Air conditioner load prediction method and device based on genetic algorithm, equipment and medium

Abstract

The invention discloses an air conditioner load prediction method and device based on a genetic algorithm, equipment and a medium. The method comprises the steps that a historical parameter data set corresponding to a target air conditioning system is obtained; performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set; training a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model; and performing data prediction on the real-time parameter data corresponding to the target air conditioning system based on the target deep learning model to obtain a load prediction result corresponding to the target air conditioning system. According to the technical scheme, the air conditioner load can be predicted through the real-time data, and the prediction efficiency and precision of air conditioner load prediction are improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device, equipment and medium for predicting air conditioning load based on a genetic algorithm. Background Art

[0002] With the rapid development of computer technology, it has become very important to predict the load of air conditioning systems so that they can be stopped when they are not needed or when the load demand is low, thus avoiding unnecessary energy consumption.

[0003] In the prior art, air conditioning load forecasting is usually performed based on machine learning. However, if basic methods such as machine learning are used for air conditioning load forecasting, the performance is poor in time series forecasting. At the same time, data feature redundancy affects the training of the model, reducing the prediction efficiency and accuracy of air conditioning load forecasting. Therefore, how to use real-time data to forecast air conditioning load and improve the prediction efficiency and accuracy of air conditioning load forecasting is a problem that needs to be solved urgently. Summary of the invention

[0004] The present invention provides a method, device, equipment and medium for air conditioning load prediction based on a genetic algorithm, which can solve the problem of low prediction efficiency and accuracy of air conditioning load prediction.

[0005] According to one aspect of the present invention, there is provided an air conditioning load prediction method based on a genetic algorithm, comprising:

[0006] Obtain a historical parameter data set corresponding to the target air-conditioning system;

[0007] Performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set;

[0008] Training a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model;

[0009] Based on the target deep learning model, data prediction is performed on the real-time parameter data corresponding to the target air-conditioning system to obtain a load prediction result corresponding to the target air-conditioning system.

[0010] According to another aspect of the present invention, there is provided an air conditioning load prediction device based on a genetic algorithm, comprising:

[0011] A data acquisition module is used to obtain a historical parameter data set corresponding to the target air-conditioning system;

[0012] A data preprocessing module, used for performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set;

[0013] A model training module, used to train a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model;

[0014] The load prediction module is used to perform data prediction on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain the load prediction result corresponding to the target air-conditioning system.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the air conditioning load prediction method based on genetic algorithm described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the air conditioning load prediction method based on genetic algorithm described in any embodiment of the present invention when executed.

[0020] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the air conditioning load prediction method based on a genetic algorithm described in any embodiment of the present invention is implemented.

[0021] The technical solution of the embodiment of the present invention performs feature selection on the historical parameter data set corresponding to the target air-conditioning system through a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set. Furthermore, a preset deep learning model is trained based on the historical data feature subset to obtain a trained target deep learning model. Finally, data prediction is performed on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain a load prediction result corresponding to the target air-conditioning system. Since a genetic algorithm is used for feature selection and a deep learning model is used for time-series air-conditioning load prediction, the problem of low prediction efficiency and accuracy of air-conditioning load prediction is solved, and real-time data can be used for air-conditioning load prediction, thereby improving the prediction efficiency and accuracy of air-conditioning load prediction.

[0022] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 is a flow chart of an air conditioning load prediction method based on a genetic algorithm according to Embodiment 1 of the present invention;

[0025] Figure 2 is a flow chart of an air conditioning load prediction method based on a genetic algorithm according to Embodiment 2 of the present invention;

[0026] Figure 3 is a flowchart of a process for generating a feature subset of historical data provided in Embodiment 2 of the present invention;

[0027] Figure 4 is a schematic diagram of the structure of a target deep learning model provided according to Embodiment 2 of the present invention;

[0028] Figure 5 is a schematic diagram of the principle of a long short-term memory layer provided according to the second embodiment of the present invention;

[0029] Figure 6 is a flow chart of an optional air conditioning load prediction method based on genetic algorithm provided according to the second embodiment of the present invention;

[0030] Figure 7 is a flowchart of a decision adjustment process provided according to Embodiment 2 of the present invention;

[0031] Figure 8 is a structural schematic diagram of an air conditioning load prediction device based on a genetic algorithm according to Embodiment 3 of the present invention;

[0032] Fig. 9 It is a structural schematic diagram of an electronic device for implementing the air conditioning load prediction method based on a genetic algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "goal", "history", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] Embodiment 1

[0036] Figure 1 This is a flowchart of a method for predicting air conditioning load based on a genetic algorithm provided in the first embodiment of the present invention. This embodiment is applicable to the case of real-time prediction of air conditioning load. The method can be executed by an air conditioning load prediction device based on a genetic algorithm. The air conditioning load prediction device based on a genetic algorithm can be implemented in the form of hardware and / or software. The air conditioning load prediction device based on a genetic algorithm can be configured in an electronic device. Figure 1 As shown, the method includes:

[0037] S110: Obtain a historical parameter data set corresponding to the target air-conditioning system.

[0038] The target air conditioning system may refer to an air conditioning system for which load forecasting is required. For example, the target air conditioning system may be an air conditioning system in a certain indoor environment. The parameter data may refer to parameters related to the operation of the air conditioning system and the environment in which it is located. For example, the parameter data may be indoor temperature, outdoor temperature at several different locations, indoor humidity, the time when the air conditioner starts working, air supply volume, working time of the air conditioner, working status of the air conditioner and power consumption. Usually, the parameter data can be collected by sensors such as temperature and humidity sensors or wind speed sensors.

[0039] The historical parameter data may refer to the parameter data corresponding to the target air conditioning system in the historical time period. For example, the historical parameter data may be the parameter data corresponding to the target air conditioning system in the historical month. The historical parameter data set may refer to a set of historical parameter data in the same historical time period. It is worth noting that in order to ensure the accuracy of the subsequent training model, in the embodiment of the present invention, the amount of data in the historical parameter data set should be large enough.

[0040] S120. Perform feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set.

[0041] The preset genetic algorithm may refer to a pre-selected search algorithm for solving the optimization problem. Exemplarily, the preset genetic algorithm may be a Sharing Evolution Genetic Algorithms (SEGA). Feature selection may refer to an operation of screening each parameter data in a historical parameter data set. By feature selection, features in the historical parameter data set that are strongly correlated with the air conditioning load condition may be determined.

[0042] The historical data feature may refer to a feature in the historical parameter data set that is strongly related to the air conditioning load condition. The historical data feature subset may refer to a set of historical data features corresponding to the same historical parameter data set.

[0043] S130: Training a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model.

[0044] The preset deep learning model may refer to a pre-set model for predicting air conditioning load. Exemplarily, the preset deep learning model may be a long short-term memory network (LSTM). The target deep learning model may refer to a trained deep learning model obtained by training the preset deep learning model. Usually, the preset deep learning model has the same model architecture as the target deep learning model.

[0045] S140. Perform data prediction on real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain a load prediction result corresponding to the target air-conditioning system.

[0046] Among them, the real-time parameter data may refer to the parameter data corresponding to the target air-conditioning system collected at the current moment. The air-conditioning load may refer to the heat input or discharged by the indoor air per unit time to maintain the design conditions of the indoor air. Exemplarily, the air-conditioning load may include heat load and cooling load. The heat load may refer to the heat supplied to the building by the heating system per unit time to achieve the required indoor temperature at a certain outdoor temperature in winter. The cooling load may refer to the heat that needs to be removed from the room per unit time while keeping the indoor air thermal and humidity parameters within a certain required range. The load prediction result may refer to the air-conditioning load prediction value corresponding to the target air-conditioning system under the real-time parameter data.

[0047] The technical solution of the embodiment of the present invention performs feature selection on the historical parameter data set corresponding to the target air-conditioning system through a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set. Furthermore, a preset deep learning model is trained based on the historical data feature subset to obtain a trained target deep learning model. Finally, data prediction is performed on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain a load prediction result corresponding to the target air-conditioning system. Since a genetic algorithm is used for feature selection and a deep learning model is used for time-series air-conditioning load prediction, the problem of low prediction efficiency and accuracy of air-conditioning load prediction is solved, and real-time data can be used for air-conditioning load prediction, thereby improving the prediction efficiency and accuracy of air-conditioning load prediction.

[0048] Embodiment 2

[0049] Figure 2 A flowchart of an air conditioning load prediction method based on a genetic algorithm is provided in the second embodiment of the present invention. This embodiment is refined based on the above embodiment. In this embodiment, the operation of performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set is specifically refined. Specifically, it may include: merging and processing the historical parameter data set based on a target timestamp to generate a historical multi-feature data set corresponding to the historical parameter data set; standardizing the historical multi-feature data set based on preset standardization rules to generate a historical standardized data set corresponding to the historical parameter data set; performing feature selection on the historical standardized data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set. Figure 2 As shown, the method includes:

[0050] S210: Obtain a historical parameter data set corresponding to the target air-conditioning system.

[0051] S220 , merging and processing the historical parameter data set based on the target timestamp to generate a historical multi-feature data set corresponding to the historical parameter data set.

[0052] The target timestamp may refer to the time point information selected for data merging. Exemplarily, the target timestamp may be the moment when each historical parameter data in the historical parameter data set is collected. The historical multi-feature data may refer to a data group obtained by merging the historical parameter data corresponding to the same target timestamp in the historical parameter data set. The historical multi-feature data set may refer to a set of historical multi-feature data corresponding to the same historical parameter data set.

[0053] Specifically, after obtaining the historical parameter data set corresponding to the target air conditioning system, the historical parameter data corresponding to the same target timestamp can be merged to generate historical multi-feature data in the form of (target timestamp, indoor temperature, outdoor temperature at several different locations, indoor humidity, time when the air conditioner starts working, air supply volume, working time of the air conditioner, working status of the air conditioner, power consumption), and the historical multi-feature data corresponding to the same historical parameter data set can be combined. Thus, a historical multi-feature data set corresponding to the historical parameter data set is generated, providing an effective basis for subsequent operations.

[0054] S230: Standardize the historical multi-feature data set based on preset standardization rules to generate a historical standardized data set corresponding to the historical parameter data set.

[0055] The preset standardization rules may refer to pre-set conditions for constraining the standardization process. Exemplarily, the preset standardization rules may be missing value completion and normalization operations for the historical multi-feature data set. The historical standardized data may refer to the standard data obtained after the historical multi-feature data is standardized according to the preset standardization rules. The historical standardized data set may refer to a set of historical standardized data corresponding to the same historical multi-feature data set.

[0056] Specifically, after generating the historical multi-feature data set corresponding to the historical parameter data set, the interpolation method can be used to fill in the missing data points in the historical multi-feature data to ensure the integrity of the data. Then, all the features with missing value filling are normalized so that the data of different features have the same dimension, which is convenient for subsequent analysis.

[0057] It is worth noting that in the embodiment of the present invention, after generating the historical standardized data set corresponding to the historical parameter data set, the historical standardized data set can also be divided into a training set and a test set according to a preset data division rule. Exemplarily, the historical standardized data set is divided into 80% training set and 20% test set to ensure the effectiveness of model training and evaluation.

[0058] S240: Perform feature selection on the historical standardized data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set.

[0059] In an optional embodiment, feature selection is performed on the historical standardized data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set, including: initializing the historical standardized data set to obtain a basic individual cluster corresponding to the historical standardized data set; wherein the basic individual cluster contains individuals, and each individual represents a feature combination; incrementally processing the basic individual cluster based on a preset cross-increment rule to generate incremental individuals corresponding to the historical standardized data set; merging the basic individual cluster and the incremental individuals to generate a target individual cluster corresponding to the historical parameter data set; and screening and processing the target individual cluster based on preset optimization conditions to generate a historical data feature subset corresponding to the historical parameter data set.

[0060] Among them, the basic individual cluster may refer to a cluster composed of individuals randomly generated after initialization processing of the historical standardized data set. Usually, the basic individual cluster contains individual individuals, and each individual represents a feature combination. The preset crossover incremental rule may refer to a pre-set rule for constraining the crossover and mutation process. Exemplarily, the preset crossover incremental rule may be to perform a crossover operation first and then a mutation operation. The incremental individual may refer to a newly generated individual after incremental processing according to the preset crossover incremental rule. The target individual cluster may refer to a cluster obtained after merging each incremental individual into the basic individual cluster. The preset optimization condition may refer to a pre-set condition for optimizing and screening the target individual cluster. Exemplarily, the preset optimization condition may be the number of screened individuals.

[0061] In an optional embodiment, the basic individual cluster is incrementally processed based on a preset crossover incremental rule to generate incremental individuals corresponding to the historical standardized data set, including: performing fitness evaluation on the target individuals in the basic individual cluster based on a preset fitness function to generate an evaluation result; determining a parent set corresponding to the historical standardized data set in the historical standardized data set based on the evaluation result; and performing a crossover mutation operation on the parent set based on a preset crossover mutation strategy to obtain incremental individuals corresponding to the historical standardized data set.

[0062] The preset fitness function may refer to a pre-set function for performing fitness calculation. The performance of each individual may be evaluated by the preset fitness function. Exemplarily, the preset fitness function may be a formula: Among them, w a and w F It can represent the preset weight, which can usually be set according to the actual application requirements; F can represent the number of feature subsets of the selected individual; P can represent the total number of all features in the selected individual; It can represent the complexity of feature subsets; RMSE can represent mean square error. Specifically, RMSE can be expressed as: Among them, y can represent the real value; can represent the predicted value; n can represent the number of features in the selected individual.

[0063] Among them, the target individual may refer to an individual selected from the basic individual cluster for fitness evaluation. Exemplarily, the target individual may be any individual in the basic individual cluster. The evaluation result may refer to the calculation result generated after the fitness evaluation of the target individual is performed according to the preset fitness function. The parent may refer to an individual selected from the basic individual cluster for use as an incremental reference. The parent set may refer to a set composed of each parent corresponding to the same basic individual cluster. The preset crossover mutation strategy may refer to a preset strategy for limiting the crossover mutation operation. Exemplarily, the preset crossover mutation strategy may include the specific process of the crossover operation and the specific process of the mutation operation.

[0064] Figure 3 The flowchart of a historical data feature subset generation process provided by an embodiment of the present invention is shown. Specifically, after generating a historical standardized data set corresponding to a historical parameter data set, the historical standardized data set can be initialized and processed to generate a basic individual cluster containing N individuals. Afterwards, the fitness of the target individual in the basic individual cluster is evaluated according to a preset fitness function to generate an evaluation result, and N mothers are independently selected from the basic individual cluster based on the evaluation result to form a mother set. Further, a corresponding crossover operator is used to perform a crossover operation on each mother in the mother set to generate a new individual, and a mutation operation is performed on the individuals after the crossover to generate an incremental individual, thereby increasing the diversity of the basic individual cluster. Finally, the basic individual cluster and the incremental individual are merged into a population to form a target individual cluster of size 2N, and N individuals are selected from the target individual cluster according to the evaluation result to obtain a new population. The above operations are repeated until a preset stop condition is met, such as meeting the maximum number of iterations or the fitness is stable, and a historical data feature subset is generated. Thus, by continuously optimizing the feature subset, the optimal combination is finally converged to provide high-quality input data for the subsequent air conditioning load prediction model, thereby significantly improving the prediction performance of the model.

[0065] S250 , converting and processing the historical data feature subset based on a preset data format to generate a target data feature subset corresponding to the historical data feature subset.

[0066] Among them, the preset data format may refer to a preset model input data format. Usually, different preset deep learning models correspond to different preset data formats, and the preset data format can be determined according to the input data format specified by the preset deep learning model. The target data feature may refer to a data group obtained by formatting the historical data feature according to the preset data format. The target data feature subset may refer to a set of target data features corresponding to the same historical data feature subset.

[0067] S260: Perform data segmentation on the target data feature subset based on a preset sliding window rule to generate a target data feature subsequence corresponding to the target data feature subset.

[0068] The preset sliding window rule may refer to a pre-set rule for constraining a window sliding operation. Exemplarily, the preset sliding window rule may be a single window sliding. The target data feature subsequence may refer to a fixed-length subsequence obtained after data segmentation of the target data feature subset according to the preset sliding window rule.

[0069] S270: Input the target data feature subsequence into a preset deep learning model, train the preset deep learning model, and obtain a trained target deep learning model.

[0070] Specifically, after generating a target data feature subsequence corresponding to a target data feature subset, first, input the training set in the target data feature subsequence into a preset deep learning model, train the preset deep learning model, and obtain a trained candidate deep learning model. Then, input the test set in the target data feature subsequence into the candidate deep learning model, and test the candidate deep learning model, thereby obtaining a target deep learning model that meets the preset test requirements.

[0071] Figure 4 The figure shows a schematic diagram of the structure of a target deep learning model provided by an embodiment of the present invention. In an optional embodiment, the target deep learning model may include: an input layer, a long short-term memory layer, a regularization layer, a fully connected layer, and an output layer. Among them, the input layer is used for data input. The long short-term memory layer is used to effectively solve the problem of gradient disappearance or gradient explosion in traditional recurrent neural networks (RNNs) by introducing a gating mechanism. Figure 5 FIG. 1 is a schematic diagram of the principle of a long short-term memory layer provided by an embodiment of the present invention. Specifically, first, the target data feature subsequence (of 1 ,of 2 ,...,of n ) together constitute the input x of the current time step t, and the hidden state h of the previous time step t-1 concatenated together as the input for the subsequent steps. Afterwards, the forget gate f is calculated t , which is used to determine the cell state c from the previous moment t-1 The information that needs to be forgotten in the forget gate can be calculated using the formula: t =σ(W f ·[h t-1 ,x t ]+b f ) indicates that W f represents the weight matrix of the forget gate, b f It can represent the bias vector, σ is the sigmoid activation function. Then, calculate the input gate i t , which controls the input x of the current time step t What information needs to be written into the cell state? The input gate can use the formula: t =σ(W i ·[h t-1 ,x t ]+b i ) indicates that W i It can represent the weight matrix of the input gate, b i Can represent the bias vector. Further, generate candidate cell state m t , which is calculated by the current input and the hidden state controlled by the reset gate, and can be obtained using the formula: t =tanh(W c ·[h t-1 ,x t ]+b c ) represents. Where W c The weight matrix that can represent the candidate state, b c It can represent the bias vector, and tanh is the hyperbolic tangent activation function. Then, according to the output of the forget gate and the input gate, the current cell state c is updated. t , the formula can be used: c t =f t ·c t-1 +i t ·m t Represents, where · represents element-by-element multiplication. Finally, the output gate o is calculated t , the formula can be used: t =σ(W o ·[h t-1 ,x t ]+b o ) indicates that W o It can represent the weight matrix of the output gate, b o It can represent the bias vector, combined with the output o of the output gate tand the current cell state c t , calculate the hidden state h of the current time step t , the formula can be used: h t =o t ·tanh(c t ) is represented by the regularization (Dropout) layer. The regularization (Dropout) layer is used to temporarily discard some neural network units from the network according to a certain probability, thereby finding a thinner network from the original network. The fully connected layer is used to integrate and transform the feature information of the previous layer and map it to the sample label space, thereby achieving the classification or regression task of the input data. For example, the hidden state h t The input is sent to the fully connected layer, which learns the complex relationship between features to achieve the mapping between input features and final prediction results. The output layer is used to output the prediction results.

[0072] S280. Perform data prediction on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain a load prediction result corresponding to the target air-conditioning system.

[0073] S290: Make decision adjustments to the load forecast result based on preset expert decision information to generate a start / stop judgment result corresponding to the target air-conditioning system.

[0074] Among them, the preset expert decision information may refer to the pre-set knowledge content used to assist the output results of the target deep learning model. By presetting the expert decision information, it can be ensured that the prediction results are not only based on data, but also combined with the experience and rules in actual operation. Exemplarily, the preset expert decision information may include the operating rules of the air-conditioning system, energy-saving strategies and the impact of the external environment on the indoor load. Among them, the impact of the external environment on the indoor load may include the load demand pattern of the air-conditioning in certain specific time periods, the operating mode that reduces energy consumption under the premise of process temperature and humidity requirements, and the impact characteristics of weather changes, holidays or personnel flow on the air-conditioning load. Usually, the preset expert decision information can exist in the form of a rule base, a decision tree, fuzzy logic or other forms.

[0075] The start / stop judgment result may refer to the air conditioner start / stop decision made based on the load forecast result output by the target deep learning model and the preset expert decision information. For example, the start / stop judgment result may be the start time of the air conditioner determined based on the load forecast result; the stop time of the air conditioner determined based on the load forecast result to avoid continued operation when the load demand is low, resulting in energy waste; or the air conditioner operation mode adjusted based on the load forecast result, such as cooling, heating or air supply, to optimize energy use.

[0076] The technical solution of the embodiment of the present invention is to merge and process the historical parameter data set corresponding to the target air-conditioning system through the target timestamp, generate the historical multi-feature data set corresponding to the historical parameter data set, standardize the historical multi-feature data set based on the preset standardization rules, generate the historical standardized data set corresponding to the historical parameter data set, and perform feature selection on the historical standardized data set based on the preset genetic algorithm to obtain the historical data feature subset corresponding to the historical parameter data set. Furthermore, based on the preset data format conversion, the historical data feature subset is processed to generate the target data feature subset corresponding to the historical data feature subset, the target data feature subset is segmented based on the preset sliding window rule, and the target data feature subsequence corresponding to the target data feature subset is generated, and the target data feature subsequence is input into the preset deep learning model, and the preset deep learning model is trained to obtain the trained target deep learning model. Finally, the real-time parameter data corresponding to the target air-conditioning system is predicted based on the target deep learning model to obtain the load prediction result corresponding to the target air-conditioning system, and the load prediction result is adjusted based on the preset expert decision information to generate the start-stop judgment result corresponding to the target air-conditioning system. By using genetic algorithm for feature selection and deep learning model for time-series air-conditioning load forecasting, the problems of low prediction efficiency and accuracy of air-conditioning load forecasting are solved. Real-time data can be used for air-conditioning load forecasting, which improves the prediction efficiency and accuracy of air-conditioning load forecasting.

[0077] Figure 6 The flowchart of an optional air conditioning load prediction method based on a genetic algorithm provided by an embodiment of the present invention is shown. Specifically, first, data collection is performed on the historical parameter data set and the real-time parameter data corresponding to the target air conditioning system. After that, the historical parameter data set is merged and processed based on the target timestamp to generate a historical multi-feature data set corresponding to the historical parameter data set, and the historical multi-feature data set is standardized based on the preset standardization rules to generate a historical standardized data set corresponding to the historical parameter data set, and the historical standardized data set is feature selected based on the preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set, thereby realizing data preprocessing. Further, the preset deep learning model is trained using the historical data feature subset to obtain a trained target deep learning model, and the real-time parameter data corresponding to the target air conditioning system is predicted based on the target deep learning model to obtain the load prediction result corresponding to the target air conditioning system, thereby realizing model training and data prediction. Finally, the load prediction result is adjusted based on the preset expert decision information to generate the start-stop judgment result corresponding to the target air conditioning system, thereby realizing expert decision-making. Thus, by adopting the preset genetic algorithm for feature selection, the problem of redundant air conditioning load data features is solved. The time-series air conditioning load forecasting is performed through the target deep learning model, which solves the problem that time-series data forecasting is difficult and traditional machine learning methods are difficult to capture long time series.

[0078] Figure 7 The figure shows a flow chart of a decision adjustment process provided by an embodiment of the present invention. Specifically, data collection is performed on the real-time parameter data corresponding to the target air-conditioning system. Afterwards, the real-time parameter data is input into the trained target deep learning model, and data prediction is performed on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain the load prediction result corresponding to the target air-conditioning system. Finally, the load prediction result is adjusted based on the preset expert decision information, and the start-stop judgment result corresponding to the target air-conditioning system is generated to realize expert decision-making.

[0079] Embodiment 3

[0080] Figure 8 This is a schematic diagram of the structure of an air conditioning load prediction device based on a genetic algorithm provided in the third embodiment of the present invention. Figure 8 As shown, the device includes: a data acquisition module 310, a data preprocessing module 320, a model training module 330 and a load prediction module 340;

[0081] The data acquisition module 310 is used to obtain a historical parameter data set corresponding to the target air conditioning system;

[0082] A data preprocessing module 320 is used to perform feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set;

[0083] A model training module 330 is used to train a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model;

[0084] The load prediction module 340 is used to perform data prediction on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain the load prediction result corresponding to the target air-conditioning system.

[0085] The technical solution of the embodiment of the present invention performs feature selection on the historical parameter data set corresponding to the target air-conditioning system through a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set. Furthermore, a preset deep learning model is trained based on the historical data feature subset to obtain a trained target deep learning model. Finally, data prediction is performed on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain a load prediction result corresponding to the target air-conditioning system. Since a genetic algorithm is used for feature selection and a deep learning model is used for time-series air-conditioning load prediction, the problem of low prediction efficiency and accuracy of air-conditioning load prediction is solved, and real-time data can be used for air-conditioning load prediction, thereby improving the prediction efficiency and accuracy of air-conditioning load prediction.

[0086] Optionally, the data preprocessing module 320 may be specifically used for:

[0087] Merging and processing the historical parameter data set based on the target timestamp to generate a historical multi-feature data set corresponding to the historical parameter data set;

[0088] The historical multi-feature data set is standardized based on a preset standardization rule to generate a historical standardized data set corresponding to the historical parameter data set;

[0089] Based on a preset genetic algorithm, feature selection is performed on the historical standardized data set to obtain a historical data feature subset corresponding to the historical parameter data set.

[0090] Optionally, the data preprocessing module 320 may be specifically used for:

[0091] Initialize and process the historical standardized data set to obtain a basic individual cluster corresponding to the historical standardized data set; wherein the basic individual cluster includes each individual, and each individual represents a feature combination;

[0092] Performing incremental processing on the basic individual cluster based on a preset cross-increment rule to generate incremental individuals corresponding to the historical standardized data set;

[0093] Merging and processing the basic individual cluster and the incremental individual to generate a target individual cluster corresponding to the historical parameter data set;

[0094] The target individual cluster is screened and processed based on preset optimization conditions to generate a historical data feature subset corresponding to the historical parameter data set.

[0095] Optionally, the data preprocessing module 320 may be specifically used for:

[0096] Perform fitness evaluation on the target individuals in the basic individual cluster based on a preset fitness function and generate evaluation results;

[0097] Determining, in the historical standardized data set, a parent set corresponding to the historical standardized data set based on the evaluation result;

[0098] A crossover mutation operation is performed on the parent set based on a preset crossover mutation strategy to obtain incremental individuals corresponding to the historical standardized data set.

[0099] Optionally, the model training module 330 may be specifically used for:

[0100] The historical data feature subset is converted and processed based on a preset data format to generate a target data feature subset corresponding to the historical data feature subset;

[0101] Performing data segmentation on the target data feature subset based on a preset sliding window rule to generate a target data feature subsequence corresponding to the target data feature subset;

[0102] The target data feature subsequence is input into a preset deep learning model, and the preset deep learning model is trained to obtain a trained target deep learning model.

[0103] Optionally, the target deep learning model includes: input layer, long short-term memory layer, regularization layer, fully connected layer, and output layer.

[0104] Optionally, the air-conditioning load prediction device based on genetic algorithm may further include: an expert decision module, which is used to perform data prediction on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain the load prediction result corresponding to the target air-conditioning system, and then make decision adjustments to the load prediction result based on preset expert decision information to generate a start-stop judgment result corresponding to the target air-conditioning system.

[0105] The air conditioning load prediction device based on genetic algorithm provided in the embodiment of the present invention can execute the air conditioning load prediction method based on genetic algorithm provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0106] Embodiment 4

[0107] Fig. 9 A schematic diagram of an electronic device 410 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0108] like Fig. 9As shown, the electronic device 410 includes at least one processor 420, and a memory connected to the at least one processor 420 in communication, such as a read-only memory (ROM) 430, a random access memory (RAM) 440, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 420 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 430 or the computer program loaded from the storage unit 490 to the random access memory (RAM) 440. In the RAM 440, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 420, the ROM 430, and the RAM 440 are connected to each other via a bus 450. An input / output (I / O) interface 460 is also connected to the bus 450.

[0109] Multiple components in the electronic device 410 are connected to the I / O interface 460, including: an input unit 470, such as a keyboard, a mouse, etc.; an output unit 480, such as various types of displays, speakers, etc.; a storage unit 490, such as a disk, an optical disk, etc.; and a communication unit 4100, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 4100 allows the electronic device 410 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0110] The processor 420 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 420 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 420 executes the various methods and processes described above, such as the air conditioning load prediction method based on the genetic algorithm.

[0111] The method includes:

[0112] Obtain a historical parameter data set corresponding to the target air-conditioning system;

[0113] Performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set;

[0114] Training a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model;

[0115] Based on the target deep learning model, data prediction is performed on the real-time parameter data corresponding to the target air-conditioning system to obtain a load prediction result corresponding to the target air-conditioning system.

[0116] In some embodiments, the air conditioning load prediction method based on the genetic algorithm can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 490. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 410 via the ROM 430 and / or the communication unit 4100. When the computer program is loaded into the RAM 440 and executed by the processor 420, one or more steps of the air conditioning load prediction method based on the genetic algorithm described above can be performed. Alternatively, in other embodiments, the processor 420 can be configured to execute the air conditioning load prediction method based on the genetic algorithm by any other appropriate means (e.g., by means of firmware).

[0117] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0118] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0119] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0120] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0121] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0122] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0123] The embodiment of the present application also discloses a computer program product, which includes a computer program, and when the computer program is executed by a processor, the air conditioning load prediction method based on a genetic algorithm provided in any embodiment of the present application is implemented. The program product and the air conditioning load prediction method based on a genetic algorithm disclosed in each embodiment of the present application belong to the same inventive concept, so it will not be described here.

[0124] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0125] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. An air conditioning load prediction method based on genetic algorithm, characterized in that: include: Obtain a historical parameter data set corresponding to the target air-conditioning system; Performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set; Training a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model; Based on the target deep learning model, data prediction is performed on the real-time parameter data corresponding to the target air-conditioning system to obtain a load prediction result corresponding to the target air-conditioning system.

2. The method according to claim 1, characterized in that The performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set includes: Merging and processing the historical parameter data set based on the target timestamp to generate a historical multi-feature data set corresponding to the historical parameter data set; The historical multi-feature data set is standardized based on a preset standardization rule to generate a historical standardized data set corresponding to the historical parameter data set; Based on a preset genetic algorithm, feature selection is performed on the historical standardized data set to obtain a historical data feature subset corresponding to the historical parameter data set.

3. The method according to claim 2, characterized in that The performing feature selection on the historical standardized data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set includes: Initialize and process the historical standardized data set to obtain a basic individual cluster corresponding to the historical standardized data set; wherein the basic individual cluster includes each individual, and each individual represents a feature combination; Performing incremental processing on the basic individual cluster based on a preset cross-increment rule to generate incremental individuals corresponding to the historical standardized data set; Merging and processing the basic individual cluster and the incremental individual to generate a target individual cluster corresponding to the historical parameter data set; The target individual cluster is screened and processed based on preset optimization conditions to generate a historical data feature subset corresponding to the historical parameter data set.

4. The method according to claim 3, characterized in that: The step of performing incremental processing on the basic individual cluster based on a preset cross-increment rule to generate incremental individuals corresponding to the historical standardized data set includes: Perform fitness evaluation on the target individuals in the basic individual cluster based on a preset fitness function and generate evaluation results; Determining, in the historical standardized data set, a parent set corresponding to the historical standardized data set based on the evaluation result; A crossover mutation operation is performed on the parent set based on a preset crossover mutation strategy to obtain incremental individuals corresponding to the historical standardized data set.

5. The method according to claim 1, characterized in that The step of training a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model includes: The historical data feature subset is converted and processed based on a preset data format to generate a target data feature subset corresponding to the historical data feature subset; Performing data segmentation on the target data feature subset based on a preset sliding window rule to generate a target data feature subsequence corresponding to the target data feature subset; The target data feature subsequence is input into a preset deep learning model, and the preset deep learning model is trained to obtain a trained target deep learning model.

6. The method according to claim 1, characterized in that The target deep learning model includes: an input layer, a long short-term memory layer, a regularization layer, a fully connected layer and an output layer.

7. The method according to claim 1, characterized in that After performing data prediction on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain the load prediction result corresponding to the target air-conditioning system, the method further includes: The load forecast result is adjusted based on the preset expert decision information to generate a start / stop judgment result corresponding to the target air-conditioning system.

8. An air conditioning load prediction device based on genetic algorithm, characterized in that: include: A data acquisition module is used to obtain a historical parameter data set corresponding to the target air-conditioning system; A data preprocessing module, used for performing feature selection on the historical parameter data set based on a preset genetic algorithm to obtain a historical data feature subset corresponding to the historical parameter data set; A model training module, used to train a preset deep learning model based on the historical data feature subset to obtain a trained target deep learning model; The load prediction module is used to perform data prediction on the real-time parameter data corresponding to the target air-conditioning system based on the target deep learning model to obtain the load prediction result corresponding to the target air-conditioning system.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the air conditioning load prediction method based on a genetic algorithm according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the air conditioning load prediction method based on a genetic algorithm according to any one of claims 1 to 7 when executed.

Images