City-level micro-grid synchronous grid connection method

Abstract

The invention relates to the technical field of power grid dispatching, in particular to a city-level micro-grid synchronous grid connection method which comprises the following steps: S1, synchronous measurement; s2, obtaining a final objective function; s3, optimizing scheduling; s4, fault diagnosis and processing; s5, evaluating power supply recovery; and S6, optimizing and adjusting. According to the method, the optimal scheme meeting all targets is searched by using the multi-population genetic algorithm according to the monitored data, the optimal scheme is continuously optimized and adjusted in actual operation, and the scheduling rule and the optimal strategy of the micro-grid are found by analyzing the historical data, so that the scheduling efficiency and quality are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid dispatching, and in particular to a method for synchronously connecting a city-level microgrid to the grid. Background Art

[0002] The city-level microgrid is a new type of power system structure that combines distributed energy resources, energy storage equipment and traditional energy systems to form an efficient, clean and low-carbon city-level power supply system; synchronous grid connection means that the microgrid and the main grid remain synchronized in terms of phase, frequency, amplitude, etc. to achieve mutual exchange and complementarity of electricity; the city-level microgrid synchronous grid connection method is an important part of the city-level microgrid technology, which is of great significance to improving the reliability, operation efficiency and safety of urban power supply. At the same time, factors such as safety, stability and economy need to be considered during the grid connection process. Summary of the invention

[0003] The purpose of the present invention is to provide a method for synchronous grid connection of a city-level microgrid to solve the problems raised in the above-mentioned background technology.

[0004] In order to solve the above technical problems, the present invention aims to provide a method for synchronously connecting a city-level microgrid to the grid, comprising the following steps: S1. Synchronous measurement: Install synchronous measurement equipment between the microgrid and the target grid to obtain various parameters required for the optimized dispatching of the microgrid connected to the grid; S2. Obtaining the final objective function: establishing the final objective function of the microgrid optimal dispatching according to the microgrid grid connection constraints and various parameters; S3, Optimization and dispatching: Solve the objective function based on multi-population genetic algorithm to control the grid connection process; S4. Fault diagnosis and processing: Analyze real-time monitoring data and fault characteristics to identify the fault type and location, and take appropriate emergency measures; S5. Power supply restoration assessment: Evaluate the power supply restoration situation, analyze real-time monitoring data and power supply demand, predict the power supply restoration time, and optimize and adjust the key equipment and links in the restoration process; S6. Optimization and adjustment: Based on the deep learning algorithm, the historical data of the microgrid is learned to predict the power demand and power generation of the microgrid, and the power supply and demand of the microgrid is balanced by optimizing the power generation and load forecast of the microgrid.

[0005] As a further improvement of the present technical solution, the types of microgrids include photovoltaic, energy storage and diesel generators, and the synchronous measurement equipment in S1 includes sensors, monitoring equipment and communication systems installed on the microgrid and the main grid.

[0006] As a further improvement of the technical solution, the microgrid grid connection constraint conditions in S2 include: Balance constraints within the microgrid: in, is the microgrid load, are the output power and electrical load of the i-th microgrid at time t, is the battery charging and discharging power, discharge is positive, discharge is negative; Output power constraints of each microgrid source: in, is the output power of the i-th microgrid, and are the minimum output power and the maximum output power respectively; Interaction power constraints between microgrid and main grid: in, is the interaction power between the i-th microgrid and the main grid, and are the corresponding minimum power and maximum power respectively; Operation constraints of energy storage batteries: in, and are the minimum power and maximum power of the battery respectively; Remaining capacity constraints of the battery: in, is the remaining capacity of the battery, and They are the minimum and maximum values ​​allowed for the battery state of charge.

[0007] As a further improvement of this technical solution, the final objective function in S2 is: The time range for microgrid integration is ,exist Incorporate at all times Microgrids, is the total number of microgrids that need to be connected to the main grid, Assume that the total voltage difference generated by the microgrid being connected to the main grid is U: in, for time The total voltage difference generated when a microgrid is connected to the main grid is The maximum pressure difference that the main network can withstand; Assume that the total frequency difference caused by the microgrid being connected to the main grid is f: in, for time The total frequency difference generated when a microgrid is connected to the main grid is The maximum frequency difference that the main network can bear; Assume that the total power exchange generated by the microgrid being connected to the main grid is P: in, for time The total frequency difference generated when a microgrid is connected to the main grid is The maximum power exchange value that the main network can withstand; Microgrid The power generation cost C generated after being connected to the main grid in, for time The average operating cost of each microgrid after it is connected to the main grid; The objective function is: in, , , and is the weight set, , and The numerical value is determined by the analytic hierarchy process. , , and is the normalization coefficient.

[0008] As a further improvement of the technical solution, the step of solving the objective function based on a multi-population genetic algorithm in S3 includes: S31, Population initialization: set a series of parameters, including population size, population number, the number of generations to terminate evolution, and the initial position and fitness value of each individual, and perform chromosome encoding at the same time, with real number encoding genes corresponding to variables one by one; S32, fitness value evaluation: calculate the fitness value of each individual and evaluate the individual's quality; S33, Selection: Set the group P of size s = Sort by individual fitness in descending order , using the nonlinear normalized geometric ranking function, the selection probability of individual w is: Among them, r is the probability of selecting the best individual, b is the sequence number of individual w, and q is the adjustment value; then the roulette wheel selection is performed based on this probability; S34, Crossover: Suppose the two chromosomes are and , randomly select two points for crossover, implement crossover at point o and point p, and generate offspring as and ; S35, mutation: uniform mutation, reverse mutation and exchange mutation in coding individuals; S36, elite retention: If the fitness value of the best individual in the next generation group is less than the fitness value of the best individual in the current group, the best individual in the current group or multiple individuals with fitness values ​​greater than the fitness value of the best individual in the next generation are directly copied to the next generation, and the corresponding number of individuals in the worst next generation group are randomly replaced or replaced; S37, Immigration: Compare the fitness values ​​of the best individual in the current sub-population and the weakest individual in the target sub-population; S38, update population: recalculate individual fitness and re-evaluate individual pros and cons; S39, termination condition judgment: check whether the preset termination condition is met. If the maximum number of evolutionary iterations is reached or the change of the optimal solution is less than the preset threshold, output the current optimal solution. Otherwise, return to S33 to continue execution.

[0009] As a further improvement of the technical solution, the step of learning the historical data of the microgrid based on the deep learning algorithm in S6 to predict the power demand and power generation of the microgrid includes: S61, data collection and preprocessing: collect historical data of microgrid, clean and preprocess the data; S62. Extract features: extract relevant features according to business requirements and data characteristics; S63, Model training and optimization: Use the collected data to train the deep learning model and optimize the model's prediction performance by adjusting the model's parameters and structure; S64, prediction and optimization: Use the trained deep learning model to predict the future power demand and power generation of the microgrid. Based on the prediction results, optimize the power generation and load prediction to achieve the balance of power supply and demand in the microgrid. S65. Model evaluation and iteration: Evaluate the prediction effect and optimization results of the model to improve prediction accuracy and practical application value.

[0010] As a further improvement of the present technical solution, the features extracted in S62 include historical electricity demand, historical power generation, weather conditions and time periods. The historical electricity demand and historical power generation features are extracted based on an autoregressive moving average model, the weather condition features are extracted based on a random forest algorithm, and the time periods are classified based on a K-means algorithm.

[0011] As a further improvement of the technical solution, the future power demand and power generation of the microgrid are predicted based on a convolutional neural network in S63, and the indicators of the evaluation model in S65 include mean square error and root mean square error.

[0012] As a further improvement of the present technical solution, when predicting the future power demand and power generation of the microgrid based on the convolutional neural network in S63, historical data is used as input features and the future power demand and power generation are used as output labels.

[0013] As a further improvement of the technical solution, the mean square error calculation formula in the index of the evaluation model in S65 is: The root mean square error calculation formula in the evaluation model index in S65 is: Among them, n represents the total number of data, is the ith actual value, is the i-th predicted value or estimated value.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: in the city-level microgrid synchronous grid-connected method, a multi-population genetic algorithm is used to find the best solution that meets all objectives based on the monitored data, and it is continuously optimized and adjusted in actual operation. In addition, through the analysis of historical data, the laws and optimal strategies of microgrid scheduling are found, which greatly improves the efficiency and quality of scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is an exemplary overall flow chart of the present invention; Figure 2 is a flowchart of an exemplary optimization scheduling in the present invention; Figure 3The figure is a flowchart of exemplary optimization and adjustment in the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] like Figure 1-Figure 3 As shown, this embodiment provides a method for synchronously connecting a city-level microgrid to the grid, comprising the following steps: S1. Synchronous measurement: Synchronous measurement equipment is installed between the microgrid and the target grid to obtain various parameters required for the optimized dispatch of the grid-connected microgrid, including the type, scale, and location of the microgrid, as well as phase, frequency, amplitude and other parameters during the grid-connection process, as well as the operating status of distributed energy resources and energy storage equipment; S2. Obtaining the final objective function: establishing the final objective function of the microgrid optimal dispatching according to the microgrid grid connection constraints and various parameters; S3. Optimization and dispatching: Solve the objective function based on multi-population genetic algorithm to control the grid connection process. For the harmonic problems and power fluctuations in the grid connection process, the type and capacity of distributed power sources can be reasonably configured, and the coordinated control between power sources can be optimized to reduce the generation of harmonics and suppress power fluctuations. The use of energy storage devices such as batteries can be used to reduce peak loads and fill valleys when the power grid fluctuates, reducing the impact of power fluctuations. The use of advanced control strategies such as droop control and coordinated control can better coordinate the operation of various power sources and loads in the microgrid, reduce the impact of harmonics and power fluctuations, and timely discover harmonic and power fluctuation problems by real-time monitoring of the operation status of the microgrid, and take corresponding adjustment measures; S4, Fault diagnosis and processing: By analyzing real-time monitoring data and fault characteristics, the fault type and location can be identified, and corresponding emergency measures can be taken. For example, if a power electronic equipment fails, it can switch to backup equipment to ensure the stable operation of the microgrid; S5. Power supply restoration assessment: Evaluate the power supply restoration situation, analyze real-time monitoring data and power supply demand, predict the power supply restoration time, and optimize and adjust the key equipment and links in the restoration process; S6. Optimization and adjustment: Based on the deep learning algorithm, the historical data of the microgrid is learned to predict the power demand and power generation of the microgrid, and the power supply and demand of the microgrid is balanced by optimizing the power generation and load forecast of the microgrid.

[0018] In this embodiment, the types of microgrids include photovoltaic, energy storage and diesel generators, and the synchronous measurement equipment in S1 includes sensors, monitoring equipment and communication systems installed on the microgrid and the main grid.

[0019] Among them, the grid-connected constraints of the microgrid in S2 include: Balance constraints within the microgrid: in, is the microgrid load, are the output power and electrical load of the i-th microgrid at time t, is the battery charging and discharging power, discharge is positive, discharge is negative; Output power constraints of each microgrid source: in, is the output power of the i-th microgrid, and are the minimum output power and the maximum output power respectively; Interaction power constraints between microgrid and main grid: in, is the interaction power between the i-th microgrid and the main grid, and are the corresponding minimum power and maximum power respectively; Operation constraints of energy storage batteries: in, and are the minimum power and maximum power of the battery respectively; Remaining capacity constraints of the battery: in, is the remaining capacity of the battery, and are the minimum and maximum values ​​allowed for the battery state of charge respectively; The remaining capacity of the battery at time t and the battery The remaining power at 1 moment, It is related to the charge and discharge amount of the battery from 1 to t; When the battery is discharged, , The remaining capacity at time t is ; When the battery is charging, , The remaining capacity at time t is ; in, is the remaining capacity of the battery at time t, For charging efficiency, is the discharge efficiency.

[0020] Furthermore, the final objective function in S2 is: The time range for microgrid integration is ,exist Incorporate at all times Microgrids, is the total number of microgrids that need to be connected to the main grid, Assume that the total voltage difference generated by the microgrid being connected to the main grid is U: in, for time The total voltage difference generated when a microgrid is connected to the main grid is The maximum pressure difference that the main network can withstand; Assume that the total frequency difference caused by the microgrid being connected to the main grid is f: in, for time The total frequency difference generated when a microgrid is connected to the main grid is The maximum frequency difference that the main network can bear; Assume that the total power exchange generated by the microgrid being connected to the main grid is P: in, for time The total frequency difference generated when a microgrid is connected to the main grid is The maximum power exchange value that the main network can withstand; Microgrid The power generation cost C generated after being connected to the main grid in, for time The average operating cost of each microgrid after it is connected to the main grid; The objective function is: in, , , and is the weight set, , and The numerical value is determined by the analytic hierarchy process. , , and is the normalization coefficient.

[0021] In this embodiment, Figure 2 As shown, the steps of solving the objective function based on the multi-population genetic algorithm in S3 include: S31, population initialization: set a series of parameters, including population size, population number, the number of generations to terminate evolution, and the initial position and fitness value of each individual, and perform chromosome encoding at the same time. The real number encoding gene corresponds to the variable one by one. The variable is the time of microgrid integration and the number of microgrids integrated each time. S32, fitness value evaluation: calculate the fitness value of each individual and evaluate the individual's quality; S33, Selection: Set the group P of size s = Sort by individual fitness in descending order , using the nonlinear normalized geometric ranking function, the selection probability of individual w is: Among them, r is the probability of selecting the best individual, b is the sequence number of individual w, and q is the adjustment value; then the roulette wheel selection is performed based on this probability; S34, Crossover: Suppose the two chromosomes are and , randomly select two points for crossover, implement crossover at point o and point p, and generate offspring as and , in addition, other crossovers can be performed, such as single-point crossover, multi-point crossover, one-segment crossover, random crossover and sequential crossover; S35, mutation: uniform mutation, reverse mutation and exchange mutation in coding individuals; S36, elite retention: If the fitness value of the best individual in the next generation group is less than the fitness value of the best individual in the current group, the best individual in the current group or multiple individuals with fitness values ​​greater than the fitness value of the best individual in the next generation are directly copied to the next generation, and the corresponding number of individuals in the worst next generation group are randomly replaced or replaced; S37, Immigration: Compare the fitness values ​​of the best individual in the current sub-population and the weakest individual in the target sub-population; S38, update population: recalculate individual fitness and re-evaluate individual pros and cons; S39, termination condition judgment: check whether the preset termination condition is met. If the maximum number of evolutionary iterations is reached or the change of the optimal solution is less than the preset threshold, output the current optimal solution. Otherwise, return to S33 to continue execution.

[0022] In this embodiment, Figure 3 As shown, the steps of learning the historical data of the microgrid based on the deep learning algorithm and predicting the power demand and power generation of the microgrid in S6 include: S61, Data collection and preprocessing: Collect historical data of the microgrid, including power demand, power generation, and weather conditions, and clean and preprocess the information data, such as filling missing values, removing outliers, and normalizing; S62. Extract features: extract relevant features according to business requirements and data characteristics; S63, Model training and optimization: Use the collected data to train the deep learning model and optimize the model's prediction performance by adjusting the model's parameters and structure; S64, Prediction and Optimization: Use the trained deep learning model to predict the future power demand and power generation of the microgrid. Based on the prediction results, optimize the power generation and load prediction to achieve the balance of power supply and demand in the microgrid. Cross-validation and other methods can be used to evaluate the generalization ability of the model to ensure that the model has good performance in practical applications. S65. Model evaluation and iteration: Evaluate the prediction effect and optimization results of the model to improve prediction accuracy and practical application value.

[0023] Among them, the features extracted in S62 include historical electricity demand, historical power generation, weather conditions and time periods. The historical electricity demand and historical power generation features are extracted based on the autoregressive moving average model. The historical electricity demand and power generation data are used as input features, and the time period is used as the output label for model training and prediction. The weather condition features are extracted based on the random forest algorithm to predict the impact of different weather on circuit demand and power generation. The historical weather data is used as input features, and the corresponding circuit demand and power generation are used as output labels for model training and prediction. The time periods are classified based on the K-means algorithm. The historical electricity demand and power generation data are also used as input features, and the time period is used as the output label for model training and prediction.

[0024] Furthermore, in S63, the future power demand and power generation of the microgrid are predicted based on the convolutional neural network. A suitable convolutional neural network structure is designed, including convolutional layer, pooling layer and fully connected layer. Appropriate activation function and loss function are selected. The neural network is trained using historical data. The network weights are adjusted through the optimization algorithm until the model converges. The indicators for evaluating the model in S65 include mean square error and .

[0025] Furthermore, when predicting the future power demand and power generation of the microgrid based on the convolutional neural network in S63, the historical data is used as input features, and the future power demand and power generation are used as output labels.

[0026] Furthermore, the mean square error calculation formula in the evaluation model index in S65 is: The root mean square error calculation formula in the evaluation model in S65 is: Among them, n represents the total number of data, is the ith actual value, is the i-th predicted value or estimated value.

[0027] A person of ordinary skill in the art can understand that the process of implementing all or part of the steps of the above-mentioned embodiments can be completed by hardware, or can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium, and the above-mentioned storage medium can be a read-only memory, a disk or an optical disk, etc.

[0028] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for synchronously connecting a city-level microgrid to the grid, characterized in that: The steps include: S1. Synchronous measurement: Install synchronous measurement equipment between the microgrid and the target grid to obtain various parameters required for the optimized dispatching of the microgrid connected to the grid; S2. Obtaining the final objective function: establishing the final objective function of the microgrid optimal dispatching according to the microgrid grid connection constraints and various parameters; S3, Optimization and dispatching: Solve the objective function based on multi-population genetic algorithm to control the grid connection process; S4. Fault diagnosis and processing: Analyze real-time monitoring data and fault characteristics to identify the fault type and location, and take appropriate emergency measures; S5. Power supply restoration assessment: Evaluate the power supply restoration situation, analyze real-time monitoring data and power supply demand, predict the power supply restoration time, and optimize and adjust the key equipment and links in the restoration process; S6. Optimization and adjustment: Based on the deep learning algorithm, the historical data of the microgrid is learned to predict the power demand and power generation of the microgrid, and the power supply and demand of the microgrid is balanced by optimizing the power generation and load forecast of the microgrid.

2. The method for synchronously connecting a city-level microgrid to the grid according to claim 1, characterized in that: Microgrid types include photovoltaic, energy storage and diesel generators. The synchronous measurement equipment in S1 includes sensors, monitoring equipment and communication systems installed on the microgrid and the main grid.

3. The method for synchronously connecting a city-level microgrid to the grid according to claim 1, characterized in that: The microgrid grid connection constraints in S2 include: Balance constraints within the microgrid: in, is the microgrid load, are the output power and electrical load of the i-th microgrid at time t, is the battery charging and discharging power, discharge is positive, discharge is negative; Output power constraints of each microgrid source: in, is the output power of the i-th microgrid, and are the minimum output power and the maximum output power respectively; Interaction power constraints between microgrid and main grid: in, is the interaction power between the i-th microgrid and the main grid, and are the corresponding minimum power and maximum power respectively; Operation constraints of energy storage batteries: in, and are the minimum power and maximum power of the battery respectively; Remaining capacity constraints of the battery: in, is the remaining capacity of the battery, and They are the minimum and maximum values ​​allowed for the battery state of charge.

4. The method for synchronously connecting a city-level microgrid to the grid according to claim 1, characterized in that: The final objective function in S2 is: The time range for microgrid integration is ,exist Incorporate at all times Microgrids, is the total number of microgrids that need to be connected to the main grid, Assume that the total voltage difference generated by the microgrid being connected to the main grid is U: in, for time The total voltage difference generated when a microgrid is connected to the main grid is The maximum pressure difference that the main network can withstand; Assume that the total frequency difference caused by the microgrid being connected to the main grid is f: in, for time The total frequency difference generated when a microgrid is connected to the main grid is The maximum frequency difference that the main network can bear; Assume that the total power exchange generated by the microgrid being connected to the main grid is P: in, for time The total frequency difference generated when a microgrid is connected to the main grid is The maximum power exchange value that the main network can withstand; Microgrid The power generation cost C generated after being connected to the main grid: in, for time The average operating cost of each microgrid after it is connected to the main grid; The objective function is: in, , , and is the weight set, , and The numerical value is determined by the analytic hierarchy process. , , and is the normalization coefficient.

5. The method for synchronously connecting a city-level microgrid to the grid according to claim 1, characterized in that: The step of solving the objective function based on the multi-population genetic algorithm in S3 includes: S31, Population initialization: set a series of parameters, including population size, population number, the number of generations to terminate evolution, and the initial position and fitness value of each individual, and perform chromosome encoding at the same time, with real number encoding genes corresponding to variables one by one; S32, fitness value evaluation: calculate the fitness value of each individual and evaluate the individual's quality; S33, Selection: Set the group P of size s = Sort by individual fitness in descending order , using the nonlinear normalized geometric ranking function, the selection probability of individual w is: Among them, r is the probability of selecting the best individual, b is the sequence number of individual w, and q is the adjustment value; then the roulette wheel selection is performed based on this probability; S34, Crossover: Suppose the two chromosomes are and , randomly select two points for crossover, implement crossover at point o and point p, and generate offspring as and ; S35, mutation: uniform mutation, reverse mutation and exchange mutation in coding individuals; S36, elite retention: If the fitness value of the best individual in the next generation group is less than the fitness value of the best individual in the current group, the best individual in the current group or multiple individuals with fitness values ​​greater than the fitness value of the best individual in the next generation are directly copied to the next generation, and the corresponding number of individuals in the worst next generation group are randomly replaced or replaced; S37, Immigration: Compare the fitness values ​​of the best individual in the current sub-population and the weakest individual in the target sub-population; S38, update population: recalculate individual fitness and re-evaluate individual pros and cons; S39, termination condition judgment: check whether the preset termination condition is met. If the maximum number of evolutionary iterations is reached or the change of the optimal solution is less than the preset threshold, output the current optimal solution. Otherwise, return to S33 to continue execution.

6. The method for synchronously connecting a city-level microgrid to the grid according to claim 1, characterized in that: The step of learning the historical data of the microgrid based on the deep learning algorithm in S6 to predict the power demand and power generation of the microgrid includes: S61, data collection and preprocessing: collect historical data of microgrid, clean and preprocess the data; S62. Extract features: extract relevant features according to business requirements and data characteristics; S63, Model training and optimization: Use the collected data to train the deep learning model and optimize the model's prediction performance by adjusting the model's parameters and structure; S64, prediction and optimization: Use the trained deep learning model to predict the future power demand and power generation of the microgrid. Based on the prediction results, optimize the power generation and load prediction to achieve the balance of power supply and demand in the microgrid. S65. Model evaluation and iteration: Evaluate the model’s prediction effect and optimization results to improve prediction accuracy and practical application value.

7. The method for synchronously connecting a city-level microgrid to the grid according to claim 6, characterized in that: The features extracted in S62 include historical electricity demand, historical power generation, weather conditions and time periods. The historical electricity demand and historical power generation features are extracted based on an autoregressive moving average model, the weather condition features are extracted based on a random forest algorithm, and the time periods are classified based on a K-means algorithm.

8. The method for synchronously connecting a city-level microgrid to the grid according to claim 6, characterized in that: In S63, the future power demand and power generation of the microgrid are predicted based on the convolutional neural network, and the indicators of the evaluation model in S65 include mean square error and root mean square error.

9. The method for synchronously connecting a city-level microgrid to the grid according to claim 8, characterized in that: When predicting the future power demand and power generation of the microgrid based on the convolutional neural network in S63, historical data is used as input features and the future power demand and power generation are used as output labels.

10. The method for synchronously connecting a city-level microgrid to the grid according to claim 8, characterized in that: The mean square error calculation formula in the evaluation model index in S65 is: The root mean square error calculation formula in the evaluation model in S65 is: Among them, n represents the total number of data, is the ith actual value, is the i-th predicted value or estimated value.

Images