prediction method and system based on transfer learning and time sequence | |
| Xue et al. | Deep anomaly detection for industrial systems: a case study | |
| CN117516939A (en) | Bearing cross-working condition fault detection method and system based on improved EfficientNetV2 | |
| CN114626426A (en) | Industrial equipment behavior detection method based on K-means optimization algorithm | |
| CN117113139A (en) | Training method and device for fault detection model, computer equipment and storage medium | |
| CN116482491A (en) | Transformer partial discharge fault diagnosis method based on Bayesian neural network |
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 |