Control method for participation of hybrid energy storage system in primary frequency modulation under deep learning

Abstract

The invention relates to the technical field of energy storage system control, in particular to a method for controlling a hybrid energy storage system to participate in primary frequency modulation under deep learning, which comprises the following steps: establishing a hybrid energy storage system which comprises a battery energy storage system and a supercapacitor energy storage system; and acquiring an action strategy of the hybrid energy storage system, and controlling frequency modulation output of the battery energy storage system and the supercapacitor energy storage system through the model. According to the method, the adaptive factors of the virtual inertia output and the virtual droop output, the virtual inertia frequency modulation coefficient and the virtual droop frequency modulation coefficient are jointly determined, the weight coefficient in the adaptive factors is obtained through a deep learning algorithm, and the method is suitable for a nonlinear time-varying system and high in reliability.

Description

Technical Field

[0001] The present invention relates to the technical field of energy storage system control, and in particular to a control method for a hybrid energy storage system participating in primary frequency modulation under deep learning. Background Art

[0002] With the rapid development of new energy and the continuous expansion of the scale of power systems, the stability and reliability of power systems are facing more and more challenges. In power systems, primary frequency regulation refers to an important means of maintaining the stability of power grid frequency by quickly responding to frequency deviations by adjusting the output power of power generation equipment. However, traditional thermal power and hydropower frequency regulation resources have certain limitations in response speed and regulation capabilities, and it is difficult to meet the high requirements of modern power systems for frequency regulation.

[0003] In recent years, the development of energy storage technology has provided a new solution for the frequency regulation of power systems. Energy storage systems have the characteristics of fast response speed, high regulation accuracy and strong flexibility, which can effectively make up for the shortcomings of traditional frequency regulation resources. The control methods of energy storage participating in the primary frequency regulation of the power grid can be mainly divided into virtual droop output and virtual inertia output. Virtual droop output can effectively control the frequency deviation, and virtual inertia output can effectively control the frequency deviation change rate. Therefore, in the process of energy storage participating in the primary frequency regulation, a control strategy combining virtual droop output and virtual inertia output is adopted to ensure that the output of energy storage at any time of frequency regulation is composed of virtual droop and virtual inertia. In the current research, some use constructed functions to allocate the proportion of virtual droop and virtual inertia, and some use fuzzy control theory to calculate and allocate the proportion of virtual droop and virtual inertia, but the reliability of the former function construction is not high, and the latter relies on the fuzzy control rule table formulated by experts. In addition, in the current research, energy storage participating in the primary frequency regulation of the power grid mostly adopts a single type of battery energy storage system, but the battery energy storage system has a small number of cycles, and frequent charging and discharging when participating in the primary frequency regulation will reduce the service life of the battery. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose a control method for a hybrid energy storage system participating in primary frequency modulation under deep learning, so as to solve the problems that the current control strategy combining virtual droop output and virtual inertia output has low reliability and relies on fuzzy control rule tables formulated by experts.

[0005] Based on the above purpose, the present invention provides a control method for a hybrid energy storage system participating in primary frequency modulation under deep learning, comprising:

[0006] Establishing a hybrid energy storage system, the hybrid energy storage system comprising a battery energy storage system and a supercapacitor energy storage system;

[0007] Obtain the action strategy of the hybrid energy storage system and control the frequency modulation output of the battery energy storage system and the supercapacitor energy storage system through the following model:

[0008]

[0009] Where: ΔP b (s) is the frequency modulation output of the battery energy storage system or supercapacitor energy storage system, ΔP b1 (s) is the virtual inertia output; ΔP b2 (s) is the virtual droop force; M E is the virtual inertia frequency modulation coefficient; K E is the virtual droop frequency modulation coefficient; is the frequency deviation change rate of the power grid; Δf(s) is the frequency deviation of the power grid; μ1 and μ2 are the adaptive factors of virtual inertia output and virtual droop output, respectively; α1 and α2 are the weight coefficients of virtual inertia output and virtual droop output, respectively, and α1+α2=1, where α1 and α2 are obtained based on the deep learning algorithm; β1 and β2 are the feedback coefficients of virtual inertia output and virtual droop output when the hybrid energy storage system is restored.

[0010] Optionally, the action strategy of the hybrid energy storage system includes:

[0011] In SOC c ≥SOC clow When SOC cmin ≤SOC c <SOC clow When SOC b ≥SOC bmin At this time, both the supercapacitor energy storage system and the battery energy storage system are in operation; at SOC cmin ≤SOC c <SOC clow When SOC b <SOC bmin When SOC c <SOC cmin And SOC b ≥SOC bmin When SOC c <SOC cmin And SOC b <SOC bmin At this time, the supercapacitor energy storage system and the battery energy storage system do not operate;

[0012] Among them, SOC b and SOC cThey are the state of charge of the battery energy storage system and the supercapacitor energy storage system; SOC bmin SOC is the minimum value of the state of charge of the battery energy storage system; cmin and SOC clow They are the minimum and minimum values ​​of the state of charge of the supercapacitor energy storage system, respectively.

[0013] Optionally, the M E =K E , the K E Including charging virtual droop frequency modulation coefficient K c1 And the discharge virtual droop frequency modulation coefficient K d1 , the K c1 and K d1 Obtained through the linear constraint relationship constructed by the following linear function:

[0014]

[0015] Among them, S min , S low , S high , S max The minimum SOC of the hybrid energy storage system is 0.1, the minimum is 0.45, the maximum is 0.55 and the maximum is 0.9, respectively. E,max The value of is 20.

[0016] Optionally, the linear constraint relationship constructed by the linear function is replaced by a regression function, and the replaced K c1 and K d1 K c2 and K d2 Indicates that the calculation formula after replacement is as follows:

[0017]

[0018] Optionally, the β1 and β2 are calculated by the following formula:

[0019]

[0020] When controlling the proportion of virtual inertia output and virtual droop output, this method jointly determines the adaptive factors of virtual inertia output and virtual droop output, virtual inertia frequency modulation coefficient and virtual droop frequency modulation coefficient, wherein the weight coefficient in the adaptive factor is obtained through a deep learning algorithm, which is suitable for nonlinear time-varying systems, has high reliability, and does not rely on a fuzzy control rule table formulated by experts, thereby overall solving the problems of low reliability and reliance on a fuzzy control rule table formulated by experts in the current control strategy combining virtual droop output and virtual inertia output. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 The embodiment of the present invention is a model block diagram of a hybrid energy storage system participating in primary frequency regulation of a power grid;

[0023] Figure 2 It is an action strategy diagram of the hybrid energy storage system according to an embodiment of the present invention;

[0024] Figure 3 It is a comparison chart of frequency deviation step response; DETAILED DESCRIPTION

[0025] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0026] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0027] like Figure 1 As shown, a control method for a hybrid energy storage system participating in primary frequency modulation under deep learning includes:

[0028] Establishing a hybrid energy storage system, the hybrid energy storage system comprising a battery energy storage system and a supercapacitor energy storage system;

[0029] Obtain the action strategy of the hybrid energy storage system and control the frequency modulation output of the battery energy storage system and the supercapacitor energy storage system through the following model:

[0030]

[0031] Where: ΔP b(s) is the frequency modulation output of the battery energy storage system or supercapacitor energy storage system, ΔP b1 (s) is the virtual inertia output; ΔP b2 (s) is the virtual droop force; M E is the virtual inertia frequency modulation coefficient; K E is the virtual droop frequency modulation coefficient, M E and K E Used to control the frequency modulation output according to the charge state of the battery energy storage system or the supercapacitor energy storage system; is the frequency deviation change rate of the power grid; Δf(s) is the frequency deviation of the power grid; μ1 and μ2 are the adaptive factors of virtual inertia output and virtual droop output respectively; α1 and α2 are the weight coefficients of virtual inertia output and virtual droop output respectively, and α1+α2=1, where α1 and α2 are obtained based on the deep learning algorithm. Specifically, since the frequency deviation change rate is The frequency deviation Δf(s) gradually increases from 0. At this time, the weight coefficient α1 of the virtual inertia control follows the frequency deviation change rate. The change trend of , gradually decreases from a large value, and the weight coefficient α2 of the virtual droop control follows the change trend of the frequency deviation Δf(s), gradually increases from 0, and in the change process of the weight coefficients of the two, α1+α2=1;

[0032] Frequency recovery phase, frequency deviation change rate It increases first and then decreases, and the frequency deviation Δf(s) decreases first and then increases. At this time, the weight coefficient α1 of the virtual inertia control follows the frequency deviation change rate. The change trend of , first increases and then decreases, the weight coefficient α2 of the virtual droop control follows the change trend of the frequency deviation Δf(s), first decreases and then increases, and in the change process of the weight coefficients of the two, α1+α2=1;

[0033] Based on the above analysis, it can be seen that the weight coefficient α1 of virtual inertia control and the weight coefficient α2 of virtual droop control change with the frequency deviation curve, which has the characteristics of nonlinear time-varying systems whose dynamic characteristics are not easy to grasp, and it is difficult to accurately quantify and estimate. Therefore, it is impossible to use equation constraints to output the specific values ​​of α1 and α2. The deep learning algorithm does not require an accurate mathematical model and is suitable for nonlinear time-varying systems. The deep learning algorithm is used to calculate the weight coefficients, and the frequency deviation and frequency deviation change rate in the frequency modulation process are input into the deep learning network to establish a suitable network structure, and the weight coefficients of virtual inertia control and virtual droop control are adaptively output; β1 and β2 are the feedback coefficients of virtual inertia output and virtual droop output when the hybrid energy storage system is restored.

[0034] like Figure 3As shown, the deep learning algorithm makes the frequency change more stable and reliable. It can be seen from the above model that when controlling the proportion of virtual inertia output and virtual droop output, this method is determined by the adaptive factor of virtual inertia output and virtual droop output, the virtual inertia frequency modulation coefficient and the virtual droop frequency modulation coefficient. The weight coefficient in the adaptive factor is obtained by the deep learning algorithm, which is suitable for nonlinear time-varying systems, has high reliability, and does not rely on the fuzzy control rule table formulated by experts. Therefore, it solves the problem of low reliability and reliance on the fuzzy control rule table formulated by experts in the current control strategy combining virtual droop output and virtual inertia output.

[0035] like Figure 2 As shown, in some embodiments, the action strategy of the hybrid energy storage system includes:

[0036] In SOC c ≥SOC clow When SOC cmin ≤SOC c <SOC clow When SOC b ≥SOC bmin At this time, both the supercapacitor energy storage system and the battery energy storage system are in operation; at SOC cmin ≤SOC c <SOC clow When SOC b <SOC bmin When SOC c <SOC cmin And SOC b ≥SOC bmin When SOC c <SOC cmin And SOC b <SOC bmin At this time, the supercapacitor energy storage system and the battery energy storage system do not operate;

[0037] Among them, SOC b and SOC c They are the state of charge of the battery energy storage system and the supercapacitor energy storage system; SOC bmin SOC is the minimum value of the state of charge of the battery energy storage system; cmin and SOC clow They are the minimum and minimum values ​​of the state of charge of the supercapacitor energy storage system, respectively.

[0038] Through the above-mentioned action strategy of the energy storage system, the number of charge and discharge times of the battery energy storage system can be reduced, thereby extending the service life of the battery energy storage system.

[0039] In some embodiments, the M E =K E , the K E Including charging virtual droop frequency modulation coefficient K c1 And the discharge virtual droop frequency modulation coefficient K d1 , the K c1 and K d1 Obtained through the linear constraint relationship constructed by the following linear function:

[0040]

[0041] Among them, S min , S low , S high , S max The minimum SOC of the hybrid energy storage system is 0.1, the minimum SOC is 0.45, the maximum SOC is 0.55, and the maximum SOC is 0.9. E,max The value of is 20.

[0042] The virtual inertia frequency modulation coefficient and virtual droop frequency modulation coefficient are not fixed values ​​in frequency modulation. The influence of SOC on its output should be considered. Too high or too low SOC of hybrid energy storage will affect the frequency modulation capability. In order to maintain the energy storage SOC in an ideal state, this part combines the output capacity of energy storage to construct the charging virtual droop frequency modulation coefficient K. c1 And the discharge virtual droop frequency modulation coefficient K d1 , by controlling the frequency modulation coefficient to maintain the energy storage SOC in an ideal state.

[0043] In some embodiments, the linear constraint relationship constructed by the linear function is replaced by a regression function, and the replaced K c1 and K d1 K c2 and K d2 Indicates that the calculation formula after replacement is as follows:

[0044]

[0045] Compared with the linear constraint relationship constructed by the linear function, the regression function is closer to the relationship between the SOC of the energy storage system and the charge and discharge coefficient, making the calculation results more accurate.

[0046] In some embodiments, the β1 and β2 are calculated by the following formula:

[0047]

[0048] It should be understood by those skilled in the art that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention (including the claims) is limited to these examples. Within the scope of the present invention, the technical features in the above embodiments or in different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the above aspects of the present invention, which are not provided in detail for the sake of simplicity.

[0049] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A control method for a hybrid energy storage system participating in primary frequency modulation under deep learning, characterized in that: include: Establishing a hybrid energy storage system, the hybrid energy storage system comprising a battery energy storage system and a supercapacitor energy storage system; Obtain the action strategy of the hybrid energy storage system and control the frequency modulation output of the battery energy storage system and the supercapacitor energy storage system through the following model: Where: ΔP b (s) is the frequency modulation output of the battery energy storage system or supercapacitor energy storage system, ΔPb1(s) is the virtual inertia output; ΔP b2 (s) is the virtual droop force; M E is the virtual inertia frequency modulation coefficient; K E is the virtual droop frequency modulation coefficient; is the frequency deviation change rate of the power grid; Δf(s) is the frequency deviation of the power grid; μ1 and μ2 are the adaptive factors of virtual inertia output and virtual droop output, respectively; α1 and α2 are the weight coefficients of virtual inertia output and virtual droop output, respectively, and α1+α2=1, where α1 and α2 are obtained based on the deep learning algorithm; β1 and β2 are the feedback coefficients of virtual inertia output and virtual droop output when the hybrid energy storage system is restored.

2. According to the control method of a hybrid energy storage system participating in primary frequency modulation under deep learning according to claim 1, it is characterized in that: The action strategy of the hybrid energy storage system includes: In SOC c ≥SOC clow When SOC cmin ≤SOC c <SOC clow When SOC b ≥SOC bmin At this time, both the supercapacitor energy storage system and the battery energy storage system are in operation; at SOC cmin ≤SOC c <SOC clow When SOC b <SOC bmin When SOC c <SOC cmin And SOC b ≥SOC bmin When SOC c <SOC cmin And SOC b <SOC bmin At this time, the supercapacitor energy storage system and the battery energy storage system do not operate; Among them, SOC b and SOC c They are the state of charge of the battery energy storage system and the supercapacitor energy storage system; SOC bmin SOC is the minimum value of the state of charge of the battery energy storage system; cmin and SOC clow They are the minimum and minimum values ​​of the state of charge of the supercapacitor energy storage system, respectively.

3. According to the control method of a hybrid energy storage system participating in primary frequency modulation under deep learning in claim 1, it is characterized in that: The M E =K E , the K E Including charging virtual droop frequency modulation coefficient K c1 And the discharge virtual droop frequency modulation coefficient K d1 , the K c1 and K d1 Obtained through the linear constraint relationship constructed by the following linear function: Among them, S min , S low , S high , S max The minimum SOC of the hybrid energy storage system is 0.1, the minimum is 0.45, the maximum is 0.55 and the maximum is 0.9, respectively. E,max The value of is 20.

4. According to the control method of a hybrid energy storage system participating in primary frequency modulation under deep learning in claim 3, it is characterized in that: The linear constraint relationship constructed by the linear function is replaced by the regression function, and the replaced K c1 and K d1 K c2 and K d2 Indicates that the calculation formula after replacement is as follows:

5. According to the control method of a hybrid energy storage system participating in primary frequency modulation under deep learning according to claim 4, it is characterized in that: The β1 and β2 are calculated by the following formula:

Images