Industrial load self-discipline constraint evaluation method under multiple time scales

Abstract

The invention provides an industrial load self-discipline constraint evaluation method under multiple time scales, and relates to the technical field of electrical engineering. According to the method, the self-discipline performance comprehensive quantitative index SDI is provided, the index fuses the confidence level and the self-discipline degree, comprehensive evaluation can be carried out on the control effect aspect of the control strategy under different time scales, then it is ensured that the control strategy has robustness and reflects the fine degree, and the balance between the accuracy and the robustness of the control effect is achieved. The self-discipline performance evaluation method under different time scales is provided, the corresponding self-discipline performance evaluation method is provided for real-time control, secondary real-time control, intra-day scheduling and day-ahead scheduling, and the execution efficiency, robustness and accuracy of a control strategy are considered, so that the self-discipline level of the system under different time scales is comprehensively evaluated. A non-parametric kernel density estimation method is adopted, industrial load self-discipline deviation distribution characteristics are effectively captured and depicted, and the method is not influenced by a specific probability distribution form.

Description

Technical Field

[0001] The invention relates to the field of electrical engineering technology, and in particular to an industrial load self-discipline constraint evaluation method under multiple time scales. Background Art

[0002] In the era of intelligent power system and Industry 4.0, refined management and control of industrial loads have become the core tasks for optimizing power system operation. Traditional modeling methods can no longer meet the requirements of precise balance of power supply and demand and instantaneous stability control of power grid frequency because they fail to fully consider the dynamic characteristics of industrial loads, adjustability in the production process and random uncertainty. Thanks to the deep integration of information technology and advanced sensing technology, industrial users can obtain more detailed energy consumption data, which provides a valuable information basis for creating accurate and adaptive industrial load models. However, how to select appropriate data to be effectively integrated into the load modeling process, realize the self-adjustment and self-constraint of the load model, and accurately evaluate the behavioral characteristics of actual industrial loads and the immediate response to power system dispatch instructions is a technical challenge that needs to be solved urgently.

[0003] At present, the research on power system stability assessment mainly focuses on the evaluation and calculation of stability margin. The article "Calculation Method of Static Voltage Stability Margin of Power System Based on AQ Node" in "Power System Technology" Vol. 43, No. 2, 2019, pp. 714-721, introduces AQ nodes and maps the Jacobian matrix to a modified Jacobian matrix to avoid the singularity of the matrix to more accurately calculate the stability margin. This method is efficient and intuitive, but lacks consideration of dynamic characteristics and problems of different time scales.

[0004] There are also some studies on power system stability assessment that revolve around frequency. The article "Frequency Stability Assessment of Power Systems Including Wind Power Based on Lightweight Gradient Boosting Machine and Generative Adversarial Network" on pages 3181-3193 of Volume 46, Issue 8 of "Power System Technology" in 2022, introduces lightweight gradient boosting machine (LightGBM) and generative adversarial network (GAN) to establish an evaluation model for three key indicators: frequency change rate, transient frequency extreme value and quasi-steady-state frequency. This method performs well in dealing with frequency stability assessment of power systems containing wind power, but it still has shortcomings in dealing with the dynamic characteristics of industrial loads and problems of different time scales. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides an industrial load self-discipline constraint evaluation method under multiple time scales, aiming to fill the deficiencies in the research on self-discipline constraint conditions and evaluation methods in the existing power system load side management methods, fully tap the self-discipline characteristics of industrial loads, thereby effectively improving the stability, reliability and economy of the power system, and solving the problem that traditional methods mainly focus on the calculation of stability margin and frequency, but lack in-depth research on the self-regulation ability of industrial loads and their impact on the operation of the power system.

[0006] The technical solution of the present invention is:

[0007] A method for evaluating industrial load self-discipline constraints under multiple time scales includes the following steps:

[0008] Step 1: Define self-discipline bias and define self-discipline based on self-discipline bias;

[0009] The autonomous deviation is a relative value obtained by calculating the difference between the actual instantaneous power value of the industrial load at a specified moment and the target power value expected to be achieved by the load control instruction at the immediately previous moment, and comparing the difference with the target power value expected to be achieved by the load control instruction at the immediately previous moment. The autonomous deviation is expressed in the form of a percentage.

[0010]

[0011] Among them, SDDP(t) is the self-discipline deviation; P ACI (t) is the target power value expected to be achieved by the load control instruction; P(t+T0) is the actual instantaneous power value at the specified time; t is the immediately previous time, and T0 is the response time of the load control instruction;

[0012] The process of defining the degree of self-discipline according to the self-discipline deviation is specifically as follows:

[0013] According to the set confidence level α%, construct a confidence interval based on this confidence level:

[0014] [-SDDP maxpre (t),SDDP maxpre (t)],t∈[0,T]

[0015] Among them, SDDP maxpre (t) is the maximum self-discipline deviation in the confidence interval with a built-in confidence level of α%; T represents the time period; the width of the confidence interval is:

[0016] Width(t)=2×SDDP maxpre (t)

[0017] Where Width(t) is the confidence interval width;

[0018] Define autonomy as:

[0019] SDP(t)=1-Width(t),t∈[0,T]

[0020] Among them, SDP(t) is the degree of self-discipline in the time period T;

[0021] Step 2: Based on the autonomy deviation and the autonomy degree, the autonomy indexes of different time scales are defined as comprehensive quantitative indicators of the autonomy performance of industrial loads; the autonomy indexes of different time scales include real-time autonomy index, sub-real-time autonomy index, intraday scheduling autonomy index and day-ahead autonomy index, which correspond to the control instructions of real-time control, sub-real-time control, intraday scheduling and day-ahead scheduling respectively;

[0022] The calculation formula of real-time self-discipline index is:

[0023]

[0024] RSDI(t) is the real-time self-discipline index, which is a value between 0 and 1. The larger the value, the higher the self-discipline level of the control system. R (t) is the real-time autonomy, CLP R (t) is the real-time confidence level, T2 is the time scale corresponding to the control instruction, SDDP(t+T0) is the autonomy deviation after time T2, and T1 is the time it takes for the autonomy deviation to fall back to the corresponding confidence interval;

[0025] The calculation formula of the sub-real-time self-discipline index is:

[0026]

[0027] Among them, SSDI(t) is the sub-real-time autonomy index, which is a value between 0 and 1. The larger the value, the higher the autonomy level of the control system; SDP S (t) is the real-time autonomy, CLP S (t) is the sub-real-time confidence level, a0 is the frequency stability coefficient, the larger a0 is, the more sensitive the sub-real-time autonomy index is to frequency fluctuations; Freq(t) is the voltage frequency at time t after the control command is issued;

[0028] The calculation formula of the intraday scheduling self-discipline index is:

[0029]

[0030] Among them, ISDI(t) is the intraday self-discipline index, which is a value between 0 and 1. The larger the value, the higher the level of self-discipline of the control system. I (t) is the intraday self-discipline; CLP I(t) is the intraday confidence level, a1 is the load stability coefficient, the larger a1 is, the more sensitive the intraday self-discipline index is to the fluctuation of self-discipline deviation;

[0031] The calculation formula of the day-ahead self-discipline index is:

[0032]

[0033] Among them, DSDI(t) is the day-ahead self-discipline index, which is a value between 0 and 1. The larger the value, the higher the level of self-discipline of the control system; SDP D (t) is the day-ahead autonomy, CLP D (t) is the day-ahead confidence level, a2 is the maximum autonomy deviation peak amplification factor, the larger a2 is, the more sensitive the day-ahead autonomy index is to the maximum autonomy deviation peak; max(SDDP(t)) is the maximum autonomy deviation peak, i.e., the maximum value of autonomy deviation;

[0034] Step 3: Collect the self-discipline deviation data of industrial load equipment within n historical days; the self-discipline deviation data includes voltage, current, power factor and frequency;

[0035] Step 4: preprocessing the collected autonomous deviation data to obtain preprocessed autonomous deviation data;

[0036] The preprocessing method is specifically as follows: firstly, the self-discipline deviation data is cleaned, the existing outliers are deleted and the missing values ​​are supplemented by the interpolation method, and then the self-discipline deviation data is normalized or standardized;

[0037] Step 5: Calculate the instantaneous value of actual power using the preprocessed autonomous deviation data, and then calculate the autonomous deviation;

[0038] Step 6: Based on the time scale of the control command, the autonomy index of the corresponding scale in step 2 is calculated according to the autonomy deviation calculated in step 5, and then the autonomy level of the control system of the industrial load is evaluated;

[0039] Specifically, a Gaussian kernel function is selected to obtain the probability density distribution of the self-discipline deviation, and a confidence interval under a set confidence level is determined according to the probability density distribution of the self-discipline deviation, and then the width of the confidence interval is determined to calculate the degree of self-discipline, and then the self-discipline index of the corresponding scale is calculated according to the calculation formula of the self-discipline index of the corresponding scale in step 2;

[0040]

[0041] Among them, f(x) is the probability density function of the autonomous deviation, h is the bandwidth of the Gaussian kernel function, and x is the independent variable.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] 1. A comprehensive quantitative index of autonomous performance, SDI, is proposed. This index combines the confidence level and autonomy, and can comprehensively evaluate the control effect of the control strategy at different time scales, thereby ensuring that the control strategy is both robust and sophisticated, achieving a balance between the accuracy and robustness of the control effect.

[0044] 2. An autonomous performance evaluation method at different time scales is proposed. Corresponding autonomous performance evaluation methods are proposed for real-time control, sub-real-time control, intraday scheduling and day-ahead scheduling. The execution efficiency, robustness and accuracy of the control strategy are taken into consideration, so as to comprehensively evaluate the autonomous level of the system at different time scales.

[0045] 3. The non-parametric kernel density estimation method is used to effectively capture and characterize the distribution characteristics of industrial load self-discipline deviation, which is not affected by the specific probability distribution form. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of a method for evaluating industrial load self-discipline constraints under multiple time scales in an embodiment of the present invention;

[0047] Figure 2 It is an actual load fluctuation diagram of a certain place on a certain day in an embodiment of the present invention;

[0048] Figure 3 It is an expected load fluctuation diagram of a certain place on a certain day in an embodiment of the present invention;

[0049] Figure 4 This is a schematic diagram of the fluctuation of the self-discipline deviation within 24 hours at a certain place on a certain day in an embodiment of the present invention;

[0050] Figure 5 A schematic diagram of a probability density function of self-discipline deviation generated according to self-discipline deviations of several days in an embodiment of the present invention;

[0051] Figure 6 It is a tree diagram of self-discipline indicators in an embodiment of the present invention;

[0052] Figure 7 This is a schematic diagram of the 24-hour fluctuation of the autonomy deviation of a certain day with different confidence zones marked in an embodiment of the present invention;

[0053] Figure 8 The figure is a schematic diagram of the probability density distribution of the self-discipline deviation of a certain day for 24 hours with different confidence intervals and different confidence levels marked in the embodiment of the present invention;

[0054] Fig. 9 The hourly fluctuation of the self-discipline deviation of a certain place on a certain day in the embodiment of the present invention;

[0055] Fig.10It is the maximum autonomous deviation peak per hour at a certain place on a certain day in the embodiment of the present invention. DETAILED DESCRIPTION

[0056] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0057] The key technical points of the present invention are:

[0058] 1. Definition of autonomy deviation and autonomy: The autonomy deviation is defined to quantify the gap between the actual power of industrial load and the preset control instructions in different time periods. The concepts of autonomy and confidence level are introduced to comprehensively consider the accuracy and consistency of the load model when executing control instructions.

[0059] 2. Use non-parametric kernel density estimation technology: It can effectively capture and characterize the distribution characteristics of industrial load self-discipline deviation and is not restricted by specific probability distribution forms.

[0060] 3. Introduction of the self-discipline index: A comprehensive quantitative indicator of self-discipline performance, the self-discipline index (SDI), was introduced, which comprehensively considers the confidence level and self-discipline, and balances the accuracy and robustness of the control effect.

[0061] 4. Evaluation of autonomous performance at different time scales: Corresponding autonomous performance evaluation methods are proposed for real-time control, sub-real-time control, intraday scheduling and day-ahead scheduling, taking into account the execution efficiency, robustness and accuracy of the control strategy, so as to ensure the autonomous level of the system at different time scales.

[0062] The present invention provides an embodiment, taking the load of a factory in a certain place as an example, such as Figure 1 As shown, a method for evaluating industrial load self-discipline constraints under multiple time scales includes the following steps:

[0063] Step 1: Define self-discipline bias and define self-discipline based on self-discipline bias;

[0064] In the field of industrial load control, the volatility of industrial load is caused by factors such as process changes, grid connection of new energy, equipment start-up and shutdown, state changes, and various random events. Under the premise of ensuring that the basic constraints such as product output, quality, and process stability in the industrial production process are met, it is particularly important to quantify the changes in actual load power in different time periods. To this end, we introduced a new performance indicator - self-discipline deviation.

[0065] Specifically, the self-regulatory bias is calculated by Figure 2 The difference between the actual instantaneous power value of the industrial load at a specified moment and the target power value expected to be achieved by the load control instruction at the immediately previous moment is calculated. Figure 3Compared with the target power value expected to be achieved by the load control instruction at the immediately previous moment, the following is obtained: Figure 4 The relative value shown reflects the response degree and execution efficiency of the industrial load to the load control instruction at that moment. The self-discipline deviation is expressed in the form of percentage:

[0066]

[0067] Among them, SDDP(t) is the Self-Discipline Deviation Percentage; P ACI (t) is the target power value expected to be achieved by the load control instruction (Actual Control Instruction); P(t+T0) is the actual instantaneous power value at the specified time; t is the immediately previous time, and T0 is the response time of the load control instruction;

[0068] The introduction of this self-discipline deviation provides a more sophisticated and flexible analytical method for industrial load control, enabling the control system to respond more quickly to dynamic changes in industrial production. By monitoring and adjusting the self-discipline deviation, the load balance and stability can be achieved more effectively, thereby ensuring that the key indicators in the production process are optimally met and maintained.

[0069] Furthermore, in order to more accurately quantify the credibility level of the industrial load self-discipline deviation at a given moment, the definition of confidence interval in statistics is introduced.

[0070] Specifically, according to the set confidence level α%, a confidence interval based on the confidence level is constructed, which means that in repeated observations and multiple experiments, there is at least a probability of finding that the actual measured autonomous deviation will fall within a predetermined interval. This interval defined by the probability α% is the confidence interval, and it is clearly expressed in the technical solution as follows:

[0071] [-SDDP maxpre (t),SDDP maxpre (t)],t∈[0,T]

[0072] Among them, SDDP maxpre (t) is the maximum self-discipline deviation in the confidence interval with a built-in confidence level of α%; T represents the sampling time period;

[0073] The width of the confidence interval is equal to twice the maximum autonomous deviation, that is:

[0074] Width(t)=2×SDDP maxpre (t)

[0075] Where Width(t) is the confidence interval width;

[0076] Therefore, the confidence level of the confidence interval in a certain time period T is α% and is recorded as:

[0077] p(SDDP(t)∈[-SDDP maxpre (t),SDDP maxpre (t)])=α%,t∈[0,T])

[0078] Among them, p(SDDP(t)∈[-SDDP maxpre (t),SDDP maxpre (t)]) is the probability that the self-regulatory deviation falls into the confidence interval;

[0079] The confidence interval at the confidence level α% is as follows Figure 5 As shown;

[0080] Furthermore, in order to measure the autonomous regulation capability of the power system and the degree to which it meets the preset goals, the present invention introduces a new concept that reflects the level of autonomous regulation of the control system - autonomy. In a certain period of time, the smaller the confidence interval width of the autonomy deviation, the higher the degree of autonomy of the power system; otherwise, the lower the degree of autonomy.

[0081] Therefore, the degree of autonomy is defined as:

[0082] SDP(t)=1-Width(t),t∈[0,T]

[0083] Among them, SDP(t) is the degree of self-discipline in the time period T, in percentage form;

[0084] This indicator quantifies the accuracy and consistency of the control system's response to changes in load control instructions. On the basis of ensuring that the output and product quality required by the production task meet the standards, when the self-discipline value approaches the ideal state of 1, it means that the control system exhibits a higher degree of self-discipline, that is, the industrial load can more accurately follow and adapt to the changing dynamics of the load control instructions, thereby reflecting excellent load self-adjustment capabilities and overall operating efficiency.

[0085] Step 2: Based on the autonomy deviation and the autonomy degree, the autonomy indexes of different time scales are defined as comprehensive quantitative indicators of the autonomy performance of industrial loads; the autonomy indexes of different time scales include real-time autonomy index, sub-real-time autonomy index, intraday scheduling autonomy index and day-ahead autonomy index, which correspond to the control instructions of real-time control, sub-real-time control, intraday scheduling and day-ahead scheduling respectively;

[0086] like Figure 6As shown in the figure, the Self-discipline Index (SDI): uses the F1 score as the core evaluation indicator to measure the self-discipline level of industrial loads. It not only considers the confidence level to reflect the robustness of control, but also combines the degree of self-discipline to reflect the fineness of control, and achieves a balance between accuracy and robustness through weighted harmonic average;

[0087]

[0088] Among them, SDI(t) is the autonomy index, which is a value between 0 and 1. The larger the value of SDI(t), the higher the level of autonomy of the control system;

[0089] Figure 7 Different confidence intervals are marked. Figure 8 The confidence levels of different confidence intervals are marked, and the self-discipline index under the corresponding self-discipline degree is shown in Table 1:

[0090] Table 1 Confidence level and self-discipline under different confidence intervals

[0091]

[0092] The evaluation of the autonomous performance of real-time control is aimed at the control time scale between milliseconds and seconds. The main goal of real-time control is to quickly suppress the accidental transient disturbances of the power grid and ensure the dynamic stability and voltage stability of the control system. Therefore, when conducting the evaluation, attention should be paid to the execution efficiency of the control strategy in practical applications, such as the suppression effect on negative indicators such as frequency fluctuations and voltage drops, and the time required for the control system to return to the expected normal state.

[0093] The calculation formula of real-time self-discipline index is:

[0094]

[0095] RSDI(t) is the real-time self-discipline index, which is a value between 0 and 1. The larger the value, the higher the self-discipline level of the control system. R (t) is the real-time autonomy, CLP R (t) is the real-time confidence level, T2 is the time scale corresponding to the control instruction, SDDP(t+T0) is the autonomous deviation after time T2, and T1 is the time it takes for the autonomous deviation to fall back to the corresponding confidence interval (after a disturbance or unexpected situation occurs, the autonomous deviation will exceed the confidence interval, and then the time it takes from exceeding the confidence interval to falling back to the confidence interval);

[0096] The evaluation of the autonomous performance of sub-real-time control is aimed at the control time scale between seconds and minutes. Sub-real-time control is mainly responsible for achieving short-term power balance and frequency regulation. The evaluation needs to examine the adaptability of the control strategy in the face of load forecast errors or changes in market conditions. Specifically, it includes whether it can effectively maintain frequency stability and optimize power generation costs.

[0097] The calculation formula of the sub-real-time self-discipline index is:

[0098]

[0099] Among them, SSDI(t) is the sub-real-time autonomy index, which is a value between 0 and 1. The larger the value, the higher the autonomy level of the control system; SDP S (t) is the real-time autonomy, CLP S (t) is the sub-real-time confidence level, a0 is the frequency stability coefficient, the larger a0 is, the more sensitive the sub-real-time autonomy index is to frequency fluctuations; Freq(t) is the voltage frequency at time t after the control command is issued;

[0100] The evaluation of the autonomous performance of intraday dispatch is aimed at the control time scale between minutes and hours. The primary goal of intraday dispatch is to effectively respond to changes in daily load demand and the uncertainty of load forecast errors. During the evaluation process, it is necessary to analyze in detail whether the proposed dispatch scheme has the ability to make timely adjustments based on the rolling updated load forecast and whether these adjustments can be implemented with a rapid response speed; pay attention to whether the control system can achieve accurate tracking and control of the load to ensure efficient load management and stable power supply services in daily operations. Evaluating the effectiveness of the autonomous performance of intraday dispatch requires comprehensive consideration of the flexibility and response speed of the dispatch system to ensure that it can maintain stable and reliable operation in a dynamic load environment. The hourly autonomous deviation fluctuations are as follows: Fig. 9 shown.

[0101] The calculation formula of the intraday scheduling self-discipline index is:

[0102]

[0103] Among them, ISDI(t) is the intraday self-discipline index, which is a value between 0 and 1. The larger the value, the higher the level of self-discipline of the control system. I (t) is the intraday self-discipline; CLP I (t) is the intraday confidence level, a1 is the load stability coefficient, the larger a1 is, the more sensitive the intraday self-discipline index is to the fluctuation of self-discipline deviation;

[0104] The evaluation of the autonomous performance of the day-ahead dispatch is aimed at the control time scale between the hourly level and the daily level. The main goal of the day-ahead dispatch is to ensure that the load demand is met while the supply and demand are balanced by scientifically and rationally arranging the operation plans of various types of generating units, energy storage facilities and grid resources based on the forecast of the power demand and supply situation for the next day or a specific time in the future, and to ensure that the power supply of the power system in each time period accurately matches the actual load demand, so as to avoid frequency fluctuations and grid stability problems caused by power shortages or surpluses. To this end, the maximum autonomous deviation peak is added to comprehensively reflect the autonomous performance. The maximum autonomous deviation peak per hour is as follows: Fig.10 shown.

[0105] The calculation formula of the day-ahead self-discipline index is:

[0106]

[0107] Among them, DSDI(t) is the day-ahead self-discipline index, which is a value between 0 and 1. The larger the value, the higher the level of self-discipline of the control system; SDP D (t) is the day-ahead autonomy, CLP D (t) is the day-ahead confidence level, a2 is the maximum autonomy deviation peak amplification factor, the larger a2 is, the more sensitive the day-ahead autonomy index is to the maximum autonomy deviation peak; max(SDDP(t)) is the maximum autonomy deviation peak, i.e., the maximum value of autonomy deviation;

[0108] Step 3: Collect the self-discipline deviation data of industrial load equipment within n historical days; the self-discipline deviation data includes voltage, current, power factor and frequency;

[0109] In this implementation, self-discipline deviation data is obtained through means such as system logs, smart sensors, and real-time monitoring equipment;

[0110] Step 4: preprocessing the collected autonomous deviation data to obtain preprocessed autonomous deviation data;

[0111] The preprocessing method is specifically as follows: firstly, the self-discipline deviation data is cleaned, the existing outliers are deleted and the missing values ​​are supplemented by the interpolation method to ensure the consistency and availability of the data, and then the self-discipline deviation data is normalized or standardized to facilitate subsequent statistical analysis;

[0112] Step 5: Calculate the instantaneous value of actual power using the preprocessed autonomous deviation data, and then calculate the autonomous deviation;

[0113] Step 6: Based on the time scale of the control command, the autonomy index of the corresponding scale in step 2 is calculated according to the autonomy deviation calculated in step 5, and then the autonomy level of the control system of the industrial load is evaluated;

[0114] Specifically: adopt non-parametric kernel density estimation technology, select Gaussian kernel function to obtain the probability density distribution of self-discipline deviation, and determine the confidence interval under the set confidence level according to the probability density distribution of self-discipline deviation, and then determine the width of the confidence interval to calculate the degree of self-discipline, and then calculate the self-discipline index of the corresponding scale according to the calculation formula of the self-discipline index of the corresponding scale in step 2;

[0115] The non-parametric kernel density estimation technology does not rely on the prior assumptions about the data distribution form, and is more flexible to adapt to the complex error distribution that may exist in actual industrial systems;

[0116]

[0117] Where f(x) is the probability density function of the autonomous deviation, h is the bandwidth of the Gaussian kernel function, and x is the independent variable;

[0118] Through the above steps, the control system obtains the probability density distribution of the industrial load self-discipline deviation. This method has strong adaptability and feasibility for solving the error problem caused by complexity in industrial systems.

[0119] In this embodiment, based on the selected Gaussian kernel function, the preprocessed historical n-day autonomous deviation data is used to estimate the probability density distribution of the autonomous deviation for each time period within 24 hours. After determining the probability density distribution of the autonomous deviation, the confidence interval at the confidence level α% is obtained.

Claims

1. A method for evaluating industrial load self-discipline constraints under multiple time scales, characterized in that: The steps include: Step 1: Define self-discipline bias and define self-discipline based on self-discipline bias; Step 2: Based on the self-discipline deviation and self-discipline degree, define the self-discipline index of different time scales as the comprehensive quantitative index of the self-discipline performance of industrial loads; Step 3: Collect the self-discipline deviation data of industrial load equipment within n historical days; Step 4: preprocessing the collected autonomous deviation data to obtain preprocessed autonomous deviation data; Step 5: Calculate the instantaneous value of actual power using the preprocessed autonomous deviation data, and then calculate the autonomous deviation; Step 6: Based on the time scale of the control instruction, the autonomy index of the corresponding scale in step 2 is calculated according to the autonomy deviation calculated in step 5, and then the autonomy level of the control system of the industrial load is evaluated.

2. According to the multi-time scale industrial load self-discipline constraint evaluation method of claim 1, it is characterized in that: The autonomous deviation in step 1 is a relative value obtained by calculating the difference between the actual instantaneous power value of the industrial load at a specified moment and the target power value expected to be achieved by the load control instruction at the immediately previous moment, and comparing the difference with the target power value expected to be achieved by the load control instruction at the immediately previous moment. The autonomous deviation is expressed in the form of a percentage. Among them, SDDP(t) is the self-discipline deviation; P ACI (t) is the target power value expected to be achieved by the load control instruction; P(t+T0) is the actual instantaneous power value at the specified time; t is the immediately previous time, and T0 is the response time of the load control instruction; The process of defining the degree of self-discipline according to the self-discipline deviation is specifically as follows: According to the set confidence level α%, construct a confidence interval based on this confidence level: [-SDDP maxpre (t),SDDP maxpre (t)],t∈[0,T] Among them, SDDP maxpre (t) is the maximum self-discipline deviation in the confidence interval with a built-in confidence level of α%; T represents the time period; The confidence interval width is: Width(t)=2×SDDP maxpre (t) Where Width(t) is the confidence interval width; Define autonomy as: SDP(t)=1-Width(t),t∈[0,T] Among them, SDP(t) is the degree of self-discipline in the time period T.

3. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 2 is characterized in that: The autonomy indexes of different time scales described in step 2 include real-time autonomy index, sub-real-time autonomy index, intraday scheduling autonomy index and day-ahead autonomy index, which correspond to the control instructions of real-time control, sub-real-time control, intraday scheduling and day-ahead scheduling respectively.

4. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 1 is characterized in that: The autonomous deviation data in step 3 includes voltage, current, power factor and frequency.

5. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 1 is characterized in that: The preprocessing method described in step 4 is specifically as follows: first, the autonomous deviation data is cleaned, the existing outliers are deleted and the missing values ​​are supplemented by the interpolation method, and then the autonomous deviation data is normalized or standardized.

6. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 3 is characterized in that: The calculation formula of the real-time self-discipline index is: RSDI(t) is the real-time self-discipline index, which is a value between 0 and 1. The larger the value, the higher the self-discipline level of the control system. R (t) is the real-time autonomy, CLP R (t) is the real-time confidence level, T2 is the time scale corresponding to the control instruction, SDDP(t+T0) is the autonomy deviation after time T2, and T1 is the time it takes for the autonomy deviation to fall back to the corresponding confidence interval.

7. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 3 is characterized in that: The calculation formula of the sub-real-time autonomy index is: Among them, SSDI(t) is the sub-real-time autonomy index, which is a value between 0 and 1. The larger the value, the higher the autonomy level of the control system; SDP S (t) is the real-time autonomy, CLP S (t) is the sub-real-time confidence level, a0 is the frequency stability coefficient, the larger a0 is, the more sensitive the sub-real-time autonomy index is to frequency fluctuations; Freq(t) is the voltage frequency at time t after the control command is issued.

8. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 3 is characterized in that: The calculation formula of the intraday scheduling self-discipline index is: Among them, ISDI(t) is the intraday self-discipline index, which is a value between 0 and 1. The larger the value, the higher the level of self-discipline of the control system. I (t) is the intraday self-discipline; CLP I (t) is the intraday confidence level, a1 is the load stability coefficient, and the larger a1 is, the more sensitive the intraday self-discipline index is to the fluctuation of self-discipline deviation.

9. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 3 is characterized in that: The calculation formula of the day-ahead self-discipline index is: Among them, DSDI(t) is the day-ahead self-discipline index, which is a value between 0 and 1. The larger the value, the higher the level of self-discipline of the control system; SDP D (t) is the day-ahead autonomy, CLP D (t) is the day-ahead confidence level, a2 is the maximum autonomy deviation peak amplification coefficient, the larger a2 is, the more sensitive the day-ahead autonomy index is to the maximum autonomy deviation peak; max(SDDP(t)) is the maximum autonomy deviation peak, that is, the maximum value of the autonomy deviation.

10. The method for evaluating industrial load self-discipline constraints under multiple time scales according to claim 1 is characterized in that: The step 6 specifically comprises: selecting a Gaussian kernel function to obtain a probability density distribution of the self-discipline deviation, and determining a confidence interval at a set confidence level according to the probability density distribution of the self-discipline deviation, and then determining the width of the confidence interval to calculate the degree of self-discipline, and then calculating the self-discipline index of the corresponding scale according to the calculation formula of the self-discipline index of the corresponding scale in step 2; Among them, f(x) is the probability density function of the autonomous deviation, h is the bandwidth of the Gaussian kernel function, and x is the independent variable.

Images