A wind energy regulation hybrid energy storage system and control method thereof

Abstract

The present invention discloses a wind energy regulation hybrid energy storage system and a control method thereof, including a wind energy data acquisition module, a prediction module, an energy storage management module, a hybrid energy storage module and an intelligent controller for high-voltage electric power metering; the energy storage management module also includes a control system, an energy router and a BMS; the real-time wind energy sampling data collected by the wind energy acquisition module is predicted by the prediction module and input into the control system, and the control system formulates energy storage and grid connection strategies corresponding to the real-time electricity price and wind energy load, so that the energy storage system can be switched flexibly and the benefits can be maximized. The present invention provides an intelligent wind energy regulation scheme suitable for coastal and windy factories, which acquires the time series and trend series of power generation by collecting data on factors such as wind speed and wind direction of the factory; and finds the best fitting performance model through algorithm iteration. It effectively solves the problem that wind energy is easy to collect but difficult to use efficiently, provides good benefits, and has long-term application prospects.

Description

Technical Field

[0001] The present invention relates to a wind power generation system and a control method thereof, and in particular to a wind energy regulation hybrid energy storage system and a control method thereof. Background Art

[0002] With the increasing attention paid to renewable energy around the world, wind energy, as a clean and renewable energy source, has become increasingly prominent in the energy structure. There are many factories distributed along the coast, and wind power generation has broad development prospects. Therefore, how to effectively store, regulate and utilize wind power generation has become an important research direction in the field of wind energy. With the continuous development of smart grids and demand response technologies, energy storage systems often face problems such as low energy utilization and poor profitability. Being able to predict the demand on the user side, adjust its power consumption within a certain range, participate in the operation and regulation of the power grid, and achieve peak shaving and valley filling can greatly promote the efficient use of energy and reduce resource consumption. However, under the current application conditions in the existing technology, there are still few prediction schemes that comprehensively consider the load side and electricity prices, and the use of optimization algorithms cannot combine electricity price factors with energy regulation.

[0003] On the other hand, hybrid energy storage systems are currently unable to measure the power of lithium batteries and supercapacitor devices, measure the capacity of negative hybrid energy storage modules in real time, and connect and network them in a timely manner to ensure the stable operation of the power grid. Summary of the invention

[0004] Purpose of the invention: In view of the above problems, the present invention proposes a wind energy regulated hybrid energy storage system and a control method thereof, which can reasonably regulate the charging and discharging of the hybrid energy storage system based on wind energy generation and electricity prices, thereby ensuring the fullness of load energy supply, saving power supply costs, and improving energy utilization.

[0005] Technical solution: The technical solution adopted by the present invention is a wind energy regulation hybrid energy storage system and a control method thereof, including:

[0006] A data acquisition module is used to collect real-time wind energy load data through an electric high-voltage metering intelligent controller;

[0007] A prediction module, including a wind load prediction module and an electricity price prediction module, wherein the wind load prediction module obtains a wind load prediction result based on real-time wind load data, and the electricity price prediction module obtains an electricity price prediction result based on historical electricity price data;

[0008] Hybrid energy storage modules, including lithium battery modules and supercapacitor modules, are used to store and release wind energy;

[0009] Electric high-voltage metering intelligent controller, used to monitor the power at the load end of the power grid line;

[0010] The energy storage management module includes a control system, an energy router and a BMS; the control system obtains the prediction results of the prediction module and obtains the charging and discharging decision results through the charging and discharging control strategy; the energy router controls the charging and discharging actions of the hybrid energy storage module according to the charging and discharging decision results; the BMS is used to obtain the status information of the battery pack and send it to the energy router.

[0011] The power at the load end of the monitoring power grid line includes active power, reactive power, apparent power and power factor.

[0012] The control system obtains a charge and discharge decision result through a charge and discharge control strategy according to the wind energy load prediction result, the electricity price prediction result, the average electricity price and the remaining capacity of the hybrid energy storage module sent by the energy router.

[0013] The present invention proposes a control method for the wind energy regulation hybrid energy storage system. The charge and discharge control strategy in the control system includes:

[0014] Case 1: When P pre >C hess When the hybrid energy storage module discharges and the grid supplies power;

[0015] Case 2: When P pre ≤C hess When the hybrid energy storage module discharges, the grid stops supplying power;

[0016] Case 3: When When the power grid only charges the lithium battery module;

[0017] Case 4: When When the power grid charges the hybrid energy storage module.

[0018] Obtaining wind load forecast results based on real-time wind load data includes the following steps:

[0019] Step 1: Preprocess the real-time wind load data and historical load data;

[0020] Step 2: Extract each modal component and a residual component of the wind load data;

[0021] Step 3: Normalize each modal component and divide it into training data set and test data set;

[0022] Step 4: Take the training data set as input x and train the KELM prediction model. The output of the hidden layer of the KELM prediction model enters the output layer and introduces the kernel function. The output of the KELM prediction model is:

[0023]

[0024] In the formula, is the output of the KELM prediction model, y is the output of the hidden layer, C represents the penalty coefficient, Ω=HH T ,K(x i ,x j ) is the Gaussian kernel function, expressed as,

[0025]

[0026] ψ is the kernel function; x i 、x j is the i-th and j-th item of input x;

[0027] Step 5: Solve the minimum output error;

[0028] Step 6: Use the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model.

[0029] The data preprocessing in step 1 is calculated as follows:

[0030] Perform Fourier Transform on the data, the transformation formula is as follows:

[0031]

[0032] in, T' is the period, the fundamental frequency is ω, B n is the data set, n is the number of modal components, and n tends to infinity;

[0033]

[0034] Among them, f n (t) is the modal component, x is the load data ordinal number, n is the number of modal components, δ is the Dirac function, and w is the frequency.

[0035] The IGW algorithm is used to optimize the kernel function and penalty coefficient in the KELM model. The objective function in the IGW algorithm is:

[0036]

[0037] e fit =k1e1+k2e2

[0038] C fit =f(g fit ,e fit )

[0039] Among them, g fit is the kernel function to be evaluated, e fit is the penalty coefficient to be evaluated, k1 and k2 are parameter coefficients, d is the fitness value, and gnow is the current kernel function, e1 is the adaptive weight, e2 is a random number in [0, 1], C fit =f(·) is the fitness function.

[0040] The IGW algorithm is used to optimize the kernel function and penalty coefficient in the KELM model. The search step calculation formula for the kernel function and penalty coefficient value in the IGW algorithm is as follows:

[0041] St i (t+1)=T·St i (t)+C1·cos(r1)·(best i -x i (t))+C2·sin(r2)·(best i -x i (t))

[0042] Among them, St i (t+1) is the step size for setting the time t+1, St i (t) is the step size for setting the time t, best i is the best time, x i (t) is the load data sequence number at time t, C1 and C2 are random variables between [0, 2], r1 and r2 are random numbers between [0, 1], and T is the transfer factor;

[0043] The IGW algorithm is used to optimize the kernel function and penalty coefficient in the KELM model. The update formula of the population position is:

[0044]

[0045] Among them, g(t i )e(t i ), g(t i+1 )e(t i+1 ) represent the solution vectors of the function of the i-th time period in the i-th and i+1-th time period iterations respectively; is the adaptive weight, f(g i ,f i ) is the target function, r is a random number in [0, 1]; η * is the optimal solution in the current iteration; H i Run the power ranking set for the global search for the i-th period.

[0046] The electricity price forecast result is obtained based on the historical electricity price data, which is similar to the method of wind energy load forecast result, including the following steps:

[0047] Step 1: Input historical electricity price data, perform data preprocessing, and perform Fourier Transform on the data;

[0048] Step 2: Extract the modal components and a residual component of the electricity price data;

[0049] Step 3: Normalize each component and divide it into training data set and test data set;

[0050] Step 4: Input the training data set into the IGW prediction model for training. When the output of the hidden layer of the KELM prediction model enters the output layer, the kernel function is introduced;

[0051] Step 5: Solve the minimum output error, solve the equation through RMSE and then find the minimum training error;

[0052] Step 6: Use the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model.

[0053] Beneficial effects: Compared with the prior art, the present invention has the following advantages: The present invention can reasonably regulate the charging and discharging of the hybrid energy storage system based on wind power generation and electricity prices by formulating reasonable regulation strategies for electricity prices and wind load energy supply. The VMD-IGW-KELM optimization algorithm proposed in the present invention accurately predicts electricity prices and wind loads, and supports the energy storage management module to formulate reasonable regulation strategies. The present invention not only ensures the fullness of load energy supply, but also saves the cost of power supply and improves energy utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 It is an architectural diagram of the wind energy regulation hybrid energy storage system described in the present invention. DETAILED DESCRIPTION

[0055] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0056] Example 1

[0057] The wind energy control hybrid energy storage system described in the present invention has a structure diagram as shown in FIG. Figure 1 As shown, it includes: a data acquisition module, a prediction module, an energy storage management module, a hybrid energy storage module, and an intelligent controller for high-voltage power metering.

[0058] The data collection module collects wind energy data and real-time electricity price data. The collection targets are the monthly wind energy data of a factory area in the coastal area of ​​Jiangsu and the monthly historical electricity price data published by the local government.

[0059] The prediction module includes a wind load prediction module and an electricity price prediction module. The wind load prediction module obtains load prediction results based on real-time collected wind load data, and the electricity price prediction module obtains electricity price prediction results based on monthly historical electricity price data.

[0060] The energy storage management module includes a control system, an ER energy router and a BMS; the energy router is a new type of energy management device that can realize the conversion, distribution and management of multiple energy sources, such as controlling the storage and release of energy; the control system makes decisions and controls the charging and discharging actions of the hybrid energy storage module through the charging and discharging control strategy. Specifically, the control system controls the charging and discharging of the hybrid energy storage module by controlling the ER energy router. The BMS works closely with the ER energy router. The BMS provides battery pack status information to prevent overcharging and over-discharging of the battery to ensure the safe operation of the battery. At the same time, it monitors the battery status and is responsible for monitoring the health of the energy storage module. The BMS measures the power of lithium batteries and supercapacitor devices and measures the capacity of negative hybrid energy storage modules in real time.

[0061] The hybrid energy storage module includes a lithium battery module and a supercapacitor module, which are used for storing and releasing wind energy.

[0062] The intelligent controller for high-voltage power metering is used to monitor the power at the load end of the power grid line, including active power, reactive power, apparent power, and power factor. In addition, the intelligent controller for high-voltage power metering also has a regulating function, which can control the operating status of high-voltage equipment according to real-time monitoring data to ensure the stable operation of the power grid.

[0063] Example 2

[0064] The control method of wind energy regulation hybrid energy storage described in the present invention is applied in a wind energy regulation hybrid energy storage system. The control system in the energy storage management module makes decisions and controls the charging and discharging of the hybrid energy storage module through the charging and discharging control strategy. The specific charging and discharging control strategy principle is:

[0065] The predicted wind energy load data P is obtained through the load VMD-IGW-KELM prediction module pre , predict electricity price data M ec , remember the local average electricity price The remaining capacity of the hybrid energy storage module is monitored by BMS as C hess .

[0066] Execute the discharging / charging strategy according to the wind energy load and electricity price, including 4 strategies;

[0067] Case 1: When P pre >C hess When the hybrid energy storage module discharges and the grid supplies power;

[0068] Case 2: When P pre ≤C hess When the hybrid energy storage module discharges and the grid supplies power;

[0069] Case 3: When When the power grid only charges the lithium battery module;

[0070] Case 4: When When the power grid charges the hybrid energy storage module.

[0071] The prediction steps of the load VMD-IGW-KELM prediction module of the present invention are as follows:

[0072] Step 1: Input the real-time wind load data and use the historical load data as historical data for data preprocessing. Perform Fourier Transform on the data. The transformation formula is as follows:

[0073]

[0074] in, The period is T, T tends to infinity, the fundamental frequency is ω, Bn is the data set, n is the number of modal components, and n tends to infinity.

[0075]

[0076] Among them, f n (t) is the modal component, x is the load data ordinal number, n is the number of modal components, δ is the Dirac function, and w is the frequency.

[0077] Step 2: Extract the modal components and a residual component of the wind load data.

[0078] Step 3: Normalize each component and divide it into training data set and test data set.

[0079] Step 4: Take the training data set as input x and bring it into the KELM prediction model for training to determine the output of the jth hidden layer node. The model output is the electric energy required for the predicted load.

[0080] Determine the output H(x) of the hidden layer of the KELM prediction model. The calculation formula is as follows:

[0081] H(x)=[h1(x),h2(x),…,h L (x)].

[0082] When the signal passes through the hidden layer and enters the output layer, the kernel function is introduced to replace h(x). The output of the KELM prediction model is recorded as The expression formula is as follows:

[0083]

[0084] Where Ω=HH T , h(x)=H is the output matrix of the hidden layer, H + is the generalized inverse matrix of matrix H, K(x i ,x j ) is the Gaussian kernel function, expressed as:

[0085]

[0086] ψ is the kernel function, and its value directly affects the generalization ability of KELM. In addition, the penalty coefficient C will affect the prediction accuracy of the KELM model.

[0087] Step 5: Solve the minimum output error and find the minimum training error by solving the equation using the RMSE root mean square error. The objective function to be solved is as follows:

[0088]

[0089] Among them, ξ represents the training error, Num is the total number of samples, P Mi represents the true value of wind energy load at time i, P prei Represents the predicted value of wind energy load at time i.

[0090] Step 6: Use the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model.

[0091] The IGW algorithm is used to optimize the kernel function and penalty coefficient in the KELM model, which specifically includes the following steps:

[0092] 1) Population initialization: Set the maximum number of iterations of the algorithm, the upper and lower limits, and set the search space of the algorithm according to the upper and lower limits. The IGW algorithm randomly generates a group of populations in the search space, sets the population size to P, and then starts the search process. Each population corresponds to a set of kernel functions and penalty coefficient values;

[0093] 2) Evaluate the objective function, input the kernel function and penalty coefficient parameters in the collected KELM model into the evaluated objective function, and the objective function formula is as follows:

[0094]

[0095] e fit =k1e1+k2e2

[0096] C fit =f(g fit ,e fit )

[0097] Among them, gfit is the kernel function (equal to ψ above), e fit is the penalty coefficient (equal to C above), k1 and k2 are parameter coefficients, d is the fitness value, g now is the current kernel function, e1 is the adaptive weight, e2 is a random number in [0, 1], C fit =f(·) is the fitness function, that is, the value of the objective function;

[0098] 3) Bring the initial population into the objective function for evaluation, sort the population, and take the group value of the kernel function and penalty coefficient corresponding to the population with the highest ranking as the best;

[0099] 4) Global search phase:

[0100] The randomly generated positions are fed into the objective function for evaluation, so that each of them generates a set of corresponding operating powers and sorts them:

[0101] H(t)=[f(g1,e2),f(g2,e2)......f(g n ,e n )]

[0102]

[0103] Where f(t) is the fitness value matrix, f(g n ,e n ) expresses the objective function, t and iterNum are the current iteration number and the maximum number of iterations respectively; For adaptive weights, the operating power value corresponding to the best fitness value ranking is used as the optimal value, and the operating power value corresponding to the second best fitness value ranking is used as the second best value. The position is updated by updating the positions of the optimal value and the second best value.

[0104] 5) Evaluate the step size of each kernel function and penalty coefficient value, the formula is as follows:

[0105] St i (t+1)=T·St i (t)+C1·cos(r1)·(best i -x i (t))+C2·sin(r2)·(best i -x i (t))

[0106] Where C1 and C2 are random variables between [0,2], r1 and r2 are random numbers between [0,1], and T is the transfer factor;

[0107] 6) Limit the step range and set the maximum step size; the purpose of limiting the step range is to prevent the solved kernel function and penalty coefficient values ​​from oscillating up and down, resulting in divergence of the solved group value, and introduce a reasonable interval to limit the movement of the population; setting the maximum step size is to define the maximum step size that the population can experience in its position coordinates during the iteration process;

[0108] 7) Determine whether the step length exceeds the specified position. If so, replace its position with the respective upper and lower limits; if not, go to step 8);

[0109] 8) Update the position of the population so that the target moves to the kernel function and penalty coefficient value with higher ranking in the search space;

[0110]

[0111] Among them, g(t i )e(t i ), g(t i+1 )e(t i+1 ) represent the solution vectors of the function of the i-th time period in the i-th and i+1-th time period iterations respectively; r is a random number in [0, 1]; η * is the optimal solution in the current iteration; H i is the function of the i-th period;

[0112] 9) Determine whether the set maximum number of iterations has been reached. If not, return to step 3). If yes, go to step 8);

[0113] 10) Output the value of the optimal kernel function and penalty coefficient.

[0114] The prediction steps of the electricity price VMD-IGW-KELM prediction module are similar to those of the load VMD-IGW-KELM prediction module. The specific process is as follows:

[0115] Step 1: Input historical electricity price data, perform data preprocessing, and perform Fourier Transform on the data.

[0116] Step 2: Extract the modal components and a residual component of the electricity price data.

[0117] Step 3: Normalize each component and divide it into training data set and test data set.

[0118] Step 4: Input the training data set into the IGW prediction model for training, determine the output of the jth hidden layer node, and introduce the kernel function to replace h(x) when the output of the hidden layer of the KELM prediction model enters the output layer. The output of the output layer is the predicted electricity price.

[0119] Step 5: Solve the minimum output error and solve the equation through RMSE to find the minimum training error.

[0120] Step 6: Use the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model.

Claims

1. A wind energy regulation hybrid energy storage system, characterized in that: include: A data acquisition module is used to collect real-time wind energy load data through an electric high-voltage metering intelligent controller; The prediction module includes a wind load prediction module and an electricity price prediction module. The wind load prediction module obtains a wind load prediction result based on real-time wind load data, and the electricity price prediction module obtains an electricity price prediction result based on historical electricity price data; the VMD-IGW-KELM optimization algorithm is used to predict electricity prices and wind loads; The VMD-IGW-KELM optimization algorithm uses the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model. The objective function in the IGW algorithm is: and fit =k1e1+k2e2 C fit =f(g fit ,e fit ) Among them, g fit is the kernel function to be evaluated, e fit is the penalty coefficient to be evaluated, k1 and k2 are parameter coefficients, d is the fitness value, and g now is the current kernel function, e1 is the adaptive weight, e2 is a random number in [0, 1], C fit =f(·) is the fitness function; The VMD-IGW-KELM optimization algorithm uses the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model. The search step calculation formula for the kernel function and penalty coefficient value in the IGW algorithm is as follows: St i (t+1)=T·St i (t)+C1·cos(r1)·(best i -x i (t))+C2·sin(r2)·(best i -x i (t)) Among them, St i (t+1) is the step size for setting the time t+1, St i (t) is the step size for setting the time t, best i is the best time, x i (t) is the load data sequence number at time t, C1 and C2 are random variables between [0, 2], r1 and r2 are random numbers between [0, 1], and T is the transfer factor; The update formula of population position is: Among them, g(t i )e(t i ), g(t i+1 )e(t i+1 ) represent the solution vectors of the function of the i-th time period in the i-th and i+1-th time period iterations respectively; is the adaptive weight, f(g i ,f i ) is the objective function, r is a random number in [0, 1]; η * is the optimal solution in the current iteration; H i The power ranking set for the global search operation during the i-th period; Hybrid energy storage modules, including lithium battery modules and supercapacitor modules, are used to store and release wind energy; Electric high-voltage metering intelligent controller, used to monitor the power at the load end of the power grid line; The energy storage management module includes a control system, an energy router and a BMS; the control system obtains the prediction results of the prediction module and obtains the charging and discharging decision results through the charging and discharging control strategy; the energy router controls the charging and discharging actions of the hybrid energy storage module according to the charging and discharging decision results; the BMS is used to obtain the status information of the battery pack and send it to the energy router.

2. The wind energy control hybrid energy storage system according to claim 1 is characterized in that: The power at the load end of the monitoring power grid line includes active power, reactive power, apparent power and power factor.

3. The wind energy control hybrid energy storage system according to claim 1 is characterized in that: The control system obtains a charge and discharge decision result through a charge and discharge control strategy according to the wind energy load prediction result, the electricity price prediction result, the average electricity price and the remaining capacity of the hybrid energy storage module sent by the energy router.

4. A control method for the wind energy regulation hybrid energy storage system according to claim 1, characterized in that: The charging and discharging control strategies in the control system include: Case 1: When P pre >C hess When the hybrid energy storage module discharges and the grid supplies power; Case 2: When P pre ≤C hess When the hybrid energy storage module discharges, the grid stops supplying power; Case 3: When When the power grid only charges the lithium battery module; Case 4: When When the power grid charges the hybrid energy storage module; P pre To predict wind load data, M ec To predict electricity price data, is the local average electricity price, C hess To use BMS to monitor the remaining capacity of the hybrid energy storage module.

5. The control method of the wind energy regulation hybrid energy storage system according to claim 4 is characterized in that: Obtaining wind load forecast results based on real-time wind load data includes the following steps: Step 1: Preprocess the real-time wind load data and historical load data; Step 2: Extract each modal component and a residual component of the wind load data; Step 3: Normalize each modal component and divide it into training data set and test data set; Step 4: Take the training data set as input x and train the KELM prediction model. The output of the hidden layer of the KELM prediction model enters the output layer and introduces the kernel function. The output of the KELM prediction model is: In the formula, is the output of the KELM prediction model, y is the output of the hidden layer, C represents the penalty coefficient, Ω=HH T ,K(x i ,x j ) is the Gaussian kernel function, expressed as, ψ is the kernel function; x i 、x j is the i-th and j-th item of input x; Step 5: Solve the minimum output error; Step 6: Use the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model.

6. The control method of the wind energy regulation hybrid energy storage system according to claim 5 is characterized by: The data preprocessing in step 1 is calculated as follows: Perform Fourier Transform on the data, the transformation formula is as follows: in, T' is the period, the fundamental frequency is ω, B n is the data set, n is the number of modal components, and n tends to infinity; Among them, f n (t) is the modal component, x is the load data sequence number, n is the number of modal components, δ is the Dirac function, and w is the frequency.

7. The control method of the wind energy regulation hybrid energy storage system according to claim 5 is characterized by: The electricity price forecast result is obtained based on the historical electricity price data, including the following steps: Step 1: Input historical electricity price data, perform data preprocessing, and perform Fourier Transform on the data; Step 2: Extract the modal components and a residual component of the electricity price data; Step 3: Normalize each component and divide it into training data set and test data set; Step 4: Input the training data set into the IGW prediction model for training. When the output of the hidden layer of the KELM prediction model enters the output layer, the kernel function is introduced; Step 5: Solve the minimum output error, solve the equation through RMSE and then find the minimum training error; Step 6: Use the IGW algorithm to optimize the kernel function and penalty coefficient in the KELM model.

Images