Smart city power consumption data statistical planning analysis method and system

Abstract

The invention discloses a smart city power consumption data statistical planning analysis method and system, and belongs to the technical field of energy systems. According to the method, electricity consumption data of a smart city in set time are collected, electricity consumption data of one smart city in set time are collected, the collection period is 24 hours, data are loaded through functions in the pandas, the number and accuracy of the data can meet the requirements, and then training of subsequent data on a model is facilitated; by analyzing the rule of data changing along with time and analyzing time sequence data distribution, a corresponding mathematical model is established to analyze and obtain a future electricity consumption prediction result of the smart city, statistics and operation of future electricity consumption in operation of the smart city are achieved, and allocation of power generation resources such as coal based on budget of the future electricity consumption is facilitated.

Description

Technical Field

[0001] The present invention belongs to the field of energy systems, and particularly relates to a method and system for statistical planning analysis of electricity consumption data in a smart city. Background Art

[0002] A smart city refers to the application of intelligent computing technologies such as the Internet of Things, cloud computing, big data, and spatial geographic information integration in the fields of urban planning, design, construction, management, and operation, making the key infrastructure components and services of the city, such as urban management, education, medical care, real estate, transportation, public utilities, and public safety, more interconnected, efficient, and intelligent.

[0003] Smart cities often intersect with regional development concepts such as digital cities, perception cities, wireless cities, intelligent cities, ecological cities, and low-carbon cities, and even get mixed up with industry informatization concepts such as e-government, intelligent transportation, and smart grids. Interpretations of the concept of smart cities often have different focuses. Some views believe that the key lies in technology application, some views believe that the key lies in network construction, some views believe that the key lies in human participation, some views believe that the key lies in the wisdom effect, and some leading cities in urban informatization construction emphasize people orientation and sustainable innovation. In short, wisdom is not just intelligence. A smart city is by no means just another term for an intelligent city or the intelligent application of information technology, but also includes the connotations of human wisdom participation, people orientation, and sustainable development.

[0004] In the process of the operation of a smart city, electricity is an essential resource. Traditional electricity prediction methods mainly rely on linear regression or simple time series analysis of historical electricity consumption data. However, the electricity demand in a smart city presents multi-dimensional dynamic characteristics: on the one hand, the urban electricity load is affected by multiple non-linear factors such as industrial structure, population flow, meteorological conditions, and emergencies (such as major events and disaster weather); on the other hand, the popularization of smart grid technologies such as distributed energy access and demand-side response further exacerbates the volatility and uncertainty of the grid load. Once the power system is difficult to actively allocate, it will affect electricity consumption on a large scale. Summary of the Invention

[0005] The purpose of the present invention is to overcome the above-mentioned disadvantages of the prior art and provide a method and system for statistical planning analysis of electricity consumption data in a smart city to solve the problems in the prior art that the volatility and uncertainty of electricity consumption in a smart city are relatively large and the power load is difficult to allocate in a timely and effective manner.

[0006] To achieve the above purpose, the present invention adopts the following technical solutions: In the first aspect, the present invention provides a method for statistical planning analysis of electricity consumption data in a smart city, including the following steps: Collect electricity consumption data within a set time in the smart city; Load power consumption data through pandas, remove abnormal data, and obtain a power consumption dataset, where the power consumption in the power consumption dataset is related to time; Based on the power consumption dataset, obtain a power prediction model through the Gaussian Naive Bayes method or neural network; Predict the power consumption of a smart city in a day through the power prediction model; Allocate power resources based on the predicted power consumption.

[0007] A further improvement of the present invention lies in: Preferably, the unit of the set time is year; the period for collecting power consumption data is 24h.

[0008] Preferably, the specific process of loading power consumption data through pandas and removing abnormal data includes the following steps: (1) Load data using a function in pandas; (2) Obtain the dimension of the data; (3) Obtain the missing values and abnormal values of the data, and remove the abnormal values; (4) Add new columns at the missing positions of the data and the positions where abnormal values are removed to complete the data loading process.

[0009] Preferably, the power consumption in the power consumption dataset has a period of 24h.

[0010] Preferably, the process of obtaining a power prediction model through the Gaussian Naive Bayes method is as follows: (1) Import the dataset into the Gaussian Naive Bayes classifier; (2) Train the Gaussian Naive Bayes classifier with the data in the dataset to obtain a power prediction model.

[0011] Preferably, the output of the power prediction model of the Gaussian Naive Bayes classifier is the power value with the highest probability within the set time.

[0012] Preferably, the process of obtaining a power prediction model through the neural network is as follows: (1) Input data related to power to obtain the mapping relationship between features and power; (2) Use the features as the input of the neural network and the power as the output of the neural network; (3) Train the neural network to obtain a trained neural network model.

[0013] Preferably, the training process of the neural network model is as follows: (1) Construct a neural network model; (2) Compile the neural network model; (3) Train the neural network using the training set and debug the hyperparameters to obtain the trained neural network.

[0014] Preferably, the compilation of the neural network includes setting the learning objective and optimization algorithm.

[0015] In a second aspect, the present invention provides a smart city electricity consumption data statistics, planning and analysis system, comprising: A data acquisition module for acquiring electricity consumption data within a set time of the smart city; A data processing module for loading the electricity consumption data through pandas, removing abnormal data, and obtaining an electricity consumption data set, wherein the electricity consumption in the electricity consumption data set is related to time; A model construction module for obtaining an electricity consumption prediction model based on the electricity consumption data set through the Gaussian Naive Bayes method or a neural network; An electricity consumption prediction module for predicting the electricity consumption of the smart city for one day through the electricity consumption prediction model; A resource allocation module for allocating power resources based on the predicted electricity consumption.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention discloses a method for smart city electricity consumption data statistics, planning and analysis. This method collects electricity consumption data within a set time of the smart city. The electricity consumption data of a smart city within a set time is collected, and the collection period is 24 hours. By using the functions in pandas to load the data, the quantity and accuracy of the data can meet the requirements, which is conducive to the subsequent training of the model by the data. By analyzing the law of data change over time and analyzing the time series data distribution, a corresponding mathematical model is established to analyze and obtain the predicted result of the future electricity consumption of the smart city, realizing the statistics and calculation of the future electricity consumption during the operation of the smart city, facilitating the allocation of power generation resources such as coal based on the budget of the future electricity consumption, ensuring the normal operation of the smart city, and avoiding the situation of excessive or too low power generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of the smart city electricity consumption data statistics, planning and analysis system of the present invention; Figure 2 It is a flowchart of data loading and processing of the present invention; Figure 3 It is a flowchart of the Gaussian Naive Bayes method of the present invention; Figure 4 It is a flowchart of the neural network method of the present invention; Figure 5 It is a system diagram of the smart city electricity consumption data statistics, planning and analysis system of the present invention. Detailed implementation manners

[0018] Hereinafter, the terms "first", "second", "third", and "fourth" are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", "third", and "fourth" may explicitly or implicitly include one or more of such features.

[0019] The co - shooting method provided by the embodiments of this application can be applied to terminal devices such as mobile phones, tablet computers, wearable devices, in - vehicle devices, augmented reality (AR) / virtual reality (VR) devices, laptop computers, ultra - mobile personal computers (UMPCs), netbooks, and personal digital assistants (PDAs). The embodiments of this application do not impose any restrictions on the specific types of terminal devices.

[0020] It should be noted that the terms "first", "second", etc. in the description and drawings of the present invention are used to distinguish similar objects and do not necessarily have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non - exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0021] Please refer to Figures 1 to 5 , the present invention provides a technical solution: a smart city electricity consumption data statistics, planning and analysis system. The smart city electricity consumption data statistics, planning and analysis system includes the following steps: S1. Collect data, and collect the electricity consumption data of the smart city within a set time. S2. Load the electricity consumption data through pandas, remove abnormal data, analyze the law of data change over time and analyze the time - series data distribution to obtain an electricity consumption data set, where the electricity consumption in the electricity consumption data set is related to time. S3. Based on the electricity consumption data set, obtain an electricity consumption prediction model through the Gaussian Naive Bayes method or a neural network. S4. Predict the electricity consumption of the smart city for one day through the electricity consumption prediction model. S5, allocate power resources according to the predicted electricity consumption.

[0022] In this embodiment, in S1, collect the electricity consumption data of a smart city in the past five years, and the collection period is 24 hours; in this step, the collection time is in years, and 24 hours is used as the sampling period, so as to obtain a sufficient amount of data. Through a large amount of data, it is prepared for improving the training accuracy of the later model.

[0023] See Figure 2 , in this embodiment, the steps of data loading and processing in S2 include: S21, use the functions in pandas to load data, implement data cleaning and sorting to ensure that the data can be correctly read into the memory.

[0024] S22, view the shape of the data to quickly verify whether the dimension of the data meets the expectation during the data processing. In this step, the first ten rows of the data need to be viewed; S23, find missing values and outlier values with default markers in the data, and process the outlier values with default markers; S24, add new columns at the missing places and the positions where outliers are removed to complete the data loading process.

[0025] In this step, due to the collection period of the electricity consumption data and the large amount of collected data, introducing the data through pandas can remove outlier values, and the quantity of the data can meet the requirements of subsequent model training, so that a large amount of data can be effectively applied, improving the efficiency of subsequent model training.

[0026] In this embodiment, in S3, the method for establishing the corresponding data model specifically includes the Gaussian Naive Bayes method and the neural network model.

[0027] Naive Bayes (NB) is a classification algorithm in machine learning based on Bayes' theorem. It assumes that the input features are independent of each other and have the same impact on the classification result, so it is called Naive Bayes. Specifically, it determines the classification to which the input sample belongs by calculating the prior probability and conditional probability. The prior probability refers to the probability of each classification appearing in the entire dataset, and the conditional probability refers to the probability distribution of the input sample on each feature given a certain classification. In practical applications, Naive Bayes is often used in tasks such as text classification and spam filtering, and has the advantages of fast calculation speed and insensitivity to the amount of data. Bayes' formula is: ; p(A, B): represents the probability that event A and event B occur simultaneously; p(B): represents the probability that event B occurs, which is called the prior probability; p(A): represents the probability that event A occurs; p(A|B): represents the probability that event A occurs under the condition that event B occurs, which is called the posterior probability; p(B|A): represents the probability that event B occurs under the condition that event A occurs.

[0028] See Figure 3 , in this embodiment, in S3, the Gaussian Naive Bayes method includes: S311, import the data set into the Gaussian Naive Bayes classifier; S312, perform classification calculation using Gaussian Naive Bayes to obtain the power prediction model.

[0029] S313, model prediction. Assume that each feature follows a Gaussian distribution, and obtain the mean estimate and variance estimate. Two probabilities need to be calculated, namely: the conditional probability: P(X(i)=x(i)|Y=ck ) and the prior probability of class ck: P(Y=ck). Through analysis, it is found that the training data is numerical data. Here, it is assumed that each feature follows a Gaussian distribution. Therefore, Gaussian Naive Bayes is selected for classification calculation. Gaussian Naive Bayes assumes that each feature follows a Gaussian distribution, and a random variable X follows a data distribution with a mathematical expectation of, and a variance of is called a Gaussian distribution. For each feature, the mean is generally used to estimate u and the variance of all features is estimated .

[0030] In the above process, since the electricity consumption stored in the imported data set is related to time, after being imported into the Gaussian Naive Bayes classifier, the Gaussian Naive Bayes classifier can obtain See Figure 4 , in this embodiment, the process of obtaining the power prediction model through the neural network includes: S321, input the data related to electricity, obtain the corresponding mapping relationship by analyzing the features, obtain the features related to electricity, and then obtain the mapping relationship between the features and electricity; S322, use the features as the input of the neural network and the electricity as the output of the neural network; S323, train the neural network to obtain the trained neural network model.

[0031] In the above process, in S311 and S421, during the process of analyzing features or introducing data, it also includes the processes of data selection and data analysis. Although in the S2 process, data has been sorted through pandas, for complex models such as deep learning, too much sample data is likely to lead to poor data quality and lack of representativeness, resulting in poor model fitting effects. Therefore, in this step, the scope of the data table related to the task will also be clarified to avoid missing representative data or introducing a large amount of irrelevant data as noise. For the feature variables and labels of supervised learning, if they are related to the sequence in time, it is necessary to delimit the data time window, otherwise it may lead to common data leakage problems, that is, there is a situation where the causality between features and labels is reversed; through data analysis, the internal structure and rules of the data itself can be understood.

[0032] In this embodiment, the model training in S323 includes: (1)Construct the model structure, including the design of the neural network structure, the selection of activation functions, the initialization of model weights, layer normalization, and the setting of regularization strategies; the neural network consists of an input layer, hidden layers, and an output layer. The performance of models with different numbers of layers and neurons (computing units) also varies. Input layer: It is the input layer for data features. The dimensionality of the input data features corresponds to the number of neurons in the network. Hidden layers: They are the intermediate layers of the network (there can be many layers). Their function is to accept the output of the previous layer of the network as the current input value and calculate and output the current result to the next layer. The number of hidden layers and neurons directly affects the fitting ability of the model. Output layer: It is the network layer for the final result output. The number of neurons in the output layer represents the number of classification categories (Note: In the case of binary classification, it is a bit special. If the activation function of the output layer uses sigmoid, the number of neurons in the output layer is 1; if it uses softmax, the number of neurons in the output layer is 2, which corresponds to the number of classification categories). For the number of neurons in the model structure, the number of neurons in the input layer and output layer is usually determined. The main thing to consider is the depth and width of the hidden layers. On the premise of ignoring the problem of network degradation, generally, the more neurons in the hidden layer, the more capacity the model has to achieve a better fitting effect (and it is also more likely to overfit). To search for the appropriate network depth and width, common methods include manual empirical tuning, random / grid search, Bayesian optimization, etc. Empirically, one can refer to the structure of neural network models with good performance in similar tasks, and then make some fine-tuning in combination with the actual task. Activation functions: According to the universal approximation theorem, simply put, a neural network with "sufficiently deep network layers" and "at least one hidden layer with an activation function" can fit any function, which shows the importance of activation functions. They play a role in the non-linear transformation of the feature space. For the empirical approach to the selection of activation functions, for the output layer, the activation function of the output layer for binary classification often selects the sigmoid function, and for multi-classification, it selects softmax; for regression tasks, it is determined whether to use an activation function according to the range of the output value. For the activation functions of the hidden layers, the ReLU function is usually selected to ensure the learning efficiency. The initialization of weight parameters can accelerate the convergence speed of the model and affect the model results.Common initialization methods include normal Gaussian distribution initialization. It should be noted that the weights cannot be initialized to 0, which will cause the functions of multiple hidden neurons to be equivalent to that of 1 neuron and cannot converge. Batch normalization is a commonly used optimization method for neural network models. Its principle is very simple, that is, to standardize the original values: batch normalization eliminates the distribution differences between layers while retaining the input information, has the effect of accelerating convergence and is similar to introducing noise regularization. It can be applied to the input layer or hidden layer of the network. When used for the input layer, it is the feature standardization processing commonly used in linear models. Regularization is a method that aims to reduce the generalization error at the cost of (possibly) increasing the empirical loss and suppress overfitting to improve the generalization ability of the model. Empirically, for complex tasks, deep learning models prefer more complex models with regularization to achieve better learning effects. Common regularization strategies include: dropout, L1, L2, earlystop methods. Machine deep learning makes decisions by learning a "good" model. "Good" is the learning goal of machine / deep learning, usually that is, the error between the predicted value and the target value is as low as possible. The function for measuring this error is called the cost function or loss function. More specifically, the goal of machine / deep learning is to maximize the reduction of the loss function. For different tasks, different loss functions are often needed to measure. Classic loss functions include the mean squared error loss function for regression tasks and the cross-entropy loss function for binary classification tasks, etc.; (2) Model compilation, including setting the learning goal and optimization algorithm; through the optimization algorithm (such as gradient descent, stochastic gradient descent, Adam, etc.), the model parameters are optimized by a finite number of iterations to minimize the value of the loss function as much as possible and obtain better parameter values. For most tasks, usually, Adam and SGD can be tried directly first, and then the effects of different optimizers can be verified on specific tasks; (3) Model training and hyperparameter debugging, including dividing the dataset and adjusting and training hyperparameters.

[0033] In this embodiment, the data for hyperparameter debugging includes the validation set ratio, batch size, number of neurons in a single layer, network depth, selection of activation function type, dropout rate, selection of loss function type, regularization term penalty coefficient, selection of gradient algorithm type, and initial learning rate, etc.

[0034] In this embodiment, before model training, the HoldOut validation method (in addition to methods such as the leave-one-out method and k-fold cross-validation) is used to divide the dataset into a training set and a test set, and the training set can be further subdivided into a training set and a validation set to facilitate the evaluation of the model's performance. ① Training set: Used to run the learning algorithm to train the model. ② Development set: Used to adjust the model hyperparameters, EarlyStopping, select features, etc. to select a suitable model. ③ Test set: Only used to evaluate the performance of the selected model, but will not change the learning algorithm or parameters based on this.

[0035] In this embodiment, it also includes a model evaluation and optimization process.

[0036] See Figure 5 , the second aspect of the present invention discloses a smart city electricity consumption data statistics, planning and analysis system, including: A data collection module that collects electricity consumption data within a set time in the smart city; A data processing module that loads the electricity consumption data through pandas, removes abnormal data, and obtains an electricity consumption dataset, where the electricity consumption in the electricity consumption dataset is related to time; A model construction module that obtains an electricity consumption prediction model based on the electricity consumption dataset through the Gaussian Naive Bayes method or neural network; An electricity consumption prediction module that predicts the electricity consumption of the smart city for one day through the electricity consumption prediction model; A resource allocation module that allocates power resources based on the predicted electricity consumption.

[0037] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0038] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of the processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementation in the processFigure 1 one process or multiple processes and / or blocks Figure 1 a device for the functions specified in one block or multiple blocks.

[0039] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device that implements the functions in the process Figure 1 one process or multiple processes and / or blocks Figure 1 specified in one block or multiple blocks.

[0040] These computer program instructions may also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in the process Figure 1 one process or multiple processes and / or blocks Figure 1 specified in one block or multiple blocks.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: it is still possible to modify the specific implementation manners of the present invention or make equivalent replacements, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A method for statistical planning analysis of electricity consumption data in a smart city, characterized in that, It includes the following steps: Collect the power consumption data of the smart city within the set time; Load the power consumption data through pandas, remove the abnormal data, and obtain the power consumption dataset, where the power consumption in the power consumption dataset is related to time; Based on the power consumption dataset, obtain the power prediction model through the Gaussian Naive Bayes method or neural network; Predict the power consumption of the smart city for one day through the power prediction model; Allocate the power resources according to the predicted power consumption.

2. The statistical planning analysis method for electricity consumption data in a smart city according to claim 1, wherein, The unit of the set time is year; the period for collecting the power consumption data is 24h.

3. A method for statistical planning and analysis of electricity consumption data in a smart city according to claim 1, characterized in that, The specific process of loading the power consumption data through pandas and removing the abnormal data includes the following steps: (1) Use the function in pandas to load the data; (2) Obtain the dimension of the data; (3) Obtain the missing values and abnormal values of the data, and remove the abnormal values; (4) Add new columns at the missing places of the data and the positions where the abnormal values are removed to complete the data loading process.

4. A method for statistical planning and analysis of power consumption data in a smart city according to claim 1, characterized in that The power consumption in the power consumption dataset is in a 24h cycle.

5. A method for statistical planning analysis of electricity consumption data in a smart city according to claim 1, characterized in that, The process of obtaining the power prediction model through the Gaussian Naive Bayes method is as follows: (1) Import the dataset into the Gaussian Naive Bayes classifier; (2) Train the Gaussian Naive Bayes classifier with the data in the dataset to obtain the power prediction model.

6. The statistical planning analysis method for electricity consumption data in a smart city according to claim 1, wherein The output of the power prediction model of the Gaussian Naive Bayes classifier is the power value with the highest probability within the set time.

7. A method for statistical planning analysis of electricity consumption data in a smart city according to claim 1, characterized in that, The process of obtaining the power prediction model by the neural network is as follows: (1) Input the data related to power to obtain the mapping relationship between features and power; (2) Use the features as the input of the neural network and the power as the output of the neural network; (3) Train the neural network to obtain the trained neural network model.

8. A method for statistical planning analysis of electricity consumption data in a smart city according to claim 7, characterized in that The training process of the neural network model is as follows: (1) Construct the neural network model; (2) Compile the neural network model; (3) Train the neural network with the training set and debug the hyperparameters to obtain the trained neural network.

9. The statistical planning analysis method for electricity consumption data in a smart city according to claim 8, characterized in that The compilation of the neural network includes the setting of the learning objective and optimization algorithm.

10. A smart city power consumption data statistics, planning and analysis system, characterized in that, It includes A data collection module, which is used to collect the power consumption data of the smart city within the set time; A data processing module, which is used to load the power consumption data through pandas, remove the abnormal data, and obtain the power consumption dataset, where the power consumption in the power consumption dataset is related to time; A model construction module, which is used to obtain the power prediction model based on the power consumption dataset through the Gaussian Naive Bayes method or neural network; A power prediction module, which is used to predict the power consumption of the smart city for one day through the power prediction model; A resource allocation module, which is used to allocate the power resources according to the predicted power consumption.

Images