Multi-model fusion resident electricity consumption prediction method

Abstract

The invention provides a multi-model fusion resident electricity consumption prediction method, which comprises a data acquisition step, a first model construction step, an input data judgment step, a first prediction step, a second prediction step and an electricity consumption calibration step. The power consumption of the electric appliance can be predicted through the fuzzy prediction model, and when a user only inputs necessary option data and a specific time period, the power consumption of the electric appliance can be predicted through the fine prediction model, so that the data required to be manually input by the user is greatly reduced, the automation is improved, and the user experience is improved. Different models can accurately predict the electricity consumption for different input data, and in addition, for the electric appliance which is used for more than 15 minutes each time and has constant power, the electricity consumption of the electric appliance can be calibrated by acquiring 24-hour 96-point load data of the electric appliance. The method has the advantages that the automation level of residential electricity consumption prediction and the accuracy of energy consumption estimation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electricity utilization, in particular to a multi-model fusion resident electricity consumption prediction method. BACKGROUND

[0002] At present, in view of the situation that some residents have doubts about the household electricity consumption, the power company or the related platform usually provides an electricity consumption estimation function based on user's active input. This function aims to provide a convenient and transparent electricity query means for users, to help users reasonably judge whether the electricity consumption is abnormal, and to assist in solving the electricity fee dispute. Specifically, the user can manually input the opening time and closing time of a certain electrical appliance in the application, select the electrical appliance name (such as air conditioner, refrigerator, washing machine, etc.), and the system estimates the electricity consumption of the electrical appliance in the time period in combination with the preset algorithm or the database standard power data.

[0003] Although the existing method improves the user's perception of electricity data and has good user interaction, it still has limitations: the recognition of electrical appliances depends on user's manual input, the automatic level is low; the power estimation accuracy is limited, and it cannot fully reflect the energy consumption difference brought by different brands, models and use environment; it cannot flexibly adapt to complex or diversified electricity consumption scenarios. Therefore, how to further improve the estimation accuracy, the automatic level and adapt to more electricity consumption scenarios is the technical problem and development direction to be solved at present. SUMMARY

[0004] The present application provides a multi-model fusion resident electricity consumption prediction method to solve the problem of low automatic level and low accuracy of energy consumption estimation of the existing resident electricity consumption prediction method.

[0005] The present application provides a multi-model fusion resident electricity consumption prediction method, which specifically includes a data acquisition step, a first model construction step, an input data discrimination step, a first prediction step, a second prediction step and an electricity consumption calibration step.

[0006] The data acquisition step is used to acquire the mandatory data and the optional data of the electrical appliances, and the 96-point load data of the electrical appliances and the environmental factors are acquired through the power grid for 24 hours; the first model construction step is used to construct a first model for fine prediction of the power consumption and a second model for fuzzy prediction of the power consumption; the input data discrimination step is used to discriminate the category of the input data, when the optional data includes the operation time period of the electrical appliances, and the mandatory data and the optional data are used as the input data, a first prediction step is performed, and the input data is defined as first input data, when only the mandatory data is used as the input data, a second prediction step is performed, and the input data is defined as second input data; the first prediction step is to input the first input data into the first model, and the power consumption of the residents is predicted through the first model; the second prediction step is to input the second input data into the second model, and the power consumption of the residents is predicted through the second model; the power consumption calibration step is to preset an electrical appliance calibration condition, when there is an electrical appliance meeting the electrical appliance calibration condition, the power consumption of the electrical appliance is calibrated based on the obtained 96-point load data, and the calibrated power consumption is output.

[0007] Further, the first prediction step includes a first fine calculation step, when the electrical appliance is an electrical appliance in linear relationship with power, the predicted power consumption of the electrical appliance is defined as a first power consumption, and the calculation formula of the first power consumption is

[0008] wherein E represents the first power consumption, K is a power consumption coefficient of each electrical appliance determined by experiment, and P is the rated power of the electrical appliance.

[0009] Further, the first prediction step includes a second fine calculation step, when the electrical appliance is a refrigerator type electrical appliance, the predicted power consumption of the electrical appliance is defined as a second power consumption, and the calculation formula of the second power consumption is

[0010] wherein E represents the second power consumption, F represents the monthly power factor, P represents the user input power or the reference power, and T represents the use time. fridge month base

[0011] Further, the first prediction step includes a second model construction step, a model training step and a third fine calculation step.

[0012] ​​​The second model construction step is used to construct a random forest regression model; the model training step is to train the mapping relationship between air conditioner power consumption and environmental factors with historical sample data, and the random forest regression model is trained with training set data; the third fine calculation step is to define the predicted power consumption of the electric appliance as a third power consumption when the electric appliance is a variable frequency air conditioner, and the calculation formula of the third power consumption is

[0013]

[0014] Wherein, train_RF represents the random forest regression model training stage, represents the training set data, and represents the feature vector, represents the actual power consumption of the corresponding historical record under the feature vector, represents the output feature vector, represents the trained random forest regression model, and w represents the dimension of the feature vector, represents the input feature vector, represents the third power consumption.

[0015] Further, the first prediction step includes a fourth fine calculation step, when the electric appliance is a mode type electric appliance, the predicted power consumption of the electric appliance is defined as a fourth power consumption, and the calculation formula of the fourth power consumption is

[0016] Wherein, represents the fourth power consumption, E modle represents the power consumption of the mode type electric appliance in the corresponding mode.

[0017] Further, the second prediction step includes a first fuzzy calculation step, when the electric appliance is a short-time electric appliance, the predicted power consumption of the electric appliance is defined as a fifth power consumption, and the calculation formula of the fifth power consumption is

[0018] Wherein, E single represents the fifth power consumption, C single represents the average power consumption of the electric appliance for single use.

[0019] Further, the second prediction step includes a second fuzzy calculation step, when the electric appliance is a daily average electric appliance, the predicted power consumption of the electric appliance is defined as a sixth power consumption, and the calculation formula of the sixth power consumption is

[0020] Wherein, E daily represents the sixth power consumption, C monthE represents the average power consumption of the electrical appliance in the current month.

[0021] Further, the second prediction step comprises a third fuzzy calculation step, when the electrical appliance is a mode type electrical appliance, defining the predicted power consumption of the electrical appliance as a fourth power consumption, and the calculation formula of the fourth power consumption is

[0022] wherein, E represents the fourth power consumption, E modle represents the power consumption of the mode type electrical appliance in the corresponding mode.

[0023] Further, the second prediction step comprises a fourth fuzzy calculation step, when the electrical appliance is a time period dependent electrical appliance, defining the predicted power consumption of the electrical appliance as a first power consumption, and the calculation formula of the first power consumption is

[0024] wherein, E represents the first power consumption, K is a power consumption coefficient of each electrical appliance determined by experiment, and P is the rated power of the electrical appliance.

[0025] Further, the power consumption calibration step specifically comprises a normalization step, a data sequence screening step, a power value division step, a power value prediction step and an energy consumption prediction step.

[0026] The normalization step is to normalize the power data, and the formula is

[0027] wherein, represents the normalized power data, represents the minimum power in the 96-point load power data collected by the smart meter within 24 hours, represents the load power data.

[0028] The data sequence screening step is to screen the data sequence in the specified time period based on the obtained normalized power data and power consumption data, and the formula is

[0029] wherein, represents the data sequence, represents any one of the 96-point load power data, represents the power consumption data, and represents the time period specified by the user.

[0030] The power value division step is to divide the power value by a preset power interval width threshold, and the formula is

[0031] wherein, represents a power interval, represents a preset power interval width threshold.

[0032] The power value prediction step is based on obtaining a power interval, selecting the mean value of all power values in the interval with the highest frequency and power consistent with the power of the electrical appliance as the predicted power value of the electrical appliance, and the formula is

[0033] wherein, represents the predicted power value of the electrical appliance, and Nr represents the number of power values of the electrical appliance consistent with the selected power interval, represents the selected power interval.

[0034] The energy consumption prediction step is based on the predicted power value of the electrical appliance to predict the energy consumption of the electrical appliance, and the formula is

[0035] wherein, represents the predicted energy consumption of the electrical appliance, i.e. the calibrated electricity consumption, represents the use duration of the electrical appliance input by the user.

[0036] The present application provides a multi-model fusion resident electricity consumption prediction method, which realizes accurate electricity consumption prediction of different input data of users by constructing a fine prediction model and a fuzzy prediction model. When the user only inputs the mandatory item data, the mandatory item data only includes the model and use mode of the electrical appliance, and the electricity consumption of the electrical appliance can be predicted by the fuzzy prediction model. When the user only inputs the mandatory item data and the specific time period, the specific time period is the optional item data, and the electricity consumption of the electrical appliance can be predicted by the fine prediction model. The data that the user needs to manually input is greatly reduced, the automation is improved, the electricity consumption of different input data can be accurately predicted by different models, and in addition, for the electrical appliance with a use duration of more than 15 minutes and constant power each time, the electricity consumption of the electrical appliance is calibrated by collecting 24-hour 96-point load data of the electrical appliance, further improving the accuracy of the electricity consumption estimation of the electrical appliance. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Figure 1 is a flow chart of the multi-model fusion resident electricity consumption prediction method described in the embodiment; Figure 2 is a flow chart of the electricity consumption prediction of the variable frequency air conditioner appliance described in the embodiment; Figure 3 is a flow chart of the electricity consumption calibration step described in the embodiment. DETAILED DESCRIPTION

[0039] The technical solutions in the embodiments of the present application will be clearly and completely described in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0040] As shown in Figure 1 The present application provides a multi-model fusion resident electricity consumption prediction method, which specifically comprises the following steps: S1) a data acquisition step, S2) a first model construction step, S3) an input data discrimination step, S4) a first prediction step, S5) a second prediction step, and S6) an electricity consumption calibration step.

[0041] Step S1) the data acquisition step acquires the mandatory item data and the optional item data of the appliance, and acquires the 96-point load data and environmental factors of the appliance for 24 hours through the power grid.

[0042] In the embodiment, the mandatory item data of the appliance includes the model and the use mode of the appliance, and the optional item data of the appliance includes the appliance name, the operation time period, the rated power, etc., the 96-point load data includes 96-point power data and electricity data, and the environmental factors include the external temperature, the humidity, etc. when the appliance is used.

[0043] Step S2) the first model construction step constructs a first model for fine prediction of electricity consumption and a second model for fuzzy prediction of electricity consumption.

[0044] In the embodiment, the first model is a fine prediction model, and the second model is a fuzzy prediction model.

[0045] Step S3) the input data discrimination step discriminates the category of the input data. When the operation time period of the appliance is included in the optional item data, and the mandatory item data and the optional item data are used as the input data, the first prediction step is executed, and the input data is defined as the first input data. When only the mandatory item data is used as the input data, the second prediction step is executed, and the input data is defined as the second input data.

[0046] In this embodiment, when a user simultaneously inputs both required and optional data, and the optional data includes the appliance's operating time period, the power consumption of the appliance can be predicted using the refined prediction model. Specifically, when inputting optional data, the appliance's operating time period must be entered, while other optional data can be entered according to the user's preference. When the user only inputs required data, the power consumption of the appliance can be predicted using the fuzzy prediction model.

[0047] Step S4) First prediction step: Input the first input data into the first model and predict residential electricity consumption through the first model.

[0048] Further, step S4) the first prediction step includes step S41) the first fine calculation step, whereby, when the appliance is linearly related to its power, the predicted power consumption of the appliance is defined as the first power consumption, and the calculation formula for the first power consumption is:

[0049] Where E represents the first power consumption, K is the power consumption coefficient for each appliance determined by experiments, and P is the rated power of the appliance.

[0050] In this embodiment, K is a coefficient representing the relationship between the electricity consumption, duration, and power of various electrical appliances, obtained from statistics in the laboratory's database.

[0051] Further, step S4) the first prediction step includes step S42) the second fine calculation step. When the appliance is a refrigerator, the predicted power consumption of the appliance is defined as the second power consumption, and the calculation formula for the second power consumption is as follows:

[0052] Among them, E fridge F represents the second electricity consumption. month P represents the monthly power factor. base This indicates the user input power or reference power, and T represents the usage time, calculated in 24-hour increments.

[0053] like Figure 2 As shown, step S4) the first prediction step includes step S43) the second model building step, step S44) the model training step, and step S45) the third fine calculation step.

[0054] Step S43) Second model construction step: Construct a random forest regression model.

[0055] Step S44) Model training step: Use historical sample data to train the mapping relationship between air conditioner power consumption and environmental factors, and train the random forest regression model using training set data; Step S45) Third fine calculation step: When the appliance is an inverter air conditioner, the predicted power consumption of the appliance is defined as the third power consumption, and the calculation formula for the third power consumption is as follows:

[0056]

[0057] Where train_RF represents the training phase of the random forest regression model. The training set data is represented by [symbol], and the feature vector is represented by [symbol]. This indicates the actual power consumption of the corresponding historical data under this feature vector. This represents the output feature vector. This represents a trained random forest regression model, where w represents the dimension of the feature vectors. This represents the input feature vector. This indicates the third type of electricity consumption.

[0058] In this embodiment, the power consumption of the inverter air conditioner varies with a series of factors such as ambient temperature and set temperature. Therefore, for calculating the power consumption of the inverter air conditioner, a random forest regression model is constructed, and a mapping relationship between the air conditioner's power consumption and environmental factors is trained using historical sample data. Training a random forest regressor In order to predict the third electricity consumption.

[0059] Further, step S4) the first prediction step includes step S46) the fourth fine calculation step. When the appliance is a mode-type appliance, different modes of the appliance correspond to different power consumption. The predicted power consumption of the appliance is defined as the fourth power consumption, and the calculation formula for the fourth power consumption is as follows:

[0060] in, E represents the fourth type of electricity consumption. modle This indicates the power consumption of the appliance under the corresponding mode.

[0061] Step S5) Second prediction step: Input the second input data into the second model and predict residential electricity consumption through the second model.

[0062] Further, step S5) the second prediction step includes step S51) the first fuzzy calculation step. When the electrical appliance is a short-term appliance, the single usage time of this type of appliance is short, the time is relatively concentrated, and the power consumption is relatively fixed. The predicted power consumption of this appliance is defined as the fifth power consumption, and the calculation formula for the fifth power consumption is as follows:

[0063] Among them, E single C represents the fifth electricity consumption. single This indicates the average power consumption of the appliance per use.

[0064] Further, step S5) the second prediction step includes step S52) the second fuzzy calculation step. When the daily average power consumption of the electrical appliance is relatively stable, the daily usage time and power consumption of this type of appliance vary with the season or month. The predicted power consumption of this appliance is defined as the sixth power consumption, and the calculation formula for the sixth power consumption is as follows:

[0065] Among them, E daily C represents the sixth electricity consumption. month This indicates the average power consumption of the appliance in the current month.

[0066] Further, step S5) the second prediction step includes step S53) the third fuzzy calculation step. When the appliance is a pattern-type appliance, the predicted power consumption of the appliance is defined as the fourth power consumption, and the calculation formula for the fourth power consumption is as follows:

[0067] in, E represents the fourth type of electricity consumption. modle This indicates the power consumption of the appliance under the corresponding mode.

[0068] Further, step S5) the second prediction step includes step S54) the fourth fuzzy calculation step. When the appliance is a time-dependent appliance, the duration of use for different users is not fixed, thus relying heavily on time to calculate electricity consumption. For air conditioners, inverter air conditioners only need to input the time period for electricity consumption prediction. The predicted electricity consumption of this appliance is defined as the first electricity consumption, and the calculation formula for the first electricity consumption is as follows:

[0069] Where E represents the first power consumption, K is the power consumption coefficient for each appliance determined by experiments, and P is the rated power of the appliance.

[0070] Step S6) Power consumption calibration step: By pre-setting an appliance calibration condition, when an appliance meets the appliance calibration condition, the power consumption of the appliance is calibrated based on the obtained 96 load data points, and the calibrated power consumption is output.

[0071] In this embodiment, when there is an appliance with a usage time of more than 15 minutes and a constant power, the actual energy consumption of the target appliance can be accurately estimated by normalizing the 96 load data points, statistically analyzing the frequency of power range occurrences, and combining energy ratio inference. The power consumption of the appliance is inferred based on the appliance operating time period and the 96 power and power data points obtained in the optional data obtained in step S1).

[0072] like Figure 3 As shown, step S6) power consumption calibration step specifically includes step S61) normalization step, step S62) data series filtering step, step S63) power value division step, step S64) power value prediction step and step S65) energy consumption prediction step.

[0073] Step S61) Normalization step: Normalize the power data using the following formula:

[0074] in, This represents the normalized power data. This indicates the minimum power among the 96 load power data points of the appliance collected by the smart meter over 24 hours. This represents the load power data.

[0075] Step S62) Data Series Filtering Step: Based on the obtained normalized power data and electricity data, filter out the data series within a specified time period. The formula is as follows:

[0076] in, Represents a data sequence. This represents any one of the 96 load power data points. This indicates battery power data. [Indicates the operating time of the appliance.]

[0077] Step S63) Power value division step: By presetting a power interval width threshold, the power value is divided into intervals, and the formula is as follows:

[0078] in, Indicates the power range. This indicates the preset power range width threshold.

[0079] Step S64) Power value prediction step: Based on the obtained power range, the average of all power values ​​in the range with the highest frequency and power matching the power level of the appliance is selected as the predicted power value of the appliance. The formula is as follows:

[0080] in, Nr represents the predicted power value of the electrical appliance, and Nr represents the number of power values ​​of the electrical appliance that fall within the selected power range. This indicates the selected power range.

[0081] Step S65) and the energy consumption prediction step: Based on the predicted power value of the appliance, predict the energy consumption of the appliance, using the following formula:

[0082] in, This represents the predicted energy consumption of the appliance, i.e., the calibrated power consumption. This indicates the duration of use of the appliance as input by the user.

[0083] In this embodiment, regardless of whether the power consumption of an appliance is predicted by the first model or the second model, if the appliance's usage meets the appliance calibration conditions, step S6) power consumption calibration will be performed to re-estimate the appliance's power consumption, and the power consumption will be calculated. As the final output of the power consumption of the appliance, the present invention adopts a more suitable method to predict the power consumption of different appliances and their respective usage conditions, thereby improving the accuracy of power consumption prediction.

[0084] This application provides a multi-model fusion method for predicting residential electricity consumption. By constructing a refined prediction model and a fuzzy prediction model, it achieves accurate electricity consumption prediction for different user input data. When the user only inputs required data, including only the appliance model and usage mode, the fuzzy prediction model can predict the appliance's electricity consumption. When the user only inputs required data and a specific time period, where the specific time period is optional, the refined prediction model can predict the appliance's electricity consumption. This greatly reduces the amount of data that users need to manually input, improving automation. Different models can accurately predict electricity consumption for different input data. In addition, for appliances with a usage time of more than 15 minutes each time and a constant power, this method will also calibrate the appliance's electricity consumption by collecting 24-hour 96-point load data, further improving the accuracy of the appliance's electricity consumption estimation.

[0085] The above provides a detailed description of the multi-model fusion method for predicting residential electricity consumption. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1.A method for predicting resident electricity consumption by multi-model fusion, characterized in that, Specifically comprising the following steps comprising: The data acquisition step, acquiring the mandatory data and optional data of the electrical appliance, and acquiring the 96-point load data of the electrical appliance and environmental factors through the power grid for 24 hours; The first model construction step, constructing a first model for fine prediction of power consumption and a second model for fuzzy prediction of power consumption; The input data discrimination step, judging the category of the input data, when the electrical appliance operating time period is included in the optional data, and the mandatory data and the optional data are used as input data, the first prediction step is executed, and the input data is defined as first input data, when only the mandatory data is used as input data, the second prediction step is executed, and the input data is defined as second input data; The first prediction step, inputting the first input data into the first model, and predicting the residential power consumption through the first model; The second prediction step, inputting the second input data into the second model, and predicting the residential power consumption through the second model; And The power consumption calibration step, presetting an electrical appliance calibration condition, when there is an electrical appliance meeting the electrical appliance calibration condition, calibrating the power consumption of the electrical appliance based on the obtained 96-point load data, and outputting the calibrated power consumption. 2.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The first prediction step comprises: The first fine calculation step, when the electrical appliance is an electrical appliance with linear relationship with power, defining the predicted power consumption of the electrical appliance as the first power consumption, and the calculation formula of the first power consumption is Wherein, E represents the first power consumption, K is the power consumption coefficient of each electrical appliance determined by experiment, and P is the rated power of the electrical appliance. 3.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The first prediction step comprises: The second fine calculation step, when the electrical appliance is a refrigerator type electrical appliance, defining the predicted power consumption of the electrical appliance as the second power consumption, and the calculation formula of the second power consumption is wherein E fridge represents the second power consumption, F month represents the monthly power factor, P base represents the user input power or reference power, and T represents the use time. 4.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The first prediction step comprises: The second model construction step, constructing a random forest regression model; The model training step, training the mapping relationship between air conditioner power consumption and environmental factors with historical sample data, and training the random forest regression model with training set data; The third fine calculation step, when the electrical appliance is a variable frequency air conditioner, defining the predicted power consumption of the electrical appliance as the third power consumption, and the calculation formula of the third power consumption is wherein train_RF represents the random forest regression model training stage, represents the training set data, represents the feature vector, represents the actual power consumption of the corresponding historical record under the feature vector, represents the output feature vector, represents the trained random forest regression model, w represents the dimension of the feature vector, represents the input feature vector, represents the third power consumption. 5.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The first prediction step comprises: The fourth fine calculation step, when the electrical appliance is a mode type electrical appliance, defining the predicted power consumption of the electrical appliance as the fourth power consumption, and the calculation formula of the fourth power consumption is wherein, represents the fourth power consumption, E modle represents the power consumption of the mode class appliance in the corresponding mode. 6.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The second prediction step comprises: The first fuzzy calculation step, when the electrical appliance is a short-time electrical appliance, defining the predicted power consumption of the electrical appliance as the fifth power consumption, and the calculation formula of the fifth power consumption is wherein E single represents the fifth power consumption, C single represents the average power consumption of the electric appliance for a single use. 7.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The second prediction step comprises: The second fuzzy calculation step, when the electrical appliance is a daily average electrical appliance, defining the predicted power consumption of the electrical appliance as the sixth power consumption, and the calculation formula of the sixth power consumption is Wherein, E daily represents the sixth power consumption, C month represents the average power consumption of the current month of the electrical appliance. 8.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The second prediction step comprises: The third fuzzy calculation step, when the electrical appliance is a mode type electrical appliance, defining the predicted power consumption of the electrical appliance as the fourth power consumption, and the calculation formula of the fourth power consumption is wherein, represents the fourth power consumption, E modle represents the power consumption of the mode class appliance in the corresponding mode. 9.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The second prediction step comprises: A fourth fuzzy calculation step, when the electrical appliance is a time period dependent electrical appliance, defining the predicted power consumption of the electrical appliance as a first power consumption, and the calculation formula of the first power consumption is Wherein, E represents the first power consumption, K is a power consumption coefficient of each electrical appliance determined by experiment, and P is the rated power of the electrical appliance. 10.The multi-model fused resident electricity consumption prediction method of claim 1, wherein, The power consumption calibration step specifically includes the following steps: A normalization step, performing normalization processing on the power data, and the formula is wherein, represents the normalized power data, represents the minimum power of the 96-point load power data collected by the smart meter for 24 hours, represents the load power data; A data sequence screening step, screening out data sequences in a specified time period based on the obtained normalized power data and power consumption data, and the formula is wherein, represents a data sequence, represents any one of 96-point load power data, represents electric quantity data, represents a time period designated by a user; A power value division step, presetting a power interval width threshold, and dividing the power value according to intervals, and the formula is wherein, represents a power interval, represents a preset power interval width threshold; A power value prediction step, based on the obtained power interval, selecting the mean value of all power values in the interval with the highest frequency and power consistent with the power size of the electrical appliance as the predicted power value of the electrical appliance, and the formula is wherein, represents the predicted power value of the appliance, Nr represents the number of the appliance power values that fit into the selected power interval, represents the selected power interval; and An energy consumption prediction step, predicting the energy consumption of the electrical appliance based on the predicted power value of the electrical appliance, and the formula is An energy consumption prediction step, predicting the energy consumption of the electrical appliance based on the predicted power value of the electrical appliance, and the formula is wherein, represents the predicted energy consumption of the electrical appliance, i.e. the calibrated power consumption, represents the user inputted usage duration of the electrical appliance.

Images