Hierarchical load response potential evaluation method considering power consumption behaviors of users
By building a device demand response potential model and a multi-level load response potential evaluation model, and combining information entropy theory to analyze the controllability of equipment, the problem of lack of considering the physical characteristics and behavioral logic of large-scale industrial and commercial loads in the existing technology is solved, and a more accurate load response potential evaluation is achieved.
Patent Information
- Application Number
- CN202411939868.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-26
- Publication Date
- 2025-04-15
AI Technical Summary
The existing technology lacks the physical characteristics, production timing and behavioral logic of large-scale industrial and commercial loads in load aggregation modeling, resulting in insufficient evaluation methods for hierarchical load response potential of users' electricity use behavior.
Build a device demand response potential model, including device status model, user intention model and behavior prediction time scale model, use information entropy theory to analyze device controllability, calculate the demand response potential within any response period, and build a multi-level load response potential evaluation model based on this.
The accuracy of equipment status prediction during the response period is improved, and the response potential is evaluated in combination with the controllability of the equipment, and the degree of user willingness is fully considered, which solves the problem of missing load resource aggregation models in different industries.
Smart Images
Figure CN119834222A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of power grid user load regulation, and in particular relates to a hierarchical load response potential assessment method taking into account user electricity consumption behavior. Background Art
[0002] The potential analysis of existing technology demand response is mainly aimed at a certain type of load or industry. When considering the user's willingness, the user's willingness is linked to a certain indicator, such as the user's willingness to adjust the air-conditioning load is defined by the indoor and outdoor temperature difference. Generally, intelligent algorithms such as neural network models are used to predict the load. When modeling load aggregation, single or multi-type load aggregation modeling methods are mostly studied from the perspectives of controllable loads on the resident side, building temperature control loads, and electric vehicle charging and discharging loads. There is a lack of controllable load aggregation models that take into account the physical characteristics, production timing, and behavioral logic of large industrial and commercial loads.
[0003] Therefore, a hierarchical load response potential assessment method that takes into account user electricity consumption behavior is urgently needed. The accuracy of equipment status prediction during the response period is improved, and the response potential is evaluated in combination with equipment controllability, which fully considers the user's willingness; it provides a reference solution to solve the current problem of lack of load resource aggregation models in different industries. Summary of the invention
[0004] The object of the present invention is to provide a hierarchical load response potential assessment method considering user power consumption behavior, characterized in that it comprises the following steps:
[0005] Step S1: constructing a device demand response potential model and analyzing factors affecting the device demand response potential;
[0006] Step S2: Combine the factors affecting the equipment demand response potential, build a user electricity consumption behavior prediction model, use information entropy theory to analyze the controllability of the equipment, and calculate the demand response potential within any response period;
[0007] Step S3: Based on the demand response potential, a multi-level load response potential assessment model is constructed to achieve hierarchical load response potential assessment.
[0008] The equipment demand response potential model includes: an equipment state model, a user willingness model and a behavior prediction time scale model;
[0009] The equipment state model includes: a cyclic working equipment state model and a random logic equipment model;
[0010] The user willingness model is: the intensity of the user's willingness to respond to grid demand by actively changing the operating status of certain equipment during the demand response period;
[0011] The behavior prediction time scale model is as follows: the user's electricity consumption behavior is modeled as a discrete time series in which the device group state changes over time; and the future value of the target is predicted by predicting the time series.
[0012] The state model of the cyclic working equipment is:
[0013] In a certain demand response period T res The maximum response potential for all devices of this type at T res The total power consumption of the maximum number of operations within the time period:
[0014]
[0015] Where W i is the power consumed in one working process, m is the number of fixed logic devices, k r For a device in T res Maximum number of runs within ;
[0016] The random logic device model is: once the random logic device is turned on, the operating power remains unchanged, and the device energy consumption is mainly affected by factors such as the duration of the turn-on time that are closely related to the user's random behavior. Its corresponding demand potential is:
[0017]
[0018] Where P i is the rated power of the equipment, t k For this device in T res The duration of the opening time in the time period.
[0019] The user electricity consumption behavior prediction model includes:
[0020] The input variables are: the device status of the current window and the device characteristics of the next window;
[0021] The device status of the current window is:
[0022] S app =[S1,S2,...S m ],S i =[0,1,...,s T ],s t ∈[stati1,...,stati k ]
[0023] In the formula, m is the number of devices, T is the window length, and k is the number of states owned by the i-th device;
[0024] The next window device characteristics are:
[0025] C win =[T win ,D win ]
[0026] Where, T win+1 =[Mon,D,H,M], which are the encoded month, date, hour, minute and other time information, respectively. win =[AP, TH, RAD], which are the corresponding influencing factors such as temperature, relative humidity, and light intensity in the next window;
[0027] The output is: the device status of the next time window:
[0028] S app =[S1,S2,...S m ],S i =[s T+1 ,...,s T+window ].
[0029] The use of information entropy theory to analyze the controllability of the device includes: defining the user device startup process as B s ; The user equipment shutdown process is B e ; Equipment startup controllability C on ; Equipment shutdown controllability C off ;
[0030] The probability density of the device on and off time is the probability per minute in a day, and the controllability of the device on is C on and equipment shutdown controllability C off The device controllability index calculated using information entropy is:
[0031]
[0032] In the formula, B(x i ) is the probability density at a certain point.
[0033] The calculation of the demand response potential in any response period includes: any response period T res The power consumption of the devices within and the user's willingness to respond;
[0034] Given a response period T res After that, the historical state of device i is known to be The status during this period is Where T p is the length of the historical data window, T res The length of the window for predicting electricity usage behavior;
[0035] Device i in T res The energy that can be reduced during the period is:
[0036] Rs i =R1+R2+R3 (5)
[0037] Where R1 is the res The energy consumed by the complete operation process within the time period, that is, the opening and closing processes of the equipment are both within T res During the time period, R1 is:
[0038]
[0039] Where k is T res The number of complete runs within The energy consumed for each complete process;
[0040] R2 and R3 are respectively res The energy consumed in the operation process from running to closing and from closing to opening. R2 and R3 do not contain a complete operation curve. The definitions of R2 and R3 are:
[0041]
[0042] Where t0 is T res The starting time of the time period, te i The time when the equipment ends operation, ts i For equipment in T res The opening time within, t res T res The end time of the period;
[0043] If in T res If the device is predicted to be in the off state during the time period, the device has a turn-on response characteristic, and the energy consumption can be increased as follows:
[0044]
[0045] Where k is T res The number of processes that can be fully run within
[0046] Considering the information entropy, the device adjustable potential is:
[0047]
[0048] The construction of a multi-level load response potential assessment model includes:
[0049] (1) Predict device behavior;
[0050] (2) Consider the adjustable capacity of logic and potential;
[0051] (3) Establish a hierarchical response potential model.
[0052] The device behavior prediction comprises:
[0053] Collect the equipment status before demand response and update it to the historical status database;
[0054] If a demand response instruction is received at this time, the status of each device is predicted during the win_size_pre period starting from time t:
[0055]
[0056] In the formula, is the predicted state of the i-th device in the corresponding period, spre() is the state prediction model proposed above, and its inputs are S i (t), the device state of the input window length, C i (t+win_size_pre), the covariate of the output window length.
[0057] The adjustable capacity considering logic and potential includes:
[0058] By adjusting the electricity consumption behavior to reduce the load to achieve participation in grid regulation, the participation in grid regulation includes: interrupting the current load and shifting the load operation time period;
[0059] Consider the working status of the device within the response period, the constraints imposed by the device on its own or other devices’ operating logic, and the operating logic between devices.
[0060] The establishment of a hierarchical response potential model comprises:
[0061] After obtaining the detailed power consumption information of each type of load and the adjustable potential information within the response period, the adjustable potential information belonging to the same category is aggregated, and the weighted adjustable potential is calculated according to the equipment information entropy as the criterion for selecting user response priority; the above process is repeated for all users, and finally a multi-level response potential model of equipment-user-industry-region is formed.
[0062] The beneficial effects of the present invention are:
[0063] The present invention discloses a hierarchical load response potential assessment method that considers user electricity consumption behavior, proposes a user electricity consumption behavior prediction model, and predicts the load state within the response period; uses information entropy theory to analyze the controllability of the equipment, and proposes a response potential assessment method that combines the controllability of the equipment, providing a theoretical basis for the response potential calculation within the demand response period.
[0064] By analyzing the factors affecting the equipment demand response potential; establishing a user electricity consumption behavior prediction model based on temporal convolutional networks and attention mechanisms; predicting electricity consumption behavior within the response period through user historical data; and evaluating the response potential of different industries and multi-level loads after considering the response characteristics and controllability of the equipment within the response period.
[0065] Based on the analysis of factors affecting the equipment response potential, a power consumption behavior prediction model considering user habits is proposed to improve the accuracy of equipment status prediction during the response period; the response potential is evaluated in combination with the controllability of the equipment; the user's willingness is fully considered; and a reference solution is provided for the current lack of load resource aggregation models for different industries. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1 A schematic flow chart of a hierarchical load response potential assessment method considering user electricity consumption behavior according to the present invention;
[0067] Figure 2 A schematic diagram of a user electricity consumption behavior prediction model according to an embodiment of the present invention;
[0068] Figure 3 Schematic diagram of four equipment operating states within a response period according to an embodiment of the present invention. DETAILED DESCRIPTION
[0069] The present invention provides a hierarchical load response potential assessment method taking into account user electricity consumption behavior. The present invention is further described in detail below in conjunction with the accompanying drawings.
[0070] like Figure 1 The embodiment of the present invention shown discloses a hierarchical load response potential assessment method considering user power consumption behavior, comprising the following steps:
[0071] Step S1: constructing a device demand response potential model and analyzing factors affecting the device demand response potential;
[0072] Step S2: Combine the factors affecting the equipment demand response potential, build a user electricity consumption behavior prediction model, use information entropy theory to analyze the controllability of the equipment, and calculate the demand response potential within any response period;
[0073] Step S3: Based on the demand response potential, a multi-level load response potential assessment model is constructed to achieve hierarchical load response potential assessment.
[0074] Each step is described in detail below.
[0075] 1. Factors affecting equipment demand response potential
[0076] The equipment demand response potential model includes: an equipment state model, a user willingness model and a behavior prediction time scale model;
[0077] The equipment state model includes: a cyclic working equipment state model and a random logic equipment model;
[0078] The user willingness model is: the intensity of the user's willingness to respond to grid demand by actively changing the operating status of certain equipment during the demand response period;
[0079] The behavior prediction time scale model is as follows: the user's electricity consumption behavior is modeled as a discrete time series in which the device group state changes over time; and the future value of the target is predicted by predicting the time series.
[0080] (1) Device status
[0081] During the operation of the equipment, the transition between different operating states of each device causes the total power to fluctuate up and down. The rules of state transition of different types of equipment are very different, and their demand response potential should also be discussed in categories.
[0082] The state model of the cyclic working equipment is:
[0083] Cyclic working devices are also called fixed logic devices. They are mostly multi-state devices and continuous state devices. res The maximum response potential for all devices of this type at T res The total power consumption of the maximum number of operations within the time period:
[0084]
[0085] W i is the power consumed in one working process, m is the number of fixed logic devices, k r For a device in T res The maximum number of runs within .
[0086] The random logic device model is:
[0087] Once a random logic device is turned on, its operating state is relatively stable, and its operating power generally remains unchanged. The energy consumption of the device is mainly affected by factors such as the duration of the turn-on time that are closely related to the user's random behavior. Its corresponding demand potential can be calculated by the following formula:
[0088]
[0089] Where P i is the rated power of the equipment, t k For this device in T res The duration of the opening time in the time period.
[0090] (2) User Intentions
[0091] The user willingness model is:
[0092] In demand response, the user's willingness to respond refers to the degree of willingness to respond to grid demand by actively changing the operating status of certain equipment during the demand response period. If a user is accustomed to using a certain device during a certain period, the regulation potential of the device during this period needs to be lowered.
[0093] The power consumption environment is also an important factor affecting the response potential. For example, the most typical air conditioning load, its response potential is affected by both the ambient temperature and the user's comfort temperature.
[0094] (3) Behavior prediction time scale
[0095] The behavior prediction time scale model is:
[0096] The user's electricity consumption behavior data belongs to a time series. The prediction of the time series is to predict the future value of the target through the development trend of historical data.
[0097] In this embodiment, the user's electricity consumption behavior is modeled as a discrete time series in which the state of the device group changes over time. The state sequence of most of the devices is a periodic sequence. Over time, their operating states show a certain degree of periodic changes, ranging from hours to days, and the state changes of individual devices show random characteristics. The choice of time step depends on many factors. The larger the time step, the more information the model receives, and the better it can capture the long-term correlation of the data, but it will also increase the complexity of the model and increase the risk of overfitting. The smaller the time step, the fewer computing resources the model requires, but the model may pay more attention to noise and cause underfitting.
[0098] 2. Electricity consumption behavior prediction model and response potential calculation
[0099] (1) User electricity consumption behavior prediction model
[0100] like Figure 2 As shown, the user electricity consumption behavior prediction model includes:
[0101] The user electricity consumption behavior prediction model is trained through long-term data to learn the user's electricity consumption habits. The input variables of the model are the device status S of the current window app =[S1,S2,...S m ],S i =[0,1,...,s T ],s t ∈[stati1,...,stati k], where m is the number of devices, T is the window length, and k is the number of states owned by the i-th device. And the next window device feature, C win =[T win ,D win ], where T win+1 =[Mon,D,H,M], which are the encoded month, date, hour, minute and other time information, respectively. win =[AP, TH, RAD], which are the corresponding temperature, relative humidity, light intensity and other influencing factors in the next window. The output is the device status S of the next time window app =[S1,S2,...S m ],S i =[s T+1 ,...,s T+window ].
[0102] After determining the equipment status within the response period, it is divided into four typical curves. Figure 3 As shown, there are four possible operating processes of the device during the response period: remaining turned on during the response period, having a complete operating process, being turned off at a certain moment, or being turned on at a certain moment, which are state 1, state 2, state 3, and state 4 respectively.
[0103] (2) Equipment controllability analysis
[0104] For a certain device, the methods of participating in demand response are divided into two categories: controllable on and controllable off. They respectively represent the degree of user's willingness to actively turn on or off the device during a certain period of time. The higher the controllability of the device, the more users tend to actively change the device status to participate in demand response.
[0105] The controllability of the device analyzed using information entropy theory includes:
[0106] Define the process of turning on and off the user equipment as B s , B e , the corresponding device open controllability and closed controllability C on and C off .
[0107] The controllability of the on-behavior refers to the degree to which the user's behavioral habits are violated by delaying or advancing the start-up time of the appliance. If the user is accustomed to turning on a certain appliance state at a fixed time period, the variance of the probability distribution of the start-up time of the corresponding appliance state is small. Since changing the user's behavioral habits will deviate from the user's inherent habits, the controllability of the on-behavior of the appliance state should be low at this time. Similarly, a small variance in the probability distribution of the controllability of the off-behavior means that the off-behavior is less controllable, and it is difficult to achieve demand response by shutting down the device in advance.
[0108] In order to quantify the controllability of device behavior based on probability density, the concept of information entropy is introduced. In information theory, information entropy is widely used in the measurement of uncertainty. The higher the information entropy, the greater the uncertainty of the information.
[0109] H(X)=-∑ i P(x i )logP(x i ) (3)
[0110] The probability density of the device on and off time is the probability per minute in a day, C on and C off They represent the device controllability index calculated using information entropy. The calculation formula is as follows:
[0111]
[0112] B(x i ) is the probability density at a certain point. In information entropy, the greater the probability, the less "information" it contains, which means the uncertainty of the event is smaller, and vice versa. on and C off In the data, if the time period during which users use a certain device is relatively concentrated, the probability density presents a clustered bimodal characteristic, and its overall information entropy is relatively small. If the time period during which users use the device is relatively dispersed, its information entropy is relatively large.
[0113] For the same device, C on and C off There may not be a strong correlation between the size of C on Larger device C off Not necessarily big. on The larger the value, the more likely it is that users are willing to adjust the device start-up time to participate in demand response. off The larger the value is, the more likely it is that users are willing to adjust the shutdown time of their equipment to participate in demand response. on Larger devices are called open response devices, C off The larger ones are called closed response devices.
[0114] (3) Response potential calculation
[0115] The calculation of demand response potential within any response period includes:
[0116] In any response period T res The demand response potential within a period should comprehensively consider the power consumption of the equipment and the user's response willingness during this period. res After that, the historical state of device i is known to be The status during this period is Where Tp is the length of the historical data window, T res is the window length for predicting power consumption behavior. Then device i is res The energy that can be reduced during a period is shown in the following formula:
[0117] Rs i =R1+R2+R3 (5)
[0118] In formula (5), R1 is res The energy consumed by the complete operation process within the time period, that is, the opening and closing processes of the equipment are both within T res Within the time period, R1 is calculated as shown in the formula:
[0119]
[0120] In formula (6), k is T res The number of complete runs within The energy consumed for each complete process;
[0121] R2 and R3 are respectively res The energy consumed in the operation process from running to closing and from closing to opening, that is, R2 and R3 do not contain a complete operation curve. The specific calculation formula is as follows:
[0122]
[0123] In formula (7), t0 is T res The starting time of the time period, te i The time when the equipment ends operation, ts i For equipment in T res The opening time within, t res T res The end time of the period;
[0124] If in T res If the device is predicted to be in the off state during the time period, the device has a turn-on response characteristic, and the energy consumption that can be increased is shown in the following formula:
[0125]
[0126] Where k is T res The number of processes that can be fully run within.
[0127] Since users have different usage habits for different devices, the actual device response potential should take into account the information entropy of user usage habits. A device with high information entropy means that users are more inclined to change the state of the device to participate in the response. Therefore, even if a device with high information entropy has a smaller amount of power available for response, the user may have a higher priority in selecting this device for response, and its response potential should be higher after comprehensive consideration.
[0128] Considering the information entropy, the device adjustable potential is:
[0129]
[0130] 3. Multi-level load response potential assessment
[0131] The construction of a multi-level load response potential assessment model includes:
[0132] (1) Predict device behavior;
[0133] (2) Consider the adjustable capacity of logic and potential;
[0134] (3) Establish a hierarchical response potential model.
[0135] The device behavior prediction comprises:
[0136] Collect the device status before demand response and update it to the historical status database. If a demand response instruction is received at this time, the status of each device is predicted in the win_size_pre period starting from time t, which can be obtained by the following formula:
[0137]
[0138] in is the predicted state of the i-th device in the corresponding period, spre() is the state prediction model proposed above, and its inputs are S i (t), the device state of the input window length, C i (t+win_size_pre), output window length covariate, in this embodiment, such as time, temperature and other information.
[0139] The adjustable capacity considering logic and potential includes:
[0140] There are two ways for users to reduce load by adjusting their electricity consumption behavior to participate in grid regulation: interrupting the current load and shifting the load operation time period. For a certain device, the operation status during the response period can also be divided into four types: Figure 3As shown in the figure. As mentioned above, states 2 and 4 can participate in the response by shifting the operating time period. States 1 and 3 can only participate in the response by interrupting the load. Therefore, the calculation of demand response potential should also consider the working status of the equipment during the response period.
[0141] In addition to the impact of the working state of the equipment during the response period on the adjustable characteristics of the equipment, some equipment is also constrained by its own or other equipment's operating logic. For example, some heating and molding processes in industrial loads do not have the characteristics of being interruptible. In addition, the operating logic between equipment is also an important factor that restricts the adjustable characteristics of equipment.
[0142] The establishment of a hierarchical response potential model comprises:
[0143] After obtaining the detailed power consumption information of each type of load and the adjustable potential information within the response period, the adjustable potential information belonging to the same category is aggregated, and the weighted adjustable potential is calculated according to the device information entropy as the standard for selecting the user response priority. The above process is repeated for all users, and finally a multi-level response potential model of device-user-industry-region is formed.
Claims
1. A hierarchical load response potential assessment method considering user electricity consumption behavior, characterized in that: The steps include: Step S1: constructing a device demand response potential model and analyzing factors affecting the device demand response potential; Step S2: Combine the factors affecting the equipment demand response potential, build a user electricity consumption behavior prediction model, use information entropy theory to analyze the controllability of the equipment, and calculate the demand response potential within any response period; Step S3: Based on the demand response potential, a multi-level load response potential assessment model is constructed to achieve hierarchical load response potential assessment.
2. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 1 is characterized in that: The equipment demand response potential model includes: an equipment state model, a user willingness model and a behavior prediction time scale model; The equipment state model includes: a cyclic working equipment state model and a random logic equipment model; The user willingness model is: the intensity of the user's willingness to respond to grid demand by actively changing the operating status of certain equipment during the demand response period; The behavior prediction time scale model is as follows: the user's electricity consumption behavior is modeled as a discrete time series in which the device group state changes over time; and the future value of the target is predicted by predicting the time series.
3. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 2 is characterized in that: The state model of the cyclic working equipment is: In a certain demand response period T res The maximum response potential for all devices of this type at T res The total power consumption of the maximum number of operations within the time period: Where W i is the power consumed in one working process, m is the number of fixed logic devices, k r For a device in T res Maximum number of runs within ; The random logic device model is: once the random logic device is turned on, the operating power remains unchanged, and the device energy consumption is mainly affected by factors such as the duration of the turn-on time that are closely related to the user's random behavior. Its corresponding demand potential is: Where P i is the rated power of the equipment, t k For this device in T res The duration of the opening time in the time period.
4. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 1 is characterized in that: The user electricity consumption behavior prediction model includes: The input variables are: the device status of the current window and the device characteristics of the next window; The device status of the current window is: S app =[S1,S2,...S m ],S i =[0,1,...,s T ],s t ∈[stati1,...,stati k ] In the formula, m is the number of devices, T is the window length, and k is the number of states owned by the i-th device; The next window device characteristics are: C win =[T win ,D win ] Where, T win+1 =[Mon,D,H,M], which are the encoded month, date, hour, minute and other time information, respectively. win =[AP, TH, RAD], which are the corresponding influencing factors such as temperature, relative humidity, and light intensity in the next window; The output is: the device status of the next time window: S app =[S1,S2,...S m ],S i =[s T+1 ,...,s T+window ]。 5. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 1 is characterized in that: The use of information entropy theory to analyze the controllability of the device includes: defining the user device startup process as B s ; The user equipment shutdown process is B e ; Equipment start controllability C on ; Equipment shutdown controllability C off ; The probability density of the device on and off time is the probability per minute in a day, and the controllability of the device on is C on and equipment shutdown controllability C off The device controllability index calculated using information entropy is: In the formula, B(x i ) is the probability density at a certain point.
6. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 1 is characterized in that: The calculation of the demand response potential in any response period includes: any response period T res The power consumption of the devices within and the user's willingness to respond; Given a response period T res After that, the historical state of device i is known to be The status during this period is Where T p is the length of the historical data window, T res The length of the window for predicting electricity usage behavior; Device i in T res The energy that can be reduced during the period is: Rs i =R1+R2+R3 (5) Where R1 is the res The energy consumed by the complete operation process within the time period, that is, the opening and closing processes of the equipment are both within T res During the time period, R1 is: Where k is T res The number of complete runs within The energy consumed for each complete process; R2 and R3 are respectively res The energy consumed in the operation process from running to closing and from closing to opening. R2 and R3 do not contain a complete operation curve. The definitions of R2 and R3 are: Where t0 is T res The starting time of the time period, te i The time when the equipment ends operation, ts i For equipment in T res The opening time within, t res T res The end time of the period; If in T res If the device is predicted to be in the off state during the time period, the device has a turn-on response characteristic, and the energy consumption can be increased as follows: Where k is T res The number of processes that can be fully run within Considering the information entropy, the device adjustable potential is:
7. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 1 is characterized in that: The construction of a multi-level load response potential assessment model includes: (1) Predict device behavior; (2) Consider the adjustable capacity of logic and potential; (3) Establish a hierarchical response potential model.
8. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 7 is characterized in that: The device behavior prediction comprises: Collect the equipment status before demand response and update it to the historical status database; If a demand response instruction is received at this time, the status of each device is predicted during the win_size_pre period starting from time t: In the formula, is the predicted state of the i-th device in the corresponding period, spre() is the state prediction model proposed above, and its inputs are S i (t), the device state of the input window length, C i (t+win_size_pre), the covariate of the output window length.
9. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 7 is characterized in that: The adjustable capacity considering logic and potential includes: By adjusting the electricity consumption behavior to reduce the load to achieve participation in grid regulation, the participation in grid regulation includes: interrupting the current load and shifting the load operation time period; Consider the working status of the device within the response period, the constraints imposed by the device on its own or other devices’ operating logic, and the operating logic between devices.
10. The hierarchical load response potential assessment method considering user electricity consumption behavior according to claim 7, characterized in that: The establishment of the hierarchical response potential model comprises: After obtaining the detailed power consumption information of each type of load and the adjustable potential information within the response period, the adjustable potential information belonging to the same category is aggregated, and the weighted adjustable potential is calculated according to the equipment information entropy as the criterion for selecting user response priority; the above process is repeated for all users, and finally a multi-level response potential model of equipment-user-industry-region is formed.