Disclosure of Invention
In order to solve the technical problem, the invention provides a comprehensive evaluation method for the icing state of the high-voltage power transmission line, which can more accurately determine the icing thickness of the power transmission line through the data value acquired by the tension sensor. The method can not only evaluate the reliability of the current state, but also predict the future state, thereby providing powerful technical support for preventing icing accidents.
The technical scheme adopted by the invention is as follows:
a comprehensive evaluation method for icing state of a high-voltage transmission line comprises the following steps:
step 1: acquiring the ambient temperature, the relative humidity, the wind speed and the wind direction through a microclimate monitoring collector, judging whether the icing condition is met, and if so, executing the step 2; if not, continuing monitoring.
Step 2: inputting real-time monitored temperature, relative humidity, wind speed and wind direction data into a power transmission line icing combination prediction model considering a time accumulation effect, predicting icing thickness, judging whether the icing thickness reaches a set threshold value, and if so, executing a step 3; if not, continuing monitoring.
And step 3: and after the icing thickness of the power transmission line reaches a set threshold value, giving out early warning, starting to acquire data on the power transmission line by the tension sensor at the moment, and carrying out inversion on the icing thickness of the power transmission line based on the tension value measured by the tension sensor.
In the step 1, the icing conditions are as follows:
wherein: t is ambient temperature, unit: DEG C; h is ambient relative humidity, unit: percent; s is the ambient wind speed, unit: m/s; d is the included angle of wind direction and circuit, unit: degree.
The step 2 comprises the following steps:
s2.1: optimizing parameters of a multi-core related vector machine by adopting a self-adaptive parallel Jaya algorithm, and establishing an icing growth rate combined prediction model;
s2.2: on the basis of the icing growth rate combination prediction model, considering the time accumulation effect of icing growth and the initial thickness of icing of the transmission line at different stages to obtain an icing thickness prediction result;
s2.3: and finally, comparing the predicted ice coating thickness with a set threshold value, and judging whether the predicted ice coating thickness reaches the set threshold value.
The adaptive parallel Jaya algorithm divides the total into a plurality of sub-populations, then executes the Jaya algorithm aiming at the plurality of sub-populations in a parallel sequencing mode and finds the overall optimal solution.
In S2.1: establishing an icing growth rate combination prediction model, which comprises the following specific steps:
firstly, preprocessing an original icing data set to obtain an icing growth rate data set;
secondly, 30% of samples are extracted from the icing growth rate data set by adopting a layered sampling method to serve as a test set, and the rest 70% of samples serve as a training set.
Optimizing the parameters of the prediction model by using a self-adaptive parallel Jaya algorithm, and improving the prediction accuracy and generalization capability of the prediction model;
and fourthly, after the optimal parameters of the prediction model are determined, training the model according to the training set, and verifying on the test set.
In the step 3, the tension sensor acquires data of the power transmission line, the icing thickness of the power transmission line is obtained through the tension measured by the tension sensor and the data of sag and the like, and a derivation formula is as follows:
wherein σ
A 、σ
B The tension, σ, measured for the point of suspension
0 Is a tension force in the horizontal direction, beta is an included angle between a connecting line between wire suspension points of the two base tower towers and the horizontal direction, gamma is a specific load of the transmission line in an ice coating state, l is a span, h is a height difference of the two base tower towers, f is a height difference of the two base tower towers
m The maximum sag of the wire.
Indicates when the left variable is σ
A When the right side is calculated as a sign, the left side is given as a variable
B The sign is calculated to the right.
Indicating the sag at the midpoint of A, B.
Wherein f is x Is any point on the wire sag, x is the horizontal distance from the suspension point A to any point on the wire, f m The maximum sag of the wire, and the span.
The following equations (2) to (4) are derived:
mγ 2 -nγ+k=0 (9);
m, n and k are three algebraic variables, and the meanings of the variables are listed in the formulas (10) to (12).
Wherein,
wherein a is the distance from the lowest point of the sag to the suspension point A, and sigma A The tension measured for the suspension point, θ A Is the angle between the outlet of the transmission line and the horizontal direction, sigma 0 Is a pulling force in the horizontal direction, beta is an included angle between a connecting line between two base rod tower power transmission line suspension points and the horizontal direction, f 0 The lowest point sag of the power transmission line, and h is the height difference of the two base towers.
By solving the equation (5), γ can be obtained, and γ is the specific load of the transmission line in the ice-coated state.
Transmission line icing thickness b, unit: mm, determined by the following formula:
in the formula: d is the outer diameter of the transmission line, unit: mm;
gamma is the specific load of the transmission line in an icing state, and A is the sectional area of the transmission line; beta is a c The wind load adjustment coefficient of the overhead line of 500kV or above; alpha is alpha f Is the wind speed unevenness coefficient;
μ sc the form factor and the diameter of the overhead line<Taking 1.2 when the thickness is 17 mm; wire diameter>Taking 1.1 when the thickness is 17 mm; during ice coating, no matter the wire diameter, 1.2 is taken;
b is the icing wind load increasing coefficient, and 1.1 is taken in a 5mm ice area; 1.2 mm of ice area is taken; taking 1.3 in a 15mm ice area; taking 1.5-2.0 of ice areas with the thickness of 20mm and above;
theta is an included angle between the wind direction and the line direction;
W V for designing the wind pressure under the benchmark wind speed, unit: pa.
The invention discloses a comprehensive evaluation method for icing state of a high-voltage transmission line, which has the following technical effects:
1) the evaluation method of the invention can not only evaluate the reliability of the current state, but also predict the future state, thereby providing powerful technical support for preventing icing accidents and having certain application prospect.
2) The method adopts a microclimate monitoring system as an auxiliary means for monitoring the icing thickness of the transmission line, judges whether the environmental temperature, the relative humidity, the wind speed and the wind direction reach the icing condition of the transmission line, simultaneously considers that the icing thickness continuously and cumulatively changes along with time, sends out early warning after the icing thickness of the transmission line reaches the thickness needing early warning, then starts to record the tension value of the icing transmission line by a tension sensor, and obtains the icing thickness of the transmission line by inversion. So, can avoid force sensor to record a large amount of invalid data when the power transmission line does not have the icing state, reduce data processing volume, force sensor and record appearance begin operating condition after only reporting to the police simultaneously will reduce the power consumption, slow down the instrument and roll over the damage.
Detailed Description
The invention is described in further detail below with reference to the accompanying drawings:
a comprehensive assessment method for the icing state of a high-voltage power transmission line predicts the icing thickness by monitoring microclimate conditions and considering time accumulation effect, judges whether the icing thickness reaches a set threshold value or not, and finally adopts a tension sensor to collect data of the power transmission line to assess the icing state. The method flow is shown in fig. 1, and specifically comprises the following steps:
step 1: acquiring the ambient temperature, the ambient relative humidity, the wind speed and the wind direction through a ZL6 miniature meteorological monitoring system, judging whether an icing condition is met, and if so, executing a step 2; if not, continuing monitoring.
Wherein the icing conditions are as follows:
wherein: t is ambient temperature, unit: DEG C; h is ambient relative humidity, unit: percent; s is the ambient wind speed, unit: m/s; d is the included angle of wind direction and circuit, unit: degree.
Step 2: inputting real-time monitored temperature, relative humidity, wind speed and wind direction data into a power transmission line icing combination prediction model considering a time accumulation effect, predicting icing thickness, judging whether the icing thickness reaches a set threshold value, and if so, executing a step 3; if not, continuing monitoring.
The method adopts a self-adaptive parallel Jaya algorithm to optimize parameters of a multi-core related vector machine and establishes an icing growth rate combined prediction model; then, on the basis of the combined prediction model, considering the time accumulation effect of icing growth and the initial thicknesses of different stages to obtain an icing thickness prediction result; and finally, comparing the predicted ice coating thickness with a set threshold value, and judging whether the predicted ice coating thickness reaches the set threshold value.
Firstly, preprocessing original icing data and microclimate data (temperature, relative humidity, wind speed and wind direction) to obtain an icing growth rate data set and modeling; and then, on the basis of the growth rate prediction model, the influence of the time accumulation effect on the icing thickness is brought into consideration, and an icing thickness combination prediction model considering the time accumulation effect is constructed.
In order to reduce the influence of abnormal data on the model prediction performance and accelerate the training speed of the model, the invention preprocesses the original icing and microclimate data in three aspects of abnormal data elimination, first-order difference and data normalization.
Suppose training sample set D { (x) 1 ,y 1 ),(x 2 ,y 2 ),…,(x N ,y N ) Where N denotes the number of samples, x i A column vector, y, of all input eigenvalues at a time i Is represented by x i Rate of ice accretion under conditions. The regression equation for the RVM model can be expressed as
In the formula, K (x, x) i ) Is the kernel function of RVM model, omega i The weight vector is written as the weight vector form of (omega) for the kernel function weight 0 ,ω 1 ,…,ω N ),ε i 、ω 0 Representing additive noise and bias constants. Epsilon i Obeying a zero mean gaussian distribution.
The invention carries out weighted combination on the polynomial kernel function with the global response characteristic and the Gaussian kernel function with the local response characteristic to obtain a new combined kernel function K C (x,x i ) And the expression is as follows
In the formulaλ is a weight parameter for adjusting the influence weights of different kernels; r is 1 ,r 2 ,r 3 ,r g Respectively, the parameters of the corresponding kernel.
As shown in the formula (3), the multi-nuclear RVM model has λ and r 1 、r 2 、r 3 、r g The total number of the parameters to be optimized is 5, and the selection of the parameters has a crucial influence on the performance of the icing growth rate combined prediction model.
In order to reduce the influence of improper parameter selection on the model performance, the invention optimizes the model parameters by adopting an adaptive parallel Jaya algorithm. The core idea of the Jaya algorithm is to modify the population optimal solution according to equation (4) during each iteration.
S′ j,k,i =S j,k,i +r 1,j,i (S j,best,i -|S j,k,i |)-r 2,j,i (S j,worst,i -|S j,k,i |) (4);
In the formula, r 1,j,i ,r 2,j,i Is two random numbers uniformly distributed in a 0-1 interval j,k,i Is the value of the jth variable of the kth solution in the current population in the ith iteration, S j,best,i Is the j variable value of the optimal solution in the i iteration, S j,worst,i In the same way, S' j,k,i To correspond to the updated value.
The adaptive parallel Jaya algorithm divides the population into a plurality of sub-populations, then executes the Jaya algorithm in a parallel sequential mode aiming at the plurality of sub-populations and finds the overall optimal solution. In order to prevent the overfitting phenomenon in the optimization process of the model, the average root mean square error of the model after the K-fold cross validation is used as a fitness function on the basis of a K-fold cross validation algorithm, so that the parameter value which enables the generalization performance of the model to be optimal is ensured to be found.
The fitness value expression of a certain sub-population solution S is as follows,
wherein K is the number of turns and m is the number of samples in a certain subgroupThis number, y j,i Actual icing growth Rate, y 'of ith sample at jth Cross-validation' j,i The predicted icing growth rate for the ith sample at the jth cross-validation.
The steps of optimizing the model parameters by using the adaptive parallel Jaya algorithm are as follows:
firstly, initializing optimization algorithm parameters: the number of the parameters to be optimized is 5; the number of the initial sub-populations is 8, the maximum value of the number of the sub-populations is 15, and the minimum value is 1; the maximum number of iterations is 500; the convergence accuracy is 0.001; the cross-validation fold is 10.
② randomly generating 40 sets of solution vectors [ lambda, r ] g ,r 1 ,r 2 ,r 3 ]And randomly dividing the whole solution vector into 8 sub-populations.
Thirdly, in the sub-population, calculating the fitness values of different solution vectors according to the training results of the formula (5) and the MKRVM, wherein the minimum fitness value is the current optimal solution S best Otherwise, it is the current worst solution S worst 。
And fourthly, updating and iterating according to the equation (4) to obtain a correction solution according to the obtained current optimal solution vector and the worst solution vector. If the correction solution is better than the current optimal solution, the correction solution is reserved, meanwhile, the number of the sub-populations is reduced by 1, otherwise, the current optimal solution is reserved, and the number of the sub-populations is increased by 1. All the retained solutions are taken as input for the next iteration.
Fifthly, executing the steps III and IV in a parallel sequencing mode aiming at all the sub-populations, and judging whether an iteration termination condition is reached; if so, finishing optimization and outputting the optimal solution vector as a parameter of the prediction model, otherwise, continuing to execute.
The concrete steps of establishing the icing growth rate prediction model are as follows:
the method comprises the following steps: firstly, an original icing data set is preprocessed to obtain an icing growth rate data set.
And step two: and (3) performing feature selection on the icing growth rate data set according to a random forest algorithm to obtain the most main microclimate factors influencing the icing growth as input features of the prediction model.
③: the data set is divided into a training set and a test set. In order to avoid the influence on the final result caused by the deviation introduced by data division, a hierarchical sampling method is adopted for data set division so as to keep the consistency of data distribution as much as possible. The specific method comprises the following steps: 30% of samples are extracted from the ice coating growth rate data set by a hierarchical sampling method to serve as a test set, and the rest 70% of samples serve as a training set.
Fourthly, the preparation method comprises the following steps: and optimizing the model parameters by using a self-adaptive parallel Jaya algorithm, and improving the prediction precision and generalization capability of the model.
Fifth, the fifth step: and after the optimal parameters of the model are determined, training the model according to a training set and verifying on a test set.
The method for constructing the icing thickness prediction model considering the time accumulation effect comprises the following steps of:
step 1, giving the initial ice coating thickness at the current moment and relevant microclimate data in a future period of time.
And 2, predicting the icing growth rate in a future period of time according to the relevant microclimate data and the icing growth rate combination model.
And 3, predicting the icing thickness change of the power transmission line within a period of time in the future according to the formula (5) and by combining the initial icing thickness with the predicted icing growth rate.
The construction process of the icing thickness combination prediction model considering the time accumulation effect is shown in FIG. 2.
And step 3: and after the icing thickness of the power transmission line reaches a set threshold value, giving out early warning, starting to acquire data on the power transmission line by the tension sensor at the moment, and carrying out inversion on the icing thickness of the power transmission line based on the tension value measured by the tension sensor.
As shown in fig. 3, the icing thickness of the power transmission line can be obtained through data such as the tension measured by the tension sensor, sag and the like, and the derivation formula is as follows:
wherein σ A 、σ B Stress (tensile force), σ, measured for the suspension point 0 Is horizontalStress (tensile force) in the direction, beta is an included angle between a connecting line between suspension points of the transmission lines of the two base tower towers and the horizontal direction, gamma is specific load of the transmission lines in an ice coating state, l is a span, h is a height difference of the two base tower towers, and f is m The maximum sag of the transmission line.
Wherein f is m The maximum sag of the transmission line, gamma is the specific load of the transmission line in an icing state, l is the span, and sigma is 0 Is stress (pulling force) in the horizontal direction, beta is an included angle between a connecting line between two base rod tower power transmission line suspension points and the horizontal direction, f x Is the sag of any point on the transmission line, and x is the horizontal distance from the suspension point A to any point on the transmission line.
The following equations are derived from equations (6) to (8):
mγ 2 -nγ+k=0 (9)
wherein,
wherein a is the distance from the lowest point of the sag to the suspension point A, and sigma A Stress (tensile force), theta, measured for the point of suspension A Angle of the outlet of the transmission line to the horizontal, sigma 0 Is stress (tensile force) in the horizontal direction, and beta is a clamp between a connecting line between two base tower transmission line suspension points and the horizontal directionCorner, f 0 The lowest point sag of the power transmission line, and h is the height difference of the two base towers.
By solving the equation (9), γ can be obtained, and γ is the specific load of the transmission line in the ice-coated state.
The thickness b (mm) of the coated ice on the transmission line can be obtained by the following formula:
in the formula: d is the outer diameter of the transmission line, unit: mm;
gamma is the specific load of the transmission line in an icing state, and A is the sectional area of the transmission line;
β c the wind load adjustment coefficient of the overhead line of 500kV or above; alpha is alpha f Is the wind speed unevenness coefficient; mu.s sc The form factor and the diameter of the overhead line<Taking 1.2 when the thickness is 17 mm; wire diameter>Taking 1.1 when the thickness is 17 mm; during ice coating, no matter the size of the wire diameter, 1.2 is taken;
b is the icing wind load increasing coefficient, and 1.1 is taken in a 5mm ice area; 1.2 in a 10mm ice area; taking 1.3 in a 15mm ice area; taking 1.5-2.0 of ice areas with the thickness of 20mm and above;
theta is an included angle between the wind direction and the line direction; w V For designing the wind pressure under the benchmark wind speed, unit: pa.
From the above, the icing thickness b can be obtained.