Method and device for predicting delay risk of photovoltaic power generation project

Abstract

The invention belongs to the field of photovoltaic technology, and discloses a photovoltaic power generation project delay risk prediction method and device, and the method comprises the steps: obtaining a construction increment value; updating the time difference sequence according to the construction increment value; inputting the time difference sequence into a progress prediction model to obtain a first delay day number; obtaining a preset standard progress sequence and a plurality of nonlinear construction sequences, and performing dimensionless processing; calculating a correlation coefficient set of the preset standard progress sequence and each nonlinear construction sequence; calculating a comprehensive risk value according to each correlation coefficient set and a corresponding preset weight; inputting the comprehensive risk value into a risk prediction model to obtain a second delay day number; and obtaining a delay risk prediction result according to the first delay day number and the second delay day number. According to the method, the information value in the nonlinear construction sequence can be fully utilized, and the accuracy of photovoltaic power generation project delay risk prediction is greatly improved.

Description

Technical Field

[0001] This application relates to the field of photovoltaic technology, and in particular, to a method and device for predicting the risk of delay in photovoltaic power generation projects. Background Art

[0002] With the development of renewable energy such as solar energy, the installed capacity of photovoltaic power generation has grown rapidly. During the construction process of photovoltaic power plants, in order to ensure that the photovoltaic project can be completed on schedule according to the project requirements, it is particularly important to predict the risk of construction period delay; in the current prior art, the methods for predicting the construction progress delay of traditional energy projects mainly use statistical models for prediction, such as linear regression, time series analysis and other models.

[0003] However, the construction progress is affected by various factors, and these factors do not all have linear characteristics. For example, the changes in data such as the attendance rate and attendance time of workers, and the operating efficiency of equipment cannot form a standard linear pattern, making it difficult for traditional statistical models to capture the relationship between non-linear data, and the prediction accuracy is poor. Summary of the Invention

[0004] This application provides a method and device for predicting the risk of delay in photovoltaic power generation projects, which can make full use of the information value in the non-linear construction sequence and greatly improve the accuracy of predicting the risk of delay in photovoltaic power generation projects.

[0005] In the first aspect, an embodiment of this application provides a method for predicting the risk of delay in a photovoltaic power generation project, including:

[0006] Obtain the construction increment value; update the time difference sequence according to the construction increment value;

[0007] Input the time difference sequence into the progress prediction model to obtain the first delay days;

[0008] Obtain the preset standard progress sequence and multiple non-linear construction sequences, and perform dimensionless processing;

[0009] Calculate the set of correlation coefficients between the preset standard progress sequence and each non-linear construction sequence;

[0010] Calculate the comprehensive risk value according to each set of correlation coefficients and the corresponding preset weights;

[0011] Input the comprehensive risk value into the risk prediction model to obtain the second delay days;

[0012] Obtain the delay risk prediction result according to the first delay days and the second delay days.

[0013] Further, the construction increment value is the difference between the construction data at the current moment and the construction data at the previous moment;

[0014] The construction data includes meteorological data and construction progress data;

[0015] The construction progress data includes the number of installed photovoltaic modules, the length of laid cables, and the number of brackets.

[0016] Furthermore, the non-linear construction sequence includes the sequence of the number of worker attendance, the sequence of equipment operation efficiency, and the sequence of equipment operation quantity; the preset standard progress sequence includes multiple time nodes and the corresponding standard construction progress at each time node.

[0017] Furthermore, the method further includes:

[0018] Obtaining the historical difference sequences and historical delay days of multiple historical photovoltaic power generation projects;

[0019] Determining multiple groups of model orders according to the autocorrelation function and partial autocorrelation function graphs;

[0020] Adopting the Bayesian criterion to determine the target order among each group of model orders;

[0021] Constructing an autoregressive integrated moving average model based on the target order;

[0022] Training the autoregressive integrated moving average model according to each historical difference sequence and the corresponding historical delay days until the residuals of the autoregressive integrated moving average model are white noise, and obtaining the trained progress prediction model.

[0023] Furthermore, the method further includes:

[0024] After obtaining the historical difference sequences, removing the outliers that exceed the preset standard range;

[0025] Performing linear interpolation to supplement the missing values in each historical difference sequence.

[0026] Furthermore, the above-mentioned adopting the Bayesian criterion to determine the target order among each group of model orders includes:

[0027] Fitting an autoregressive integrated moving average model according to the model order and calculating the maximum likelihood estimate value;

[0028] Calculating the Bayesian value according to the dimension of the historical difference sequence, the model order, and the maximum likelihood estimate value;

[0029] Selecting the model order corresponding to the smallest Bayesian value as the target order.

[0030] Furthermore, the model order includes the autoregressive term order, the differencing order, and the moving average term order.

[0031] Furthermore, the above-mentioned calculating the set of correlation coefficients between the preset standard progress sequence and each non-linear construction sequence includes:

[0032] Calculate the differences between the preset standard progress sequence and the non-linear construction sequence at the same sampling moment;

[0033] Replace each difference with its corresponding absolute value, and determine the maximum absolute value and the minimum absolute value;

[0034] Calculate the set of correlation coefficients based on the preset discrimination coefficient, the maximum absolute value, and the minimum absolute value.

[0035] Furthermore, calculating the comprehensive risk value according to each set of correlation coefficients and the corresponding preset weights includes:

[0036] Calculate the average value of each element in the set of correlation coefficients as the correlation degree of the non-linear construction sequence;

[0037] Multiply the correlation degree by the preset weight of the corresponding non-linear construction sequence to obtain the sequence risk value;

[0038] Add up the sequence risk values of each non-linear construction sequence to obtain the comprehensive risk value.

[0039] Furthermore, the method further includes:

[0040] Calculate the standard deviation of the dimensionless non-linear construction sequence;

[0041] Calculate the correlation coefficient between the non-linear construction sequence and each other non-linear construction sequence;

[0042] Calculate the conflict value of the non-linear construction sequence according to each correlation coefficient;

[0043] Multiply the conflict value by the standard deviation to obtain the information carrying capacity;

[0044] Obtain the preset weight of the non-linear construction sequence according to the information carrying capacity.

[0045] Furthermore, the method further includes:

[0046] Obtain the historical construction sequence corresponding to each non-linear construction sequence;

[0047] Calculate the coefficient of variation of each historical construction sequence based on the preset number of sliding window periods;

[0048] Determine the preset weight of the non-linear construction sequence corresponding to the historical construction sequence according to the coefficient of variation.

[0049] Furthermore, the method further includes:

[0050] After obtaining the delay risk prediction result, determine whether the delay risk prediction result is greater than the preset delay threshold;

[0051] If so, risk response suggestions are generated according to the sequence risk values of each non-linear construction sequence.

[0052] Further, the generation of risk response suggestions according to the sequence risk values of each non-linear construction sequence includes:

[0053] Sort each non-linear construction sequence in descending order of the sequence risk value;

[0054] Obtain the last element of the non-linear construction sequence ranked first;

[0055] Difference calculation step, calculate the difference between the last element and the preset maximum value of the non-linear construction sequence;

[0056] Judge whether the ratio of the difference to the preset maximum value is greater than the preset threshold;

[0057] If so, put the sequence name of the non-linear construction sequence into the risk response suggestions;

[0058] Obtain the non-linear construction sequence ranked next, and return to the difference calculation step until the number of sequence names in the risk response suggestions is equal to the preset number.

[0059] In a second aspect, another embodiment of the present application provides a prediction device for the risk of photovoltaic power generation project delay, including:

[0060] The first acquisition module is used to acquire the construction increment value; update the time difference sequence according to the construction increment value;

[0061] The first prediction module is used to input the time difference sequence into the progress prediction model to obtain the first delay days;

[0062] The second acquisition module is used to acquire the preset standard progress sequence and multiple non-linear construction sequences, and perform dimensionless processing;

[0063] The correlation module is used to calculate the correlation coefficient set of the preset standard progress sequence and each non-linear construction sequence;

[0064] The risk calculation module is used to calculate the comprehensive risk value according to each correlation coefficient set and the corresponding preset weight;

[0065] The second prediction module is used to input the comprehensive risk value into the risk prediction model to obtain the second delay days;

[0066] The comprehensive calculation module is used to obtain the delay risk prediction result according to the first delay days and the second delay days.

[0067] Further, the device further includes:

[0068] A historical acquisition module, configured to acquire the historical difference sequences and historical delay days of multiple historical photovoltaic power generation projects;

[0069] An order generation module, configured to determine multiple groups of model orders according to the autocorrelation function and partial autocorrelation function graphs;

[0070] An order determination module, configured to determine the target order among the groups of model orders by using the Bayesian criterion;

[0071] A model construction module, configured to construct an autoregressive integrated moving average model based on the target order;

[0072] A training module, configured to train the autoregressive integrated moving average model according to each historical difference sequence and the corresponding historical delay days until the residuals of the autoregressive integrated moving average model are white noise, and obtain the trained progress prediction model.

[0073] Furthermore, the device further includes:

[0074] A standard deviation module, configured to calculate the standard deviation of the dimensionless nonlinear construction sequence;

[0075] A correlation module, configured to calculate the correlation coefficients between the nonlinear construction sequence and each other nonlinear construction sequence;

[0076] A conflict module, configured to calculate the conflict value of the nonlinear construction sequence according to each correlation coefficient;

[0077] An information module, configured to multiply the conflict value by the standard deviation to obtain the information carrying capacity;

[0078] A first weight module, configured to obtain the preset weight of the nonlinear construction sequence according to the information carrying capacity.

[0079] Furthermore, the device further includes:

[0080] A historical sequence module, configured to acquire the historical construction sequences corresponding to each nonlinear construction sequence;

[0081] A variation calculation module, configured to calculate the variation coefficient of each historical construction sequence based on the preset sliding window period number;

[0082] A second weight module, configured to determine the preset weight of the nonlinear construction sequence corresponding to the historical construction sequence according to the variation coefficient.

[0083] Furthermore, the device further includes:

[0084] A suggestion generation module, configured to, after obtaining the delay risk prediction result, determine whether the delay risk prediction result is greater than the preset delay threshold; if so, generate risk response suggestions according to the sequence risk values of each nonlinear construction sequence.

[0085] In a third aspect, an embodiment of the present application provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it performs the steps of the method for predicting the risk of delay in a photovoltaic power generation project according to any one of the above embodiments.

[0086] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the method for predicting the risk of delay in a photovoltaic power generation project according to any one of the above embodiments.

[0087] In summary, compared with the prior art, the beneficial effects brought by the technical solution provided by the embodiment of the present application at least include:

[0088] For the method for predicting the risk of delay in a photovoltaic power generation project provided by the embodiment of the present application, first, for the construction data with a linear relationship, the present application uses a time series method for prediction. The time series is constructed with the construction increment value at each moment and the previous moment, which can reduce the amount of data and time required for calculation by the progress prediction model. Second, for non-linear data, the present application uses the grey relational analysis method, that is, the non-linear construction sequences that cannot be represented by a linear relationship are correlated with a preset standard progress sequence for calculation. After obtaining the comprehensive risk value, the linear relationship between the comprehensive risk value and the number of days of delay is used for model prediction. Finally, the final prediction result of the delay risk is obtained according to the prediction results of the two models. The above method constructs a linear relationship between the non-linear construction sequence and the number of days of delay through the grey relational analysis method, fully utilizes the information value in the non-linear construction sequence, and greatly improves the accuracy of predicting the risk of delay in a photovoltaic power generation project. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Figure 1 It is a flowchart of a method for predicting the risk of delay in a photovoltaic power generation project provided by an exemplary embodiment of the present application.

[0090] Figure 2 It is a flowchart of the training steps of a progress prediction model provided by an exemplary embodiment of the present application.

[0091] Figure 3 It is a flowchart of the preset weight calculation steps provided by an exemplary embodiment of the present application.

[0092] Figure 4 It is a flowchart of the steps for generating risk response suggestions provided by an exemplary embodiment of the present application.

[0093] Figure 5 It is a structural diagram of a device for predicting the risk of delay in a photovoltaic power generation project provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0094] Next, in combination with the accompanying drawings in the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0095] All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0096] Please refer to Figure 1 , the embodiments of the present application provide a method for predicting the risk of photovoltaic power generation project delay, including:

[0097] Step S11, obtaining the construction increment value; updating the time difference sequence according to the construction increment value.

[0098] Among them, the construction increment value is the difference between the construction data at the current moment and the construction data at the previous moment.

[0099] The construction data includes meteorological data and construction progress data.

[0100] The construction progress data includes the number of installed photovoltaic modules, the length of laid cables, and the number of brackets.

[0101] The meteorological data includes the temperature, humidity, wind speed, precipitation amount, and rainfall amount at the photovoltaic power generation construction site.

[0102] Specifically, the time difference sequence is composed of the differences between the construction data at each moment and the previous moment:

[0103] ΔY t =Y t -Y t-1

[0104] Among them, t and t-1 are two adjacent sampling moments. Let the construction data corresponding to the two adjacent sampling moments be subtracted, and the obtained difference forms a time difference sequence according to the sampling moments.

[0105] Step S12, inputting the time difference sequence into the progress prediction model to obtain the first number of days of delay.

[0106] Step S13, obtaining a preset standard progress sequence and a plurality of non-linear construction sequences, and performing dimensionless processing.

[0107] Among them, the non-linear construction sequences include the sequence of the number of workers on duty, the sequence of equipment operation efficiency, and the sequence of the number of operating equipment;

[0108] The equipment in the equipment operation efficiency sequence or the equipment operation quantity sequence is determined by the construction stage. For example, the most important tasks in a photovoltaic power generation project are "four connections", namely water connection, power connection, road connection, and communication. During the water connection stage, water pipes, cooling water pipes, tap water pipes, etc. need to be set up, and the corresponding construction equipment includes excavators and welding equipment; during the road connection stage, the construction equipment used includes loaders, excavators, and bulldozers, and during the power connection stage, the construction equipment used includes cranes, welding equipment, electrical testing equipment, etc.; taking the road connection stage as an example, the equipment operation efficiency sequence can include the excavation volume sequence of excavators, where each element is the total excavation cubic meters of the excavator between the current acquisition time and the previous acquisition time.

[0109] The preset standard progress sequence includes multiple time nodes and the standard construction progress corresponding to each time node.

[0110] Specifically, the dimensionless processing can select the initial value method or the mean value method.

[0111] The initial value method is to use the first data of each sequence as the reference value, and the subsequent data is divided by the reference value.

[0112] The mean value method is to divide the data of each sequence by the average value of the sequence.

[0113] Step S14, calculate the set of correlation coefficients of the preset standard progress sequence and each non-linear construction sequence.

[0114] Step S15, calculate the comprehensive risk value according to each set of correlation coefficients and the corresponding preset weights.

[0115] Step S16, input the comprehensive risk value into the risk prediction model to obtain the second delay days.

[0116] Specifically, the comprehensive risk value obtained after processing and calculating the non-linear construction sequence is an embodiment of the information value of the non-linear data, and a linear relationship can be formed with the delay days. Therefore, after obtaining the comprehensive risk value in this application, the model prediction method in the prior art can be used, and the second delay days can be predicted by training the model.

[0117] Step S17, obtain the delay risk prediction result according to the first delay days and the second delay days.

[0118] Specifically, the first delay days and the second delay days can be directly added together, or multiplied by the corresponding weights respectively (the weight of the first delay days can be determined by the natural disaster risk degree of the site selection) and then added together to obtain the delay risk prediction result.

[0119] A method for predicting the delay risk of a photovoltaic power generation project provided by the above embodiment. First, for the construction data with a linear relationship, the present application uses a time series method for prediction. The time series is constructed based on the construction increment value at each moment and the previous moment, which can reduce the amount of data and time required for calculation by the progress prediction model. Second, for non-linear data, the present application uses the grey relational analysis method, that is, the non-linear construction sequences that cannot be expressed by a linear relationship are correlated with a preset standard progress sequence for calculation. After obtaining the comprehensive risk value, the linear relationship between the comprehensive risk value and the number of delay days is used for model prediction. Finally, the final delay risk prediction result is obtained based on the prediction results of the two models. The above method constructs a linear relationship between the non-linear construction sequence and the number of delay days through the grey relational analysis method, fully utilizes the information value in the non-linear construction sequence, and greatly improves the accuracy of predicting the delay risk of a photovoltaic power generation project.

[0120] Please refer to Figure 2 , in some embodiments, the method further includes:

[0121] Step S21, obtaining the historical difference sequence and the historical number of delay days of multiple historical photovoltaic power generation projects.

[0122] Among them, the historical difference sequence is a time difference sequence composed of various linear data in the historical photovoltaic power generation project, and the historical number of delay days is the difference between the expected completion time after the completion of the historical photovoltaic power generation project and the start time.

[0123] Step S22, determining multiple groups of model orders according to the autocorrelation function and the partial autocorrelation function graph.

[0124] Among them, the model order includes the autoregressive term order, the difference order, and the moving average term order.

[0125] Specifically, by plotting the autocorrelation function (ACF) and the partial autocorrelation function (PACF) graph, the parameters p (autoregressive term order), d (difference order), and q (moving average term order) of the autoregressive integrated moving average model are initially determined.

[0126] Step S23, determining the target order among each group of model orders by using the Bayesian criterion.

[0127] Specifically, first, the autoregressive integrated moving average model is fitted according to the model order, and the maximum likelihood estimate value is calculated; then, the Bayesian value is calculated according to the dimension of the historical difference sequence, the model order, and the maximum likelihood estimate value; the formula is expressed as:

[0128] BIC = ln(n)k - 2ln(L)

[0129] Among them, n is the number of data points, that is, the number of data types in the historical difference sequence. If the meteorological data can constitute 5 time difference sequences and the construction progress data can constitute 3 time difference sequences, then n here is 8; k is the number of parameters in the model order. The model order of this application includes 3 parameters, then k=3; L is the maximum likelihood estimate of the model.

[0130] The above steps are performed for each group of model orders to obtain the BIC value corresponding to each group of model orders, that is, the Bayesian value, and the model order corresponding to the smallest Bayesian value is selected as the target order.

[0131] In addition to the Bayesian criterion, the equatorial information criterion can also be used to determine the target order.

[0132] Step S24, constructing an autoregressive integrated moving average model based on the target order.

[0133] Specifically, the autoregressive integrated moving average model can be built using the statsmodels library in Python.

[0134] Step S25, training an autoregressive integrated moving average model according to each historical difference sequence and the corresponding historical extension days until the residual of the autoregressive integrated moving average model is white noise, thereby obtaining a trained progress prediction model.

[0135] Specifically, check whether the residual of the model is white noise, that is, the ACF and PACF diagrams of the residuals have no significant lag terms within the confidence interval. At this time, it is considered that the model has been fitted and can be formally applied for prediction.

[0136] The above embodiment adopts the autoregressive integrated moving average model that is good at analyzing multivariate time series data for training, allowing each variable to be both a dependent variable and an independent variable in the model, so as to capture the dynamic relationship between variables.

[0137] In some embodiments, the method further comprises:

[0138] Step S31, after obtaining the historical difference sequence, remove the abnormal values ​​that exceed the preset standard range.

[0139] Step S32, performing linear interpolation to supplement the missing values ​​in each historical difference sequence.

[0140] Since the sensors that collect data on site may be damaged or malfunction, resulting in abnormal data or no data being collected, the data must be preprocessed before training the model to remove outliers and missing values ​​(missing values ​​can be replaced by the average of the left and right normal values) to ensure the training accuracy of the model.

[0141] In one embodiment, calculating the set of correlation coefficients between the above-mentioned preset standard progress sequence and each non-linear construction sequence includes:

[0142] Step S141, calculating each difference between the preset standard progress sequence and the non-linear construction sequence at the same sampling moment.

[0143] Specifically, the time nodes in the preset standard progress sequence are consistent with the sampling moments of each data in the non-linear construction sequence. Therefore, a one-to-one subtraction is performed on the preset standard progress sequence and the non-linear construction sequence with the same length to obtain a plurality of differences.

[0144] Step S142, replacing each difference with its corresponding absolute value, and determining the maximum absolute value and the minimum absolute value.

[0145] Step S143, calculating the set of correlation coefficients based on the preset resolution coefficient, the maximum absolute value, and the minimum absolute value.

[0146] The set of correlation coefficients includes the correlation coefficient corresponding to the absolute value of each difference, and the correlation coefficient reflects the degree of association between the non-linear construction sequence and the preset standard progress sequence at a certain moment. The calculation formula is:

[0147]

[0148] Among them, Δ min is the minimum absolute value of the difference between the two sequences, Δ max is the maximum absolute value of the difference between the two sequences, Δ t(i,0) is the difference between the data of the two sequences at the moment t(i,0), and ρ is the preset resolution coefficient, generally taking 0.5.

[0149] The correlation coefficients ξ i,0 at different moments constitute the set of correlation coefficients of this non-linear construction sequence.

[0150] In some embodiments, calculating the comprehensive risk value according to each set of correlation coefficients and the corresponding preset weights includes:

[0151] Step S151, calculating the average value of each element in the set of correlation coefficients as the degree of association of the non-linear construction sequence.

[0152] The greater the degree of association, the higher the similarity between the non-linear construction sequence and the preset standard progress sequence. The calculation formula is:

[0153]

[0154] Among them, R is the degree of association of the non-linear construction sequence, and n is the number of data in the non-linear construction sequence.

[0155] Step S152: Multiply the correlation degree by the preset weight corresponding to the non - linear construction sequence to obtain the sequence risk value.

[0156] Among them, the preset weight represents the degree of influence of the non - linear construction sequence on project delay.

[0157] Step S153: Add up the sequence risk values of each non - linear construction sequence to obtain the comprehensive risk value.

[0158] It can be considered that by weighting the correlation degree, the comprehensive risk value of delay caused by these non - linear data is obtained, and then according to the relationship between the comprehensive risk value and the number of delay days, a risk prediction model is established.

[0159] Through the grey correlation analysis method in the above - mentioned embodiments, the analysis of non - linear data in photovoltaic power generation construction can be realized, the correlation degree between each factor and construction delay can be obtained, so as to predict the project delay risk.

[0160] Please refer to Figure 3 , in some embodiments, the method further includes:

[0161] Step S41: Calculate the standard deviation of the dimensionless non - linear construction sequence.

[0162] Specifically, this step is to calculate the comparison intensity of the non - linear construction sequence, which is expressed in the form of standard deviation:

[0163]

[0164] Among them, m is the number of data in the non - linear construction sequence, x i ′ j is the i - th data in the j - th non - linear construction sequence, is the mean value of each data in the j - th non - linear construction sequence, and σ j is the standard deviation of the j - th non - linear construction sequence.

[0165] Step S42: Calculate the correlation coefficient between the non - linear construction sequence and each other non - linear construction sequence.

[0166] Step S43: Calculate the conflict value of the non - linear construction sequence according to each correlation coefficient.

[0167] The conflict value reflects the correlation degree between different non - linear construction sequences. If there is a significant positive correlation, the smaller the conflict value. The formula for calculating the conflict value between the j - th non - linear construction sequence and other non - linear construction sequences is:

[0168]

[0169] Among them, rtj is the correlation coefficient between the t-th non-linear construction sequence and the j-th non-linear construction sequence. The Pearson correlation coefficient can be used, and the calculation method of this coefficient is prior art and will not be elaborated here.

[0170] Step S44: Multiply the conflict value by the standard deviation to obtain the information carrying capacity.

[0171] Step S45: Obtain the preset weight of the non-linear construction sequence according to the information carrying capacity.

[0172] The information carrying capacity synthesizes the comparison strength and conflict. Calculate the preset weight of each sequence according to the information carrying capacity:

[0173]

[0174] C j = σ j f j

[0175] where C j is the information carrying capacity of the j-th non-linear construction sequence, and w j is the corresponding preset weight.

[0176] The objective weight assignment method given in the above embodiments is applicable to multi-attribute decision analysis, avoiding inaccurate weights caused by artificially setting preset weights subjectively, and improving the calculation accuracy of the comprehensive risk value.

[0177] In some embodiments, the method further includes:

[0178] Step S51: Obtain the historical construction sequences corresponding to each non-linear construction sequence.

[0179] Here, the correspondence means that the data types and sequence lengths of the historical construction sequence and the non-linear construction sequence are the same.

[0180] Step S52: Calculate the coefficient of variation of each historical construction sequence based on the preset number of sliding window periods.

[0181] Among them, if the sampling moments in the sequence are recorded by day, the preset window period can be set to 7 days (one week) or 30 days (one month); if it is recorded by hour, the window period can be set to 24 hours (one day).

[0182] The coefficient of variation is a statistic for measuring the degree of data dispersion, and the specific calculation steps are as follows.

[0183] First, calculate the average value of the historical construction sequence within the sliding window:

[0184]

[0185] where n is the number of data within the sliding window, y ij is the value of the i-th data within the window, and j is the label of the historical construction sequence.

[0186] Then, for each historical construction sequence, calculate its standard deviation S within the sliding window j :

[0187]

[0188] Finally, for the j-th historical construction sequence, calculate its coefficient of variation V j :

[0189]

[0190] Step S53: Determine the preset weight of the non-linear construction sequence corresponding to the historical construction sequence according to the coefficient of variation.

[0191] The formula for determining the preset weight of the j-th non-linear construction sequence according to the coefficient of variation is as follows:

[0192]

[0193] where m is the number of non-linear construction sequences, which is also the number of historical construction sequences.

[0194] The above embodiment determines the weight by performing a sliding window analysis on the historical construction sequence, which is more objective than the objective weight assignment method and avoids inaccurate weights caused by anomalies in the current construction data. However, due to the need for sliding window calculation, its computational amount and calculation time are slightly higher than those of the objective weight assignment method. Therefore, it can be selected according to actual needs in specific applications.

[0195] In some embodiments, the method further includes:

[0196] Step S61: After obtaining the delay risk prediction result, determine whether the delay risk prediction result is greater than the preset delay threshold.

[0197] Step S62: If so, generate risk response suggestions according to the sequence risk values of each non-linear construction sequence.

[0198] Although it is possible to simply use the sequence name of the non-linear construction sequence with a high sequence risk value as the risk response suggestion, a high sequence risk value does not necessarily mean that there is still room for adjustment. For example, the sequence risk value of the equipment operation quantity sequence of a certain type of equipment is relatively high, but due to many external reasons, it is impossible to call more equipment. Therefore, when giving risk response suggestions, this application also needs to consider the non-adjustable situation caused by force majeure factors.

[0199] Please refer to Figure 4, generating risk response suggestions based on the sequence risk values of each non-linear construction sequence, including:

[0200] Step S621, sort each non-linear construction sequence in descending order of the sequence risk value.

[0201] Step S622, obtain the last element of the non-linear construction sequence ranked first.

[0202] Step S623, a difference calculation step, calculate the difference between the last element and the preset maximum value of the non-linear construction sequence.

[0203] Step S624, determine whether the ratio of the difference to the preset maximum value is greater than the preset threshold.

[0204] Step S625, if so, put the sequence name of the non-linear construction sequence into the risk response suggestion.

[0205] Step S626, obtain the non-linear construction sequence ranked next, and return to the difference calculation step, i.e., Step S623, until the number of sequence names in the risk response suggestion is equal to the preset number.

[0206] First of all, sorting by the sequence risk value is because the construction sequence with a higher sequence risk value has a greater impact on the delay. Therefore, the construction sequence with a high sequence risk value is adjusted preferentially. The last element represents the data obtained from the most recent sampling of the non-linear construction sequence, that is, the latest real-time data. The preset maximum value is the upper limit that the parameters of this non-linear construction sequence can reach. For example, the maximum number of equipment operations is the maximum number that can be mobilized for this equipment set by humans, and the equipment operation efficiency is the product of the maximum number of this equipment and the maximum workload. The ratio of the difference of the last element and the preset maximum value is used to judge the adjustable degree.

[0207] For example, the maximum number of workers on duty is 500, and the last element is 476, then the difference is 24; the maximum number of bulldozers in operation is 10, and the last element is 6, then the difference is 4. It can be seen that due to different dimensions, the numerical size of the difference does not represent the adjustable degree of the corresponding sequence. Therefore, this application adopts a ratio algorithm. The ratio of the difference of the number of workers on duty to the preset maximum value is 4.8%, while the ratio of the number of bulldozers in operation is 40%. The preset threshold is 20%, then "the number of bulldozers in operation" is put into the risk response suggestion until the number of suggestions reaches the preset number, and the generation of the response strategy stops.

[0208] The above embodiment can screen out the parameters with a relatively high sequence risk value and a relatively high adjustable degree by calculating the ratio of the last element and the preset maximum value, which helps the staff to make management decisions.

[0209] Please refer to Figure 5Another embodiment of the present application provides a device for predicting the risk of delay in a photovoltaic power generation project, comprising:

[0210] The first acquisition module 101 is used to acquire the construction increment value; and update the time difference sequence according to the construction increment value.

[0211] The first prediction module 102 is used to input the time difference sequence into the progress prediction model to obtain a first delay day.

[0212] The second acquisition module 103 is used to acquire a preset standard progress sequence and a plurality of nonlinear construction sequences, and perform dimensionless processing.

[0213] The association module 104 is used to calculate the association coefficient set between the preset standard progress sequence and each nonlinear construction sequence.

[0214] The risk calculation module 105 is used to calculate the comprehensive risk value according to each correlation coefficient set and the corresponding preset weight.

[0215] The second prediction module 106 is used to input the comprehensive risk value into the risk prediction model to obtain a second extension number of days.

[0216] The comprehensive calculation module 107 is used to obtain a postponement risk prediction result according to the first postponement days and the second postponement days.

[0217] Furthermore, the device also includes:

[0218] The history acquisition module is used to obtain the historical difference series and historical extension days of multiple historical photovoltaic power generation projects.

[0219] The order generation module is used to determine multiple groups of model orders according to the autocorrelation function and the partial autocorrelation function graph.

[0220] The order determination module is used to determine the target order in each group of model orders using the Bayesian criterion.

[0221] Model building module for building an autoregressive integrated moving average model based on the target order.

[0222] The training module is used to train the autoregressive integrated moving average model according to each historical difference sequence and the corresponding historical extension days until the residual of the autoregressive integrated moving average model is white noise, thereby obtaining a trained progress prediction model.

[0223] Furthermore, the device also includes:

[0224] The preprocessing module is used to remove abnormal values ​​that exceed the preset standard range after obtaining the historical difference sequence; and to perform linear difference supplement on the missing values ​​in each historical difference sequence.

[0225] Further, the above-mentioned association module 104 is specifically configured to perform the following steps:

[0226] Calculate the differences between the preset standard progress sequence and the non-linear construction sequence at the same sampling moment.

[0227] Replace each difference with its corresponding absolute value, and determine the maximum absolute value and the minimum absolute value.

[0228] Calculate the set of correlation coefficients based on the preset resolution coefficient, the maximum absolute value, and the minimum absolute value.

[0229] Further, the above-mentioned risk calculation module 105 is specifically configured to perform the following steps:

[0230] Calculate the average value of each element in the set of correlation coefficients as the correlation degree of the non-linear construction sequence.

[0231] Multiply the correlation degree by the preset weight corresponding to the non-linear construction sequence to obtain the sequence risk value.

[0232] Add up the sequence risk values of each non-linear construction sequence to obtain the comprehensive risk value.

[0233] Further, the device further includes:

[0234] A standard deviation module, configured to calculate the standard deviation of the dimensionless non-linear construction sequence.

[0235] A correlation module, configured to calculate the correlation coefficient between the non-linear construction sequence and each other non-linear construction sequence.

[0236] A conflict module, configured to calculate the conflict value of the non-linear construction sequence according to each correlation coefficient.

[0237] An information module, configured to multiply the conflict value by the standard deviation to obtain the information carrying capacity.

[0238] A first weight module, configured to obtain the preset weight of the non-linear construction sequence according to the information carrying capacity.

[0239] Further, the device further includes:

[0240] A historical sequence module, configured to obtain the historical construction sequences corresponding to each non-linear construction sequence.

[0241] A variation calculation module, configured to calculate the coefficient of variation of each historical construction sequence based on the preset number of sliding window periods.

[0242] A second weight module, configured to determine the preset weight of the non-linear construction sequence corresponding to the historical construction sequence according to the coefficient of variation.

[0243] Further, the device further includes:

[0244] A suggestion generation module, configured to, after obtaining the extension risk prediction result, determine whether the extension risk prediction result is greater than a preset extension threshold; if so, generate a risk response suggestion according to the sequence risk values of each non-linear construction sequence.

[0245] For the specific limitations of the prediction device for the extension risk of a photovoltaic power generation project provided in this embodiment, reference may be made to the embodiment of the prediction method for the extension risk of a photovoltaic power generation project in the foregoing text, which will not be elaborated herein. Each module in the above-mentioned prediction device for the extension risk of a photovoltaic power generation project can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor in the computer device in the form of hardware or be independent of it, or can be stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0246] This application embodiment provides a computer device, which may include a processor, a memory, a network interface, and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the processor executes the steps of the prediction method for the extension risk of a photovoltaic power generation project in any of the above embodiments.

[0247] For the working process, working details, and technical effects of the computer device provided in this embodiment, reference may be made to the embodiment of the prediction method for the extension risk of a photovoltaic power generation project in the foregoing text, which will not be elaborated herein.

[0248] This application embodiment provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the prediction method for the extension risk of a photovoltaic power generation project in any of the above embodiments. Among them, the computer-readable storage medium refers to a carrier for storing data, which may include, but is not limited to, a floppy disk, an optical disk, a hard disk, a flash memory, a USB flash drive, and / or a Memory Stick, etc. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. For the working process, working details, and technical effects of the computer-readable storage medium provided in this embodiment, reference may be made to the embodiment of the prediction method for the extension risk of a photovoltaic power generation project in the foregoing text, which will not be elaborated herein.

[0249] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM).

[0250] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0251] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for predicting the risk of delay in photovoltaic power generation projects, characterized in that: include: Get the construction increment value; Update the time difference sequence according to the construction increment value; Inputting the time difference sequence into a progress prediction model to obtain the first delay days; Obtain the preset standard progress sequence and multiple nonlinear construction sequences, and perform dimensionless processing; Calculating a set of correlation coefficients between the preset standard progress sequence and each of the nonlinear construction sequences; Calculate a comprehensive risk value according to each of the association coefficient sets and the corresponding preset weights; Inputting the comprehensive risk value into a risk prediction model to obtain a second extension number of days; The extension risk prediction result is obtained based on the first extension day and the second extension day.

2. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 1, characterized in that: Also includes: Obtain the historical difference series and historical extension days of multiple historical photovoltaic power generation projects; Determine the order of multiple groups of models according to the autocorrelation function and partial autocorrelation function graphs; Using the Bayesian criterion to determine the target order in each group of model orders; Constructing an autoregressive integrated moving average model based on the target order; The autoregressive integrated moving average model is trained according to each of the historical difference sequences and the corresponding historical extension days until the residual of the autoregressive integrated moving average model is white noise, thereby obtaining the trained progress prediction model.

3. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 2, characterized in that: The method of using the Bayesian criterion to determine the target order in each group of model orders includes: Fitting an autoregressive integrated moving average model according to the model order and calculating a maximum likelihood estimate; Calculating a Bayesian value based on the dimension of the historical difference sequence, the model order and the maximum likelihood estimate; The model order corresponding to the minimum Bayesian value is selected as the target order.

4. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 1, characterized in that: The calculating of the set of correlation coefficients between the preset standard progress sequence and each of the nonlinear construction sequences includes: Calculate the differences between the preset standard progress sequence and the nonlinear construction sequence at the same sampling time; Replace each difference with the corresponding absolute value and determine the maximum absolute value and the minimum absolute value; The set of correlation coefficients is calculated based on a preset resolution coefficient, a maximum absolute value, and a minimum absolute value.

5. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 4, characterized in that: The calculating of the comprehensive risk value according to each of the association coefficient sets and the corresponding preset weights includes: Calculating the average value of each element in the correlation coefficient set as the correlation degree of the nonlinear construction sequence; The correlation degree is multiplied by the preset weight of the corresponding nonlinear construction sequence to obtain a sequence risk value; The sequence risk values ​​of each of the nonlinear construction sequences are added together to obtain the comprehensive risk value.

6. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 5, characterized in that: Also includes: Calculate the standard deviation of the dimensionless nonlinear construction sequence; Calculating the correlation coefficient between the nonlinear construction sequence and other nonlinear construction sequences; Calculating the conflict value of the nonlinear construction sequence according to each correlation coefficient; Multiplying the conflict value and the standard deviation to obtain information carrying capacity; A preset weight of the nonlinear construction sequence is obtained according to the information carrying capacity.

7. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 5, characterized in that: Also includes: Acquire a historical construction sequence corresponding to each of the nonlinear construction sequences; Calculate the coefficient of variation of each historical construction sequence based on the preset sliding window period; The preset weight of the nonlinear construction sequence corresponding to the historical construction sequence is determined according to the coefficient of variation.

8. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 5, characterized in that: Also includes: After obtaining the delay risk prediction result, determining whether the delay risk prediction result is greater than a preset delay threshold; If so, risk response suggestions are generated according to the sequence risk values ​​of each of the nonlinear construction sequences.

9. The method for predicting the risk of delay in photovoltaic power generation projects according to claim 8, characterized in that: Generating risk response suggestions according to the sequence risk values ​​of each nonlinear construction sequence includes: Sorting each of the nonlinear construction sequences in descending order of the sequence risk value; Obtaining the last element of the first-order nonlinear construction sequence; A difference calculation step, calculating the difference between the last element and a preset maximum value of the nonlinear construction sequence; Determine whether the ratio of the difference to the preset maximum value is greater than a preset threshold; If yes, then putting the sequence name of the nonlinear construction sequence into the risk response suggestion; Obtain the nonlinear construction sequence ranked next, and return to the difference calculation step until the number of sequence names in the risk response proposal is equal to a preset number.

10. A device for predicting the risk of delay in photovoltaic power generation projects, characterized in that: include: The first acquisition module is used to obtain the construction increment value; Update the time difference sequence according to the construction increment value; A first prediction module, used for inputting the time difference sequence into a progress prediction model to obtain a first delay day; The second acquisition module is used to acquire a preset standard progress sequence and multiple nonlinear construction sequences and perform dimensionless processing; A correlation module, used for calculating a set of correlation coefficients between the preset standard progress sequence and each of the nonlinear construction sequences; A risk calculation module, used to calculate a comprehensive risk value according to each of the association coefficient sets and the corresponding preset weights; A second prediction module, used for inputting the comprehensive risk value into a risk prediction model to obtain a second extension day; The comprehensive calculation module is used to obtain the extension risk prediction result according to the first extension days and the second extension days.

Images