A novel power system planning and construction phase business weight analysis method

Abstract

The application discloses a novel power system planning and construction stage business weight analysis method, relates to the technical field of power systems, adopts PCA to screen influence factors, obtains a group of linearly independent influence factors, reduces the repetitiveness of calculation, reduces unnecessary troubles for subsequent research, and makes the comprehensive algorithm have the advantages of fast training speed and good convergence. Data analysis is carried out on the main influence factors, the influence weight of business on voltage quality problems is obtained by using PSO_BP, problem decision of a power grid enterprise is assisted according to the weight result, and thus the voltage quality is improved. The PSO_BP algorithm combines a particle swarm algorithm and a neural network, optimizes initial weight values and threshold values, improves the performance of the neural network, makes the network not fall into a local optimal solution in a retraining process, obtains a global optimal solution, makes the voltage qualification rate maximum, and optimizes the voltage quality.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of power systems, and particularly relates to a novel power system planning and construction phase business weight analysis method. BACKGROUND

[0002] With the large access of wind, light and other new energy, the power grid gradually changes from the traditional power system form to the new power system form with more randomness, volatility and intermittency. On the one hand, the nonlinear load in the new power system gradually increases, which continuously increases the influence on the voltage quality and easily threatens the economic and safe operation of the power system. On the other hand, more and more wind power, hydroelectric power and photovoltaic power are connected to the power grid, which leads to the poor stability of the new power system and the voltage prone to flicker, fluctuation, harmonic and other power quality problems.

[0003] The planning and construction of the power system is an important prerequisite to ensure that the power grid has a strong physical structure and reliable operation capability. The advantages and disadvantages of the planning and construction results of the new power system directly determine the load characteristics and operation characteristics of the power system in a long time scale, which has an extremely important influence on the voltage quality of the power system. The business in the planning and construction phase of the new power system mainly includes load forecasting, network planning, reactive power planning and project approval, which respectively undertake different tasks and have important influence on the planning and construction results and then have a key influence on the voltage quality of the power system. However, the influence of each business on the voltage quality is different, and the corresponding planning and construction measures and adjustment degree are different. Therefore, it is urgent to propose a novel power system planning and construction phase business weight analysis method to quantitatively analyze the influence weight of each business, to assist the power grid enterprise in making decisions according to the weight, and to focus on the business with a larger weight.

[0004] Most of the traditional business evaluation relies on subjective methods such as analytic hierarchy process, without considering the influence of each business on the voltage quality, and the objective analysis of each business in the actual power system planning and construction is not sufficient, and the decisions made are mostly difficult to consider the influence of objective data; most of the existing business evaluation models independently evaluate a single business, rarely evaluate the whole process of the business system, and also do not consider the influence and guiding role of the weight on the decision of the power grid enterprise, therefore, the application proposes a novel power system planning and construction phase business weight analysis method based on an improved PSO_BP algorithm, which optimizes the neural network by using principal component analysis and particle swarm algorithm, avoids the local optimal problem, and improves the speed of the algorithm and the accuracy of quantitative evaluation. SUMMARY

[0005] The present application aims to provide a novel power system planning and construction phase business weight analysis method, so as to overcome the shortcomings that the existing business evaluation model mostly independently evaluates a single business, rarely evaluates the whole process business system, and does not consider the influence and guiding role of weight on the decision of the power grid enterprise.

[0006] To achieve the above-mentioned purpose, the present application provides a novel power system planning and construction phase business weight analysis method, comprising the following steps:

[0007] The main quantitative factors of the business of the power grid planning and construction dimension are extracted by principal component analysis;

[0008] The initial weight and threshold of the BP neural network are optimized by the PSO algorithm, and the PSO_BP neural network is constructed according to the optimized initial weight and threshold;

[0009] The main quantitative factors are input into the PSO_BP neural network for training, so as to obtain the influence weight of each business of the power grid planning and construction dimension on the voltage quality problem.

[0010] Preferably, the business of the power grid planning and construction dimension is divided into two stages: the distribution network planning and design stage and the engineering procurement and construction stage; the distribution network planning and design stage comprises load forecasting, network frame planning, reactive power planning and scheme evaluation, and the engineering procurement and construction stage comprises project approval, equipment selection, budget and engineering construction.

[0011] Preferably, the main quantitative factors of the business of the power grid planning and construction dimension are extracted by principal component analysis, and specifically comprise:

[0012] All influence factors of the business of the power grid planning and construction dimension are recorded as quantitative factors;

[0013] The quantitative factors are standardized;

[0014] The correlation coefficient is calculated, and the correlation coefficient matrix is calculated in combination with the standardized quantitative factors;

[0015] The eigenvector and eigenvalue of the correlation coefficient matrix are calculated;

[0016] The variance contribution rate and cumulative variance contribution rate are calculated according to the eigenvalue, and the eigenvector of each eigenvalue is calculated;

[0017] The principal component is calculated according to the eigenvalue;

[0018] The principal component coefficient is calculated, and the weight of all quantitative factors is calculated according to the principal component coefficient, variance contribution rate and cumulative variance contribution rate;

[0019] The main quantitative factors are screened according to the weight.

[0020] Preferably, the main quantitative factors include: load characteristics, grid structure, power supply radius, power supply line diameter, tie-in switch distribution, reactive equipment quantity and specification, reactive equipment location, compensation mode, line loss, investment and maintenance cost, optimal planning scheme, investment cost, equipment operation and inspection cost, voltage control effect, engineering construction period, construction expansion timing, main equipment selection, incoming and outgoing line scheme, annual capital construction investment plan, annual technical transformation investment plan, annual operation and maintenance investment plan, equipment procurement, capital construction, equipment installation and acceptance, and pre-operation preparation work.

[0021] Preferably, the initial weight and threshold of the BP neural network are optimized by using the PSO algorithm, and the method specifically includes the following steps:

[0022] The speed and position of each particle are initialized, and the particle is the initial connection weight and threshold of each layer of the BP neural network;

[0023] The fitness value of each particle is calculated;

[0024] The position and speed of each particle are adjusted;

[0025] It is judged whether the particle is in the feasible region, if yes, the next step is entered, and if not, the previous step is returned;

[0026] It is judged whether the particle meets the end condition, if yes, the optimization is ended, otherwise, the fitness value of the particle is recalculated and judged until the optimization is ended, and the optimal initial weight and threshold of the BP neural network are obtained.

[0027] Preferably, the PSO_BP neural network is constructed according to the optimized initial weight and threshold, and the method specifically includes:

[0028] The BP neural network is established, and a learning mode is provided to the PSO_BP neural network, the PSO_BP neural network including an input layer, a hidden layer and an output layer;

[0029] The net input and output of each unit of the hidden layer and the output layer are calculated;

[0030] The generalized error of each unit of the hidden layer and the output layer is calculated;

[0031] The initial weight and threshold of the BP neural network are optimized according to the PSO algorithm, the connection weight between the hidden layer and the output layer and the threshold of each unit of the hidden layer are adjusted, and the connection weight between the input layer and the hidden layer and the threshold of each unit of the hidden layer are adjusted;

[0032] The iteration learning times of the BP neural network are updated, and the iteration times are increased by one each time the iteration is performed once;

[0033] Judge whether all the training of the iteration learning number is completed, when the training is completed, the iteration learning number is updated again, the next step is entered, otherwise the learning module mode is acquired again to train the PSO_BP neural network;

[0034] Judge whether the voltage qualified rate of the output data of the BP neural network meets the condition, or whether the learning iteration number reaches the upper limit, if both are, the training of the PSO_BP neural network is completed, the PSO_BP neural network is obtained, otherwise the learning module mode is acquired again to train the PSO_BP neural network.

[0035] Preferably, the BP neural network adopts RMSE.

[0036] Preferably, in the BP neural network, the BP neurons are distributed in the input layer, and the radial basis function RBF is distributed in the hidden layer.

[0037] Compared with the prior art, the present application has the following beneficial effects:

[0038] The novel power system planning and construction stage business weight analysis method provided by the present application adopts PCA to screen the influence factors, obtains a group of linearly independent influence factors, reduces the repetitiveness of calculation, reduces unnecessary troubles for subsequent research, makes the comprehensive algorithm have the advantages of fast training speed and good convergence, analyzes the main influence factors, obtains the influence weight of the business on the voltage quality problem by using PSO_BP, and assists the power grid enterprise in problem decision-making according to the weight result, so as to improve the voltage quality. The PSO_BP algorithm combines the particle swarm algorithm and the neural network, optimizes the initial weight value and the threshold value, improves the performance of the neural network, makes the network not fall into a local optimal solution in the retraining process, obtains a global optimal solution, makes the voltage qualified rate maximum, and optimizes the voltage quality. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the present application, the drawings needed in the embodiment description will be briefly introduced below, and obviously, the drawings in the following description are only one embodiment of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0040] Figure 1 is a flow chart of a novel power system planning and construction stage business weight analysis method of the present application;

[0041] Figure 2 is a flow chart of the present application for optimizing the initial weight value and the threshold value of the BP neural network by using the PSO algorithm;

[0042] Figure 3This is a schematic diagram of the BP neural network structure of the present invention;

[0043] Figure 4 This is a flowchart of the present invention for constructing a PSO_BP neural network based on optimized initial weights and thresholds. Detailed Implementation

[0044] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] like Figure 1 As shown, the novel business weight analysis method for the planning and construction phase of a power system provided by this invention includes the following steps:

[0046] S1. Principal component analysis is used to extract the main quantitative factors of business in the power grid planning and construction dimension;

[0047] The business of power grid planning and construction is divided into two stages: distribution network planning and design stage and engineering procurement and construction stage. The distribution network planning and design stage includes load forecasting, grid planning, reactive power planning and scheme evaluation. The engineering procurement and construction stage includes project initiation, equipment selection, budgeting and engineering construction.

[0048] Principal Component Analysis (PCA), as a dimensionality reduction algorithm, is based on orthogonal transformation. Taking two quantization factors, α and β, as an example, rotating them by an angle θ reduces their dispersion. The transformation process is as follows:

[0049]

[0050] That is to say

[0051]

[0052] in,

[0053]

[0054] Equation (3) shows that U is an orthogonal matrix. The basic principle of PCA principal component analysis is to transform the originally discrete influencing factor data into regular and concentrated data through this orthogonal matrix, thereby achieving dimensionality reduction.

[0055] Therefore, the main quantitative factors for extracting business aspects of power grid planning and construction using principal component analysis specifically include:

[0056] S11, the quantification of the business model of the power grid planning and construction dimension is affected by many factors, and the influencing factors intersect with each other. Therefore, the influencing factors of the business need to be screened, and all the influencing factors of the power grid planning and construction dimension business are denoted as quantification factors;

[0057] S12, the quantification factors are standardized;

[0058]

[0059] In the above formula, Z is the standardized result matrix of the quantification factor, and a and β are two quantification factors, a i , β i is the i-th group of data of the quantification factors a and β;

[0060] S13, the correlation coefficient is calculated, and the correlation coefficient matrix is calculated combined with the standardized quantification factors;

[0061] The calculation formula of the correlation coefficient r is as follows:

[0062]

[0063] In the above formula, x i is the i-th quantification factor, y i is a quantification factor different from x i , and n is the total number of quantification factors;

[0064] The correlation coefficient matrix R Z T is the transpose matrix of the quantification factor standardized matrix Z;

[0065] S14, the eigenvectors and eigenvalues of the correlation coefficient matrix are calculated, and the eigenvalues are set as λ i (λ i > 0, i = 1, 2,..., p), which satisfies |R- λI p | = 0, and is arranged in descending order, wherein I p is a p-order unit matrix;

[0066] S15, according to the eigenvalues, the variance contribution rate and the cumulative variance contribution rate K are calculated, and the eigenvectors of each eigenvalue are calculated;

[0067] Specifically, the cumulative variance contribution rate K is generally not less than 85%, which is expressed as:

[0068]

[0069] For each eigenvalue, Rx = λx, x is the eigenvector to be solved;

[0070] S16, calculating principal components F according to the characteristic values i :

[0071]

[0072] S17, calculating principal component coefficients, and calculating weights of all quantization factors according to the principal component coefficients, variance contribution rates and cumulative variance contribution rates;

[0073] wherein,

[0074] S18, screening main quantization factors according to the weights.

[0075] The weights are calculated by the above method, and the main quantization factors are obtained by screening according to the weights and ignoring the influence factors with lower weights. The main quantization factors include load characteristics, grid structure, power supply radius, power supply line diameter, distribution of tie switches, number and specifications of reactive equipment, positions of reactive equipment, compensation methods, line loss, investment and maintenance costs, optimal planning schemes, investment costs, equipment operation and maintenance costs, voltage control effects, engineering construction periods, construction expansion sequences, main equipment selection, in-out line schemes, annual capital construction investment plans, annual technical transformation investment plans, annual operation and maintenance investment plans, equipment procurement, capital construction, equipment installation and acceptance, and pre-commissioning preparations. The business evaluation system shown in Tables 1 and 2 can be established.

[0076] Table 1: Business evaluation system in the distribution network planning and design stage

[0077]

[0078]

[0079] Table 2: Business evaluation system in the engineering procurement stage

[0080]

[0081] S2, using the PSO algorithm to optimize the initial weights and thresholds of the BP neural network, and constructing a PSO_BP neural network according to the optimized initial weights and thresholds. The BP neural network can be continuously adjusted and trained according to the error to obtain a neural network model containing optimal thresholds and weights, and finally make the trained network model output the desired results. The parameters of the BP network model are optimized by the PSO algorithm, initialized to the corresponding parameter values of the PSO algorithm, and the global optimal solution is obtained by calculating the fitness value of the population, the individual extreme value and the population extreme value, and then the BP network is adjusted according to the training results, thereby improving the voltage quality.

[0082] wherein, the initial weights and thresholds of the BP neural network are optimized by the PSO algorithm, as shown in Figure 2As shown, specifically comprising the following steps:

[0083] S211, initializing the speed and position of each particle, the particle being the connection weight and threshold value of each layer of the BP neural network;

[0084] S212, calculating the fitness value of each particle;

[0085] S213, adjusting the position and speed of each particle;

[0086] S214, judging whether the particle is in the feasible region, if yes, entering the next step, if not, returning to the previous step;

[0087] S215, judging whether the particle meets the end condition, if yes, ending the optimization, otherwise, recalculating the fitness value of the particle from the judgment until the optimal initial weight and threshold value of the BP neural network are obtained.

[0088] Wherein, the PSO_BP neural network is constructed according to the optimized initial weight and threshold value, as shown in Figure 4 , specifically comprising:

[0089] S221, establishing the BP neural network and setting the BP neural network, including providing the learning mode to the BP neural network, wherein the learning mode includes: tutor learning, non-tutor learning and re-energizing learning, the structure of the BP neural network is as shown in Figure 3 , the BP neural network includes: input layer, hidden layer and output layer, the BP neural network adopts RMSE, the BP neurons are distributed in the input layer, and the radial basis function RBF is distributed in the hidden layer;

[0090] Specifically, the BP neural network model adopts RMSE, and the root mean square error is used as the fitness function:

[0091]

[0092] In the above formula, y i is the actual output;

[0093] The BP neurons are distributed in the first layer, that is:

[0094]

[0095] In the above formula, m is the number of hidden layers, x j is the input value of the jth neuron, S i is the area of the ith layer of Gaussian sphere, n is the number of neurons, Y 1,i is the output of the ith hidden layer of BP neurons;

[0096] The radial basis function RBF is distributed in the second layer, that is:

[0097]

[0098] In the above formula, k is the Gaussian spherical radius, Y 2,i is the i-th hidden layer output of the radial basis function RBF.

[0099] The output layer is the third layer, that is:

[0100]

[0101] In the above formula, Y3 is the output of the third layer.

[0102] S221, calculate the net input and output of each unit of the hidden layer and the output layer;

[0103] S222, calculate the generalized error of each unit of the hidden layer and the output layer; the number of neurons of the BP neural network is n1, n2 and n3 respectively, and the calculation formula of the error E is as follows:

[0104]

[0105]

[0106] In the above formula, θ j and η k are the thresholds of the second layer and the third layer, f0 and f1 are the excitation functions of the second layer and the third layer, y k is the actual output.

[0107] S223, according to the initial weight and threshold value of the BP neural network obtained by S215, adjust the connection weight between the hidden layer and the output layer and the threshold value of each unit of the hidden layer, adjust the connection weight between the input layer and the hidden layer and the threshold value of each unit of the hidden layer;

[0108] S224, update the iteration number of the BP neural network, and increase the iteration number by one each time;

[0109] S225, judge whether the training of all iteration numbers is completed, if the training is completed, update the iteration number again, enter the next step S226, otherwise return to step S221 to reacquire the learning mode to train the BP neural network;

[0110] S226, judge whether the voltage qualification rate of the output data of the BP neural network meets the condition, and whether the iteration number reaches the upper limit, if both are, complete the training of the BP neural network, obtain the PSO_BP neural network, otherwise reacquire the learning mode to train the BP neural network.

[0111] S3, input the main quantization factor into the PSO_BP neural network for training to obtain the influence weight of each business on the voltage quality problem in the power grid planning and construction dimension.

[0112] One embodiment of the novel power system planning and construction stage business weight analysis method is described in detail to make those skilled in the art understand the present application better:

[0113] A 33-node network is simulated to build a frame, and the method of the present application is used for training to obtain the influence weight results as shown in Table 3,

[0114] Table 3: Normalized results of each business influence weight

[0115] Business Weight Net frame planning 0.2363 Load forecasting 0.2269 Reactive power planning 0.2157 Scheme evaluation 0.0911 Equipment selection 0.0631 Budget 0.0631 Project approval 0.0524 Engineering construction 0.0516

[0116] As can be seen from Table 3, the improved PSO_BP algorithm used in the present application can obtain the influence weight of each business on the voltage quality, and the results show that the weights of network planning, load forecasting and reactive power planning are larger, and the weights of scheme evaluation, equipment selection and budget are smaller. The weight of network planning business is the largest, that is, the influence of network planning on voltage quality is the largest, so in the whole process of power grid business decision analysis, network planning should be considered as the priority, the network structure should be designed reasonably in the early stage, the distribution of substations should be determined, and enough space should be reserved for later modification to prevent and improve voltage quality problems.

[0117] The above disclosure is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any skilled person in the art can easily think of changes or modifications within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A novel power system planning and construction phase business weight analysis method, characterized in that, The method comprises the following steps: The main quantitative factors of the power grid planning and construction dimension business are extracted by principal component analysis; The initial weights and thresholds of the BP neural network are optimized by the PSO algorithm, and a PSO_BP neural network is constructed according to the optimized initial weights and thresholds; The main quantitative factors are input into the PSO_BP neural network for training, and the influence weights of each business of the power grid planning and construction dimension on the voltage quality problem are obtained; The power grid planning and construction dimension business is divided into two stages: a distribution network planning and design stage and an engineering procurement and construction stage; The distribution network planning and design stage comprises load forecasting, network planning, reactive power planning and scheme evaluation, and the engineering procurement and construction stage comprises project approval, equipment selection, budget and engineering construction; The main quantitative factors comprise load characteristics, network structure, power supply radius, power supply line diameter, tie switch distribution, reactive power equipment quantity and specification, reactive power equipment position, compensation mode, line loss, investment and maintenance cost, optimal planning scheme, investment cost, equipment operation and inspection cost, voltage control effect, engineering construction period, construction expansion timing, main equipment selection, incoming and outgoing line scheme, annual capital construction investment plan, annual technical transformation investment plan, annual operation and maintenance investment plan, equipment procurement, capital construction, equipment installation and acceptance and pre-commissioning preparation work; The training convergence index of the PSO_BP neural network comprises whether the voltage qualified rate of the output data meets the condition.

2. The novel power system planning and construction phase business weight analysis method according to claim 1, characterized in that, The main quantitative factors of the power grid planning and construction dimension business are extracted by principal component analysis, specifically comprising: All influence factors of the power grid planning and construction dimension business are recorded as quantitative factors; The quantitative factors are standardized; The correlation coefficient is calculated, and the correlation coefficient matrix is calculated in combination with the standardized quantitative factors; The eigenvectors and eigenvalues of the correlation coefficient matrix are calculated; The eigenvalues are used to calculate the variance contribution rate and the cumulative variance contribution rate, and the eigenvectors of each eigenvalue are calculated; The principal components are calculated according to the eigenvalues; The principal component coefficients are calculated, and the weights of all quantitative factors are calculated according to the principal component coefficients, the variance contribution rate and the cumulative variance contribution rate; The main quantitative factors are screened according to the weights.

3. The novel power system planning and construction phase business weight analysis method according to claim 1, characterized in that, The initial weights and thresholds of the BP neural network are optimized by the PSO algorithm, specifically comprising the following steps: The speed and position of each particle are initialized, and the particles are the connection weights and thresholds of each layer of the BP neural network; The fitness value of each particle is calculated; The position and speed of each particle are adjusted; It is judged whether the particle is in the feasible region, if yes, the next step is entered, if not, the previous step is returned; It is judged whether the particle meets the end condition, if yes, the optimization is ended, otherwise, the fitness value of the particle is recalculated and judged until the optimization is ended, and the optimal initial weights and thresholds of the BP neural network are obtained.

4. The novel power system planning and construction phase business weight analysis method according to claim 1, characterized in that, The PSO_BP neural network is constructed according to the optimized initial weights and thresholds, specifically comprising: A BP neural network is established, and a learning mode is provided for the BP neural network, wherein the BP neural network comprises an input layer, a hidden layer and an output layer; The net input and output of each unit of the hidden layer and the output layer are calculated; Calculate the generalization error of each unit of the hidden layer and the output layer; Optimize the initial weight and threshold of the BP neural network according to the PSO algorithm, adjust the connection weight between the hidden layer and the output layer and the threshold of each unit of the hidden layer, and adjust the connection weight between the input layer and the hidden layer and the threshold of each unit of the hidden layer; Update the iteration number of the BP neural network; Determine whether the training of all iteration numbers is completed, when the training is completed, update the iteration number again, enter the next step, otherwise, acquire the learning mode again to train the BP neural network; Determine whether the voltage qualification rate of the output data of the BP neural network meets the condition, and determine whether the iteration number reaches the upper limit, if both are yes, the training of the BP neural network is completed, and the PSO_BP neural network is obtained, otherwise, acquire the learning mode again to train the BP neural network.

5. The novel power system planning and construction phase business weight analysis method according to claim 4, characterized in that, The BP neural network adopts the RMSE.

6. The novel power system planning and construction phase business weight analysis method according to claim 4, characterized in that, In the BP neural network, the BP neurons are distributed in the input layer, and the radial basis function RBF is distributed in the hidden layer.

Images