Park micro-grid energy prediction and correction method based on fuzzy language function

Abstract

The invention discloses a park micro-grid energy prediction and correction method based on a fuzzy language function. According to the method, a park micro-grid photovoltaic power generation capacity is taken as a prediction target, illuminance, temperature and air visibility are taken as core indexes to construct a park micro-grid photovoltaic power generation system prediction model, and the characteristics of high calculation speed and high responsiveness of a fuzzy language function are creatively utilized to track system prediction errors and error change rates. Core parameter indexes of the model are corrected in real time after fuzzy reasoning calculation, so that the model always tracks real-time environment characteristics to realize evolution, the park microgrid energy state is effectively predicted, a basis is provided for power dispatching and the like of the whole system, the charging and discharging time sequence of an energy storage system is guided, and the energy efficiency is improved. Therefore, a scientific solution is provided for the problem.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of distributed energy generation prediction in park micro-grid system, in particular to a park micro-grid energy prediction correction method based on fuzzy language function. BACKGROUND

[0002] At present, distributed energy has become an important energy development direction in the world. However, the park distributed energy has great intermittency and volatility. For example, the power generation of photovoltaic is greatly affected by weather and cannot be effectively predicted, thereby causing imbalance of park micro-grid scheduling, causing light abandonment or power supply gap, and further causing the rise of park energy consumption cost. In addition, inaccurate prediction leads to improper configuration of energy storage system capacity, waste of investment, etc. Although some photovoltaic power generation models have been put forward, the overall architecture and parameters are relatively fixed, which cannot cope with complex and variable weather and park surrounding environment, thereby cannot effectively eliminate the prediction error to cause the model to gradually produce large deviation in the use process, and further cannot be used. SUMMARY

[0003] The purpose of the present application is to provide a park micro-grid energy prediction correction method based on fuzzy language function, so as to realize the best planning effect as much as possible. To this end, the present application adopts the following technical solutions:

[0004] The park micro-grid energy prediction correction method based on fuzzy language function comprises the following steps:

[0005] Step (1): based on the illumination, temperature and air visibility, a photovoltaic power generation function is constructed;

[0006] Step (2): according to the sampling interval, the actual photovoltaic power generation data in the park micro-grid system is sampled in real time;

[0007] Step (3): comparing the model prediction value with the actual photovoltaic power generation, the error e(k) and the error change rate Δe(k) are calculated, k is the sampling time;

[0008] Step (4): taking the error e(k) and the error change rate Δe(k) as input variables, the error of the change of each core parameter value in the photovoltaic power generation function is calculated through fuzzy reasoning calculation, and each core parameter value in the function is corrected;

[0009] Step (5): the corrected photovoltaic power generation function is used to predict the photovoltaic power generation in the park micro-grid system.

[0010] On the basis of the above technical solutions, the present application can also adopt the following further technical solutions, or use these further technical solutions in combination:

[0011] The photovoltaic power generation function constructed in the step (1) is:

[0012]

[0013] Where: P is the actual power generation of the photovoltaic system; I is the real-time irradiance; C is the real-time visibility of the park; T is the real-time temperature of the park; P rated Under standard conditions (1000W / m) 2 Rated power generation of photovoltaic system (at 25℃); T std Standard temperature (25℃); I std Standard light intensity (1000W / m²) 2 ); β is the temperature coefficient; ΔI, ΔC, and ΔT are the real-time parameters to be corrected for light intensity, park visibility, and park temperature, respectively.

[0014] In step (3), the error e(k) and the rate of change of error Δe(k) can be expressed as:

[0015]

[0016] Among them, P pre (k) represents the predicted power generation at sampling time k; P real (k) represents the actual power generation at sampling time k.

[0017] In step (4), fuzzy reasoning is used to calculate the error in the value change of each core parameter in the photovoltaic power generation function, and the value of each core parameter in the function is corrected. The specific process is as follows:

[0018] (4.1) Perform fuzzification calculation on the error e(k) and the rate of change of error Δe(k) to generate fuzzy vectors E(k) and ΔE(k) of the error e(k) and the rate of change of error Δe(k). Divide the fuzzy subsets of the error e(k) and the rate of change of error Δe(k) into 5 levels and use a Gaussian curve function:

[0019]

[0020] Membership degrees are calculated for both error e(k) and error rate of change Δe(k).

[0021] Among them, u j (x) is the membership function; σ j c is the center value of the Gaussian function. j Let be the standard deviation of the Gaussian function; j represents the membership functions, j∈{NB,NS,ZO,PS,PB}, i.e. {negative large, negative small, zero, positive small, positive large}. Therefore, the fuzzy vectors E(k) and ΔE(k) are both 1×5 vectors.

[0022] (4.2) Construct a reasoning knowledge base and build a fuzzy evaluation matrix R.

[0023] (4.3) Using Mamdani reasoning method, fuzzy vector E(k) and ΔE(k) are carried out fuzzy reasoning, and fuzzy vector V of three real-time parameters to be corrected ΔI, ΔC, ΔT is calculated ΔI ΔC ΔT ; The reasoning implication operator used by Mamdani reasoning method is:

[0024] V = EoR (4)

[0025] Wherein, o represents "max-min" operator;

[0026] (4.4) The fuzzy vector V ΔI ΔC ΔT is de-fuzzied to obtain the final correction value ΔI, ΔC, ΔT; The fuzzy vector V ΔI ΔC ΔT is de-fuzzied by using weighted average method, and its expression is:

[0027]

[0028] Wherein, v i is the membership value of each position in the fuzzy vector V ΔI ΔC ΔT ; σ j is the membership center value of each position in the fuzzy vector (V ΔI ΔC ΔT ), that is, the center value of the above-mentioned Gaussian function in each membership interval; v * is the de-fuzzied value of the fuzzy vector V ΔI ΔC ΔT , that is, ΔI, ΔC, ΔT three real-time parameters to be corrected;

[0029] (4.5) The three real-time parameters to be corrected ΔI(k), ΔC(k), ΔT(k) are brought into formula (1), the prediction function is corrected, and P at k sampling time is calculated.

[0030] Further, for step (4.2):

[0031] When the error is negative and large, if the change of the error is also negative and fast, it indicates that the actual order power is much larger than the overall prediction value of formula (1), the negative error is large and has a further divergence trend, therefore, the error needs to be quickly eliminated in the opposite direction, which can be realized by increasing the prediction value P(k) at this time, and considering the one-way decreasing property of formula (1), therefore, V​​​​​​​​​​​​ΔI 、V ΔC 、V ΔT If(eis NB)and(ecis NS)then(Vis NM);

[0032] If(eis NB)and(ecis NS)then(Vis NM);

[0033] If(eis NB)and(ecis NS)then(Vis NM);

[0034] If(eis NB)and(ecis NS)then(Vis NM);

[0035] If(eis NB)and(ecis NS)then(Vis NM);

[0036] If(eis NB)and(ecis NS)then(Vis NM);

[0037] If(eis NB)and(ecis NS)then(Vis NM);

[0038]

[0039] The method of the application is based on the park micro-grid energy prediction correction method of the fuzzy language function, takes the park micro-grid photovoltaic power generation as the prediction target, constructs a park micro-grid photovoltaic power generation system prediction model with illumination, temperature and air visibility as core indexes, and creatively uses the characteristics of fast calculation speed and high responsiveness of the fuzzy language function to track the system prediction error and error change rate, and the core parameter indexes of the model are corrected in real time after fuzzy reasoning calculation, so that the model always tracks the real-time environmental characteristics to realize "evolution", thereby effectively predicting the energy state of the park micro-grid, providing a basis for power dispatching and the like of the whole system, and guiding the charging and discharging timing of the energy storage system, thereby providing a scientific solution to the problem. BRIEF DESCRIPTION OF DRAWINGS

[0040] Figure 1 The algorithm flowchart of the application is shown in the figure;

[0041] Figure 2 The Gaussian function curve is shown in the figure;

[0042] Figure 3 The energy prediction effect comparison chart before and after correction is shown in the figure; DETAILED DESCRIPTION

[0043] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application. In addition, the technical features involved in the various embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0044] The park micro-grid energy prediction correction method based on fuzzy language function provided by the present application comprises the following steps:

[0045] Step (1): Based on the illumination (I, unit: W / m 2 ), temperature (T, unit: ℃) and air visibility (C, unit: %), a photovoltaic power generation function is constructed.

[0046] The photovoltaic power generation function is:

[0047]

[0048] Wherein: P is the actual power generation of the photovoltaic system; I is the real-time illumination intensity; C is the real-time visibility of the park; T is the real-time temperature of the park; P rated is the rated power generation of the photovoltaic system under standard conditions (1000 W / m 2 , 25℃); T std is the standard temperature (25℃); I std is the standard illumination intensity (1000 W / m 2 ); β is the temperature coefficient; ΔI, ΔC, ΔT are real-time parameters to be corrected.

[0049] Step (2): Real-time sampling of the actual power generation data of the photovoltaic system in the park micro-grid system with k as the sampling interval.

[0050] Step (3): Comparing the model prediction value with the actual power generation of the photovoltaic system, the error e(k) and the error change rate Δe(k) are calculated.

[0051]

[0052] Wherein, P pre (k) is the prediction value at k sampling time; P real (k) is the actual value at k sampling time;

[0053] Step (4): Taking the error e(k) and the error change rate Δe(k) as input variables, the error of the change of each core parameter value in the photovoltaic power generation function is calculated through fuzzy reasoning calculation, and the core parameter value in the function is corrected. The specific process is:

[0054] (4.1) Fuzzy calculation is performed on error e(k) and error change rate Δe(k) to generate fuzzy vectors E(k) and ΔE(k) of error e(k) and error change rate Δe(k). Fuzzy subsets of error e(k) and error change rate Δe(k) are divided into 5 levels, and Gaussian curve function is adopted:

[0055]

[0056] Membership calculation is performed on error e(k) and error change rate Δe(k) respectively.

[0057] wherein, u j (x) is membership function; σ j is center value of Gaussian function, c j is standard deviation of Gaussian function, and the function curve is shown in Figure 2 j is each membership function, j ∈ {NB, NS, ZO, PS, PB}, i.e. {negative big, negative small, zero, positive small, positive big}. Therefore, fuzzy vectors E(k) and ΔE(k) are 1 × 5 vectors respectively.

[0058] (4.2) Reasoning knowledge base is constructed, and fuzzy evaluation matrix R is constructed.

[0059] When error is negative and big, if error change is also negative and fast, it indicates that actual order power is much greater than overall predicted value of formula (1), negative error is big and has further divergence trend, therefore, error needs to be quickly eliminated in reverse direction, which can be realized by increasing predicted value P(k) at this moment, and considering one-way decreasing property of (1), therefore, V ΔI , V ΔC , V ΔT need to be increased in negative direction; if error change is not very fast, correction amount should ensure not to produce big error change while eliminating error; when error is small, in addition to eliminating error, system stability also needs to be considered to prevent system from producing unnecessary overshoot or even turbulence. According to such principle, the following fuzzy correction rules can be obtained:

[0060] If (e is NB) and (ec is NB) then (V is NB);

[0061] If (e is NB) and (ec is NS) then (V is NM);

[0062] If (e is NB) and (ec is Z) then (V is NM);

[0063] If (e is NB) and (ec is PS) then (V is NS);

[0064] If (e is NB) and (ec is PB) then (Vis ZO); etc.

[0065] Similar to the above control rules, there are 25 rules in total, which are summarized in the following table:

[0066]

[0067] (4.3) Using the Mamdani reasoning method, the fuzzy vectors E(k) and ΔE(k) are subjected to fuzzy reasoning, and the fuzzy vectors V ΔI , V ΔC , V ΔT of the three real-time parameters to be corrected ΔI, ΔC and ΔT are calculated. The reasoning implication operator used in the Mamdani reasoning method is:

[0068] V = EoR (4)

[0069] Where o represents the "max-min" operator.

[0070] (4.4) The fuzzy vectors V ΔI , V ΔC , V ΔT are de-fuzzified to obtain the final correction values ΔI, ΔC and ΔT. The weighted average method is used to de-fuzzify the fuzzy vectors V ΔI , V ΔC , V ΔT , and the expression is:

[0071]

[0072] Where v i is the membership value of each position in the fuzzy vector (V ΔI , V ΔC , V ΔT ); σ j is the membership center value of each position in the fuzzy vector (V ΔI , V ΔC , V ΔT ), i.e. the center value of the aforementioned Gaussian function in each membership interval. v * is the de-fuzzified value of the fuzzy vector (V ΔI , V ΔC , V ΔT ), i.e. the three real-time parameters to be corrected ΔI, ΔC and ΔT.

[0073] (4.5) The three real-time parameters to be corrected ΔI(k), ΔC(k) and ΔT(k) are brought into formula (1), the prediction function is corrected, and P at k sampling time is calculated.

[0074] Step (5): predicting the photovoltaic power generation in the park micro-grid system by using the corrected photovoltaic power generation function.

[0075] Experimental and simulation conclusions

[0076] To verify the effectiveness of the method, the method is simulated and verified: the method selects the park micro-grid system for testing, in order to avoid some unexpected situations in the experiment and affect the normal power consumption and work of the park, a transformer is added downstream of the existing main transformer, the isolation transformer capacity is 160kVA, and a load monitor is installed on the added transformer to monitor the working state of the transformer; the total capacity of the energy storage unit (including PCS and battery pack) is about 110kWh. In the system, the rated power of photovoltaic power generation is 119kW, and the system is sampled and corrected 288 times each time (test period 24 hours, sampling period 5 minutes). The simulation is programmed based on MATLAB2024b (64bit) software, and communication is realized with the system PLC through the OPC protocol. Figure 1 The calculation flowchart of the prediction correction method is shown.

[0077] Figure 3 The comparison results of the energy prediction effect before and after correction in the simulation are shown. Figure 3 It can be seen from the comparison results that the prediction curve effect (black curve) produced by the correction algorithm is obviously better than the prediction effect (gray curve) without correction, which shows that the method has good practicability in actual prediction.

[0078] The above results all show the excellent effect of the method in the field of distributed energy power generation prediction in the park micro-grid system. It should be noted that those skilled in the art can easily understand that the above description is only a preferred embodiment of the present application and is not intended to limit the present application, and any modification, equivalent replacement and improvement made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for energy prediction and correction of industrial park microgrids based on fuzzy linguistic functions, characterized in that, The method includes the following steps: Step (1): Construct a photovoltaic power generation function based on illuminance, temperature, and air visibility; Step (2): Based on the sampling interval, sample the actual power generation data of the photovoltaic system in the park's microgrid system in real time; Step (3): Compare the model prediction with the actual photovoltaic power generation, and calculate the error e(k) and the error change rate Δe(k), where k is the sampling time; Step (4): Using the error e(k) and the error change rate Δe(k) as input variables, calculate the error of each core parameter value in the photovoltaic power generation function through fuzzy inference, and correct the value of each core parameter in the function; Step (5): Predict the photovoltaic power generation in the park's microgrid system using the modified photovoltaic power generation function.

2. The method as described in claim 1, characterized in that, The photovoltaic power generation function constructed in step (1) is as follows: Where: P is the actual power generation of the photovoltaic system; I is the real-time irradiance; C is the real-time visibility of the park; T is the real-time temperature of the park; P rated T represents the rated power output of the photovoltaic system under standard conditions. std Standard temperature; I std β represents the standard light intensity; β is the temperature coefficient; ΔI, ΔC, and ΔT are the real-time parameters to be corrected for light intensity, park visibility, and park temperature, respectively.

3. The method as described in claim 1, characterized in that, In step (3), the error e(k) and the rate of change of error Δe(k) are expressed as follows: Among them, P pre (k) represents the predicted power generation at sampling time k; P real (k) represents the actual power generation at sampling time k.

4. The method as described in claim 1, characterized in that, In step (4), fuzzy reasoning is used to calculate the error in the value change of each core parameter in the photovoltaic power generation function, and the value of each core parameter in the function is corrected. The specific process is as follows: (4.1) Perform fuzzification calculation on the error e(k) and the rate of change of error Δe(k) to generate fuzzy vectors E(k) and ΔE(k) of the error e(k) and the rate of change of error Δe(k). Divide the fuzzy subsets of the error e(k) and the rate of change of error Δe(k) into 5 levels and use Gaussian curve functions: The membership degrees of error e(k) and error change rate Δe(k) are calculated separately. Among them, u j (x) is the membership function; σ j c is the center value of the Gaussian function. j Let be the standard deviation of the Gaussian function; j represents the membership functions, j∈{NB,NS,ZO,PS,PB}, i.e. {negative large, negative small, zero, positive small, positive large}. Therefore, the fuzzy vectors E(k) and ΔE(k) are both 1×5 vectors. (4.2) Construct a reasoning knowledge base and build a fuzzy evaluation matrix R. (4.3) Using the Mamdani inference method, fuzzy inference is performed on the fuzzy vectors E(k) and ΔE(k) to calculate the fuzzy vectors V of the three real-time parameters to be corrected, ΔI, ΔC, and ΔT. ΔI V ΔC V ΔT The inference implication operator used in the Mamdani inference method is: V = EoR(4) Where o represents the "max-min" operator; (4.4) Fuzzy vector V ΔI V ΔC V ΔT The fuzzy calculations were performed to obtain the final corrected values ​​ΔI, ΔC, and ΔT; the weighted average method was then used to correct the fuzzy vector V. ΔI V ΔC V ΔT The expression for defuzzification is as follows: Among them, v i For the fuzzy vector V ΔI V ΔC V ΔT Membership values ​​at each position in the matrix; σ j For the fuzzy vector V ΔI V ΔC V ΔT The membership center values ​​at each position in the interval, i.e., the center values ​​of the aforementioned Gaussian function in each membership interval; v * For the fuzzy vector V ΔI V ΔC V ΔT The unfuzzy values ​​are the three real-time parameters to be corrected: ΔI, ΔC, and ΔT. (4.5) Substitute the three real-time parameters to be corrected, ΔI(k), ΔC(k), and ΔT(k), into formula (1) to correct the prediction function and calculate P at sampling time k.

Images