Demand management control method and system based on real-time rolling prediction and decision analysis

Abstract

The invention discloses a demand management control method and system based on real-time rolling prediction and decision analysis, and relates to the field of energy scheduling, comprising load prediction of a demand gateway line and power demand trend analysis of a controlled object. Determining a power demand control strategy by using a load prediction result and a power demand trend analysis result and through gate demand periodical rolling prediction and optimization calculation; constructing a sensitivity matrix between the power variation between the demand gateway and the load line and the power variation of the load line; using the sensitivity matrix to provide a demand out-of-limit value of a demand gateway line and support construction of a demand out-of-limit analysis control strategy; and performing power calculation by combining an equipment energy efficiency model and line load characteristics, and dynamically adjusting an energy distribution strategy of each piece of equipment. According to the invention, accurate prediction and dynamic adjustment of energy demands can be realized, energy distribution can be optimized, and operation cost can be reduced.

Description

Technical Field

[0001] The present invention relates to the field of energy scheduling. Specifically, it relates to a demand management control method and system based on real-time rolling prediction and decision analysis. Background Art

[0002] The basic electricity charges of industrial users can be measured by the installed capacity of transformers or according to the maximum demand. For enterprises adopting "maximum demand", they should reasonably arrange the operation modes of each process, try to avoid the simultaneous startup or operation of high-power equipment, and especially effectively control the processes with impact load characteristics to reduce the peak load of the enterprise and lower the electricity charge expenditure. Against the background of complex industrial user processes and complex electricity consumption compositions, it is difficult to continue relying on manual demand control. The demand analysis and control technology based on real-time rolling prediction can assist industrial users in improving the level of demand management control.

[0003] At the same time, with the development of smart grid and Internet of Things technologies, energy monitoring systems have become an important tool for modern energy management. However, existing systems mostly adopt static or periodic prediction models, such as time series analysis, machine learning models, etc. These methods have certain limitations in prediction accuracy and real-time performance. Especially in the face of a rapidly changing energy market and equipment status, their prediction results often lag behind the actual demand, making it difficult to achieve efficient energy management.

[0004] In the prior art, for example, Chinese Patent CN111342459B discloses a power demand decision analysis system and method, including a computational experiment unit, an evaluation and analysis unit, a control strategy execution unit, etc., which realizes the calculation of demand and over-limit early warning, and continuously optimizes the calculation of the demand over-limit values at each gateway by combining the demand change trends and historical data at each gateway. However, the above power demand decision analysis system and method still have the following deficiencies:

[0005] 1) Limitations of the feedback mechanism: There may be a certain lag in the external feedback and internal feedback mechanisms of the system. When sudden changes or abnormal situations occur in the power system, the feedback mechanism may not be able to respond and adjust in time, resulting in the lag and inaccuracy of the decision analysis results. 2) Adaptability and flexibility issues: The system is optimized for specific power systems or demand scenarios during design. However, with the continuous development and change of the power system, the adaptability and flexibility of the system may be restricted; in addition, the system may lack sufficient flexibility to meet the specific needs of different users or regions. In actual applications, it may be necessary to carry out customized and personalized development and adjustment according to specific situations. 3) Cost and investment issues: The construction and maintenance of the system require a large amount of capital, manpower, and material resources; in addition, the operation and maintenance of the system also require professional technical teams and personnel support. If the corresponding technical support and talent reserves are lacking, it may lead to problems such as unstable system operation or difficult maintenance.

[0006] In addition, the disadvantages of the existing related technologies are mainly reflected in the following aspects:

[0007] Firstly, the prediction accuracy is insufficient and it is unable to accurately reflect real-time changes.

[0008] Secondly, the response is lagged and it is difficult to adjust the energy distribution strategy in a timely manner.

[0009] Thirdly, the system complexity is high and the operation and maintenance cost is large.

[0010] For the problems in the related technologies, no effective solutions have been proposed yet. Summary of the Invention

[0011] In view of the problems in the related technologies, the present invention provides a demand management control method and system based on real-time rolling prediction and decision analysis to overcome the above-mentioned technical problems existing in the existing related technologies.

[0012] Therefore, the specific technical solutions adopted by the present invention are as follows:

[0013] According to one aspect of the present invention, a demand management control method based on real-time rolling prediction and decision analysis is provided, including:

[0014] Inputting historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction on the demand metering point line; according to the process characteristics of the controlled object, using the load prediction result as the input to perform power demand trend analysis of the controlled object; using the load prediction result and the power demand trend analysis result, and through periodic rolling prediction and optimization calculation of the metering point demand, determining the power demand control strategy;

[0015] Analyzing the power change relationship between the demand metering point and the load line through the power perturbation method and power flow calculation, and constructing a sensitivity matrix between the power change amount between the demand metering point and the load line and the power change amount of the load line; using the sensitivity matrix to provide the demand metering point line's demand exceeding the limit value and supporting the construction of a demand over-limit analysis and control strategy;

[0016] Combining the equipment energy efficiency model and the line load characteristics to perform power calculation and dynamically adjusting the energy distribution strategy of each device.

[0017] Further, inputting historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction on the demand metering point line includes:

[0018] Obtaining power demand load data, and calculating the active power of the demand load through a sliding window; calculating the real-time demand of the metering point line through the active power of the demand load; obtaining data related to power demand load characteristic factors;

[0019] Input the active power of the demand load, the real-time demand of the gateway line, the power demand load data, and the data related to the power demand load characteristic factors into the prediction model to predict the power load of the future power consumption process of the demand gateway line.

[0020] Furthermore, the construction of the prediction model includes:

[0021] According to the load characteristics and operating conditions of different processes, adopt the method of preferentially switching between multiple algorithm models and prediction schemes, so that the prediction model supports the load prediction of various equipment;

[0022] Among them, the method of preferentially switching between multiple algorithm models and prediction schemes includes:

[0023] Combine multiple models and algorithms to form a prediction model library, and use self-learning simulation training technology and artificial experience for optimization to form a comprehensive model reflecting the load change law.

[0024] Furthermore, according to the process characteristics of the controlled object, taking the load prediction result as the input, the power demand trend analysis of the controlled object includes:

[0025] Based on the process characteristics of the controlled object, divide the process of the controlled object into the rising edge, the falling edge, and the middle section;

[0026] In the rising edge and falling edge stages, use the similarity measurement method to take several groups of data with the highest similarity in the historical data of the controlled object as the prediction samples, and use the trend extrapolation method to correct the trend analysis curve of the power demand of the controlled object;

[0027] In the middle section, use the similarity measurement method to take several groups of data with the highest similarity in the historical data of the controlled object as the prediction samples, and use the linear regression method for trend analysis.

[0028] Furthermore, use the load prediction result and the power demand trend analysis result, and through the periodic rolling prediction and optimization calculation of the gateway demand, determine the power demand control strategy, including:

[0029] Through the load prediction result and the power demand trend analysis result, periodically roll-calculate the power distribution to perform periodic rolling prediction on the gateway demand;

[0030] If the periodic rolling prediction value of the gateway demand is greater than the set warning threshold, perform an alarm process;

[0031] If the periodic rolling prediction value of the gateway demand is greater than the assessment value, obtain the mapping relationship between the gateway and the load, and calculate the mapping relationship between the maximum adjustment margin of each load and the change value of the gateway demand under the condition of meeting the safety margin conditions of each load adjustment, and sort them in descending order according to the adjustment margin of each load;

[0032] When adjusting the load, accumulate the adjustment amount of the demand at the gateway until it is greater than the total adjustment amount, and generate a power demand control strategy;

[0033] According to the load adjustment amount of each line, conduct a security check through power flow calculation. After passing the security check, issue the power demand control strategy.

[0034] Furthermore, analyze the relationship between the power changes of the demand gateway and the load lines through the power perturbation method and power flow calculation, and construct a sensitivity matrix between the power change amount between the demand gateway and the load lines and the power change amount of the load lines, including:

[0035] Take the current active power of the load line as the first reference value, and use the power perturbation method to change the size of the first reference value;

[0036] After each change in the active power of the load line, enable power flow calculation to obtain the magnitude of the change in the active power of the demand gateway line;

[0037] According to the power flow calculation results, sort the influence degrees of all load lines related to the demand gateway in descending order, and optimize the demand exceeding the limit value of the demand gateway line according to the safety adjustment margin of the load line and in combination with the economic evaluation results;

[0038] Through sensitivity analysis, establish a sensitivity matrix between the power change amount between the demand gateway and the load lines and the power change amount of the load lines;

[0039] Among them, using the power perturbation method to change the size of the first reference value includes: increasing or decreasing continuously by a fixed value upward and downward respectively until reaching the upper and lower limits of the first reference value.

[0040] Furthermore, the power flow calculation includes:

[0041] Construct node variables and node power equations, and the node variables include the active power consumed by the load, the reactive power consumed by the load, the active power of the generator, the reactive power of the generator, the voltage magnitude of the node, and the phase angle of the node. The node power equations include active power equations and reactive power equations;

[0042] Construct algorithms for power flow calculation, including the Gauss - Seidel power flow calculation method, the Newton - Raphson power flow calculation method, and the PQ decomposition power flow calculation method.

[0043] Furthermore, using the sensitivity matrix to provide the demand exceeding the limit value of the demand gateway line includes:

[0044] According to the coefficients in the sensitivity matrix, combined with the adjustable upper and lower limit constraints of the load line, evaluate the adjustable ability of the load line, and judge whether the demand exceeding the limit value meets the safety margin evaluation;

[0045] When the safety margin assessment cannot be met, the power perturbation method is used to adjust the demand exceeding the limit value of the demand gateway line, and the safety margin assessment is carried out again;

[0046] Taking the current active power of the demand gateway line as the second reference value, the power perturbation method is used to change the magnitude of the second reference value; after the active power of the demand gateway line changes, the power flow calculation is enabled, and the change magnitude of the load lines related to the demand gateway line is recorded;

[0047] Taking the current active power of the load line as the third reference value, the power perturbation method is used to change the active power magnitudes of all load lines; after each change in the active power of the load line, the power flow calculation is enabled, and the change magnitude of the active power of the demand gateway line is recorded;

[0048] Based on the results of the power perturbation method and the power flow calculation, the influence degrees of all load lines related to the demand gateway line are sorted from large to small, and the specific upper and lower adjustment limits are calculated according to the correlation degree of the load line to the demand gateway line;

[0049] According to the safety adjustment margin of the load line and combined with the economic evaluation results, the demand exceeding the limit value of each demand gateway line is optimized.

[0050] Furthermore, the support for constructing the demand over-limit analysis and control strategy includes:

[0051] Using the sensitivity matrix to describe the influence degree of the change in the active power of the load line on the demand gateway power, providing technical support for the formulation of the demand over-limit analysis and control strategy.

[0052] Furthermore, using the sensitivity matrix to describe the influence degree of the change in the load line power on the demand gateway power includes:

[0053] The nonlinear equations of the power system are expanded in a Taylor series near the demand gateway to obtain a linearized system of equations, simplifying the problem into a solvable form;

[0054] In the linearized system of equations, by solving the Jacobian matrix, the sensitivity relationship between variables is obtained;

[0055] According to the relevant elements in the Jacobian matrix, a power sensitivity matrix between the demand gateway and the change in the load line power is constructed, where the element represents the influence degree on the change in the demand gateway power when the load line power changes.

[0056] Furthermore, the demand over-limit analysis and control strategy includes:

[0057] When the calculated value of the real-time demand at the demand metering point line exceeds the demand warning threshold, an overlimit warning signal is issued. At the same time, the trend analysis function is started to predict and analyze the trend changes of the load lines related to the demand metering point line;

[0058] When a single metering point has an overlimit warning, single metering point adjustment measures are started; when multiple metering points have overlimit warnings, multi-metering point adjustment measures are started;

[0059] Using power flow calculation of the power grid and adopting the incremental perturbation method, the adjustment amounts of each load line related to the demand metering point line in the demand overlimit analysis and control strategy are calculated, and the size of the adjustment amount is continuously adjusted according to the calculation results.

[0060] Furthermore, during demand management control, it also includes: inversion evaluation of demand overlimit;

[0061] The inversion evaluation of demand overlimit includes:

[0062] When the demand overlimit alarm is triggered, the historical energy consumption data, the current energy consumption pattern and the preset limit value are analyzed through the inversion module to obtain the reasons for the demand overlimit;

[0063] According to the reasons for the demand overlimit, corresponding corrective measures are formulated.

[0064] Furthermore, power calculation is carried out in combination with the equipment energy efficiency model and the line load characteristics, and dynamically adjusting the energy distribution strategy of each device includes:

[0065] Based on the energy efficiency data and historical operation data of the equipment, an equipment energy efficiency model is established;

[0066] Based on the load data of the line, a line load characteristic model is established to describe the changes of the load at different time periods;

[0067] According to the equipment energy efficiency model and the line load characteristic model, the energy consumption of each device under different loads is calculated, and the energy distribution strategy of each device is determined in combination with the energy efficiency and load demand of the device.

[0068] According to another aspect of the present invention, there is also provided a demand management control system based on real-time rolling prediction and decision analysis, which includes: a power demand control module, a demand overlimit analysis and control module, and an energy distribution module;

[0069] The power demand control module is used to input historical load data and data related to power demand load characteristic factors into a prediction model to conduct load prediction for the load lines at the demand metering points; based on the process characteristics of the controlled object, use the load prediction result as the input to conduct power demand trend analysis of the controlled object; utilize the load prediction result and the power demand trend analysis result, and through periodic rolling prediction and optimization calculation of the metering point demand, determine the power demand control strategy.

[0070] The demand over-limit analysis and control module is used to analyze the relationship between the power changes at the demand metering points and the load lines through the power perturbation method and power flow calculation, and construct a sensitivity matrix between the power change amount between the demand metering points and the load lines and the power change amount of the load lines; utilize the sensitivity matrix to provide the demand over-limit setting value for the load lines at the demand metering points and support the construction of the demand over-limit analysis and control strategy.

[0071] The energy distribution module is used to perform power calculation by combining the equipment energy efficiency model and the line load characteristics, and dynamically adjust the energy distribution strategy of each device.

[0072] The beneficial effects of the present invention are as follows:

[0073] (1) High prediction accuracy: Through real-time rolling prediction, ensure that the prediction result is always close to the actual demand, and improve the prediction accuracy.

[0074] (2) Fast response speed: The system can quickly respond to market changes and equipment status fluctuations, and timely adjust the energy distribution strategy.

[0075] (3) Energy efficiency improvement: Through accurate power calculation and dynamic optimization, achieve refined management of energy use and maximize energy efficiency.

[0076] (4) Cost reduction: Optimize the energy distribution strategy, reduce energy waste, and lower the operating cost.

[0077] (5) Strong flexibility: The system architecture is modular, easy to expand and maintain, and can adapt to the energy management system requirements of different scales and types. Description of the Drawings

[0078] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0079] Figure 1 It is the architecture diagram of the demand analysis system according to the embodiment of the present invention;

[0080] Figure 2It is a composition diagram of the demand analysis control system function modules according to an embodiment of the present invention;

[0081] Figure 3 It is an operation flowchart of the power demand analysis control system according to an embodiment of the present invention;

[0082] Figure 4 It is a technical roadmap of the power demand analysis control system according to an embodiment of the present invention;

[0083] Figure 5 It is a diagram of the data acquisition and processing module according to an embodiment of the present invention;

[0084] Figure 6 It is a schematic diagram of the data acquisition and demand calculation time interval according to an embodiment of the present invention;

[0085] Figure 7 It is a key load trend prediction diagram according to an embodiment of the present invention;

[0086] Figure 8 It is a schematic diagram of the arc furnace intermittent impact load curve model according to an embodiment of the present invention;

[0087] Figure 9 It is a flowchart of the demand analysis control strategy based on real-time rolling prediction and power calculation according to an embodiment of the present invention;

[0088] Figure 10 It is a flowchart of power calculation and energy distribution according to an embodiment of the present invention;

[0089] Figure 11 It is a function flowchart of state estimation, dispatcher's power flow, and sensitivity calculation of the power grid dispatching system according to an embodiment of the present invention;

[0090] Figure 12 It is a schematic diagram of network topology analysis according to an embodiment of the present invention;

[0091] Figure 13 It is a flowchart of the demand analysis optimization control strategy according to an embodiment of the present invention;

[0092] Figure 14 It is a flowchart of demand limit inversion according to an embodiment of the present invention;

[0093] Figure 15 It is a principle block diagram of the event recording and alarm module according to an embodiment of the present invention;

[0094] Figure 16 It is a flowchart of a demand management control method based on real-time rolling prediction and decision analysis according to an embodiment of the present invention;

[0095] Figure 17It is a block diagram of a demand management control system based on real-time rolling prediction and decision analysis according to an embodiment of the present invention.

[0096] In the figure:

[0097] 1. Power demand control module; 2. Demand limit analysis and control module; 3. Energy distribution module. Specific implementation manners

[0098] To further illustrate the embodiments, the present invention provides accompanying drawings. These drawings are part of the disclosure of the present invention, mainly used to illustrate the embodiments, and can be used in conjunction with the relevant descriptions in the specification to explain the operating principle of the embodiments. With reference to these contents, those of ordinary skill in the art should be able to understand other possible implementation manners and the advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are usually used to represent similar components.

[0099] According to an embodiment of the present invention, a demand management control method and system based on real-time rolling prediction and decision analysis are provided.

[0100] Now, the present invention will be further described in conjunction with the accompanying drawings and specific implementation manners. As Figure 16 shown, according to an embodiment of the present invention, a demand management control method based on real-time rolling prediction and decision analysis is provided, including:

[0101] S1. Input historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction on the demand checkpoint line; according to the process characteristics of the controlled object, use the load prediction result as input to perform power demand trend analysis of the controlled object; utilize the load prediction result and the power demand trend analysis result, and through periodic rolling prediction and optimization calculation of the checkpoint demand, determine the power demand control strategy.

[0102] S2. Analyze the power change relationship between the demand checkpoint and the load line through the power perturbation method and power flow calculation, and construct a sensitivity matrix between the power change amount between the demand checkpoint and the load line and the power change amount of the load line; utilize the sensitivity matrix to provide the demand limit value of the demand checkpoint line and support the construction of a demand limit analysis and control strategy.

[0103] S3. Combine the device energy efficiency model and the line load characteristic to perform power calculation, and dynamically adjust the energy distribution strategy of each device.

[0104] In one embodiment, inputting historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction on the demand checkpoint line includes:

[0105] Obtain the power demand load data, and calculate the active power of the demand load in the form of a sliding window; calculate and obtain the real-time demand of the gateway line through the active power of the demand load; obtain the data related to the power demand load characteristic factors.

[0106] Input the active power of the demand load, the real-time demand of the gateway line, the power demand load data, and the data related to the power demand load characteristic factors into the prediction model to predict the power consumption load of the future power consumption process of the demand gateway line.

[0107] In one embodiment, the construction of the prediction model includes:

[0108] According to the load characteristics and operating conditions of different processes, adopt the method of preferentially switching between multiple algorithm models and prediction schemes, so that the prediction model supports the load prediction of various equipment.

[0109] Among them, the method of preferentially switching between multiple algorithm models and prediction schemes includes:

[0110] Combine multiple models and algorithms to form a prediction model library, and use self-learning simulation training technology and artificial experience for optimization to form a comprehensive model reflecting the load change law.

[0111] In one embodiment, according to the process characteristics of the controlled object, taking the load prediction result as the input, the power demand trend analysis of the controlled object includes:

[0112] Based on the process characteristics of the controlled object, divide the process of the controlled object into the rising edge, the falling edge, and the middle section.

[0113] In the rising edge and falling edge stages, use the similarity measurement method to take several groups of data with the highest similarity in the historical data of the controlled object as the prediction samples, and use the trend extrapolation method to correct the trend analysis curve of the power demand of the controlled object.

[0114] In the middle section, use the similarity measurement method to take several groups of data with the highest similarity in the historical data of the controlled object as the prediction samples, and use the linear regression method for trend analysis.

[0115] In one embodiment, using the load prediction result and the power demand trend analysis result, and through the periodic rolling prediction and optimization calculation of the gateway demand, determining the power demand control strategy includes:

[0116] Through the load prediction result and the power demand trend analysis result, periodically roll-calculate the power distribution to perform periodic rolling prediction on the gateway demand.

[0117] If the periodic rolling prediction value of the gateway demand is greater than the set warning threshold, perform an alarm process.

[0118] If the periodic rolling prediction value of the checkpoint demand is greater than the assessment value, obtain the mapping relationship between the checkpoint and the load, and calculate the mapping relationship between the maximum adjustment margin of each load and the change value of the checkpoint demand under the condition of meeting the safety margin conditions of each load adjustment, and sort them in descending order according to the adjustment margin of each load.

[0119] When adjusting the load, accumulate the adjustment amount of the checkpoint demand until it is greater than the total adjustment amount, and generate a power demand control strategy.

[0120] According to the load adjustment amount of each line load, perform a safety check through power flow calculation. After passing the safety check, issue a power demand control strategy.

[0121] In one embodiment, analyze the relationship between the demand checkpoint and the power change of the load line through the power perturbation method and power flow calculation, and construct a sensitivity matrix between the power change amount between the demand checkpoint and the load line and the power change amount of the load line, including:

[0122] Take the current active power of the load line as the first reference value, and use the power perturbation method to change the size of the first reference value.

[0123] After each change in the active power of the load line, enable power flow calculation to obtain the change magnitude of the active power of the demand checkpoint line.

[0124] According to the power flow calculation results, sort the influence degrees of all load lines related to the demand checkpoint in descending order, and optimize the demand exceeding the limit value of the demand checkpoint line according to the safety adjustment margin of the load line and combined with the economic evaluation results.

[0125] Through sensitivity analysis, establish a sensitivity matrix between the power change amount between the demand checkpoint and the load line and the power change amount of the load line.

[0126] Among them, using the power perturbation method to change the size of the first reference value includes: increasing or decreasing respectively upward and downward by a fixed value continuously until reaching the upper and lower limits of the first reference value.

[0127] In one embodiment, the power flow calculation includes:

[0128] Construct node variables and node power equations, and the node variables include the active power consumed by the load, the reactive power consumed by the load, the active power of the generator, the reactive power of the generator, the voltage magnitude of the node, and the phase angle of the node. The node power equations include active power equations and reactive power equations.

[0129] Construct algorithms for power flow calculation, including the Gauss-Seidel power flow calculation method, the Newton-Raphson power flow calculation method, and the PQ decomposition power flow calculation method.

[0130] In one embodiment, using the sensitivity matrix to provide the demand exceeding the limit value of the demand metering point line includes:

[0131] According to the coefficients in the sensitivity matrix, combined with the adjustable upper limit and adjustable lower limit constraints of the load line, evaluate the adjustable ability of the load line, and determine whether the demand exceeding the limit value meets the safety margin evaluation.

[0132] When the safety margin evaluation cannot be satisfied, the power perturbation method is used to adjust the demand exceeding the limit value of the demand metering point line, and the safety margin evaluation is performed again.

[0133] Taking the current active power of the demand metering point line as the second reference value, use the power perturbation method to change the magnitude of the second reference value; after the active power of the demand metering point line changes, enable power flow calculation and record the change magnitude of the load line related to this demand metering point line.

[0134] Taking the current active power of the load line as the third reference value, use the power perturbation method to change the active power magnitude of all load lines; after each change in the active power of the load line, enable power flow calculation and record the change magnitude of the active power of the demand metering point line.

[0135] Based on the results of the power perturbation method and power flow calculation, sort the influence degrees of all load lines related to the demand metering point line from large to small, and calculate the specific upper and lower adjustment limits according to the correlation degree of the load line to the demand metering point line.

[0136] Optimize the demand exceeding the limit value of each demand metering point line according to the safety adjustment margin of the load line and combined with the economic evaluation results.

[0137] In one embodiment, supporting the construction of the demand exceeding limit analysis and control strategy includes:

[0138] Using the sensitivity matrix to describe the influence degree of the change in the active power of the load line on the power of the demand metering point, providing technical support for the formulation of the demand exceeding limit analysis and control strategy.

[0139] In one embodiment, using the sensitivity matrix to describe the influence degree of the change in the load line power on the power of the demand metering point includes:

[0140] Perform Taylor series expansion on the nonlinear equation of the power system near the demand metering point to obtain a linearized equation set, and simplify the problem into a solvable form.

[0141] In the linearized equation set, by solving the Jacobian matrix, obtain the sensitivity relationship between variables.

[0142] According to the relevant elements in the Jacobian matrix, a power sensitivity matrix between the demand metering point and the power change of the load line is constructed, where the element represents the influence degree of the power change of the load line on the power change of the demand metering point.

[0143] In one embodiment, the demand overlimit analysis and control strategy includes:

[0144] When the real-time demand calculation value of the demand metering point line exceeds the demand warning threshold, an overlimit warning signal is sent, and at the same time, the trend analysis function is started to predict and analyze the trend change of the load lines related to the demand metering point line.

[0145] When a single metering point overlimit warning occurs, start the single metering point adjustment measure; when multiple metering point overlimit warnings occur, start the multi-metering point adjustment measure.

[0146] Using power flow calculation of the power grid and adopting the incremental perturbation method, calculate the adjustment amount of each load line related to the demand metering point line in the demand overlimit analysis and control strategy, and continuously adjust the size of the adjustment amount according to the calculation results.

[0147] In one embodiment, during demand management control, it further includes: the inversion evaluation of demand overlimit.

[0148] The inversion evaluation of demand overlimit includes:

[0149] When the demand overlimit alarm is triggered, analyze the historical energy consumption data, the current energy consumption pattern and the preset limit value through the inversion module to obtain the reasons for the demand overlimit.

[0150] According to the reasons for the demand overlimit, formulate corresponding corrective measures.

[0151] In one embodiment, combining the equipment energy efficiency model and the line load characteristics for power calculation and dynamically adjusting the energy distribution strategy of each equipment includes:

[0152] Based on the energy efficiency data and historical operation data of the equipment, establish an equipment energy efficiency model.

[0153] Based on the load data of the line, establish a line load characteristic model to describe the change of the load in different time periods.

[0154] According to the equipment energy efficiency model and the line load characteristic model, calculate the energy consumption of each equipment under different loads, and combine the energy efficiency and load demand of the equipment to determine the energy distribution strategy of each equipment.

[0155] Such as Figure 17As shown, according to another embodiment of the present invention, a demand management control system based on real-time rolling prediction and decision analysis is further provided. The system includes: a power demand control module, a demand overlimit analysis and control module, and an energy distribution module.

[0156] The power demand control module 1 is used to input historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction on the demand checkpoint line; according to the process characteristics of the controlled object, use the load prediction result as the input to perform power demand trend analysis of the controlled object; utilize the load prediction result and the power demand trend analysis result, and through periodic rolling prediction and optimization calculation of the checkpoint demand, determine the power demand control strategy.

[0157] The demand overlimit analysis and control module 2 is used to analyze the power change relationship between the demand checkpoint and the load line through the power perturbation method and power flow calculation, and construct a sensitivity matrix between the power change amount between the demand checkpoint and the load line and the power change amount of the load line; utilize the sensitivity matrix to provide the demand overlimit value of the demand checkpoint line and support the construction of the demand overlimit analysis and control strategy.

[0158] The energy distribution module 3 is used to perform power calculation in combination with the equipment energy efficiency model and the line load characteristics, and dynamically adjust the energy distribution strategy of each device.

[0159] To facilitate the understanding of the above technical solution of the present invention, the working principle of the present invention in the actual process will be described in detail below.

[0160] The present invention aims to solve the problems of inaccurate demand prediction, response lag, and non-precise power management commonly existing in existing energy monitoring systems. Most traditional energy monitoring systems rely on historical data for simple prediction and cannot effectively cope with real-time changes in market electricity prices, equipment load fluctuations, and emergencies, resulting in low energy use efficiency and poor cost control. Therefore, the present invention proposes a demand management control method based on real-time rolling prediction and decision analysis to achieve accurate prediction and dynamic adjustment of power demand, realize peak shaving and valley filling of industrial electricity, optimize electricity charges, and ensure the power supply quality of the power grid. The purpose of the present invention is to provide a control method and system that can perform real-time rolling prediction of energy demand, combine precise power calculation, and achieve dynamic optimization of energy distribution, thereby improving energy use efficiency, reducing operating costs, reducing the peak load of enterprises through real-time rolling prediction-based demand analysis and control technology, reducing electricity charge expenditures, enhancing the energy digital management level of industrial users, and helping users build a green and low-carbon intelligent energy management system.

[0161] I. Architecture of the Demand Analysis Control System

[0162] The demand analysis control system includes physical entities, data models, connection and interaction, and application services. The architecture of the demand analysis control system is as Figure 1 shown.

[0163] 1) Physical entities: In enterprise power supply and distribution, physical entities are centralized control stations and substations, etc., that is, the actual power system of the enterprise, which is closely related to the integration with the data model, the realization of connection and interaction, and the generation of application services.

[0164] 2) Data model: Through automation and informatization means, physical objects are digitalized, systematized, and intelligentized driven by data. Based on the power grid dispatching automation system, data such as power grid monitoring, power grid model, power consumption data, load forecasting, and generation plan are integrated. The above are all based on the analysis of the power grid power flow trend. At the same time, the change of the power flow drives the operation of each part, which is the key to the realization of the function of the digital twin application for power demand decision-making analysis.

[0165] 3) Connection and interaction: Connect data, physical entities, data models, and application services into an organic whole, and through the data bus provided by the power grid dispatching system, information and data can be exchanged and transmitted between each part.

[0166] 4) Application services: Package applications into services and provide them to users in the form of application software, human-machine interfaces, etc. That is, based on the power grid dispatching system, application practices such as demand monitoring and early warning, demand decision-making and analysis are carried out.

[0167] The overall technical route of demand analysis consists of three main links: data collection and demand calculation, load forecasting and trend analysis, and optimization control and decision execution. Through data collection and demand calculation, and load forecasting and trend analysis, the functions of demand calculation and overlimit early warning are realized; then through optimization control and decision execution, combined with the demand change trend and historical data of each checkpoint, the demand overlimit values of each checkpoint are calculated, and adjustment strategies are continuously issued to achieve the effects of reducing peak demand, filling valleys, and reducing monthly demand costs.

[0168] II. Function modules of the demand analysis control system

[0169] The composition of the function modules of the demand analysis control system is as Figure 2 shown.

[0170] 1) Data acquisition and processing module: Collect electric energy meter data such as active power.

[0171] 2) Real-time calculation and centralized monitoring module: Calculate the real-time demand of the checkpoint according to the real-time value of active power in the form of a sliding window.

[0172] 3) Demand Forecasting and Trend Analysis Module: Based on the ultra-short-term load forecasting and planned load, and considering the influencing factors of load characteristics, through the algorithm model library, multi-algorithm combined forecasting and trend analysis are carried out.

[0173] 4) Demand Data Statistical Analysis and Overlimit Warning Module: Monitor the demand at the gateway through the dashboard and curve display. If the demand exceeds the limit, notify the dispatcher through an alarm.

[0174] 5) Gateway Demand Regulation Strategy and Decision Analysis Module: First, obtain the mapping relationship between the gateway and each load line, sort the load regulation margins of each line in descending order, calculate the gateway demand regulation amount, and then formulate a regulation strategy through power distribution calculation and safety check.

[0175] 6) Demand Optimization Control and Decision Execution Module: According to the demand control target value of the master station, formulate reasonable control target values for each sub-control area.

[0176] 7) Demand Overlimit Inversion Module: Analyze the process of demand change based on the inversion data analysis.

[0177] 8) Event Recording and Alarm Module: When the demand exceeds the limit and an alarm is issued, an alarm screen is launched, and the alarm event is recorded in the monitoring system database.

[0178] III. Operating Process of the Demand Analysis Control System

[0179] The operating flow chart of the demand analysis control system is as Figure 3 shown.

[0180] 1) Data Acquisition: Real-time collect data from devices such as smart grids, smart meters, and sensors, including electricity quantity, voltage, current, power factor, active and reactive power, ambient temperature, etc.

[0181] 2) Real-time Rolling Forecasting: Adopt the time series forecasting algorithm, combine external factors (such as weather forecast, market electricity price), set a time sliding window, and conduct real-time rolling load forecasting.

[0182] 3) Power Calculation and Distribution: Combine the device energy efficiency model and line load characteristics to conduct power calculation, dynamically adjust the energy distribution strategy of each device, and achieve load balancing and maximum energy efficiency.

[0183] 4) Control Strategy Execution: Convert the optimized energy distribution strategy into control instructions, and send them to each terminal device through the Internet of Things technology to achieve automated control.

[0184] 5) Feedback and Adjustment: Real-time monitor the operating status of the system, collect the data feedback after execution, evaluate the control effect, and adjust the forecasting model and control strategy as needed.

[0185] IV. Technical Route of Demand Analysis Control System

[0186] The technical route of the demand analysis control system is as Figure 4 shown.

[0187] 1) Data acquisition and demand calculation: Collect voltage, current, and power factor; Calculate the active power of the load using a sliding window; Calculate the real-time demand at the gateway; Calculate the maximum demand at the gateway.

[0188] 2) Load forecasting and trend analysis: Analyze the load characteristics of the line; Forecast the line load; Analyze the generation plan and load trend; Forecast the demand load at the gateway.

[0189] 3) Optimization control and decision execution: Obtain the mapping relationship between the gateway and the line through network topology analysis; Conduct trend analysis, overlimit detection, and overlimit warning at the gateway; Evaluate the adjustable ability of the load line using the sensitivity matrix; Optimize the demand regulation strategy at the gateway through power flow calculation.

[0190] V. Specific Applications of Demand Analysis Control System

[0191] 5.1. Data Acquisition and Demand Calculation

[0192] 5.1.1. Load and Demand at the Gateway

[0193] Demand is a typical ultra-short-term power load at the minute level, which is the average power calculated at a certain time interval, reflecting the average level of power load over a period of time. It is an important indicator for evaluating the load level of the power system and an important basis for formulating electricity prices. The maximum demand, also known as the contract electricity load, refers to the maximum value of the average power consumption per unit time of a customer within a electricity bill settlement cycle, which is the maximum active power value of the electric energy used by a power user at a certain moment. Maximum demand management is to control the load demand so that the maximum demand power does not exceed the contract electricity power, in order to improve the load rate and the utilization rate of grid equipment. In the two-part electricity price, the basic electricity charge can be calculated based on the maximum demand, which helps to regulate the electricity load and promote load balance. In actual production scenarios, the energy network structure of factories is complex, and the electricity demand load is affected by multiple devices. It is often difficult to predict the impact load with ultra-short-term fluctuations. At the same time, due to the characteristics of the demand billing mode, continuous multi-step forecasting of the demand within a certain period of time is often required to achieve effective load control, further increasing the difficulty of forecasting: when the predicted demand value is larger than the actual value, it is easy to cause false alarms of the maximum demand, affecting the user's electricity consumption experience; when the predicted value is smaller than the actual value, it is easy to cause missed reports of the maximum demand, resulting in the actual maximum demand exceeding the preset limit value and increasing the demand electricity cost. The power demand control system can adjust the electricity load, improve the electricity load rate, achieve balanced electricity consumption, reduce the basic electricity charge expenditure, and also reduce the impact on the power grid and improve the power grid power supply quality.

[0194] The substation demand is the actual power or capacity demand of users at the substation metering point (i.e., a specific point in the power grid for metering and settling electricity), within a certain time period (this time period is usually short, such as every 15 minutes or 30 minutes, depending on the metering rules of the power system and the electricity consumption characteristics of users). When measuring the substation demand, it is necessary to clarify the range of the time period, such as the demand peak value, the demand valley value, etc. The substation demand is of great significance for evaluating the electricity consumption demand of users, ensuring the stable operation of the power system, and promoting the optimal allocation of resources.

[0195] 5.1.2. Data Acquisition and Processing

[0196] The data acquisition and processing module is as Figure 5 shown.

[0197] 1) Data acquisition: Real-time data acquisition related to power demand load includes three-phase voltage, three-phase current, power factor data of the substation line, as well as generation plans, temperature and weather (sunny / cloudy / rainy, etc.) information corresponding to each hour, weekday / holiday information, etc.

[0198] 2) Data cleaning: Eliminate outliers and fill in missing values. Outliers may be caused by abnormal measurement equipment, power system failures, etc., and are eliminated through methods such as horizontal processing and vertical processing. Missing values may be caused by network problems, failures of acquisition and transmission equipment, SCADA (Supervisory Control And Data Acquisition) system failures, or power system line maintenance or load shedding power outages, etc., and can be filled in through methods such as interpolation and mean value methods.

[0199] 3) Data integration: Combine data from different channels to make it have a unified format and structure for subsequent analysis.

[0200] 4) Data transformation: Standardize and normalize the data to reduce dimensional differences and improve the stability of the model.

[0201] 5) Feature engineering: Feature selection, extract features with high correlation with the target variable (i.e., load) from the original data. These features may include time features (such as hour, week, month), environmental features (such as temperature, humidity), historical electricity consumption features (such as electricity consumption of the previous day, previous week), etc. Feature extraction, extract new features from the original data, such as calculating the average electricity consumption per hour, peak electricity consumption, etc. Feature combination, combine multiple features to generate new features to better reflect the law of load change.

[0202] 5.1.3. Power Calculation and Maximum Demand

[0203] Real-time demand load-related data includes the voltage, current, power factor, electrical load, generation plan of the gateway line, as well as the corresponding temperature and weather (sunny / cloudy / rainy, etc.) information per hour. By calculating the active power of the demand load in the form of a sliding window, the real-time electricity consumption situation of electricity users can be appropriately reflected, such as Figure 6 shown

[0204] The electricity demand calculation period is defined as 15 minutes, and the time step of the sliding window is set to 15 seconds, that is, the demand calculation is updated every 15 seconds. At any specific moment, the real-time calculated demand value represents the average power value corresponding to the completed time period, that is, the demand value at the end of the previous calculation period. The maximum demand refers to the maximum value of the measured demand usually within a specific settlement period of a month

[0205] The real-time demand of the gateway line is obtained through power calculation. The calculation formula for active power is

[0206] P = I * U * COSφ (1)

[0207] where U and I are the effective values of the AC voltage and AC current respectively, COSφ is the power factor, and φ is the phase difference between the voltage and current signals. According to the calculation rules of the two-part electricity price, the real-time demand is the average value of the active power within 15 minutes

[0208]

[0209] where D is the real-time demand calculation value, and P(m) is the active power of the demand gateway line at the mth minute

[0210] 5.2. Load Forecasting and Trend Analysis

[0211] 5.2.1. Load Characteristic Analysis

[0212] The load characteristics themselves and their prediction processes will be interfered by various factors, such as economic factors, electricity policies, electricity prices, meteorological factors, time factors, historical data, gateway lines and process equipment, and other interference factors. The influencing factors of the electricity demand load characteristics are shown in Table 1

[0213] Table 1 Analysis of Influencing Factors of Electricity Demand Load Characteristics

[0214] Influencing factors Specific analysis Economic factors Industrial structure adjustment, economic development level, industrialization and electrification level Electric power policy Power generation side power production structure, power transmission and distribution, power price formulation for power demand side management of hand-held power meters Meteorological factors Temperature, humidity, weather conditions, precipitation, wind speed Time factors Year, month, week, daily periodicity, holidays and working days Historical data Data at different historical moments and intervals Gateway lines and process equipment Key load lines, impact loads caused by different processes and equipment Other interference factors Power cuts, outage maintenance, sudden increase in order quantity

[0215] 1) Economic factors: The industrial restructuring, economic development level, and economic operation status in the location where the power system operates have a significant impact on the changes in power load characteristics. For example: the local industrial level, the load rate of general industrial users is higher than that of other ordinary residential users; the local electrification level, the load rate is basically proportional to the number of electrical equipment; the local industrial proportion, the higher the proportion of industrial users in the local power consumption structure, the faster the local power load growth rate.

[0216] 2) Power policies: It mainly includes three aspects. One is the policy centered on optimizing the power production structure on the power generation side; the second is the reform centered on "highway-izing" the power grid in power transmission and distribution; the third is the policy centered on promoting the power consumption revolution on the power sales side. These policies, from the formulation of electricity prices to the innovation of demand-side management technologies and even the construction of smart grids, have had a non-negligible impact on improving the reliability of power load characteristic prediction.

[0217] 3) Meteorological factors: It covers a lot of content, such as temperature, humidity, weather conditions, precipitation, wind speed, etc., mainly showing seasonal periodic changes. Among them, the level of temperature affects the changes of a large part of the power load.

[0218] 4) Time factors: In different time periods, the power load characteristics show different features. First is the annual, monthly, weekly, and daily periodicity. The annual and monthly periodicity is mainly the impact of seasonal changes on the air-conditioning power load; the weekly periodicity is mainly the impact of weekdays and weekends. During weekdays, the proportion of industrial production load of enterprises increases, and during weekends, the residential electricity consumption and so on increase significantly; the daily periodicity is the change of the power load during the peak and valley periods of a day. Then is the impact of legal holidays. During holidays such as the Spring Festival, Labor Day, and National Day, most enterprises will have holidays and rest, resulting in a significant decrease in the industrial production power load in the secondary industry, while the power load of the service industry in the tertiary industry increases significantly.

[0219] 5) Historical data: In the process of power load characteristic prediction modeling, a lot of historical data will be used. On the one hand, the influence degree of data at different historical moments on the current data is different. The farther the data is from the current time, the less useful value it has, and the closer the data is to the current time, the more it can reflect the current load change trend. On the other hand, which historical time interval of data can completely reflect the change law of the power load is helpful to improve the prediction accuracy.

[0220] 6) Tie-line and process equipment: Power systems of different types and scales have different operating laws, which will have different impacts on the characteristics of electrical loads. It is necessary to count the key load lines that affect the demand. The load characteristics of different processes and equipment in industrial enterprises are also different. For example, the iron and steel industry has a wide variety of processes, mainly including rolling mills, electric arc furnaces, oxygen generators, blowers, water pumps, conveyors, etc. The load characteristics are divided into stable loads, periodic fluctuating loads, continuous impact loads, and intermittent impact loads. It is crucial to analyze the production load data to obtain the laws of load nature and its impact on the tie-line demand under normal production conditions.

[0221] 7) Other interference factors: Sudden events such as power cuts and maintenance outages will cause large fluctuations in electrical loads. For industrial production users, unknown product order numbers will affect the electricity consumption load, and a sudden increase in the order quantity will cause a large instantaneous increase in the electrical load.

[0222] 5.2.2. Load forecasting

[0223] Input historical load data and relevant data on the characteristics of power demand loads into the prediction model to conduct ultra-short-term load forecasting for the tie-line of the demand, which means predicting the electricity consumption load of important electricity-consuming processes in the next 30 minutes at 1-minute intervals.

[0224] The load characteristics of different processes in industrial enterprises are different, and the operating conditions of the processes are also different. It is necessary to adopt a method of preferentially switching between multiple algorithm models and prediction schemes to achieve the load forecasting function for supporting various equipment, and predict the total load of the corresponding tie-line through the superposition of prediction results. To address the above problems, the present invention establishes an algorithm library for the load forecasting system. Existing models include LSTM, GRU, GRU-FCN, TCN-LSTM, Seq2Seq, etc. Among them, Seq2Seq is a sequence-to-sequence model, LSTM is a long short-term memory network, and GRU is a gated recurrent unit.

[0225] Conduct adaptive training according to different situations, automatically optimize the method combination of the comprehensive model, and automatically optimize the different parameter ratios of the same method, so as to ensure the accuracy and reliability of the prediction results of the system in different seasons. The prediction technology includes a variety of models and algorithms. Through various combinations, a large prediction model library is formed. The self-learning simulation training technology and artificial experience are used for optimization to form a comprehensive model that can better reflect the load change law and improve the prediction accuracy. In addition, for the comprehensive model combinations obtained through training, which are applicable to specific environments and have high prediction accuracy, the system provides a mechanism for saving according to the comprehensive prediction scheme, which is convenient for future use and is also an important prior condition for prediction.

[0226] The evaluation indicators of the prediction model are:

[0227] 1) Mean Absolute Error (MAE), which represents the average of the absolute errors between the predicted values and the observed values:

[0228]

[0229] 2) Mean Absolute Percentage Error (MAPE):

[0230]

[0231] 3) Root Mean Square Error (RMSE), which represents the sample standard deviation between the predicted values and the observed values, indicating the degree of dispersion of the samples:

[0232]

[0233] where y i is the actual value at time i, is the predicted value at time i, n is the number of samples in the test set, and y max and y min are the maximum and minimum values of the demand load in the test set samples, respectively.

[0234] 4) Peak Absolute Percentage Error (PAPE). To reflect the prediction performance for large demand loads, that is, to evaluate the fitting effect of the model on the maximum value of the power demand load curve within a specified settlement period, a new error evaluation index is proposed to analyze the prediction performance of the peak demand load:

[0235]

[0236] where T is the measurement period of the local peak demand load, defaulting to 15 minutes. l is the total time length of the test set, and by definition, is the peak demand load in the k-th statistical period, is the predicted value corresponding to the peak demand load in the k-th statistical period.

[0237] 5.2.3. Demand Trend Analysis

[0238] Taking the power demand load prediction results as the input, perform the demand trend analysis of the gateway line or process equipment, that is, predict and analyze the demand change in the next 15 minutes, as Figure 7 shown.

[0239] Demand trend analysis is to make predictions based on the process characteristics and power supply requirements of the controlled object, providing relatively accurate prediction information for optimization decisions. It is the core of the entire closed-loop control of electricity demand and a necessary condition for formulating control strategies. Taking the rated maximum demand as the standard, it analyzes and judges the current power consumption situation, and accurately and intelligently determines the best time to increase or decrease the load.

[0240] From the analysis of the load characteristics of each process in the iron and steel enterprise and the requirements of the production process, it can be seen that the electric arc furnace is the object for realizing the closed-loop control of electricity demand. Taking the electric arc furnace as an example below, the implementation process of the demand trend analysis of the electric arc furnace is introduced.

[0241] 1) Trend analysis model: The typical curve model of the electric arc furnace in a smelting cycle consists of three parts: the rising edge, the middle section, and the falling edge. The time lengths of the rising edge and the falling edge curves are determined according to statistical laws, and the time length of the middle section varies with the smelting plan. In a smelting cycle, the starting curves of steelmaking in the electric arc furnace are relatively similar. After a period of time, the starting load reaches the maximum value, and then the smelting load basically tends to be stable with small oscillations. When the smelting is about to end, the load begins to gradually decrease. Therefore, prediction models for the rising edge, the falling edge, and the middle section can be established respectively, and then the three models are combined according to the time sequence to form the prediction model of the electric arc furnace in a smelting cycle.

[0242] 2) Selection of sample data: After dividing the prediction model of the electric arc furnace into three different sub-models, how to select appropriate sample data from the historical data of the electric arc furnace becomes the key to the accuracy of the prediction. The dynamic time warping method is used to measure the similarity of the samples, and based on this, the sample data for prediction is selected.

[0243] 3) According to the real-time change information of the load of the electric arc furnace, the similarity measurement method is used to select sample data from the recent historical data of the electric arc furnace, and three different prediction models are used for trend analysis in combination with the smelting process of the electric arc furnace, meeting the requirements of the closed-loop prediction control of electricity demand.

[0244] Taking the electric arc furnace of an iron and steel industrial enterprise as an example for the electricity demand trend analysis, in the way of signal triggering, the sample data is one acquisition point every 10s, the prediction time length is one smelting cycle, and the online rolling optimization method is used to continuously correct the trend analysis curve for the next 5 minutes until the end of the entire smelting process. The specific steps are as Figure 8 shown.

[0245] 1) Trigger start. When the electric arc furnace starts smelting, close the transformer switch, and open the switch after one furnace of steel is smelted. The position information of this switch is obtained in real time through the energy monitoring system. When the switch changes from "open" to "closed", start the trend analysis of the electric arc furnace, and the trend data is the synthesis of the historical samples in the sample library with the same steel type as the smelted steel.

[0246] 2) Rising edge trend analysis. After 3 minutes of the start of smelting, using the real-time sample sequence collected in this process, the three sets of data with the highest similarity in the historical data of this electric arc furnace in the past 30 days are used as prediction samples through the similarity measurement method, and then the trend analysis curve of this smelting process is corrected by the method of trend extrapolation.

[0247] 3) Middle section trend analysis. When using the rising edge model for prediction, the middle section prediction model also makes predictions based on the load data that changes in real time of the electric arc furnace, and is ready to switch models in real time. When both models determine that a switch is needed, it immediately switches to the middle section model for prediction. Since the load change in this section is relatively small, a linear regression method is used for trend analysis during trend analysis, and the samples are also selected through the similarity measurement method.

[0248] 4) Falling edge trend analysis. When the load starts to decline near the end of smelting, switch to the falling edge model. The samples and the trend analysis method are the same as those of the rising edge. When the electric arc furnace transformer switch is separated, the load of the electric arc furnace is zero, and the entire smelting process and prediction process end.

[0249] 5.3. Optimization Control and Decision Execution

[0250] 5.3.1. Power Demand Analysis Control Strategy

[0251] Power demand optimization control devices are installed at each demand monitoring point, which receive the demand control target value from the master station, and formulate reasonable sub-control area control targets in combination with the actual operating status of the controlled objects in the jurisdiction area. The demand control device collects the on-site data of the controlled objects and uploads them to the demand optimization device.

[0252] According to the load prediction and trend analysis results, the actual electricity consumption of the whole plant, and taking into account the important production links of the enterprise, judge whether the future demand will exceed the target value, calculate in advance how much load needs to be reduced to avoid the maximum demand of the enterprise from exceeding the limit, send the control command to the demand execution device, and update the set value of the lower-level power demand optimization control device in real time to achieve closed-loop control of the controlled object.

[0253] Adopt a demand analysis control strategy based on real-time rolling prediction and power calculation, as Figure 9 shown.

[0254] Model predictive control can synthesize information and establish a predictive model in the simplest way according to the characteristics of the object and control requirements. More importantly, predictive control absorbs the idea of optimal control and uses rolling optimization within a finite time, combined with feedback information to correct the uncertain influencing factors in the control process in a timely manner. Using model predictive control to formulate the power demand control strategy can make full use of the results of load forecasting and trend analysis, determine a series of future control strategies through optimization calculations, and at the same time use feedback information to form a closed-loop control, greatly improving the accuracy of demand control.

[0255] 1) Through the ultra-short-term load forecasting data and the power generation plan for the next 30 minutes, calculate the power distribution cyclically and rollingly, and conduct cyclical and rolling forecasting of the gateway demand.

[0256] 2) When implementing the control strategy, the situation of multiple gateways exceeding the limit simultaneously should be considered, and the situation of exceeding the upper limit and the lower limit simultaneously should also be considered. If the predicted value of the gateway demand is greater than the set warning threshold (e.g., 90% of the assessment value), the dispatcher should be notified through an alarm, and the planned demand adjustment strategy should be followed.

[0257] 3) If the predicted value of a certain gateway demand is greater than the assessment value, calculate the demand target adjustment value = predicted value - assessment value, and obtain the mapping relationship between the gateway line and multiple load lines.

[0258] 4) Within the range of each load safety adjustment threshold preset, calculate the mapping relationship between the maximum adjustment margin of each load and the change value of the gateway demand, and sort them in descending order according to the adjustment margin of each load.

[0259] 5) When selecting the regulated load, accumulate the adjustment amount ∑ of the gateway demand until its value is greater than the total adjustment amount (demand difference).

[0260] 6) Based on the current real-time data, combined with the active power of each regulated load, conduct power distribution calculation and security check. If the check conditions are met, the adjustment strategy can be issued; otherwise, moderately adjust the active power of the load and conduct the check again until the check conditions are met.

[0261] 5.3.2, Sensitivity Analysis and Power Flow Calculation

[0262] Analyze the relationship between the active power changes of the demand gateway and the load lines through the power perturbation method and power flow calculation. The method idea:

[0263] 1) Taking the current active power of the load line as the reference value, use the power perturbation method to change its magnitude, increasing or decreasing it continuously by a fixed value upwards and downwards until the upper and lower limits are reached.

[0264] 2) After each change in the active power of the load line, enable the power flow calculation and record the magnitude of the active power change of the demand gateway line.

[0265] 3) According to the calculation results, sort the influence degrees of all load lines related to the demand metering point in descending order, and optimize the demand exceeding the limit value of the demand metering point line according to the safety adjustment margin of the load line and in combination with the economic evaluation results.

[0266] Power flow calculation:

[0267] 1) Node variables and their classifications: Each node has 6 variables, namely the active and reactive powers consumed by the load, the active and reactive powers of the generator, the voltage magnitude and phase angle of the node.

[0268] 2) Node power equations: Two power equations, namely the active power equation and the reactive power equation, can be listed for each node.

[0269] 3) Classification of power system nodes:

[0270] ① Slack node: The active power, reactive power, voltage magnitude and phase angle of the load are known, and the active and reactive powers of the generator are to be determined. This type of node is a large hub substation, the busbar of the power plant responsible for frequency regulation tasks, and there must be and can only be one such node set.

[0271] ② PQ node: A load node. The active power, reactive power of the load and the active and reactive powers of the generator are known, and the voltage magnitude and phase angle are to be determined. This type of node is the busbar of the power plant, the busbar of the substation with reactive power sources, and there are a large number of such nodes.

[0272] ③ PV node: The active power, reactive power of the load and the active power and voltage magnitude of the generator are known, and the reactive power of the generator and the voltage phase angle are to be determined. This type of node is the busbar of the power plant with a certain amount of reactive power reserve, the busbar of the substation with reactive power sources and can maintain the busbar voltage unchanged, and there are few or even no such nodes set.

[0273] 4) Gauss - Seidel power flow calculation method (G - S method): Directly apply the node voltage equation to solve the power flow distribution. It has low requirements for the set initial values, large memory occupation for calculation, and poor convergence.

[0274] 5) Newton - Raphson power flow calculation method (N - L method): The coefficient matrix of the correction equation is the Jacobian matrix, which is a sparse matrix and an asymmetric square matrix, and each element is a function of the node voltage and changes continuously during the iteration process; this method is a widely used computer power flow algorithm, with good convergence, fast calculation speed and small memory occupation; however, it has high requirements for the set initial values, and if the initial value is not selected properly, it will not converge, and it is generally used in combination with the G - S method.

[0275] 6) PQ Decomposition Power Flow Calculation Method: In a high-voltage AC power grid, the reactance is much greater than the resistance. The active power flowing out of a node is mainly affected by the voltage phase, and the reactive power flowing out of a node is mainly affected by the voltage amplitude. Based on this characteristic, the power flow calculation method that simplifies the calculation formula of the N-L method in polar coordinates is the PQ decomposition method. Its characteristics are that, compared with the N-L method, it has more iteration times, a shorter iteration calculation time, and a shorter total calculation time. The PQ decomposition method is a method used to calculate the power flow of a power system.

[0276] 5.3.3. Power Calculation and Energy Distribution

[0277] In the power demand analysis and control system, combining the equipment energy efficiency model and the line load characteristics for power calculation and dynamically adjusting the energy distribution strategy of each device is a key step to achieve efficient energy utilization and load balancing, as Figure 10 shown.

[0278] 1) Data collection and equipment energy efficiency model construction: Collect the energy efficiency data of each device, including rated power, actual operating power, efficiency curve, etc., and integrate the historical operating data of the device, including load changes, energy consumption, etc.; Based on the collected data, establish an energy efficiency model of the device to describe the energy consumption of the device under different loads; Consider factors such as equipment aging, maintenance status, and overhaul plan, and dynamically adjust the equipment energy efficiency model.

[0279] 2) Analysis of line load characteristics: Real-time collect the load data of the line, including active power, reactive power, voltage, current, etc., and analyze the change trend and periodic characteristics of the load; Based on the load data, establish a load characteristic model of the line to describe the load changes in different time periods, and consider the volatility and uncertainty of the load to correct and optimize the load characteristic model.

[0280] 3) Power calculation and energy distribution: According to the equipment energy efficiency model and the line load characteristic model, calculate the energy consumption of each device under different loads, consider the transmission loss of the line and the operating efficiency of the device, and calculate the total energy consumption of the entire system; According to the power calculation results, dynamically adjust the energy distribution strategy of each device, preferentially allocate energy to devices with high energy efficiency and large load demands, and consider the operating constraints and forward-looking constraints of the devices to ensure that the adjusted energy distribution strategy meets the actual needs, achieve load balancing and maximize energy efficiency, improve the stability and reliability of the power system, and reduce energy consumption and operating costs.

[0281] 5.3.4. Adjustment of the assessment set value of the demand-side gateway line

[0282] The evaluation and analysis scenarios are generated by using the real-time grid section of state estimation and the base state flow calculation of the dispatcher flow. The sensitivity of the line active power to the bus injected active power, the sensitivity of the mutual influence between the bus injected power, the branch power and the bus voltage are calculated through sensitivity calculation, and the corresponding sensitivity matrix can be formed. Finally, the assessment setting value of the demand gate line is provided to the demand specialist. Specific implementation steps, such as Figure 11 shown.

[0283] 1) Obtain the power grid section, map the data and equipment model status in it to the real-time database of the dispatcher's flow, and generate the power grid management control section by combining the real-time demand of the power grid section and the threshold demand exceeding the limit value. Perform network topology analysis on the demand threshold of the power grid management control section, such as Figure 12 shown.

[0284] A demand monitoring point is set at the incoming switch. Taking the demand gate line as the starting point, a deep search method is used to find all load lines connected to the demand gate line, and a mapping table of different demand gate lines is established. The mapping table contains the corresponding relationship between the demand gate line and all related load lines. The specific contents include the gate name, gate ObId (Object ID), the number of all loads, the ObId of all loads, the load switching status, the load operation status, the number of input loads, the name of the input load, the ObId of the input load, whether the load can be switched, whether the load is adjustable, the rated demand, the maximum warning value of the demand exceeding the limit, the minimum warning value of the demand exceeding the limit, the upper limit of the load adjustment, the lower limit of the load adjustment, the load adjustment threshold, and the mapping relationship between the load adjustment and the demand. The mapping relationship is determined by the following sensitivity analysis and power perturbation method.

[0285] 2) Through sensitivity analysis, a power sensitivity matrix is established between the power change between the gateway and the load and the power change of the load line; according to the coefficients in the power sensitivity matrix, combined with the adjustable upper and lower limit constraints of the load line, the adjustable capacity of the load line is evaluated to determine whether the demand over-limit value meets the safety constraints of the system, thereby evaluating the adjustment capacity of the demand control; when the safety margin assessment cannot be met, the power perturbation method is used to adjust the demand over-limit value of the demand gateway line, and then the analysis is repeated.

[0286] 3) Taking the current active power of the demand gateway line as the reference value, the power perturbation method is used to change the size of its active power; after the active power of the demand gateway line changes, the flow calculation is enabled to record the change size of the load line related to the demand gateway line; taking the current active power of the load line as the reference value, the power perturbation method is used to change the active power size of all load lines; after each change in the active power of the load line, the flow calculation is enabled to record the change size of the active power of the demand gateway line.

[0287] 4) Based on the above calculation results, sort the influence degrees of all load lines related to the demand gateway line from large to small, and calculate the specific upper and lower adjustment limits according to the correlation degree of the load line to the demand gateway line; optimize the demand over-limit value of each demand gateway line according to the safe adjustment margin of the load line and in combination with the economic evaluation results.

[0288] The sensitivity matrix is used to describe the influence degree of the change of a certain variable (active power of the load line) in the system on another variable (active power of the gateway). Calculation method:

[0289] 1) Expand the nonlinear equations of the power system in the vicinity of the demand gateway by Taylor series and ignore the high-order terms, so as to obtain a linearized system of equations and simplify the problem into a solvable form.

[0290] 2) In the linearized system of equations, by solving the Jacobian matrix, the sensitivity relationship between various variables in the system can be obtained.

[0291] 3) According to the relevant elements in the Jacobian matrix, a power sensitivity matrix between the power change amounts of the demand gateway and the load line can be constructed, and each element represents the influence degree of the change of the load line power on the change of the gateway power.

[0292] Specific steps:

[0293] 1) Determine the system parameters: including the power grid topology structure, line impedance and capacitive reactance parameters, generator capacity and location, power demand and distribution of the load, etc. Establish a steady-state model of the power system, including node power balance equations, voltage equations, etc. These equations are highly nonlinear and involve the parameters of various components such as generators, loads, and lines.

[0294] 2) Establish a linearized system of equations: Based on the steady-state model of the power system, establish a linearized system of equations at the demand gateway, which involves performing Taylor series expansion on the power balance equation and node voltage equation and ignoring the second-order and higher-order terms.

[0295] 3) Solve the Jacobian matrix: Use numerical methods (such as Gaussian elimination method, LU decomposition, etc.) to solve the Jacobian matrix in the linearized system of equations. For each demand gateway, one row of the Jacobian matrix will contain the partial derivatives of the node power balance equation with respect to all relevant variables, which describes the linear relationship between the system state variables (such as voltage phase angle and amplitude) and the control variables (such as generator active and reactive power outputs, load power, etc.).

[0296] 4) Construct the sensitivity matrix: Extract the relevant elements between the gateway power and the load line power from the Jacobian matrix to construct the sensitivity matrix. This process usually involves rearranging the rows and columns of the Jacobian matrix. Rearrange these rows and columns to form the sensitivity matrix, where the rows represent the changes in the load line power and the columns represent the changes in the gateway power.

[0297] Example of a simplified sensitivity matrix: Suppose there is a simple power system with two gateway nodes (G1 and G2) and three load nodes (L1, L2, L3). Focus on the impact of the active power change of the load nodes on the active power of the gateway nodes, and construct a sensitivity matrix S, where the rows represent the active power changes of the load nodes and the columns represent the active power changes of the gateway nodes.

[0298] Table 2 Power sensitivity matrix between the demand gateway and the change amount of the load line power

[0299] Power change of G1 Power change of G2 Power change of L1 S11 S12 Power change of L2 S21 S22 Power change of L3 S31 S32 Power change of L4 S41 S42

[0300] The sensitivity matrix can be expressed as:

[0301]

[0302] In this matrix, S ij represents how the active power of gateway node j will change when the active power of load node i changes by one unit. This change amount can be an increase or a decrease, depending on the structure and parameters of the system.

[0303] 5.3.5, Demand Overlimit Analysis and Control Strategy

[0304] The relevant software realizes functions such as demand overlimit analysis, real-time overlimit warning, overlimit tracking analysis, load prediction of key power consumption units, enterprise power load management, and energy-saving analysis through monitoring and predicting the real-time demand of each power consumption gateway of the enterprise. The specific implementation process is as Figure 13 shown.

[0305] 1) Overlimit detection and trend analysis: When the calculated value of the real-time demand of the demand gateway line exceeds the demand warning threshold (such as 90% of the assessment set value), the overlimit warning signal is sent through the warning subsystem to remind the dispatcher to prepare for corresponding adjustment according to the predetermined control adjustment strategy; at the same time, the trend analysis function is started to predict and analyze the trend changes of the load lines related to the demand gateway line.

[0306] 2) Single-gate over-limit adjustment strategy: When there is only one gate over-limit warning in the whole system, the single-gate adjustment measure is enabled. First, determine the required adjustment amount, and then calculate the adjustment margin according to the upper and lower limits of the relevant load lines of the demand gate line (that is, the adjustable upper limit of the line minus the current active power of the line). Sort them from large to small according to the adjustment margin, and add up the adjustment values of the load lines ranked in the front until the adjustment amount of the demand gate line is satisfied, so as to determine the names of the adjusted load lines and the corresponding adjustment amounts.

[0307] 3) Multi-gate over-limit adjustment strategy: When there are multiple gate over-limit warnings in the whole system, the multi-gate adjustment measure is enabled. First, calculate the adjustment amounts of each gate, calculate the adjustment margins of the relevant load lines, and sort them from large to small according to the adjustment margin. Add up the adjustment values of the load lines ranked in the front until the adjustment amount of the demand gate line is satisfied. When determining the adjusted load lines, the uniqueness principle should be followed to avoid the phenomenon of repeated adjustment. When a certain load line belongs to different demand gate lines at the same time, the active power sensitivity analysis of the line should be carried out to ensure the optimal adjustment effect of the load line.

[0308] 4) Control strategy output: According to the analysis and optimization of different situations by the single-gate over-limit adjustment strategy and the multi-gate over-limit adjustment strategy, formulate corresponding control strategies and display them on the special interface for demand decision analysis.

[0309] 5) Evaluation and analysis of control strategy: Use power grid power flow calculation and the incremental perturbation method to calculate the adjustment amounts of each load line related to the demand gate line in the control strategy, and continuously adjust the size of the adjustment amount according to the calculation results to ensure the accuracy of actual dispatching; send the control strategy after verification to the dispatcher's control interface, and the dispatcher makes further judgments and operates and executes reasonable control strategies.

[0310] The output of the control strategy is inseparable from the detailed analysis of the adjustable load in the power system, understanding its electricity consumption characteristics, electricity consumption patterns and adjustable potential. Identify high-energy-consuming equipment, seasonal loads and loads with obvious peak-valley characteristics. Set reasonable demand control objectives according to the supply-demand balance of the power system, and consider the safety, stability and economy of the power system. According to the load characteristics and demand control objectives, formulate a detailed load adjustment plan, including adjustment time, adjustment object, adjustment method and expected effect, etc.

[0311] Utilize smart grid technology to monitor the adjustable load changes in the power system in real time. Through data analysis, promptly detect and resolve problems arising during the load adjustment process. Regularly evaluate the effectiveness of load adjustment, including the achievement of demand control targets, the supply-demand balance of the power system, and user satisfaction, etc. According to the evaluation results, promptly adjust the load adjustment strategies and management measures to ensure the stable operation of the power system and meet the electricity demands of users.

[0312] 5.3.6. Inverse analysis of demand limit violation

[0313] The daily demand curve can be superimposed with the corresponding historical active power values at the metering points. When the calculated real-time demand value of the demand metering line exceeds the demand warning threshold (such as 90% of the assessment set value), an over-limit warning signal is sent through the alarm subsystem to remind the dispatcher to prepare for corresponding adjustments according to the predetermined control and regulation strategies. Demand limit violations include single metering point over-limit and multiple metering points over-limit.

[0314] Support the inverse analysis and evaluation of demand limit violations, which means recording the data change information of relevant loads 15 minutes before and 10 minutes after the time point of limit violation when the demand limit is violated, and being able to perform inverse analysis through a dedicated interface. Analyze the process of demand change based on the inverse analysis data changes, such as Figure 14 as shown.

[0315] 1) Demand limit violation detection: The system sets the limit value or threshold of energy consumption. Real-time monitor the energy consumption data and compare it with the preset limit value. When the energy consumption exceeds the limit value, trigger the demand limit violation alarm.

[0316] 2) Inverse analysis: After the demand limit violation alarm is triggered, the inverse analysis module starts to work. By analyzing the historical energy consumption data, the current energy consumption pattern, and the limit value set by the system, find out the reasons for the demand limit violation. The inverse analysis may involve multiple factors, such as equipment operation status, energy supply situation, user demand changes, etc.

[0317] 3) Corrective measure formulation: According to the results of the inverse analysis, formulate corresponding corrective measures. Include adjusting the operating power of equipment, optimizing the energy supply plan, adjusting user demand, etc. Send the corrective measures to the relevant equipment or systems to execute the adjustments.

[0318] 4) Effect evaluation and feedback: Monitor the energy consumption situation after implementing the corrective measures. Evaluate the effectiveness of the corrective measures and judge whether the demand limit violation problem has been successfully solved. According to the evaluation results, make necessary adjustments and optimizations to the system settings or corrective measures.

[0319] 5) Report generation and recording: Generate an inverse analysis report of demand limit violation, and detailedly record the reasons for demand limit violation, the corrective measures, and the results of their effectiveness evaluation. Store the report in the system for subsequent reference and review.

[0320] 5.3.7 Event Recording and Alarm

[0321] The program logic of the event recording and alarm module of the demand management system mainly involves real-time monitoring, recording and alarm processing of system events, such as Figure 15 shown.

[0322] 1) Event Record:

[0323] ① Event classification: According to the nature and impact of the event, the event is divided into different categories, such as normal events, abnormal events, alarm events, etc.

[0324] ② Event storage: Various events are stored in the corresponding locations of the database for subsequent processing and query.

[0325] 2) Event alarm:

[0326] ① Alarm condition setting: Users can set different alarm conditions according to actual production conditions and needs, such as power consumption exceeding the preset threshold, voltage fluctuation beyond the normal range, etc. These alarm conditions are stored in the system configuration file for real-time call by the alarm module.

[0327] ②Alarm detection and processing: The alarm module monitors system data in real time and triggers an alarm when the data meets the preset alarm conditions. The alarm module will take different processing measures according to the level and type of the alarm, such as sending alarm information to relevant personnel and starting the emergency processing process.

[0328] ③ Alarm information transmission: The alarm module sends the alarm information to relevant personnel through the system's communication module, such as through SMS, email, phone, etc. The alarm information includes detailed information such as alarm type, alarm time, alarm location, alarm reason, etc., so that relevant personnel can quickly understand and handle the alarm event.

[0329] ④Alarm confirmation and release: After receiving the alarm information, the relevant personnel need to confirm and handle the alarm. When the alarm event is handled, the relevant personnel need to perform the alarm release operation in the system to update the system's alarm status.

[0330] 5.3.8 Demand analysis control software

[0331] The design scheme of the functional modules of each sub-page of the demand analysis control software interface is as follows:

[0332] 1) Home page: Real-time demand display, using dashboard, curve, bar chart or table switching; rated demand, maximum demand of the month and number of over-limit times; event record and over-limit warning.

[0333] 2) Historical demand: By selecting the key points and time periods, the historical demand of the key points and the historical data of each related load can be displayed, and the function of report export is provided.

[0334] 3) Demand forecasting: The display of the demand forecasting results of the key points and the time of demand overlimit.

[0335] 4) Demand inversion: By selecting the demand overlimit events, the display of the changes in the demand of the key points and the changes in each load 15 minutes before and 10 minutes after the demand overlimit can be realized.

[0336] 5) Adjustment strategy: The program provides the recommended operation strategies when the demand of each key point exceeds the limit, such as the priority order of load control.

[0337] Among them, the display column of the forecasting results shows the predicted values and the true values of the demand of the key points in the past and the next 24 hours. The display column of the comparison of the demand forecasting data of multiple algorithms shows the comparison of the predicted values of the demand of each algorithm, and the display methods of the table and the curve can be switched; the display column of the algorithm weights shows the proportion of the weights of each algorithm.

[0338] The key points of the present invention can be summarized as the following points:

[0339] 1) Real-time rolling forecasting and power calculation: The present invention adopts the real-time rolling forecasting technology, combines the time series forecasting algorithm and external factors (such as weather forecast, market electricity price), and realizes the accurate forecasting of the future load. An equipment energy efficiency model and a line load characteristic model are established for power calculation, and the energy distribution strategy of each equipment is dynamically adjusted to achieve load balance and maximize energy efficiency.

[0340] 2) Multi-algorithm model and optimal switching of forecasting schemes: For the load characteristics of different processes in industrial enterprises, the present invention establishes an algorithm library for the load forecasting system, including models such as LSTM, GRU, GRU-FCN, TCN-LSTM, and Seq2Seq. It can perform adaptive training according to different situations, automatically optimize the method combination and parameter ratio of the comprehensive model, and ensure the accuracy and reliability of the forecasting results.

[0341] 3) Demand trend analysis and optimal control: Through the demand trend analysis, the present invention forecasts and analyzes the changes in the demand in the next 15 minutes, and provides accurate forecasting information for the optimization decision-making. The demand analysis and control strategy based on real-time rolling forecasting and power calculation is adopted, combined with the model predictive control idea, to realize closed-loop control and improve the accuracy of demand control.

[0342] 4) Sensitivity analysis and power flow calculation: Through the power perturbation method and power flow calculation, the relationship between the active power changes of the demand key points and the load lines is analyzed, providing a basis for adjusting the assessment setting values of the demand key point lines. The sensitivity matrix is used to describe the influence degree of the change of a certain variable in the system on another variable, providing technical support for the formulation of the demand overlimit analysis and control strategy.

[0343] 5) Demand over - limit analysis and control strategy: The present invention realizes the monitoring and prediction of the real - time demand at each power - using checkpoint of an enterprise, and has functions such as demand over - limit analysis, real - time over - limit alarm, and over - limit tracking analysis. According to the load characteristics and demand control objectives, a detailed load adjustment plan is formulated, and the change of adjustable load is monitored and adjusted in real - time through smart grid technology.

[0344] 6) Integrated energy management system: The present invention provides an integrated energy management system, which includes functions such as data collection and processing, setting of checkpoint demand optimization control strategy, interface display, and demand over - limit inversion. The system can regularly evaluate the effect of load adjustment, and timely adjust the load adjustment strategy and management measures according to the evaluation results to ensure the stable operation of the power system and meet the power consumption needs of users.

[0345] In summary, the key points of the present invention are to achieve accurate prediction and dynamic adjustment of energy demand, optimize energy distribution, and reduce operating costs through technical means such as real - time rolling prediction, optimal switching of multiple algorithm models, demand trend analysis and optimization control, sensitivity analysis and power flow calculation, and demand over - limit analysis and control strategy.

[0346] The present invention includes the following technical points:

[0347] 1) Optimize the demand analysis and control technology, and combine methods such as load prediction, power calculation, network topology analysis, sensitivity analysis, and power flow calculation, which improves the accuracy of demand prediction and the effectiveness of the implementation of load regulation control strategies.

[0348] 2) Effectively integrate various time - series prediction algorithms. By optimizing the demand prediction scheme, the prediction results of various algorithm models are weighted and fused, which improves the load prediction accuracy.

[0349] 3) Improve the demand analysis and control strategy, and effectively combine the adjustment of the assessment set value of the demand checkpoint line and the demand over - limit analysis and control strategy to more accurately perform demand over - limit adjustment and improve the effectiveness of demand control.

[0350] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A demand management control method based on real-time rolling prediction and decision analysis, characterized in that, Including: Input historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction for the demand metering point line; According to the process characteristics of the controlled object, use the load prediction result as the input to perform power demand trend analysis of the controlled object; Utilize the load prediction result and the power demand trend analysis result, and through periodic rolling prediction and optimization calculation of the metering point demand, determine the power demand control strategy; Analyze the relationship between the power changes of the metering point and the load line through the power perturbation method and power flow calculation, and construct a sensitivity matrix between the power change amount between the metering point and the load line and the power change amount of the load line; Use the sensitivity matrix to provide the demand limit value of the metering point line and support the construction of a control strategy for demand over-limit analysis; Combine the equipment energy efficiency model and the line load characteristics to perform power calculation and dynamically adjust the energy distribution strategy of each device.

2. The demand management control method based on real-time rolling prediction and decision analysis according to claim 1, characterized in that The step of inputting historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction for the demand metering point line includes: Obtain power demand load data, and calculate the active power of the demand load in the form of a sliding window; calculate the real-time demand of the metering point line through the active power of the demand load; obtain data related to power demand load characteristic factors; Input the active power of the demand load, the real-time demand of the metering point line, the power demand load data, and the data related to power demand load characteristic factors into the prediction model to predict the power consumption load of the future power consumption process of the demand metering point line.

3. The demand management control method based on real-time rolling prediction and decision analysis according to claim 2, wherein The construction of the prediction model includes: According to the load characteristics and operating conditions of different processes, adopt a method of preferentially switching between multi-algorithm models and prediction schemes, so that the prediction model supports the load prediction of various devices; Among them, the method of preferentially switching between multi-algorithm models and prediction schemes includes: Combine multiple models and algorithms to form a prediction model library, and use self-learning simulation training technology and artificial experience for optimization to form a comprehensive model reflecting the load change law.

4. A demand management control method based on real-time rolling prediction and decision analysis according to claim 1, characterized in that, The step of, according to the process characteristics of the controlled object, using the load prediction result as the input to perform power demand trend analysis of the controlled object includes: Based on the process characteristics of the controlled object, divide the process of the controlled object into a rising edge, a falling edge, and an intermediate section; In the rising edge and falling edge stages, use the similarity measurement method to take several groups of data with the highest similarity in the historical data of the controlled object as prediction samples, and use the trend extrapolation method to correct the trend analysis curve of the power demand of the controlled object; In the intermediate section, use the similarity measurement method to take several groups of data with the highest similarity in the historical data of the controlled object as prediction samples, and use the linear regression method for trend analysis.

5. A demand management control method based on real-time rolling prediction and decision analysis according to claim 1, characterized in that, The step of utilizing the load prediction result and the power demand trend analysis result, and through periodic rolling prediction and optimization calculation of the metering point demand, to determine the power demand control strategy includes: Through the load prediction result and the power demand trend analysis result, perform periodic rolling calculation of the power distribution to perform periodic rolling prediction of the metering point demand; If the periodic rolling prediction value of the metering point demand is greater than the set warning threshold, perform an alarm process; If the periodic rolling prediction value of the gateway demand is greater than the assessment value, obtain the mapping relationship between the gateway and the load, and calculate the mapping relationship between the maximum adjustment margin of each load and the change value of the gateway demand under the condition of meeting the safety margin conditions of each load adjustment, and sort them in descending order according to the adjustment margin of each load; When adjusting the load, accumulate the adjustment amount of the gateway demand until it is greater than the total adjustment amount, and generate a power demand control strategy; According to the load adjustment amount of each line load, perform a safety check through power flow calculation. After passing the safety check, issue a power demand control strategy.

6. The demand management control method based on real-time rolling prediction and decision analysis according to claim 1, wherein, The analysis of the relationship between the demand gateway and the power change of the load line by the power perturbation method and power flow calculation, and the construction of the sensitivity matrix between the power change amount between the demand gateway and the load line and the power change amount of the load line include: Take the current active power of the load line as the first reference value, and use the power perturbation method to change the size of the first reference value; After each change in the active power of the load line, enable power flow calculation to obtain the change in the active power of the demand gateway line; According to the power flow calculation results, sort the influence degrees of all load lines related to the demand gateway in descending order, and optimize the demand exceeding the limit value of the demand gateway line according to the safety adjustment margin of the load line and the economic evaluation results; Through sensitivity analysis, establish a sensitivity matrix between the power change amount between the demand gateway and the load line and the power change amount of the load line; Among them, the use of the power perturbation method to change the size of the first reference value includes: increasing or decreasing continuously by a fixed value upward and downward respectively until reaching the upper and lower limits of the first reference value.

7. A demand management control method based on real-time rolling prediction and decision analysis according to claim 6, characterized in that, The power flow calculation includes: Construct node variables and node power equations, where the node variables include the active power consumed by the load, the reactive power consumed by the load, the active power of the generator, the reactive power of the generator, the voltage magnitude of the node, and the phase angle of the node, and the node power equations include active power equations and reactive power equations; Construct algorithms for power flow calculation, including the Gauss-Seidel power flow calculation method, the Newton-Raphson power flow calculation method, and the PQ decomposition power flow calculation method.

8. A demand management control method based on real-time rolling prediction and decision analysis according to claim 6, characterized in that, The use of the sensitivity matrix to provide the demand exceeding the limit value of the demand gateway line includes: According to the coefficients in the sensitivity matrix, combined with the adjustable upper and lower limit constraints of the load line, evaluate the adjustable ability of the load line, and judge whether the demand exceeding the limit value meets the safety margin assessment; When the safety margin assessment cannot be met, use the power perturbation method to adjust the demand exceeding the limit value of the demand gateway line, and re-perform the safety margin assessment; Take the current active power of the demand gateway line as the second reference value, and use the power perturbation method to change the size of the second reference value; after the active power of the demand gateway line changes, enable power flow calculation and record the change size of the load line related to the demand gateway line; Take the current active power of the load line as the third reference value, and use the power perturbation method to change the active power size of all load lines; after each change in the active power of the load line, enable power flow calculation and record the change size of the active power of the demand gateway line; Based on the results of the comprehensive power perturbation method and power flow calculation, sort the influence degrees of all load lines related to the demand metering point line from large to small, and calculate the specific upper and lower adjustment limits according to the correlation degree of the load line to the demand metering point line; Optimize the demand overlimit value of each demand metering point line according to the safety adjustment margin of the load line and combined with the economic evaluation results.

9. A demand management control method based on real-time rolling prediction and decision analysis according to claim 1, characterized in that, The support for constructing the demand overlimit analysis and control strategy includes: Use the sensitivity matrix to describe the influence degree of the active power change of the load line on the power of the demand metering point, providing technical support for the formulation of the demand overlimit analysis and control strategy.

10. A demand management control method based on real-time rolling prediction and decision analysis according to claim 9, characterized in that The use of the sensitivity matrix to describe the influence degree of the load line power change on the power of the demand metering point includes: Perform Taylor series expansion of the nonlinear equations of the power system near the demand metering point to obtain a linearized system of equations, simplifying the problem into a solvable form; In the linearized system of equations, obtain the sensitivity relationship between variables by solving the Jacobian matrix; According to the relevant elements in the Jacobian matrix, construct a power sensitivity matrix between the power change of the demand metering point and the load line, where the element represents the influence degree on the power change of the demand metering point when the load line power changes.

11. A demand management control method based on real-time rolling prediction and decision analysis according to claim 1, characterized in that, The demand overlimit analysis and control strategy includes: When the calculated value of the real-time demand of the demand metering point line exceeds the demand warning threshold, issue an overlimit warning signal, and at the same time start the trend analysis function to predict and analyze the trend changes of the load lines related to the demand metering point line; When a single metering point overlimit warning occurs, start the single metering point adjustment measure; when multiple metering point overlimit warnings occur, start the multi-metering point adjustment measure; Use the power flow calculation of the power grid and adopt the incremental perturbation method to calculate the adjustment amount of each load line related to the demand metering point line in the demand overlimit analysis and control strategy, and continuously adjust the size of the adjustment amount according to the calculation results.

12. A demand management control method based on real-time rolling prediction and decision analysis according to claim 1, characterized in that It also includes: Inverse evaluation of demand overlimit; The inverse evaluation of demand overlimit includes: When the demand overlimit alarm is triggered, analyze the historical energy consumption data, current energy consumption pattern and preset limit values through the inversion module to obtain the reasons for the demand overlimit; Formulate corresponding corrective measures according to the reasons for the demand overlimit.

13. A demand management control method based on real-time rolling prediction and decision analysis according to claim 1, characterized in that, The combination of the equipment energy efficiency model and the line load characteristics for power calculation and dynamically adjusting the energy distribution strategy of each equipment includes: Based on the energy efficiency data and historical operation data of the equipment, establish an equipment energy efficiency model; Based on the load data of the line, establish a line load characteristic model to describe the changes of the load at different time periods; According to the equipment energy efficiency model and the line load characteristic model, calculate the energy consumption of each equipment under different loads, and combine the energy efficiency and load demand of the equipment to determine the energy distribution strategy of each equipment.

14. A demand management control system based on real-time rolling prediction and decision analysis is used to implement the demand management control method based on real-time rolling prediction and decision analysis described in claims 1-13, and is characterized in that, The system includes: a power demand control module, a demand overlimit analysis and control module, and an energy distribution module; The power demand control module is used to input historical load data and data related to power demand load characteristic factors into a prediction model to perform load prediction for the demand metering point line; according to the process characteristics of the controlled object, take the load prediction result as the input to perform power demand trend analysis of the controlled object; utilize the load prediction result and the power demand trend analysis result, and through periodic rolling prediction and optimization calculation of the metering point demand, determine the power demand control strategy. The demand overlimit analysis and control module is used to analyze the power change relationship between the demand metering point and the load line through the power perturbation method and power flow calculation, and construct a sensitivity matrix between the power change amount between the demand metering point and the load line and the power change amount of the load line; utilize the sensitivity matrix to provide the demand overlimit setting value of the demand metering point line and support the construction of the demand overlimit analysis and control strategy. The energy distribution module is used to perform power calculation in combination with the equipment energy efficiency model and the line load characteristics, and dynamically adjust the energy distribution strategy of each device.

Images