Power load management model building method and system

Abstract

The invention relates to a power load management model building method and system, and the method comprises the steps: collecting the equipment data of different equipment and the environment data of different environments, and storing the equipment data and the environment data in a resource pool; constructing a device model of the corresponding device according to the device data; constructing an environment model corresponding to the environment according to the environment data; obtaining power load data, obtaining characteristics of the power load data through a big data analysis technology, and determining the type of the power load data based on the characteristics; according to the type, the plurality of equipment models, the environment model and the data in the resource pool, a power load management model is constructed, the equipment models are constructed to reflect the performance and the power consumption of the equipment, the environment model is also constructed to simulate the influence of environment factors on the electricity consumption of the equipment, and the electricity consumption of the equipment can be predicted more accurately.

Description

Technical Field

[0001] The present invention relates to a method and system for building a power load management model, and belongs to the technical field of power management. Background Art

[0002] The power load model has an important impact on the analysis and calculation of the power system, and can be used to train personnel in power load knowledge, thereby improving professional ability. Therefore, the accuracy of the power load model is an important factor in the accuracy of power system simulation calculations and the reliability of power knowledge training results. The existing technology generally builds a load model based on relatively simple historical load data, and lacks consideration of the various load influencing factors. Therefore, the integrity of the built model is still insufficient.

[0003] Prior art, such as the Chinese patent application with patent number "CN113570277A", discloses a method and device for power capacity management. The problem with the above-mentioned prior art is that it mainly relies on the equipment information, statistical information, power consumption profile information and equipment basic information of the pre-deployed equipment to build a power consumption prediction model. This method may ignore the impact of environmental factors on the power consumption of the equipment. Using the LSTM algorithm to build a power consumption prediction model is an effective method, but it may lack consideration of other potential influencing factors (such as environmental factors), which may lead to limited prediction accuracy of the model in some cases. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention proposes a method and system for building a power load management model.

[0005] The technical solution of the present invention is as follows:

[0006] In one aspect, the present invention provides a method for building a power load management model, comprising the following steps:

[0007] Collecting device data of different devices and environmental data of different environments, and also including a resource pool, the resource pool is used to store device data, environmental data and preset configuration data;

[0008] Build a device model for the corresponding device based on the device data in the resource pool;

[0009] Build an environmental model of the corresponding environment based on the environmental data in the resource pool;

[0010] Acquire power load data, the power load data including load power, power consumption behavior and working mode, acquire the characteristics of the power load data through big data analysis technology, and determine the type of the power load data based on the characteristics, the characteristics including load power curve, load weather sensitivity, temperature sensitivity, trusted capacity analysis, power consumption behavior analysis, short-term load forecasting, power consumption trend analysis, price volatility and seasonal volatility;

[0011] A power load management model is constructed according to the type of the power load data, a plurality of device models and environment models, the device data in the resource pool, the environment data and the preset configuration data.

[0012] As a preferred implementation, the method for constructing the device model is:

[0013] Filter outliers and erroneous data from device data;

[0014] Normalize the filtered device data and divide the filtered device data into a training set and a validation set;

[0015] Select the neural network model as the base and use the mean squared error MSE as the loss function:

[0016]

[0017] Among them, y i Indicates the actual output power value, y′ i represents the output power value predicted by the model, and n represents the number of samples;

[0018] The device model training method is:

[0019] S1: Input the training set into the machine learning model for training, calculate the predicted output through forward propagation, and calculate the loss value according to the loss function;

[0020] S2: Use the optimizer Adam to perform backpropagation based on the loss value and update the weights of the machine learning model;

[0021] S3: Execute step S1 until the n+1th loss value calculated using the validation set is greater than or equal to the nth loss value, or the preset number of training rounds is reached.

[0022] As a preferred implementation, the environmental model includes an indoor temperature model and a light intensity model;

[0023] The indoor temperature model is expressed as:

[0024] Tin(t+1)=Tin(t)+α(Tset(00)_Tin(t))+β(Tout(t)_Tin(t));

[0025] Wherein, Tout(t) represents the outdoor temperature at the tth moment, Tset(00) represents the preset temperature, Tin(t) represents the indoor temperature at the tth moment, and α and β represent the adjustment coefficients;

[0026] The light intensity model is expressed as:

[0027] P(t) = g(I(t))*Pmax;

[0028] Among them, g(I(t)) represents the relationship function between light intensity and power generation efficiency, I(t) represents light intensity and power generation efficiency, Pmax represents the maximum power generation, and P(t) represents the instantaneous power generation.

[0029] As a preferred implementation, the method for determining the type of power load data is:

[0030] The features are clustered and analyzed to obtain a clustering index set:

[0031] {Index type1 ,Index type2 ,…,Index typeN};

[0032] Among them, Index typeN represents the nth clustering index;

[0033] Take out the largest clustering index and compare it with the preset type:

[0034] max({Index type1 ,Index type2 ,…,Index typeN});

[0035] Among them, max() means taking the maximum value;

[0036] The type with the closest comparison result is used as the type of the power load data.

[0037] As a preferred embodiment, the power load management model also includes a load prediction algorithm:

[0038]

[0039] in, represents the power load data at time t+t0, t0 represents the preset time, X t is the historical power load data, Y t is the historical electricity price data, Z t is the historical holiday data, W t is the historical weather data, f x 、f y、f z and f w represents the regression coefficient.

[0040] In another aspect, the present invention further provides a system for building a power load management model, comprising:

[0041] Data collection module: collects device data of different devices and environmental data of different environments, and also includes a resource pool, which is used to store device data, environmental data and preset configuration data;

[0042] Device model building module: builds the device model of the corresponding device according to the device data in the resource pool;

[0043] Environmental model building module: builds an environmental model of the corresponding environment based on the environmental data in the resource pool;

[0044] Power load data acquisition module: obtains power load data, including load power, power consumption behavior and working mode, obtains the characteristics of power load data through big data analysis technology, and determines the type of power load data based on the characteristics, including load power curve, load weather sensitivity, temperature sensitivity, trusted capacity analysis, power consumption behavior analysis, short-term load forecasting, power consumption trend analysis, price volatility and seasonal volatility;

[0045] Power load management construction module: constructs a power load management model according to the type of power load data, a number of device models and environment models, device data in a resource pool, environment data, and preset configuration data.

[0046] The present invention has the following beneficial effects:

[0047] The present invention collects equipment data of different equipment and environmental data of different environments, and stores these data in a resource pool, which can more comprehensively integrate various related data and provide a rich data foundation for subsequent analysis and modeling. Not only is a device model constructed to reflect the performance and power consumption of the equipment, but an environmental model is also constructed to simulate the impact of environmental factors on the power consumption of the equipment. This modeling method that comprehensively considers equipment and environmental factors can more accurately predict the power consumption of equipment. The characteristics of power load data are obtained through big data analysis technology, and the type of power load data is determined based on these characteristics, and a power load management model is constructed. This model can consider multiple aspects of power load more comprehensively, thereby providing more accurate load forecasting and more sophisticated power management. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention is a flowchart for implementing the method. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0050] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0051] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0052] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0053] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0054] Embodiment 1:

[0055] See also Figure 1 The present invention provides a method for building a power load management model, comprising the following steps:

[0056] Collecting device data of different devices and environmental data of different environments, and also including a resource pool, the resource pool is used to store device data, environmental data and preset configuration data;

[0057] Build a device model for the corresponding device based on the device data in the resource pool;

[0058] Build an environmental model of the corresponding environment based on the environmental data in the resource pool;

[0059] Not only is a device model built to reflect the performance and power consumption of the device, but an environmental model is also built to simulate the impact of environmental factors on the power consumption of the device. This modeling method that comprehensively considers both device and environmental factors can more accurately predict the power consumption of the device compared to existing technologies;

[0060] Acquire power load data, the power load data including load power, power consumption behavior and working mode, acquire the characteristics of the power load data through big data analysis technology, and determine the type of the power load data based on the characteristics, the characteristics including load power curve, load weather sensitivity, temperature sensitivity, trusted capacity analysis, power consumption behavior analysis, short-term load forecasting, power consumption trend analysis, price volatility and seasonal volatility;

[0061] A power load management model is constructed according to the type of the power load data, a plurality of device models and environment models, the device data in the resource pool, the environment data and the preset configuration data.

[0062] Exemplarily, in one implementation, there may be multiple corresponding devices in the entire application scenario, therefore, several device models are called. Secondly, the corresponding application scenario can be broadly understood as being in a large environment (for example, conditions where the ambient temperature is basically the same), or it can be different environmental conditions for different devices in different spaces (a simple example, such as the indoor environment where one device is located and the outdoor environment where another device is located). Therefore, at least one environmental model group is called to "render" the environment for each device, ultimately forming the entire management model architecture. Furthermore, based on the type of power load data, the corresponding configuration data in the resource pool is retrieved and imported into the management model architecture to obtain a load management model. It can be understood that the purpose of importing resource data is to generate corresponding model content according to management needs in management, to achieve prediction and other purposes, and to facilitate users to manage power loads.

[0063] The characteristics of power load data are obtained through big data analysis technology, and the type of power load data is determined based on these characteristics, and a power load management model is constructed. This model can consider multiple aspects of power load more comprehensively, thereby providing more accurate load forecasting and more sophisticated power management.

[0064] As a preferred implementation, the method for constructing the device model is:

[0065] The equipment data includes but is not limited to the equipment's operating power, voltage, current, power factor, operating efficiency, etc.;

[0066] Filter outliers and erroneous data from device data;

[0067] Data cleaning: Remove outliers and erroneous data from the data. For example, if the collected current value is an unreasonable value that exceeds the normal working range of the equipment (possibly caused by a sensor failure), it needs to be removed or corrected.

[0068] Normalize the filtered device data and divide the filtered device data into a training set and a validation set;

[0069] Normalization: Normalize input data of different ranges (such as voltage, current, etc.) to a standard range.

[0070] Select the neural network model as the base and use the mean squared error MSE as the loss function:

[0071]

[0072] Among them, y i Indicates the actual output power value, y′ i represents the output power value predicted by the model, and n represents the number of samples;

[0073] Determine the input layer and output layer of the model. Input layer: The number of input nodes should be the same as the number of input features, that is, the input is rated voltage, rated current, power factor and operating efficiency, so the number of input layer nodes is 4. Output layer: The number of output nodes depends on the target you want to predict. If the goal is to predict the output power of the equipment, the number of output layer nodes is 1.

[0074] The device model training method is:

[0075] S1: Input the training set into the machine learning model for training, calculate the predicted output through forward propagation, and calculate the loss value according to the loss function;

[0076] S2: Use the optimizer Adam to perform backpropagation based on the loss value and update the weights of the machine learning model;

[0077] S3: Execute step S1 until the n+1th loss value calculated using the validation set is greater than or equal to the nth loss value, or the preset number of training rounds is reached.

[0078] As a preferred implementation, the environmental model includes an indoor temperature model and a light intensity model;

[0079] Collect environmental data that affects the equipment's operating status and load patterns, including indoor temperature, outdoor temperature, light intensity, etc.

[0080] The indoor temperature model is expressed as:

[0081] Tin(t+1)=Tin(t)+α(Tset(00)-Tin(t))+β(Tout(t)-Tin(t));

[0082] Wherein, Tout(t) represents the outdoor temperature at the tth moment, Tset(00) represents the preset temperature, Tin(t) represents the indoor temperature at the tth moment, and α and β represent the adjustment coefficients;

[0083] The light intensity model is expressed as:

[0084] P(t) = g(I(t))*Pmax;

[0085] Among them, g(I(t)) represents the relationship function between light intensity and power generation efficiency, I(t) represents light intensity and power generation efficiency, Pmax represents the maximum power generation, and P(t) represents the instantaneous power generation.

[0086] As a preferred implementation, the method for determining the type of power load data is:

[0087] The features are clustered and analyzed to obtain a clustering index set:

[0088] {Index type1 ,Index type2 ,…,Index typeN};

[0089] Among them, Index typeN represents the nth clustering index;

[0090] Take out the largest clustering index and compare it with the preset type:

[0091] max({Index type1 ,Index type2 ,…,Index typeN});

[0092] Among them, max() means taking the maximum value;

[0093] The type with the closest comparison result is used as the type of the power load data.

[0094] As a preferred embodiment, the power load management model also includes a load prediction algorithm:

[0095]

[0096] in, represents the power load data at time t+t0, t0 represents the preset time, X t is the historical power load data, Y t is the historical electricity price data, Z t is the historical holiday data, W t is the historical weather data, f x 、f y 、f z and f w represents the regression coefficient.

[0097] In the load forecasting algorithm, the historical power load data is time series data, and the dynamic changes of the load can be captured by the autoregressive integrated moving average model ARIMA or the seasonal autoregressive integrated moving average model SARIMA. Specifically, this embodiment uses the autoregressive part of the ARIMA model to represent the change trend of the load: in, is the autoregression coefficient, and p is the order of the autoregression term. The historical electricity price data is obtained based on the electricity price policy, which is a continuous variable. The historical holiday data is obtained based on the historical holiday distribution and arrangement policy. The historical weather data is expressed by temperature, which is obtained through historical temperature.

[0098] Through the prediction in accordance with the actual needs, the load of the equipment can be accurately predicted, providing reliable data for subsequent use, prediction, maintenance, etc.; the load prediction algorithm takes into account multiple factors such as historical power load data, electricity price data, holiday data and weather data, and obtains the prediction results through regression analysis. Compared with the simple prediction method in the existing technology, this algorithm can more accurately predict the change trend of future power load.

[0099] Embodiment 2:

[0100] The present invention also provides a power load management model building system, comprising:

[0101] Data collection module: collects device data of different devices and environmental data of different environments, and also includes a resource pool, which is used to store device data, environmental data and preset configuration data;

[0102] Device model building module: builds the device model of the corresponding device according to the device data in the resource pool;

[0103] Environmental model building module: builds an environmental model of the corresponding environment based on the environmental data in the resource pool;

[0104] Power load data acquisition module: obtains power load data, including load power, power consumption behavior and working mode, obtains the characteristics of power load data through big data analysis technology, and determines the type of power load data based on the characteristics, including load power curve, load weather sensitivity, temperature sensitivity, trusted capacity analysis, power consumption behavior analysis, short-term load forecasting, power consumption trend analysis, price volatility and seasonal volatility;

[0105] Power load management construction module: constructs a power load management model according to the type of power load data, a number of device models and environment models, device data in a resource pool, environment data, and preset configuration data.

[0106] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0107] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0108] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0109] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0110] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for building a power load management model, characterized in that: The following steps are involved: Collecting device data of different devices and environmental data of different environments, and also including a resource pool, the resource pool is used to store device data, environmental data and preset configuration data; Build a device model for the corresponding device based on the device data in the resource pool; Build an environmental model of the corresponding environment based on the environmental data in the resource pool; Acquire power load data, the power load data including load power, power consumption behavior and working mode, acquire the characteristics of the power load data through big data analysis technology, and determine the type of the power load data based on the characteristics, the characteristics including load power curve, load weather sensitivity, temperature sensitivity, trusted capacity analysis, power consumption behavior analysis, short-term load forecasting, power consumption trend analysis, price volatility and seasonal volatility; A power load management model is constructed according to the type of the power load data, a plurality of device models and environment models, the device data in the resource pool, the environment data and the preset configuration data.

2. The method for building a power load management model according to claim 1, characterized in that: The method for constructing the device model is: Filter outliers and erroneous data from device data; Normalize the filtered device data and divide the filtered device data into a training set and a validation set; Select the neural network model as the base and use the mean squared error MSE as the loss function: Among them, y i Indicates the actual output power value, y′ i represents the output power value predicted by the model, and n represents the number of samples; The device model training method is: S1: Input the training set into the machine learning model for training, calculate the predicted output through forward propagation, and calculate the loss value according to the loss function; S2: Use the optimizer Adam to perform backpropagation based on the loss value and update the weights of the machine learning model; S3: Execute step S1 until the n+1th loss value calculated using the validation set is greater than or equal to the nth loss value, or the preset number of training rounds is reached.

3. The method for building a power load management model according to claim 1, characterized in that: The environmental model includes an indoor temperature model and a light intensity model; The indoor temperature model is expressed as: Tin(t+1)=Tin(t)+α(Tset(00)-Tin(t))+β(Tout(t)-Tin(t)); Wherein, Tout(t) represents the outdoor temperature at the tth moment, Tset(00) represents the preset temperature, Tin(t) represents the indoor temperature at the tth moment, and α and β represent the adjustment coefficients; The light intensity model is expressed as: P(t) = g(I(t))*Pmax; Among them, g(I(t)) represents the relationship function between light intensity and power generation efficiency, I(t) represents light intensity and power generation efficiency, Pmax represents the maximum power generation, and P(t) represents the instantaneous power generation.

4. The method for building a power load management model according to claim 1, characterized in that: The method for determining the type of power load data is as follows: The features are clustered and analyzed to obtain a clustering index set: {Index type1 ,Index type2 ,…,Index typeN }; Among them, Index typeN represents the nth clustering index; Take out the largest clustering index and compare it with the preset type: max({Index type1 ,Index type2 ,…,Index typeN }); Among them, max() means taking the maximum value; The type with the closest comparison result is used as the type of the power load data.

5. The method for building a power load management model according to claim 1, characterized in that: The power load management model also includes load forecasting algorithms: in, represents the power load data at time t+t0, t0 represents the preset time, X t is the historical power load data, Y t is the historical electricity price data, Z t is the historical holiday data, W t is the historical weather data, f x 、f y 、f z and f w represents the regression coefficient.

6. A power load management model building system, characterized in that: include: Data collection module: collects device data of different devices and environmental data of different environments, and also includes a resource pool, which is used to store device data, environmental data and preset configuration data; Device model building module: builds the device model of the corresponding device according to the device data in the resource pool; Environmental model building module: builds an environmental model of the corresponding environment based on the environmental data in the resource pool; Power load data acquisition module: obtains power load data, including load power, power consumption behavior and working mode, obtains the characteristics of power load data through big data analysis technology, and determines the type of power load data based on the characteristics, including load power curve, load weather sensitivity, temperature sensitivity, trusted capacity analysis, power consumption behavior analysis, short-term load forecasting, power consumption trend analysis, price volatility and seasonal volatility; Power load management construction module: constructs a power load management model according to the type of power load data, a number of device models and environment models, device data in a resource pool, environment data, and preset configuration data.

7. The power load management model building system according to claim 6, characterized in that: The device model construction module and the device model construction method are as follows: Filter outliers and erroneous data from device data; Normalize the filtered device data and divide the filtered device data into a training set and a validation set; Select the neural network model as the base and use the mean squared error MSE as the loss function: Among them, y i Indicates the actual output power value, y′ i represents the output power value predicted by the model, and n represents the number of samples; The device model training method is: S1: Input the training set into the machine learning model for training, calculate the predicted output through forward propagation, and calculate the loss value according to the loss function; S2: Use the optimizer Adam to perform backpropagation based on the loss value and update the weights of the machine learning model; S3: Execute step S1 until the n+1th loss value calculated using the validation set is greater than or equal to the nth loss value, or the preset number of training rounds is reached.

8. The power load management model building system according to claim 6, characterized in that: The environmental model construction module includes an indoor temperature model and a light intensity model; The indoor temperature model is expressed as: Tin(t+1)=Tin(t)+α(Tset(00)-Tin(t))+β(Tout(t)-Tin(t)); Wherein, Tout(t) represents the outdoor temperature at the tth moment, Tset(00) represents the preset temperature, Tin(t) represents the indoor temperature at the tth moment, and α and β represent the adjustment coefficients; The light intensity model is expressed as: P(t) = g(I(t))*Pmax; Among them, g(I(t)) represents the relationship function between light intensity and power generation efficiency, I(t) represents light intensity and power generation efficiency, Pmax represents the maximum power generation, and P(t) represents the instantaneous power generation.

9. The power load management model building system according to claim 6, characterized in that: The method for determining the type of power load data is as follows: The features are clustered and analyzed to obtain a clustering index set: {Index type1 ,Index type2 ,…,Index typeN }; Among them, Index typeN represents the nth clustering index; Take out the largest clustering index and compare it with the preset type: max({Index type1 ,Index type2 ,…,Index typeN }); Among them, max() means taking the maximum value; The type with the closest comparison result is used as the type of the power load data.

10. The power load management model building system according to claim 6, characterized in that: It also includes a load forecasting module, which is expressed as: in, represents the power load data at time t+t0, t0 represents the preset time, X t is the historical power load data, Y t is the historical electricity price data, Z t is the historical holiday data, W t is the historical weather data, f x 、f y 、f z and f w represents the regression coefficient.

Images