Progressive model predictive control method for hybrid energy storage based on optimal duty cycle tracking

Abstract

The present invention relates to the technical field of hybrid energy storage control strategies, and discloses a hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking. The front-end and rear-end MPCs are set to control the on-off of the switch tube in the system, and the optimal duty cycle output corresponding to the minimum cost function of the front-end MPC is selected to drive the corresponding switch tube on and off to stabilize the bus voltage, and the real-time optimal reference power of the hybrid energy storage at the next moment is calculated, and the real-time optimal reference value of the rear-end MPC is used as the real-time optimal reference value after passing through a high-pass filter; the optimal switch state output corresponding to the minimum cost function of the rear-end MPC is selected, and the corresponding switch tube is driven on and off, so that the output power of the supercapacitor tracks the real-time optimal reference value given by the front-end MPC in real time, and the power fluctuation is smoothed to further stabilize the bus voltage. Under the load power fluctuation, the present invention has smaller fluctuation of the bus voltage; and the optimal reference power transmission under model predictive control is considered to optimize the control performance.

Description

Technical Field

[0001] The present invention relates to the technical field of hybrid energy storage control strategies, and in particular to a hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking. Background Art

[0002] Energy storage is essentially a dispatchable power source. According to different electrical energy storage media, it can be divided into two categories: energy storage and power storage: ① Energy storage media represented by lead-acid batteries, lithium batteries and sodium-sulfur batteries have the characteristics of high energy density and low power density; ② Power storage media represented by supercapacitors, flywheel energy storage and superconducting energy storage have the characteristics of high power density and low energy density. Fields such as energy routers, DC microgrids and wind and solar storage require energy storage to have both high energy density and high power density to meet the response to steady-state and dynamic power during system operation. However, a single energy storage medium cannot meet the above requirements. Therefore, some scholars have proposed hybrid energy storage, that is, adding a power storage structure to energy storage to achieve complementary advantages between the functions of the two energy storage media.

[0003] Hybrid energy storage control can maintain the stability of the DC bus voltage, limit the state of charge (SoC) of the energy storage medium within a safe range, and achieve coordinated distribution of high and low frequency power between the two energy storage media to improve the stability of the system and the service life of the hybrid energy storage.

[0004] Model predictive control, also known as rolling horizon control, is a feedback control strategy that has received extensive attention in recent years and has achieved fruitful research results. The mechanism of MPC (Model predictive control) can be described as: at each sampling time k, based on the currently collected measurement information x(k), a finite time domain open-loop optimization problem is solved online, and the first element u(k) of the obtained control sequence u is applied to the controlled object, and at the next sampling time k+1, the above process is repeated using the new measurement information x(k+1). The main difference between MPC and traditional control methods is that the open-loop optimization problem is solved online to obtain the open-loop optimization sequence u. The latter usually solves a feedback control law offline and applies the obtained feedback control law to the controlled system. There are various constraints in actual systems, such as input constraints (execution structure saturation), state or output constraints (certain variables required for safe production and environmental protection, such as temperature, pressure, etc. should not exceed specific thresholds). If these constraints are simply ignored during system design, the control performance of the actual system may deteriorate, or even cause system instability. It is precisely because the optimization problems solved online can easily include various equality or inequality constraints that, unlike traditional control methods, MPC is one of the most effective methods for dealing with constraint system control problems.

[0005] Figure 1 It is represented as the basic control principle diagram of MPC. Given the current state x(k) and a mathematical model that can predict the future dynamics of the system, the state space model For example: x(k) and u(k) represent the state of the system at time k and the control input (u(t)) respectively; based on this model, the state prediction of the system starting from x(k) in the future can be calculated. , the future control input of the system is the control sequence to be solved; the control goal is generally to make the system state x(t) track the expected output r(t), and the optimization objective function is defined in this way. By solving the optimization problem, an optimal control sequence can be obtained, and the first control variable u(k) in the obtained optimal sequence is applied to the system, and this cycle continues. Therefore, the three main steps of the MPC algorithm are obtained: (1) predicting the future dynamics of the system; (2) solving the optimization problem; (3) applying the first element of the solution to the controlled system.

[0006] These three steps are repeated at each sampling moment, and no matter what mathematical model is used (continuous, discrete or hybrid), the measurements obtained at each sampling moment will serve as the initial conditions for predicting the future dynamics of the system.

[0007] The basic idea of ​​hybrid energy storage control is to divide the fluctuating power into high-frequency and low-frequency components through a low-pass / high-pass filter, then use a battery to smooth the low-frequency fluctuating power component, and at the same time use a supercapacitor to smooth the high-frequency fluctuating power component.

[0008] Based on the model predictive control, a voltage-current dual closed-loop control strategy is adopted for the bidirectional DC-DC converter in the energy storage system, such as Figure 2 As shown, it is used to control the stability of the DC bus voltage. ref Indicates the bus voltage reference value, V bus Indicates the actual value of bus voltage, I ref Represents the total reference current of hybrid energy storage, I batref Represents the battery reference current, I bat Indicates the actual output current of the battery, I scref Represents the supercapacitor reference current, I sc Indicates the actual output current of the supercapacitor, LPF stands for low-pass filter, PWM stands for pulse width modulation, and PI stands for proportional-integral control.

[0009] Existing technical measures and their shortcomings:

[0010] Measure 1: Based on the DC microgrid structure including the modular hybrid energy storage system, the MPC prediction controller is used to obtain the power demand of the hybrid energy storage system at the next moment, and the power demand at the next moment is divided equally to obtain the required smoothing power of each hybrid energy storage system; the required smoothing power of each hybrid energy storage system is divided by a low-pass filter to obtain the power demand of the supercapacitor and the battery respectively; according to the power demand of the supercapacitor and the battery, the reference value of the output current of the battery and the supercapacitor is calculated, and used as the current inner loop reference input value of the battery and supercapacitor PI control of each hybrid energy storage system. However, when the load power of this measure fluctuates, the bus voltage fluctuation is still large.

[0011] Measure 2: A dual closed-loop control structure is constructed by using an outer loop voltage control based on droop characteristics and an inner loop power control based on improved model predictive control to control the energy storage converter. The outer loop is a voltage loop, which uses the droop characteristics of the system to generate an inner loop power reference value through the outer loop, which has the tendency to prevent the DC voltage from changing. The inner loop is a power loop, which is used to track the power reference value generated by the outer loop, calculate the predicted value through the prediction model, and select the switch state corresponding to the predicted value closest to the expected value to act on the energy storage converter. However, the power reference value of this measure is fixed and single at each moment, and the optimal power reference cannot be considered. Summary of the invention

[0012] In view of the above problems, the purpose of the present invention is to provide a hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking, which effectively maintains the bus voltage stability by tracking the optimal duty cycle of the bidirectional DC-DC converter switch tube, and at the same time, the super capacitor is used to smooth the power fluctuation, enhance the robustness of the system, and further stabilize the bus voltage. The technical solution is as follows:

[0013] A hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking, the structure of the hybrid energy storage system includes a battery bat and a supercapacitor sc;

[0014] The positive pole of the battery bat passes through the front-end inductor L bat The switch tube S1 is connected to the emitter of the switch tube S1 and the collector of the switch tube S2 at the same time; the emitter of the switch tube S2 is connected to the negative electrode of the battery bat; the collector of the switch tube S1 is connected to the parallel load R and one end of the bus capacitor C at the same time; the other end of the load R and the bus capacitor C is also connected to the negative electrode of the battery bat;

[0015] The upper plate of the supercapacitor sc is connected to the rear inductor L sc It is simultaneously connected to the emitter of the switch tube S3 and the collector of the switch tube S4; the collector of the switch tube S3 is connected to the collector of the switch tube S1; the emitter of the switch tube S4 is simultaneously connected to the negative electrode of the battery bat and the lower plate of the supercapacitor sc;

[0016] It also includes a front-end MPC unit and a back-end MPC unit connected via a high-pass filter, the front-end MPC unit is used to control the on and off of the switch tube S1 and the switch tube S2; the back-end MPC unit is used to control the on and off of the switch tube S3 and the switch tube S4;

[0017] For the front-end MPC unit, the control method includes:

[0018] Step A1: Calculate the bus voltage when the switch tube S1 is turned off in the buck mode or the switch tube S2 is turned on in the boost mode, and when the switch tube S1 is turned on in the buck mode or the switch tube S2 is turned off in the boost mode, respectively, and perform discretization processing to obtain the bus voltage uniformly represented by the duty cycle;

[0019] Step A2: Calculate the hybrid energy storage reference power and the front-end MPC cost function in combination with the given initial directional duty cycle prediction set;

[0020] Step A3: Select the optimal duty cycle corresponding to the minimum front-end MPC cost function Output, drive switch tube S1 and switch tube S2 to turn on and off to stabilize the bus voltage; then calculate the real-time optimal reference power of the hybrid energy storage at the next moment according to the optimal duty cycle; filter the real-time optimal reference power through a high-pass filter to obtain the real-time optimal reference value of the rear-end MPC given by the front-end MPC;

[0021] Step A4: updating the directional duty cycle prediction set for the next sampling period according to the size of the cost function of the two cycles;

[0022] For the back-end MPC unit, the control method includes:

[0023] Step B1: respectively calculating the rear-end inductor current when the switch tube S3 is turned off in the buck mode or the switch tube S4 is turned on in the boost mode, and when the switch tube S3 is turned on in the buck mode or the switch tube S4 is turned off in the boost mode, and performing discretization processing;

[0024] Step B2: Calculate the supercapacitor output power and the back-end MPC cost function in combination with the given switch state selection matrix;

[0025] Step B3: Select the optimal switch state output corresponding to the minimum cost function of the back-end MPC, drive the switch tube S3 and the switch tube S4 to be on and off, so that the supercapacitor output power tracks the real-time optimal reference value of the back-end MPC given by the front-end MPC in real time, smoothes power fluctuations, and further stabilizes the bus voltage.

[0026] The beneficial effects of the present invention are as follows: the present invention applies optimal duty cycle tracking and optimal reference power transfer to the control of hybrid energy storage. Compared with measure 1, the control scheme of the present invention has smaller fluctuations in bus voltage under load power fluctuations, or almost no fluctuations; compared with measure 2, the present invention takes into account the optimal reference power transfer under model predictive control and optimizes the control performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is the principle diagram of model predictive control. Among them, x(k) is the state of the system at time k, r(t) is the expected output of the system, and u(t) is the control input of the system. is the state prediction quantity, and p is the prediction time.

[0028] Figure 2 This is the traditional control block diagram of hybrid energy storage in the prior art.

[0029] Figure 3 This is the structural block diagram of the hybrid energy storage system.

[0030] Figure 4 This is the flow chart of the hybrid energy storage progressive model predictive control based on optimal duty cycle tracking.

[0031] Figure 5 This is the bus voltage fluctuation using the traditional control method.

[0032] Figure 6 The bus voltage fluctuation situation using the control method designed by the present invention.

[0033] Figure 7 It is the reference power and actual output power of the supercapacitor using the control method designed by the present invention. DETAILED DESCRIPTION

[0034] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] A hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking, such as Figure 3 The structural block diagram of the hybrid energy storage system shown in the figure, R represents the load, C represents the bus capacitance, the bat in the subscript represents the battery, V bat Indicates the voltage of the battery, the subscript sc indicates the supercapacitor, V sc represents the voltage across the supercapacitor, and HPF represents high-pass filter; Indicates the output current of the battery; Represents the output current of the supercapacitor; Indicates the current injected into the busbar from the battery end; Indicates the current injected into the busbar by the supercapacitor; I cIndicates the bus capacitance current.

[0036] The positive pole of the battery bat passes through the front-end inductor L bat It is simultaneously connected to the emitter of the switch tube S1 and the collector of the switch tube S2; the emitter of the switch tube S2 is connected to the negative electrode of the battery bat; the collector of the switch tube S1 is simultaneously connected to the parallel load R and one end of the bus capacitor C; the other end of the load R and the bus capacitor C is also connected to the negative electrode of the battery bat.

[0037] The upper plate of the supercapacitor sc is connected to the rear inductor L sc It is simultaneously connected to the emitter of the switch tube S3 and the collector of the switch tube S4; the collector of the switch tube S3 is connected to the collector of the switch tube S1; the emitter of the switch tube S4 is simultaneously connected to the negative electrode of the battery bat and the lower plate of the supercapacitor sc.

[0038] It also includes a front-end MPC unit and a back-end MPC unit connected via a high-pass filter. The front-end MPC unit is used to control the on-off of switch tubes S1 and S2; the back-end MPC unit is used to control the on-off of switch tubes S3 and S4.

[0039] The control flow chart is as follows Figure 4 For the front-end MPC unit, the control method includes steps A1 to A4, which are as follows:

[0040] Step A1: Calculate the bus voltage when the switch tube S1 is turned off in the buck mode or the switch tube S2 is turned on in the boost mode, and when the switch tube S1 is turned on in the buck mode or the switch tube S2 is turned off in the boost mode, respectively, and perform discretization processing to obtain the bus voltage uniformly represented by the duty cycle.

[0041] When the switch tube S1 is turned off in buck mode or the switch tube S2 is turned on in boost mode:

[0042] (1);

[0043] In the formula, is the bus voltage, C is the bus capacitance; The current injected into the busbar at the supercapacitor end, is the load current.

[0044] After discretization, we get:

[0045] (2);

[0046] In the formula, ( k ) represents the k moment; f s is the sampling frequency.

[0047] When the switch S1 is turned on in buck mode or the switch S2 is turned off in boost mode:

[0048] (3);

[0049] In the formula, is the output current of the battery.

[0050] After discretization, we get:

[0051] (4);

[0052] Combining equations (2) and (4), the duty cycle can be uniformly expressed as:

[0053] (5);

[0054] In the formula, D is the duty cycle.

[0055] It should be noted that since the front-end MPC controls the switches S1 and S2 at the battery end by tracking the optimal duty cycle, the duty cycle is linked to the output current of the battery and then to the bus voltage through the above method to achieve the control purpose.

[0056] Step A2: Given an initial set of directional duty cycle predictions, calculate the hybrid energy storage reference power and the front-end MPC cost function.

[0057] The directional duty cycle prediction method first determines the prediction direction before each iteration, aiming to reduce the amount of calculation and optimize the system control performance.

[0058] Given an initial directional duty cycle prediction set D d :

[0059] (6);

[0060] In the formula, is discrete precision.

[0061] So we have:

[0062] (7);

[0063] In the formula, ( i ) represents the element number of the corresponding matrix.

[0064] Calculate the reference power of hybrid energy storage:

[0065] (8);

[0066] Then the front-end MPC cost function is:

[0067] (9);

[0068] In the formula, is the reference value of bus voltage.

[0069] Step A3: Select the optimal duty cycle corresponding to the minimum front-end MPC cost function Output, drive switch tube S1 and switch tube S2 to turn on and off to stabilize the bus voltage; then calculate the real-time optimal reference power of the hybrid energy storage at the next moment according to the optimal duty cycle through formula (8): ; The real-time optimal reference power is filtered through a high-pass filter to obtain the real-time optimal reference value of the back-end MPC given by the front-end MPC .

[0070] Step A4: Update the directional duty cycle prediction set for the next sampling period according to the size of the cost function of the two cycles.

[0071] Duty cycle , if the cost function of the two cycles is , indicating that the cost function calculated by the front element in the directional duty cycle prediction set is smaller, so the updated directional duty cycle prediction set is On the contrary, it means that the cost function calculated by the later elements in the directional duty cycle prediction set is smaller, so the updated directional duty cycle prediction set is .

[0072] For the back-end MPC unit, the control method includes steps B1 to B3, which are as follows:

[0073] Step B1: respectively calculating the rear-end inductor current when the switch tube S3 is turned off in the buck mode or the switch tube S4 is turned on in the boost mode, and when the switch tube S3 is turned on in the buck mode or the switch tube S4 is turned off in the boost mode, and performing discretization processing.

[0074] When the switch tube S3 is turned off in buck mode or the switch tube S4 is turned on in boost mode:

[0075] (10);

[0076] In the formula, is the back-end inductor; is the back-end inductor current, is the voltage across the supercapacitor.

[0077] After discretization, we get:

[0078] (11);

[0079] When the switch tube S3 is turned on in buck mode or the switch tube S4 is turned off in boost mode:

[0080] (12);

[0081] After discretization, we get:

[0082] (13);

[0083] Step B2: Combined with the given switch state selection matrix, calculate the supercapacitor output power and the back-end MPC cost function.

[0084] Given the switch state selection matrix :

[0085] (14);

[0086] So we have:

[0087] (15);

[0088] Calculate the supercapacitor output power:

[0089] (16);

[0090] Backend MPC cost function:

[0091] (17);

[0092] In the formula, The real-time optimal reference value of the backend MPC given by the frontend MPC.

[0093] Step B3: Select the optimal switch state output corresponding to the minimum cost function of the back-end MPC, drive the switch tube S3 and the switch tube S4 to be on and off, so that the supercapacitor output power tracks the real-time optimal reference value of the back-end MPC given by the front-end MPC in real time, smoothes power fluctuations, and further stabilizes the bus voltage.

[0094] Figure 4 In the example, index represents the ordinal value corresponding to the minimum cost function in the model prediction iteration, and s opt Indicates the optimal switching state.

[0095] As Figure 3 The hybrid energy storage model shown is taken as an example for calculation and verification.

[0096] The load power fluctuations were set at 2 seconds and 3 seconds respectively, and simulation verification was performed. The bus voltage fluctuations using the traditional control method are as follows Figure 5 As shown in FIG. 1 , it can be seen that when the load power changes, the bus voltage still fluctuates by about 5V; the bus voltage fluctuation using the control method designed by the present invention is as follows Figure 6 As shown in the figure, it can be seen that the bus voltage has almost no fluctuation when the load power changes; Figure 7 It can also be seen that the supercapacitor output power tracks the front-end optimal reference in real time, further stabilizing the bus voltage.

[0097] It can be seen that adding the designed control scheme to the hybrid energy storage can effectively maintain the stability of the bus voltage and suppress power fluctuations, which proves the effectiveness of the present invention.

Claims

1. A hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking, the structure of the hybrid energy storage system includes a battery bat and a supercapacitor sc; The positive pole of the battery bat passes through the front-end inductor L bat The switch tube S1 is connected to the emitter of the switch tube S1 and the collector of the switch tube S2 at the same time; the emitter of the switch tube S2 is connected to the negative electrode of the battery bat; the collector of the switch tube S1 is connected to the parallel load R and one end of the bus capacitor C at the same time; the other end of the load R and the bus capacitor C is also connected to the negative electrode of the battery bat; The upper plate of the supercapacitor sc is connected to the rear inductor L sc It is simultaneously connected to the emitter of the switch tube S3 and the collector of the switch tube S4; the collector of the switch tube S3 is connected to the collector of the switch tube S1; the emitter of the switch tube S4 is simultaneously connected to the negative electrode of the battery bat and the lower plate of the supercapacitor sc; It is characterized in that It also includes a front-end MPC unit and a back-end MPC unit connected via a high-pass filter, the front-end MPC unit is used to control the on and off of the switch tube S1 and the switch tube S2; the back-end MPC unit is used to control the on and off of the switch tube S3 and the switch tube S4; For the front-end MPC unit, the control method includes: Step A1: Calculate the bus voltage when the switch tube S1 is turned off in the buck mode or the switch tube S2 is turned on in the boost mode, and when the switch tube S1 is turned on in the buck mode or the switch tube S2 is turned off in the boost mode, respectively, and perform discretization processing to obtain the bus voltage uniformly represented by the duty cycle; Step A2: Calculate the hybrid energy storage reference power and the front-end MPC cost function in combination with the given initial directional duty cycle prediction set; Step A3: Select the optimal duty cycle corresponding to the minimum front-end MPC cost function Output, drive switch tube S1 and switch tube S2 to turn on and off to stabilize the bus voltage; then calculate the real-time optimal reference power of the hybrid energy storage at the next moment according to the optimal duty cycle; filter the real-time optimal reference power through a high-pass filter to obtain the real-time optimal reference value of the rear-end MPC given by the front-end MPC; Step A4: updating the directional duty cycle prediction set for the next sampling period according to the size of the cost function of the two cycles; For the back-end MPC unit, the control method includes: Step B1: respectively calculating the rear-end inductor current when the switch tube S3 is turned off in the buck mode or the switch tube S4 is turned on in the boost mode, and when the switch tube S3 is turned on in the buck mode or the switch tube S4 is turned off in the boost mode, and performing discretization processing; Step B2: Calculate the supercapacitor output power and the back-end MPC cost function in combination with the given switch state selection matrix; Step B3: Select the optimal switch state output corresponding to the minimum cost function of the back-end MPC, drive the switch tube S3 and the switch tube S4 to be on and off, so that the supercapacitor output power tracks the real-time optimal reference value of the back-end MPC given by the front-end MPC in real time, smoothes power fluctuations, and further stabilizes the bus voltage.

2. The hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking according to claim 1 is characterized in that: The step A1 is specifically as follows: When the switch tube S1 is turned off in buck mode or the switch tube S2 is turned on in boost mode: （1）； In the formula, is the bus voltage, C is the bus capacitance; The current injected into the busbar at the supercapacitor end, is the load current; After discretization, we get: （2）； In the formula, ( k ) represents the k moment; f s is the sampling frequency; When the switch S1 is turned on in buck mode or the switch S2 is turned off in boost mode: （3）； In the formula, is the output current of the battery; After discretization, we get: （4）； Combining equations (2) and (4), the duty cycle can be uniformly expressed as: （5）； In the formula, D is the duty cycle.

3. The hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking according to claim 2 is characterized in that: The step A2 is specifically as follows: Given an initial directional duty cycle prediction set D d : （6）； In the formula, is the discrete precision; So we have: （7）； In the formula, ( i ) represents the element number of the corresponding matrix; Calculate the reference power of hybrid energy storage: （8）； Then the front-end MPC cost function is: （9）； In the formula, is the reference value of bus voltage.

4. The hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking according to claim 3 is characterized in that: The specific updating of the directional duty cycle prediction set for the next sampling period in step A4 is: Duty cycle , if the cost function of the two cycles is , indicating that the cost function calculated by the front element in the directional duty cycle prediction set is smaller, so the updated directional duty cycle prediction set is ; On the contrary, it means that the cost function calculated by the later elements in the directional duty cycle prediction set is smaller, so the updated directional duty cycle prediction set is .

5. The hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking according to claim 4 is characterized in that: The step B1 is specifically as follows: When the switch tube S3 is turned off in buck mode or the switch tube S4 is turned on in boost mode: （10）； In the formula, is the back-end inductor; is the back-end inductor current, is the voltage across the supercapacitor; After discretization, we get: （11）； When the switch tube S3 is turned on in buck mode or the switch tube S4 is turned off in boost mode: （12）； After discretization, we get: （13）。 6. The hybrid energy storage progressive model predictive control method based on optimal duty cycle tracking according to claim 5 is characterized in that: The step B2 is specifically as follows: Given the switch state selection matrix : （14）； So we have: （15）； in,( j ) represents the element number of the corresponding matrix; Calculate the supercapacitor output power: （16）； Backend MPC cost function: （17）； in, The real-time optimal reference value of the backend MPC given by the frontend MPC.

Images