A method for predicting power generation of zero-carbon and zero-energy building photovoltaic systems

Abstract

The present invention relates to the technical field of power generation prediction, and specifically to a method for predicting power generation of a zero-carbon zero-energy building photovoltaic system. The present invention obtains the availability of each vector element of each particle based on the correlation between the MA output value and the time in the initial prediction model at different times, and the distribution of each vector element of each particle; the speed weight of each particle is obtained by combining the difference between each power generation influencing factor data and the power generation influencing factor prediction data at all times, and the distribution characteristics of different power generation influencing factor data at the corresponding time; according to the speed weights of different particles, the speed update formula and the position update formula are corrected to obtain the global optimal position; according to the initial prediction model corresponding to the vector of the global optimal position and the real-time moment, an optimized prediction model is obtained to obtain the optimized prediction data of power generation at the next moment. The present invention optimizes the prediction model and improves the prediction accuracy by obtaining the appropriate weight of the historical moment data at the time of prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of power generation prediction, and in particular to a method for predicting power generation of a zero-carbon and zero-energy building photovoltaic system. Background Art

[0002] The zero-carbon and zero-energy building photovoltaic system achieves energy self-sufficiency through photovoltaic technology, while ensuring that the building achieves zero carbon emissions and zero energy consumption throughout its life cycle. It requires accurate prediction of power generation.

[0003] In the prior art, the ARIMA algorithm is used to predict the power generation of zero-carbon and zero-energy buildings, in which the weight of historical data is fixed in all prediction processes. However, since there are many elements that affect power generation and have a significant impact on historical data, the power generation may change according to the standard time series. The fixed weight in traditional prediction leads to poor accuracy of power generation prediction. Summary of the invention

[0004] In order to solve the technical problem that the accuracy of power generation prediction is poor due to the fixed weight in traditional prediction, the main purpose of the present invention is to provide a method for predicting power generation of a zero-carbon zero-energy building photovoltaic system, and the technical solution is as follows:

[0005] The present invention proposes a method for predicting power generation of a zero-carbon zero-energy building photovoltaic system, the method comprising:

[0006] Obtain the power generation data of the zero-carbon and zero-energy building photovoltaic system at each moment and the data of various power generation influencing factors in time series;

[0007] Construct an ARIMA initial prediction model for power generation data to obtain the power generation prediction data corresponding to the next moment at each moment, as well as the prediction data of power generation influencing factors; randomly obtain different preset weight vectors of the data moment sequence in the initial prediction model at the real-time moment based on the preset particle swarm algorithm as the initial positions of different particles;

[0008] According to the correlation between the MA output value and the time in the initial prediction model at different times, and the distribution of each vector element of each particle, the availability of each vector element of each particle is obtained; according to the availability of different vector elements of each particle, and the difference between the power generation data at the corresponding time and the power generation prediction data, the target value of each particle is obtained;

[0009] According to the difference between each power generation influencing factor data and the power generation influencing factor prediction data at all times, the availability of all vector elements of each particle, and the distribution characteristics of different power generation influencing factor data at the corresponding time, the speed weight of each particle is obtained; according to the speed weights of different particles, the speed update formula and the position update formula are corrected, and the initial speed and initial position of the particle are updated and iterated to obtain the updated position of different particles and the new target value, and obtain the global optimal position;

[0010] According to the vector corresponding to the global optimal position and the initial prediction model at the real time, an optimized prediction model is obtained to obtain the optimized prediction data of power generation at the next moment.

[0011] Furthermore, the method for obtaining the availability includes:

[0012] Obtaining the ratio of the serial number corresponding to each vector element of each particle to the number of all vector elements as the first available coefficient;

[0013] The correlation coefficient between the sequence composed of MA output values ​​in the initial prediction model at all times and the time series, the product of the first availability coefficient and each vector element value of each particle is obtained as the availability of each vector element of each particle.

[0014] Furthermore, the method for obtaining the target value includes:

[0015] Obtain the difference between the power generation data and the power generation prediction data at the moment corresponding to each vector element of each particle, and perform negative correlation mapping on the difference as the first target coefficient;

[0016] The availability of each vector element of each particle is normalized as the second objective coefficient;

[0017] The cumulative sum of the products between the first target coefficient and the second target coefficient corresponding to all vectors of each particle is obtained as the target value of each particle.

[0018] Furthermore, the method for obtaining the speed weight includes:

[0019] According to the difference between the data of each power generation influencing factor at all times and the predicted data of the power generation influencing factor, the deviation contribution of each power generation influencing factor is obtained;

[0020] For any particle, according to the distribution characteristics of the data of each power generation influencing factor at the time corresponding to each vector element, the deviation degree of each power generation influencing factor at the time corresponding to each vector element is obtained;

[0021] Obtain the product accumulation sum of the deviation contribution and deviation degree of different power generation influencing factors of each vector element of each particle at the corresponding moment, and perform negative correlation mapping as the weight contribution of each vector element of each particle;

[0022] The product-accumulation sum of the weight contributions and availability of all vector elements of each particle is obtained as the velocity weight of each particle.

[0023] Furthermore, the method for obtaining the deviation contribution includes:

[0024] The cumulative sum of the differences between the data of each power generation influencing factor and the corresponding power generation influencing factor prediction data at all times is obtained, and negative correlation mapping is performed as the deviation contribution of each power generation influencing factor.

[0025] Furthermore, the method for obtaining the degree of deviation includes:

[0026] Obtain all moments with the same data of each power generation influencing factor at the moment corresponding to each vector element as the target moment;

[0027] Obtain the cumulative sum of the differences in the data of each power generation influencing factor at the same sequence number within the corresponding neighborhood range between the time corresponding to each vector element and each target time as the local deviation value;

[0028] The mean of the local deviation values ​​between the time corresponding to each vector element and all target times is obtained, and normalized mapping is performed as the deviation degree of each power generation influencing factor at the time corresponding to each vector element.

[0029] Furthermore, the method for obtaining the global optimal position includes:

[0030] The speed update formula is weighted according to the speed weight of each particle to obtain a new speed update correction formula and a corresponding new position update correction formula;

[0031] In each iteration of the particle swarm algorithm, the target value of each particle is analyzed based on the vector corresponding to the new updated position of each particle to obtain the new target value of each particle; the target value before the position update and the new target value are compared, and the position corresponding to the minimum target value is selected as the individual optimal position, and the individual optimal position is used as the new position of each particle until the preset iteration stop condition is reached, and the new initial position of each particle is obtained as the updated position;

[0032] The target values ​​corresponding to the updated positions of all particles are obtained, and the updated position with the highest target value is taken as the global optimal position.

[0033] Furthermore, the method for obtaining the optimization prediction model includes:

[0034] The data at the corresponding moment in the initial prediction model are weighted according to the vector elements at each moment in the vector to form an optimized prediction model.

[0035] Furthermore, the ARIMA initial prediction model for power generation data is constructed to obtain power generation prediction data corresponding to the next moment at each moment, as well as power generation influencing factor prediction data, including:

[0036] Perform ADP test on the power generation data of all historical moments to determine whether the data is stable. If the test is not stable, perform differential processing on all power generation data until the test is stable, and determine the number of differentials as the differential order;

[0037] Draw an autocorrelation function diagram and a partial autocorrelation function diagram for the power generation data after differential processing to obtain multiple autoregressive orders and moving average orders; obtain the autoregressive order and moving average order corresponding to the minimum criterion value based on the AIC or BIC criterion formula as the optimal autoregressive order and the optimal moving average order;

[0038] Based on the difference order, the optimal order of autoregression and the optimal order of moving average, an ARIMA initial prediction model of power generation data is constructed to obtain the power generation prediction data corresponding to the next moment at each moment; in the ARIMA initial prediction model, the power generation data is replaced with any power generation influencing factor data to obtain the power generation influencing factor prediction data corresponding to the next moment at each moment.

[0039] Furthermore, the correlation coefficient is obtained by using the Pearson correlation coefficient.

[0040] The present invention has the following beneficial effects:

[0041] The present invention constructs an ARIMA initial prediction model for power generation data, obtains power generation prediction data corresponding to the next moment at each moment, and prediction data of power generation influencing factors. By constructing the ARIMA model, data with time correlation can be effectively processed; different preset weight vectors of the data moment sequence in the initial prediction model at the real-time moment are randomly obtained based on a preset particle swarm algorithm, as the initial positions of different particles, and by randomly generating different weight vectors as the initial positions of the particles, the diversity of the search space can be increased, and the possibility of finding a global optimal solution can be improved; according to the correlation between the MA output value and the moment in the initial prediction model at different moments, and the distribution of each vector element of each particle, the availability of each vector element of each particle is obtained, the trend and periodic change of the data are identified, and the weight vector that has a positive contribution to the prediction result is screened out; according to the availability of different vector elements of each particle, and the distribution of each vector element of each particle, the weight vector that has a positive contribution to the prediction result is obtained. The target value of each particle is obtained based on the difference between the power generation data and the power generation prediction data at the corresponding moment, and the degree of proximity of each particle to the optimal solution can be quantified; the speed weight of each particle is obtained based on the difference between each power generation influencing factor data and the power generation influencing factor prediction data at all moments, the availability of all vector elements of each particle, and the distribution characteristics of different power generation influencing factor data at the corresponding moment, which helps the particle to move more effectively in the search space; according to the speed weights of different particles, the speed update formula and the position update formula are corrected, the initial speed and initial position of the particle are updated and iterated, the updated position of different particles and the new target value are obtained, the global optimal position is obtained, and the position and speed of the particle are updated iteratively to gradually approach the global optimal solution; according to the initial prediction model corresponding to the global optimal position vector and the real-time moment, the optimized prediction model is obtained, and the optimized prediction data of power generation at the next moment is obtained. The present invention optimizes the prediction model and improves the prediction accuracy by obtaining the appropriate weight of the historical moment data during the prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 A flowchart of a method for predicting power generation of a zero-carbon and zero-energy building photovoltaic system provided by an embodiment of the present invention.

[0044] Figure 2 A flow chart of a method for obtaining a speed weight provided by an embodiment of the present invention;

[0045] Figure 3 A flow chart of a method for obtaining a degree of deviation provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the power generation prediction method of a zero-carbon zero-energy building photovoltaic system proposed by the present invention, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0047] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0048] The specific scheme of the method for predicting power generation of a zero-carbon zero-energy building photovoltaic system provided by the present invention is described in detail below with reference to the accompanying drawings.

[0049] See also Figure 1 , which shows a flow chart of a method for predicting power generation of a zero-carbon zero-energy building photovoltaic system provided by an embodiment of the present invention, and the specific method includes:

[0050] Step S1: Obtain the power generation data and various power generation influencing factor data of the zero-carbon zero-energy building photovoltaic system at each moment in time sequence.

[0051] In an embodiment of the present invention, accurate power generation prediction helps to improve the energy management efficiency of buildings and achieve building sustainability, and it is necessary to discuss the changing trend of data in historical moments; first, through the centralized monitoring system equipped with a modern photovoltaic system, the output power of multiple photovoltaic panels at each moment is collected at a time interval of 1 hour, and it is multiplied by the time interval to obtain the power generation data at the corresponding moment; considering that there are factors such as humidity, wind speed and temperature that affect photovoltaic power generation, which have a great impact on the historical moment data, different types of sensors are installed around the building to obtain data on various power generation influencing factors at the corresponding moment for analysis, therefore, the power generation data and various power generation influencing factor data at each moment in the zero-carbon zero-energy building photovoltaic system are obtained in time series.

[0052] Step S2: construct an ARIMA initial prediction model for power generation data, obtain the power generation prediction data corresponding to the next moment at each moment, and the prediction data of power generation influencing factors; based on the preset particle swarm algorithm, randomly obtain different preset weight vectors of the data moment sequence in the initial prediction model at the real-time moment as the initial positions of different particles.

[0053] Power generation data usually presents the characteristics of a time series, that is, the data presents certain trends and periodicities as time changes. The ARIMA model, as a classic model in time series analysis, can capture the characteristics of the data and thus make accurate predictions.

[0054] Preferably, in an embodiment of the present invention, an ARIMA initial prediction model of power generation data is constructed to obtain power generation prediction data corresponding to the next moment at each moment and power generation influencing factor prediction data, including:

[0055] Perform ADP test on the power generation data of all historical moments to determine whether the data is stable. If the test is not stable, perform differential processing on all power generation data until the test is stable, and determine the number of differentials as the differential order;

[0056] Draw an autocorrelation function diagram and a partial autocorrelation function diagram for the power generation data after differential processing to obtain multiple autoregressive orders and moving average orders; obtain the autoregressive order and moving average order corresponding to the minimum criterion value based on the AIC or BIC criterion formula as the optimal autoregressive order and the optimal moving average order;

[0057] Based on the difference order, the optimal order of autoregression and the optimal order of moving average, an ARIMA initial prediction model of power generation data is constructed to obtain the power generation prediction data corresponding to the next moment at each moment; in the ARIMA initial prediction model, the power generation data is replaced with any power generation influencing factor data to obtain the power generation influencing factor prediction data corresponding to the next moment at each moment.

[0058] In one embodiment of the present invention, the formula of the ARIMA initial forecasting model is expressed as:

[0059] ;

[0060] in, Indicates The power generation forecast data at the moment; represents a constant term; represents the autoregressive order; represents the moving average order; represents the coefficient of the autoregressive model; represents the coefficients of the moving average model; Indicates The error at the moment, that is, the white noise term; Indicates Power generation data at the moment; Indicates Error in time.

[0061] It should be noted that, in the embodiment of the present invention, the coefficients of the autoregressive model and the moving average model are obtained by training the ARIMA initial prediction model using the power generation data to obtain the corresponding model coefficients; the specific ARIMA prediction is a technical means well known to those skilled in the art and will not be described in detail here. The method for obtaining the prediction data of the power generation influencing factors is to replace the power generation related data in the formula with the power generation influencing factor data to obtain the power generation impact prediction data.

[0062] In order to minimize the prediction error, the particle swarm algorithm is used to find the optimal weight distribution of the data. Based on the preset particle swarm algorithm, different preset weight vectors of the data moment sequence in the initial prediction model at the real-time moment are randomly obtained as the initial positions of different particles.

[0063] It should be noted that, in one embodiment of the present invention, the preset particle swarm algorithm includes an initialization particle swarm, the number of particles is 100, the maximum number of iterations is 1000, the learning factor is 2, and the inertia weight is 0.5; wherein, the data time sequence is a sequence of the data in the initial prediction model in descending order of time. Since the weights corresponding to the data in the initial prediction model are all 1, if there is The data at each moment, then the sum of the data weights satisfies , so the preset weight vector satisfies Analyzed on the basis of.

[0064] Step S3: According to the correlation between the MA output value and the time in the initial prediction model at different times, and the distribution of each vector element of each particle, the availability of each vector element of each particle is obtained; according to the availability of different vector elements of each particle, and the difference between the power generation data at the corresponding time and the power generation prediction data, the target value of each particle is obtained.

[0065] The larger the vector element of each particle is, the greater the impact on the prediction at each moment and the greater the availability. The MA output value represents the impact of the prediction error at the historical moment on the latest prediction. If the impact of the prediction error gradually increases with the passage of time, the greater the influence of the time factor. By analyzing the correlation between the MA output value and the moment, it is determined whether the changes are consistent. If the correlation is more consistent, the distribution of the vector element is closer to the latest vector element, and the greater the availability. According to the correlation between the MA output value and the moment in the initial prediction model at different moments, and the distribution of each vector element of each particle, the availability of each vector element of each particle is obtained.

[0066] Preferably, in one embodiment of the present invention, the method for obtaining availability includes:

[0067] Obtaining the ratio of the serial number corresponding to each vector element of each particle to the number of all vector elements as the first available coefficient;

[0068] The correlation coefficient between the sequence composed of MA output values ​​in the initial prediction model at all times and the time series, the product of the first availability coefficient and each vector element value of each particle is obtained as the availability of each vector element of each particle.

[0069] In one embodiment of the present invention, the formula for availability is expressed as:

[0070] ;

[0071] in, Indicates The particle The availability of vector elements; Indicates The particle vector elements; Represents the correlation coefficient between the sequence composed of MA output values ​​in the initial forecast model at all times and the time series; Indicates The MA output value in the preset model at each historical moment; Indicates the number of historical moments; Represents the number of vector elements;

[0072] In the availability formula, the larger the correlation coefficient, the greater the impact of time contribution on the data. The larger the sequence number of a vector element, the closer it is to the latest moment, the greater the impact of time, and the greater the availability of the vector element; the larger the vector element, the greater the weight of the corresponding moment data, the greater the impact on the prediction, and the greater the availability.

[0073] It should be noted that, in one embodiment of the present invention, the method for obtaining the correlation coefficient is the Pearson correlation coefficient. In other embodiments of the present invention, DTW can also be used to obtain the relative distance between the sequence composed of MA output values ​​and the moment sequence, and negative correlation mapping can be performed to obtain the corresponding correlation coefficient. The larger the relative distance, the smaller the correlation coefficient. The specific means are technical means well known to those skilled in the art and will not be elaborated here.

[0074] The availability of vector elements reflects the analysis of reliable and effective data. By analyzing the difference between the power generation data at the corresponding moment and the power generation forecast data, the accuracy of the forecast and the complexity of the actual power generation process can be reflected. The smaller the difference, the greater the forecast accuracy, which helps to improve the overall effect and evaluate the target value of each particle. The target value of each particle is obtained based on the availability of different vector elements of each particle and the difference between the power generation data at the corresponding moment and the power generation forecast data.

[0075] Preferably, in one embodiment of the present invention, the method for obtaining the target value includes:

[0076] Obtain the difference between the power generation data and the power generation prediction data at the moment corresponding to each vector element of each particle, and perform negative correlation mapping as the first target coefficient;

[0077] The availability of each vector element of each particle is normalized as the second objective coefficient;

[0078] The cumulative sum of the products between the first target coefficient and the second target coefficient corresponding to all vectors of each particle is obtained as the target value of each particle.

[0079] In one embodiment of the present invention, the formula of the target value is expressed as:

[0080] ;

[0081] in, Indicates The target value of each particle; Indicates The particle The availability of vector elements; Indicates The particle The power generation forecast data corresponding to the moment of each vector element; Indicates The particle The power generation data at the time corresponding to each vector element; Indicates The number of vector elements for each particle; Indicates the adjustment parameter.

[0082] In the formula for the target value, Indicates The particle The availability of the first vector element and the The ratio of the sum of the availabilities of all vector elements of the particle, that is, The particle The availability of the vector elements is normalized to obtain the second target coefficient. The larger the second target coefficient, the The particle The greater the availability of the vector elements, the more conducive it is to the overall analysis and the greater the target value; China-Canada This is to avoid the formula being 0, which makes the formula meaningless. The empirical value can be 0.01; It means that a negative correlation mapping is performed on the difference between the power generation data at the moment corresponding to each vector element of each particle and the power generation prediction data, that is, the first target coefficient. The larger the difference, the less accurate the weight of the vector element corresponding to the moment, the larger the prediction error, the smaller the first target coefficient, and the smaller the target value.

[0083] Step S4: According to the difference between each power generation influencing factor data and the power generation influencing factor prediction data at all times, the availability of all vector elements of each particle, and the distribution characteristics of different power generation influencing factor data at the corresponding time, the speed weight of each particle is obtained; according to the speed weights of different particles, the initial speed and initial position of the corresponding particle are updated and iterated to obtain the updated position of different particles and the new target value, and the global optimal position is obtained.

[0084] Since different weather conditions have different deviation contributions, it is necessary to analyze each power generation influencing factor. By analyzing the correlation between the predicted data of each power generation influencing factor and the data of the power generation influencing factor, if the predicted data is closer to the actual data, the power generation influencing factor has a smaller impact on the power generation, and the speed weight is larger. The availability reflects the impact of the vector component on the prediction result. The greater the availability, the more obvious the impact. The speed weight of each particle is obtained according to the difference between the data of each power generation influencing factor and the predicted data of the power generation influencing factor at all times, the availability of all vector elements of each particle, and the distribution characteristics of the data of different power generation influencing factors at the corresponding time.

[0085] Preferably, in one embodiment of the present invention, the method for obtaining the speed weight is as follows: Figure 2 , which shows a flow chart of a method for obtaining speed weight, including:

[0086] Step S201: Obtain the deviation contribution of each power generation influencing factor according to the difference between each power generation influencing factor data and the power generation influencing factor prediction data at all times.

[0087] The power generation process is affected by a variety of actual factors, resulting in differences between the power generation influencing factor data and the power generation influencing factor prediction data. By analyzing the differences between each power generation influencing factor data and the power generation influencing factor prediction data at different times, the degree of influence of each power generation influencing factor on the power generation process can be analyzed, and the deviation contribution of the power generation influencing factor can be evaluated; according to the differences between each power generation influencing factor data and the power generation influencing factor prediction data at different times, the deviation contribution of each power generation influencing factor can be obtained.

[0088] Preferably, in one embodiment of the present invention, the method for obtaining the deviation contribution includes:

[0089] The cumulative sum of the differences between the data of each power generation influencing factor and the corresponding power generation influencing factor prediction data at all times is obtained, and negative correlation mapping is performed as the deviation contribution of each power generation influencing factor.

[0090] In one embodiment of the present invention, the formula for the deviation contribution is expressed as:

[0091] ;

[0092] in, Indicates Deviation contribution of various power generation influencing factors; Indicates Time next Prediction data of various factors affecting power generation; Indicates Time next Data on factors affecting power generation; represents the number of all moments; Represents an exponential function with a natural constant as its base.

[0093] In the formula for the deviation contribution, Perform negative correlation mapping, It means calculating the cumulative difference between each power generation influencing factor data and the corresponding power generation influencing factor prediction data at all times. The larger the cumulative difference, the greater the difference between each power generation influencing factor data and the corresponding power generation influencing factor prediction data, the greater the prediction error, the smaller the impact of the power generation influencing factor on the power generation, and the smaller the deviation contribution.

[0094] Step S202: For any particle, according to the distribution characteristics of the data of each power generation influencing factor at the time corresponding to each vector element, the deviation degree of each power generation influencing factor at the time corresponding to each vector element is obtained.

[0095] The factors affecting power generation change over time. By analyzing the distribution of the data of each factor affecting power generation at the corresponding moment of each vector element, it is helpful to understand the fluctuation of the data of each vector element at the corresponding moment relative to the whole, and more accurately evaluate the degree of deviation of each factor affecting power generation under each vector element.

[0096] Preferably, in one embodiment of the present invention, the method for obtaining the degree of deviation is as follows: Figure 3 , which shows a flow chart of a method for obtaining the degree of deviation, including:

[0097] Step S301: For any particle, all moments with the same data of each power generation influencing factor at the moment corresponding to each vector element are obtained as the target moment.

[0098] In order to analyze the fluctuation trend of each power generation influencing factor data at the time corresponding to each vector element, the time corresponding to other data of the same power generation influencing factor is selected for comparison, which is helpful for subsequent understanding of how the power generation influencing factor changes over time.

[0099] Step S302: Obtain the cumulative sum of the differences of each power generation influencing factor data at the same sequence number within the corresponding neighborhood range between the time corresponding to each vector element and each target time as the local deviation value.

[0100] By analyzing the trend differences of the data of power generation influencing factors within the neighborhood range between the corresponding moment of each vector element and each target moment, the deviation of the data of power generation influencing factors can be reflected. The smaller the difference, the smaller the corresponding local deviation value; conversely, the larger the difference, the larger the corresponding local deviation value.

[0101] It should be noted that, in one embodiment of the present invention, the method for obtaining the neighborhood range is to take any moment as the main moment, and the range is composed of a preset number of moments adjacent to the left and right sides, wherein the preset number is 10, 10 moments are obtained on each side, and the neighborhood range includes 21 moments; in other embodiments of the present invention, the size of the neighborhood range can be set according to the specific circumstances, which is not limited or elaborated here.

[0102] Step S303: obtaining the mean of the local deviation values ​​between the time corresponding to each vector element and all target times, and performing normalized mapping as the deviation degree of each power generation influencing factor at the time corresponding to each vector element.

[0103] The overall level of the local deviation between the corresponding moment of each vector element and all target moments is quantified by taking the average value, so as to more comprehensively evaluate the deviation level of the factors affecting power generation.

[0104] In one embodiment of the present invention, for any particle, the formula for the degree of deviation is expressed as:

[0105]

[0106] in, Indicates The vector element corresponds to the The degree of deviation of various factors affecting power generation; Indicates The vector element corresponds to the time in the neighborhood of The next moment Data on factors affecting power generation; Indicates The vector element corresponds to The target time is within the neighborhood of The next moment Data on factors affecting power generation; Represents the number of moments within the neighborhood; Indicates The vector elements correspond to the number of target moments; Represents the maximum and minimum normalization function.

[0107] In the formula for the degree of deviation, Indicates The vector elements correspond to the time and The target time in the neighborhood The next moment Differences in data on factors affecting power generation; Indicates that the The vector elements correspond to the time and The cumulative sum of the differences in the data of each power generation influencing factor at the same sequence number in the neighborhood between the target moments is the local deviation value. The larger the local deviation value, the greater the data difference, the weaker the time series characteristics, and the greater the deviation of the corresponding generating point influencing factor.

[0108] Step S203: Obtain the product accumulation sum of the deviation contribution and deviation degree of different power generation influencing factors of each vector element of each particle at the corresponding moment, and perform negative correlation mapping as the weight contribution of each vector element of each particle.

[0109] The greater the deviation contribution, the greater the degree of deviation, the more inconsistent fluctuations exist, and the more it is necessary to reduce the impact of the factors and perform negative correlation mapping to make the weight contribution smaller.

[0110] Step S204: Obtain the product-accumulation sum of the weight contribution and availability of all vector elements of each particle as the velocity weight of each particle.

[0111] In one embodiment of the present invention, the formula of speed weight is expressed as:

[0112] ;

[0113] in, Indicates The velocity weight of each particle; Indicates The particle The availability of vector elements; Indicates Deviation contribution of various power generation influencing factors; Indicates The vector element The degree of deviation of various factors affecting power generation; represents an exponential function with a natural constant as base; Indicates the number of factors affecting power generation; Indicates The number of vector elements per particle.

[0114] In the formula of speed weight, the exponential function with natural constant as base is used to calculate Perform negative correlation mapping, It means to obtain the product accumulation sum of the deviation contribution and deviation degree of different power generation influencing factors of each vector element of each particle at the corresponding moment. The larger the product accumulation sum, the greater the corresponding deviation contribution and deviation degree, the smaller the contribution to the speed, and the smaller the speed weight. The particle The greater the availability of a vector element, the greater its impact on the moment data analysis, and the greater the speed weighting required.

[0115] Taking into account the speed weights of different particles, updating their positions and speeds can achieve real-time tracking and optimization; according to the speed weights of different particles, the initial speed and initial position of the corresponding particles are updated and iterated to obtain the updated positions of different particles and new target values, and obtain the global optimal position.

[0116] Preferably, in one embodiment of the present invention, the method for obtaining the global optimal position includes:

[0117] The speed update formula is weighted according to the speed weight of each particle to obtain a new speed update correction formula and a corresponding new position update correction formula;

[0118] In each iteration of the particle swarm algorithm, the target value of each particle is analyzed based on the vector corresponding to the new updated position of each particle to obtain the new target value of each particle; the target value before the position update and the new target value are compared, and the position corresponding to the minimum target value is selected as the individual optimal position, and the individual optimal position is used as the new position of each particle until the preset iteration stop condition is reached, and the new initial position of each particle is obtained as the updated position;

[0119] The target values ​​corresponding to the updated positions of all particles are obtained, and the updated position with the highest target value is taken as the global optimal position.

[0120] It should be noted that the updating formulas for speed and position in the particle swarm algorithm are well known to those skilled in the art and will not be elaborated here; the preset iteration stop condition is to stop after the number of iterations is met.

[0121] Step S5: According to the vector of the position corresponding to the global optimal position and the initial prediction model at the real time, an optimized prediction model is obtained to obtain the optimized prediction data of the power generation at the next moment.

[0122] The global optimal position represents the particle that reaches the optimal solution in the entire search space. Optimizing the initial prediction model through the global optimal position can reduce errors and help improve the accuracy and generalization ability of the prediction model.

[0123] Preferably, in one embodiment of the present invention, the method for obtaining the optimized prediction model includes:

[0124] The optimized prediction model is constructed by weighting the vector elements at each moment in the vector and the data at the corresponding moment in the initial prediction model.

[0125] In one embodiment of the present invention, the formula for optimizing the prediction model is expressed as:

[0126] ;

[0127] in, Indicates The power generation forecast data at the moment; represents a constant term; represents the autoregressive order; represents the moving average order; represents the coefficient of the autoregressive model; represents the coefficients of the moving average model; Indicates Error in time; Indicates Power generation data at the moment; Indicates Error in time; Indicates The vector element corresponding to the moment.

[0128] In summary, the present invention obtains the availability of each vector element of each particle according to the correlation between the MA output value and the time in the initial prediction model at different times, and the distribution of each vector element of each particle; obtains the speed weight of each particle by combining the difference between each power generation influencing factor data and the power generation influencing factor prediction data at all times, and the distribution characteristics of different power generation influencing factor data at the corresponding time; according to the speed weights of different particles, the speed update formula and the position update formula are corrected to obtain the global optimal position; according to the initial prediction model of the global optimal position corresponding vector and the real-time moment, an optimized prediction model is obtained to obtain the optimized prediction data of the power generation at the next moment. The present invention optimizes the prediction model and improves the prediction accuracy by obtaining the appropriate weight of the historical moment data during the prediction.

[0129] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0130] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

Claims

1. A method for predicting power generation of a zero-carbon zero-energy building photovoltaic system, characterized in that: The method comprises: Obtain the power generation data of the zero-carbon and zero-energy building photovoltaic system at each moment and the data of various power generation influencing factors in time series; Construct an ARIMA initial prediction model for power generation data to obtain the power generation prediction data corresponding to the next moment at each moment, as well as the prediction data of power generation influencing factors; randomly obtain different preset weight vectors of the data moment sequence in the initial prediction model at the real-time moment based on the preset particle swarm algorithm as the initial positions of different particles; According to the correlation between the MA output value and the time in the initial prediction model at different times, and the distribution of each vector element of each particle, the availability of each vector element of each particle is obtained; according to the availability of different vector elements of each particle, and the difference between the power generation data at the corresponding time and the power generation prediction data, the target value of each particle is obtained; According to the difference between each power generation influencing factor data and the power generation influencing factor prediction data at all times, the availability of all vector elements of each particle, and the distribution characteristics of different power generation influencing factor data at the corresponding time, the speed weight of each particle is obtained; according to the speed weights of different particles, the speed update formula and the position update formula are corrected, and the initial speed and initial position of the particle are updated and iterated to obtain the updated position of different particles and the new target value, and obtain the global optimal position; According to the vector corresponding to the global optimal position and the initial prediction model at the real time, an optimized prediction model is obtained to obtain the optimized prediction data of power generation at the next moment; The method for obtaining the speed weight includes: According to the difference between the data of each power generation influencing factor at all times and the predicted data of the power generation influencing factor, the deviation contribution of each power generation influencing factor is obtained; For any particle, according to the distribution characteristics of the data of each power generation influencing factor at the time corresponding to each vector element, the deviation degree of each power generation influencing factor at the time corresponding to each vector element is obtained; Obtain the product accumulation sum of the deviation contribution and deviation degree of different power generation influencing factors of each vector element of each particle at the corresponding moment, and perform negative correlation mapping as the weight contribution of each vector element of each particle; Obtain the product-accumulation sum of the weight contribution and availability of all vector elements of each particle as the velocity weight of each particle; The method for obtaining the deviation contribution includes: Obtain the cumulative sum of the differences between the data of each power generation influencing factor and the corresponding power generation influencing factor prediction data at all times, and perform negative correlation mapping as the deviation contribution of each power generation influencing factor; The method for obtaining the degree of deviation includes: Obtain all moments with the same data of each power generation influencing factor at the moment corresponding to each vector element as the target moment; Obtain the cumulative sum of the differences in the data of each power generation influencing factor at the same sequence number within the corresponding neighborhood range between the time corresponding to each vector element and each target time as the local deviation value; The mean of the local deviation values ​​between the time corresponding to each vector element and all target times is obtained, and normalized mapping is performed as the deviation degree of each power generation influencing factor at the time corresponding to each vector element.

2. The method for predicting power generation of a zero-carbon zero-energy building photovoltaic system according to claim 1, characterized in that: The method for obtaining the availability includes: Obtaining the ratio of the serial number corresponding to each vector element of each particle to the number of all vector elements as the first available coefficient; The correlation coefficient between the sequence composed of MA output values ​​in the initial prediction model at all times and the time series, the product of the first availability coefficient and each vector element value of each particle is obtained as the availability of each vector element of each particle.

3. The method for predicting power generation of a zero-carbon zero-energy building photovoltaic system according to claim 1, characterized in that: The method for obtaining the target value includes: Obtain the difference between the power generation data and the power generation prediction data at the moment corresponding to each vector element of each particle, and perform negative correlation mapping on the difference as the first target coefficient; The availability of each vector element of each particle is normalized as the second objective coefficient; The cumulative sum of the products between the first target coefficient and the second target coefficient corresponding to all vectors of each particle is obtained as the target value of each particle.

4. The method for predicting power generation of a zero-carbon zero-energy building photovoltaic system according to claim 1, characterized in that: The method for obtaining the global optimal position includes: The speed update formula is weighted according to the speed weight of each particle to obtain a new speed update correction formula and a corresponding new position update correction formula; In each iteration of the particle swarm algorithm, the target value of each particle is analyzed based on the vector corresponding to the new updated position of each particle to obtain the new target value of each particle; the target value before the position update and the new target value are compared, and the position corresponding to the minimum target value is selected as the individual optimal position, and the individual optimal position is used as the new position of each particle until the preset iteration stop condition is reached, and the new initial position of each particle is obtained as the updated position; The target values ​​corresponding to the updated positions of all particles are obtained, and the updated position with the highest target value is taken as the global optimal position.

5. The method for predicting power generation of a zero-carbon zero-energy building photovoltaic system according to claim 1, characterized in that: The method for obtaining the optimization prediction model includes: The data at the corresponding moment in the initial prediction model are weighted according to the vector elements at each moment in the vector to form an optimized prediction model.

6. The method for predicting power generation of a zero-carbon zero-energy building photovoltaic system according to claim 1, characterized in that: The ARIMA initial prediction model for power generation data is constructed to obtain power generation prediction data corresponding to the next moment at each moment, as well as power generation influencing factor prediction data, including: Perform ADP test on the power generation data of all historical moments to determine whether the data is stable. If the test is not stable, perform differential processing on all power generation data until the test is stable, and determine the number of differentials as the differential order; Draw an autocorrelation function diagram and a partial autocorrelation function diagram for the power generation data after differential processing to obtain multiple autoregressive orders and moving average orders; obtain the autoregressive order and moving average order corresponding to the minimum criterion value based on the AIC or BIC criterion formula as the optimal autoregressive order and the optimal moving average order; Based on the difference order, the optimal order of autoregression and the optimal order of moving average, an ARIMA initial prediction model of power generation data is constructed to obtain the power generation prediction data corresponding to the next moment at each moment; in the ARIMA initial prediction model, the power generation data is replaced with any power generation influencing factor data to obtain the power generation influencing factor prediction data corresponding to the next moment at each moment.

7. The method for predicting power generation of a zero-carbon zero-energy building photovoltaic system according to claim 2, characterized in that: The method for obtaining the correlation coefficient is the Pearson correlation coefficient.

Images