CN104484829A - Accurate early warning method for transmission line mountain fire - Google Patents

Abstract

The invention discloses an accurate early warning method for transmission line mountain fire. The method comprises four steps: collecting fire point information, performing wind speed signal decomposition, performing signal denoising and performing early warning on the transmission line mountain fire. The out-dated manual monitoring and early warning mode is changed by the accurate early warning method disclosed by the invention; the accurate early warning method has the characteristics that the calculation is convenient, the effect is remarkable, and the like, and is suitable for predicting and warning the transmission line mountain fire at different altitudes and in different regions; according to the accurate early warning method provided by the invention, help is provided for performing scientific and accurate early warning on the transmission line mountain fire; at the initial stage of mountain fire information, targeted action can be achieved, and line loss caused by blind prediction and warning is avoided.

Description

The method of the accurate early warning of a kind of transmission line forest fire

Technical field

The invention belongs to the method for the accurate early warning of a kind of transmission line forest fire in electric power project engineering.

Background technology

In recent years, along with the exploitation of electric power resource, increasing transmission line of electricity is through lofty mountains and steep hills area, topography and geomorphology, the weather conditions of these area uniquenesses very easily cause mountain fire, light then cause transmission line of electricity to trip, heavy then cause and burn steel tower, cause long expendable great electric power accident.Mountain fire is one of principal element causing transmission line of electricity to trip, and along with the continuous expansion of electrical network scale, the frequent appearance of extreme drought weather in recent years in addition, mountain fire has become one of key factor of harm power network safety operation.According to the incomplete statistics data of periodical literature " the south electric network first quarter in 2010 circuit mountain fire trip condition is analyzed ", " high voltage direct current transmission line fault Position Research summary ", " transmission line forest fire fault analysis and prevention ", mountain fire caused great threat to China's transmission line of electricity: as 2010 to 2012, the line tripping that Yunnan Power System 110kV and above transmission line of electricity cause because of mountain fire totally 25 times, especially precious 7 I, the II loop line tripping operation of Kunming Areas 500kV on March 30th, 2012, has constituted three-class power security incident.Guangxi Power Grid 110kV and the accumulative mountain fire that occurs of above transmission line of electricity trip 31, in addition, have more than 20 transmission lines of electricity to be forced to emergency outage by mountain fire affects.

Both at home and abroad relevant scholar causes the mechanism of line tripping to mountain fire, mountain fire monitoring and early warning, mountain fire cause line fault model etc. and conduct in-depth research.Domestic periodical document has been set forth " the mountain fire breakdown characteristics of transmission line of electricity and Mechanism Study " and " puncture mechanism of transmission line of alternation current model under mountain fire condition ", document " the overhead transmission line stoppage in transit probability model under mountain fire condition " have studied the overhead transmission line stoppage in transit probability model under mountain fire condition, gives the relation of transmission line of electricity stoppage in transit probability and the quality inspection such as mountain fire distance, dense smoke concentration, temperature, humidity.But utilize the sensors such as video image to achieve the anti-mountain fire comprehensive monitor system in overhead transmission line corridor, there is poor real, fail to report and report by mistake, smoke transducer sensitivity is by the problem of on-the-spot wind direction environmental impact; In addition video image sensors not easily install in forest land and manufacture, maintenance cost is very huge.Also a lot of scholar is had to carry out large quantity research to the flashover property of transmission line of electricity under mountain fire condition abroad.Document " Flashover characteristicsof vertical-type model power line in the presence of combustion flash " utilizes satellite remote sensing to monitor forest mountain fire, local space distribution and the minutia of fire can be reflected, and have studied the application of polar orbiting meteorological satellite in the anti-mountain fire monitoring of transmission line of electricity, but there is poor real, fail to report and report by mistake in satellite-based monitoring system, to the very little mountain fire of scope spreading the shortcoming such as existing defects in judgement, bring certain difficulty to accurate recognition mountain fire.

Summary of the invention

In order to solve the problem, the method for the accurate early warning of a kind of transmission line forest fire provided by the invention, comprises following 4 steps:

S1: collect fiery dot information

HJ-1B satellite data or remote sensing recognition technology is adopted to collect fiery dot information;

In order to improve accuracy and practicality, the fire point that radius is more than or equal to 1m by the present invention is judged as fiery dot information, only need collect the fiery point that radius is more than or equal to 1m, just can predict, judge and the alarm condition of a fire spread region and size;

S2: wind velocity signal decomposes

The general detection of wind speed and direction can adopt the wind direction and wind velocity sensor of prior art, is arranged on below transmission line of electricity, and obtain optimum economic benefit for arriving minimum cost, each sensor is at a distance of 10km, and every day carries out wind speed and direction sampling; The different scale of necessary being in air speed data time series or trend component decompose out by classical mountain fire mathematical model step by step, produce a series of data sequence with same characteristic features yardstick, the sequence after decomposition obtains regularity compared with wind speed original data sequence;

Wind velocity signal is divided into following 3 steps:

S2.1: utilize spline interpolation to obtain coenvelope signal x respectively according to all maximum points of the wind series x (t) collected and all minimum points ₁(t) and lower envelope signal x ₂(t), and the mean value calculating coenvelope and lower envelope signal, such as formula (1):

$m\left(t\right)=\frac{1}{2}(x_1\left(t\right)+x_2\left(t\right))---\left(1\right)$

S2.2: the difference h (t) calculating x (t) and m (t), using h (t) as new x (t), repeat above operation, until when the h (t) of gained IMF meets following two conditions: 1. for whole analysis data, extreme value is counted and is more or less the same in 1 with number at zero point; 2. at any time on, the envelope formed by Local modulus maxima and the mean value of envelope formed by local minizing point are zero, and namely the average of IMF goes to zero;

Note c ₁(t)=h (t), then c ₁t () is first IMF component, it contains the most short period component in original series;

S2.3: deduct first the IMF signal c separated from x (t) ₁t (), obtains remaining difference signal r (t), i.e. formula (2):

r(t)＝x(t)-c ₁(t) (2)

Using r (t) as new x (t), repeat above process, obtain c successively ₂(t), c ₃(t) ..., c _n(t), until r (t) substantially become monotonic trend or | r (t) |≤δ (t), δ (t) they be the standard deviation of restriction.Therefore, original series x (t) can be decomposed into formula (3):

$x\left(t\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{c}_i\left(t\right)+r\left(t\right)---\left(3\right)$

The different scale of necessary being in air speed data time series or trend component can decompose out by the method step by step, produce a series of data sequence with same characteristic features yardstick, sequence after decomposition has stronger regularity compared with wind speed original data sequence, can improve credibility and the accuracy of prediction;

S3: signal denoising

Following 5 steps are adopted to carry out signal denoising:

S3.1: a series of closed subspace { V making square integrable function of wind speed space X (R) _j} _{j ∈ Z}a Multiresolution Decomposition (or infinitely approaching) of X (R), if φ (t) ∈ X (R) is scaling function, for j metric space V _jorthonormal basis, space W _jfor V _jat V _j+1in the orthogonal complement space, namely function ψ _j,k(t)=2 ^j/2ψ (2 ^jt-k) be space W _jorthonormal basis, then this subspace can be decomposed into formula (4):

$V_J=V_{J-1}\⊕W_{J-1}=V_{J-2}\⊕W_{J-2}\⊕W_{J-1}=\·\·\·\left(4\right)$

S3.2: to arbitrary signal x (t) ∈ L (R) can be formula (5) with resolution decomposition:

Wherein, J represents the number of plies of decomposition, c ₀k () is scale coefficient, d _jk () is coefficient, if scaling function is one group of orthogonal basis, then c _j(k), d _jk () can be expressed as formula (6):

S3.3: from Double-scaling equation formula (7):

$\left.\begin{array}{c}{c}_{j-1}\left(k\right)=\underset{h}{\mathrm{\Σ}}h(n-2k)c_j\left(n\right)\\ d_{j-1}\left(k\right)=\underset{n}{\mathrm{\Σ}}g(n-2k)c_j\left(n\right)\end{array}\right\}---\left(7\right)$

Wherein, h (k), g (k) are called low-pass filter coefficients and Hi-pass filter coefficient, and meet:

g(k)＝(-1) ^kh(1-k) (8)

Initial scale coefficient c _jk () can be obtained by signal x (t) Direct Sampling.If the sample frequency of signal is greater than nyquist frequency, so c _jk () just can well approximation signal x (t), that is, and not Required coefficient d completely under this yardstick _j(k).

S3.4: by analyzing above and deriving, obtains the coefficient of each frequency content of signal decomposition, can be expressed as the form of matrix, shown in (9):

C _signal＝[c ₀,d ₀,d ₁,…,d _J-1] (9)

In formula, J represents Decomposition order.

According to the feature of measurement noises, arrange the threshold value of each coefficient, the part higher than setting threshold value retains, and the part lower than setting threshold value is set to 0, obtains the coefficient of reconstruct

C' _signal＝[c' ₀,d' ₀,d' ₁,…,d' _J-1] (11)

S3.5: formula (11) is substituted into reconstruct equation

Get final product the multinomial noise in filtering wind speed original signal.

S4: transmission line forest fire early warning.

According to current classical mountain fire mathematical model, after incendiary source is on fire, when calm, form a circular scene of a fire, its flame spread rates formula (13) is:

$v_0=\frac{I_r\mathrm{\ξ}}{{\mathrm{\ρ}}_b\mathrm{\ϵ}{Q}_{\mathrm{ig}}}---\left(13\right)$

In formula, v ₀for mountain fire rate of propagation, km/h, I _rfor response intensity, ξ be thermoflux than coefficient, ρ _b, ε, Q _igbe the constant that combustible characteristic is relevant.If but by the impact of wind speed, flame will spread along all directions with friction speed, the oval scene of a fire that take incendiary source as focus will be formed.Oval long axis direction is maximum flame spread rates direction, and consistent with wind vector direction, maximum flame spread rates is formula (14):

v＝v ₀(1+ψ _V) (14)

In formula, ψ _vfor wind speed correction factor.

Definition k is oval semi-major axis a and the ratio of minor semi-axis b, i.e. k=a/b, and be used for characterizing oval shape, its experimental formula is k=1+ λ V.λ is coefficient, and reference value is 0.25s/m; Wind speed V unit is km/h.

Therefore, as shown from the above formula, as long as by wind speed x'(t) formula expression (12) substitute into formula (14), just can calculate mountain fire propagation and the speed that spreads, accurate early warning is carried out to transmission line forest fire.

The present invention compared with the existing technology, has the following advantages and beneficial effect:

1. change fall behind, outmoded personal monitoring and modes of warning, there is the feature such as convenience of calculation, Be very effective, be applicable to Different Altitude, the transmission line forest fire prediction of different regions and alarm;

2. present invention reduces human input, electric power system can be realized and cut payroll to improve efficiency, advantage is created to the online monitoring cost of reduction, greatly reduces the operation and management cost of electric system;

3. the present invention effectively can predict that the mountain fire in several days spreads and rate of expansion, takes measures to provide foundation for power department and fire department ensure that transmission line of electricity runs, and can reduce the transmission line of electricity stoppage in transit probability that mountain fire causes;

4. for science and accurately early warning transmission line forest fire information provide help, when there is the mountain fire information initial stage, just can accomplish and shooting the arrow at the target, avoid because blindness is predicted and the line loss that causes of alarm.

Accompanying drawing explanation

Fig. 1 is flow chart of steps of the present invention;

Fig. 2 is wind speed time series described in the embodiment of the present invention;

Fig. 3 is the wind speed subsequence after empirical mode decomposition process described in the embodiment of the present invention;

Fig. 4 is the wind speed time series after denoising described in the embodiment of the present invention of the present invention;

Fig. 5 is the comparison chart of embodiment of the present invention warning data of the present invention and measured result.

Embodiment

Below in conjunction with accompanying drawing, technical scheme of the present invention is clearly and completely described.

As shown in Figure 1, the method for the accurate early warning of a kind of transmission line forest fire, comprises following 4 steps:

S1: collect fiery dot information.

Collect fiery dot information and adopt prior art, HJ-1B satellite data or remote sensing recognition technology, earth's surface normal temperature temperature is at about 300K, and the flame temperature of forest fires more can reach about 1000K.According to Si Difen-Boltzmann's law and Wien's displacement law, as long as blackbody temperature has very little change, under the relation that radiation amount is directly proportional to the biquadratic of temperature, the great changes of radiation will be caused, the temperature of flame high temperature heat source more will cause the sharply change of radiation, this change makes the monitoring of mountain fire point become very feasible, the radiation peak λ that earth's surface normal temperature is corresponding _maxat about 11.0 μm, the radiation peak λ that forest fires are corresponding _maxat about 3 ~ 5 μm, B07 and B08 of satellite HJ-1B is sitting at this two spectral coverage regions.Remote sensing recognition technology of the prior art can be adopted in addition to carry out identification fire dot information.

In order to improve accuracy and practicality, the fire point that radius is more than or equal to 1m by the present invention is judged as fiery dot information, only need collect the fiery point that radius is more than or equal to 1m, just can predict, judge and the alarm condition of a fire spread region and size.

S2: wind velocity signal decomposes.

According to current classical mountain fire mathematical model, relevant formulae discovery just can be utilized to draw mountain fire rate of propagation as long as be aware of wind speed.But wind speed just has very strong non-linear and randomness characteristic, and inevitably there is the interference of noise when gathering wind speed information, so utilize traditional method wind speed forecasting method to there is larger error.

The general detection of wind speed and direction can adopt prior art, such as: the WAA151 wind direction and wind velocity sensor etc. that the EL15-2 type that Tianjin Meteorological Instrument Factory is produced, the EZC-1 type of Changchun meteorologic instrument factory production, EC9-1 type and Vaisala company produce, sensors with auxiliary electrode is arranged on below transmission line of electricity, optimum economic benefit is obtained for arriving minimum cost, each sensor is at a distance of 10km, and every day carries out wind speed and direction sampling.Wind speed is nonlinear and non local boundary value problem, and the present invention adopts empirical mode decomposition method to decompose wind speed collection signal.Empirical mode decomposition is a kind of signal decomposition method based on signal local feature, is applicable to nonlinear and non local boundary value problem analysis, is a kind of adaptive signal decomposition method.Empirical mode decomposition method is supposed: the signal of any complexity is all made up of simple intrinsic mode function (IMF), and each IMF is separate, and wind velocity signal is divided into following 3 steps:

S2.1: utilize spline interpolation to obtain coenvelope signal x respectively according to all maximum points of the wind series x (t) collected and all minimum points ₁(t) and lower envelope signal x ₂(t), and the mean value calculating coenvelope and lower envelope signal, such as formula (1):

$m\left(t\right)=\frac{1}{2}(x_1\left(t\right)+x_2\left(t\right))---\left(1\right)$

S2.2: the difference h (t) calculating x (t) and m (t), using h (t) as new x (t), repeat above operation, until when the h (t) of gained IMF meets following two conditions: 1. for whole analysis data, extreme value is counted and is more or less the same in 1 with number at zero point; 2. at any time on, the envelope formed by Local modulus maxima and the mean value of envelope formed by local minizing point are zero, and namely the average of IMF goes to zero.

Note c ₁(t)=h (t), then c ₁t () is first IMF component, it contains the most short period component in original series.

S2.3: deduct first the IMF signal c separated from x (t) ₁t (), obtains remaining difference signal r (t), i.e. formula (2):

r(t)＝x(t)-c ₁(t) (2)

Using r (t) as new x (t), repeat above process, obtain c successively ₂(t), c ₃(t) ..., c _n(t), until r (t) substantially become monotonic trend or | r (t) |≤δ (t), δ (t) they be the standard deviation of restriction.Therefore, original series x (t) can be decomposed into formula (3):

$x\left(t\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{c}_i\left(t\right)+r\left(t\right)---\left(3\right)$

The different scale of necessary being in air speed data time series or trend component can decompose out by the method step by step, produce a series of data sequence with same characteristic features yardstick, sequence after decomposition has stronger regularity compared with wind speed original data sequence, can improve credibility and the accuracy of prediction.

S3: signal denoising.

But each wind series of decomposing out certainly exists measurement noises, and therefore, also need to carry out denoising to each wind series, consider the non-linear of wind series and randomness, the present invention adopts following 5 steps to carry out signal denoising:

S3.1: a series of closed subspace { V making square integrable function of wind speed space X (R) _j} _{j ∈ Z}a Multiresolution Decomposition (or infinitely approaching) of X (R), if φ (t) ∈ X (R) is scaling function, for j metric space V _jorthonormal basis, space W _jfor V _jat V _j+1in the orthogonal complement space, namely function ψ _j,k(t)=2 ^j/2ψ (2 ^jt-k) be space W _jorthonormal basis, then this subspace can be decomposed into formula (4):

$V_J=V_{J-1}\⊕W_{J-1}=V_{J-2}\⊕W_{J-2}\⊕W_{J-1}=\·\·\·\left(4\right)$

S3.2: to arbitrary signal x (t) ∈ L (R) can be formula (5) with resolution decomposition:

Wherein, J represents the number of plies of decomposition, c ₀k () is scale coefficient, d _jk () is coefficient, if scaling function is one group of orthogonal basis, then c _j(k), d _jk () can be expressed as formula (6):

S3.3: from Double-scaling equation formula (7):

$\left.\begin{array}{c}{c}_{j-1}\left(k\right)=\underset{h}{\mathrm{\Σ}}h(n-2k)c_j\left(n\right)\\ d_{j-1}\left(k\right)=\underset{n}{\mathrm{\Σ}}g(n-2k)c_j\left(n\right)\end{array}\right\}---\left(7\right)$

Wherein, h (k), g (k) are called low-pass filter coefficients and Hi-pass filter coefficient, and meet:

g(k)＝(-1) ^kh(1-k) (8)

Initial scale coefficient c _jk () can be obtained by signal x (t) Direct Sampling.If the sample frequency of signal is greater than nyquist frequency, so c _jk () just can well approximation signal x (t), that is, and not Required coefficient d completely under this yardstick _j(k).

S3.4: by analyzing above and deriving, obtains the coefficient of each frequency content of signal decomposition, can be expressed as the form of matrix, shown in (9):

C _signal＝[c ₀,d ₀,d ₁,…,d _J-1] (9)

In formula, J represents Decomposition order.

According to the feature of measurement noises, arrange the threshold value of each coefficient, the part higher than setting threshold value retains, and the part lower than setting threshold value is set to 0, obtains the coefficient of reconstruct

C' _signal＝[c' ₀,d' ₀,d' ₁,…,d' _J-1] (11)

S3.5: formula (11) is substituted into reconstruct equation

Get final product the multinomial noise in filtering wind speed original signal.

S4: transmission line forest fire early warning.

According to current classical mountain fire mathematical model, after incendiary source is on fire, when calm, form a circular scene of a fire, its flame spread rates formula (13) is:

$v_0=\frac{I_r\mathrm{\ξ}}{{\mathrm{\ρ}}_b\mathrm{\ϵ}{Q}_{\mathrm{ig}}}---\left(13\right)$

In formula, v ₀for mountain fire rate of propagation, km/h, I _rfor response intensity, ξ be thermoflux than coefficient, ρ _b, ε, Q _igbe the constant that combustible characteristic is relevant.If but by the impact of wind speed, flame will spread along all directions with friction speed, the oval scene of a fire that take incendiary source as focus will be formed.Oval long axis direction is maximum flame spread rates direction, and consistent with wind vector direction, maximum flame spread rates is formula (14):

v＝v ₀(1+ψ _V) (14)

In formula, ψ _vfor wind speed correction factor.

Definition k is oval semi-major axis a and the ratio of minor semi-axis b, i.e. k=a/b, and be used for characterizing oval shape, its experimental formula is k=1+ λ V.λ is coefficient, and reference value is 0.25s/m; Wind speed V unit is km/h.

Therefore, as shown from the above formula, as long as by wind speed x'(t) formula expression (12) substitute into formula (14), just can calculate mountain fire propagation and the speed that spreads, accurate early warning is carried out to transmission line forest fire.

In order to verify correctness of the present invention, adopt following examples to carry out checking and illustrate, the present embodiment only for verifying technical scheme of the present invention, and can not limit the scope of the invention with this.

To 720 groups of intervals wind speed of 1 hour that certain 500kV transmission line of electricity gathers, original wind speed time series as shown in Figure 2.

Wind velocity signal decomposes: wind speed time series adopts empirical modal to decompose it, obtains each IMF as shown in Figure 3.

Utilize signal denoising to carry out denoising to above-mentioned wind speed time series, obtain reconstruct wind speed time series as shown in Figure 4.

The noise-removed threshold value coefficient adopted is as shown in table 1.

Table 1 noise-removed threshold value coefficient

Decomposition layer	1	2	3	4	5	6	7	8	9
										Coefficient	3.6166	3.6387	1.4601	1.1100	0.6066	1.0020	0.4690	1.8984	1.7459

Utilize formula (12) forecasting wind speed value to substitute into formula (14) and calculate mountain fire rate of propagation early warning result and measured result as shown in Figure 5, the method for the invention is consistent with actual mountain fire speed.In addition, contrast shaft tower mountain fire early warning information record, according to associated video monitored condition and on-the-spot meteorological condition, the dense degree of smog and burnt area variation tendency, also demonstrate correctness of the present invention.

Claims (1)

1. a method for the accurate early warning of transmission line forest fire, is characterized in that, comprises and collects fiery dot information, wind velocity signal decomposition, signal denoising and transmission line forest fire early warning 4 steps:

S1: adopt HJ-1B satellite data or remote sensing recognition technology to collect fiery dot information

Fire point radius being more than or equal to 1m is judged as fiery dot information;

S2: wind velocity signal decomposes

The general of wind speed and direction detects the wind direction and wind velocity sensor adopting prior art, is arranged on below transmission line of electricity, and obtain optimum economic benefit for arriving minimum cost, each sensor is at a distance of 10km, and every day carries out wind speed and direction sampling; Classical mountain fire mathematical model is adopted to decompose out step by step by the different scale of necessary being in air speed data time series or trend component, produce a series of data sequence with same characteristic features yardstick, the sequence after decomposition obtains regularity compared with wind speed original data sequence;

Wind velocity signal is divided into following 3 steps:

S2.1: utilize spline interpolation to obtain coenvelope signal x respectively according to all maximum points of the wind series x (t) collected and all minimum points ₁(t) and lower envelope signal x ₂(t), and the mean value calculating coenvelope and lower envelope signal, such as formula (1):

$m\left(t\right)=\frac{1}{2}(x_1\left(t\right)+x_2\left(t\right))---\left(1\right)$

S2.2: the difference h (t) calculating x (t) and m (t), using h (t) as new x (t), repeat above operation, until when the h (t) of gained IMF meets following two conditions: 1. for whole analysis data, extreme value is counted and is more or less the same in 1 with number at zero point; 2. at any time on, the envelope formed by Local modulus maxima and the mean value of envelope formed by local minizing point are zero, and namely the average of IMF goes to zero;

Note c ₁(t)=h (t), then c ₁t () is first IMF component, it contains the most short period component in original series;

S2.3: deduct first the IMF signal c separated from x (t) ₁t (), obtains remaining difference signal r (t), i.e. formula (2):

r(t)＝x(t)-c ₁(t) (2)

Using r (t) as new x (t), repeat above process, obtain c successively ₂(t), c ₃(t) ..., c _n(t), until r (t) substantially become monotonic trend or | r (t) |≤δ (t), δ (t) they be the standard deviation of restriction; Therefore, original series x (t) can be decomposed into formula (3):

$x\left(t\right)=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{c}_i\left(t\right)+r\left(t\right)---\left(3\right)$

The different scale of necessary being in air speed data time series or trend component can decompose out by the method step by step, produce a series of data sequence with same characteristic features yardstick, sequence after decomposition has stronger regularity compared with wind speed original data sequence, can improve credibility and the accuracy of prediction;

S3: signal denoising

Following 5 steps are adopted to carry out signal denoising:

S3.1: a series of closed subspace { V making square integrable function of wind speed space X (R) _j} _{j ∈ Z}a Multiresolution Decomposition (or infinitely approaching) of X (R), if φ (t) ∈ X (R) is scaling function, for j metric space V _jorthonormal basis, space W _jfor V _jat V _j+1in the orthogonal complement space, namely function for space W _jorthonormal basis, then this subspace can be decomposed into formula (4):

$V_J=V_{J-1}\⊕W_{J-1}=V_{J-2}\⊕W_{J-2}\⊕W_{J-1}=\·\·\·---\left(4\right)$

S3.2: to arbitrary signal x (t) ∈ L (R) can be formula (5) with resolution decomposition:

Wherein, J represents the number of plies of decomposition, c ₀k () is scale coefficient, d _jk () is coefficient, if scaling function is one group of orthogonal basis, then c _j(k), d _jk () can be expressed as formula (6):

S3.3: from Double-scaling equation formula (7):

$\left.\begin{array}{c}{c}_{j-1}\left(k\right)=\underset{n}{\mathrm{\Σ}}h(n-2k)c_j\left(n\right)\\ d_{j-1}\left(k\right)=\underset{n}{\mathrm{\Σ}}g(n-2k)c_j\left(n\right)\end{array}\right\}---\left(7\right)$

Wherein, h (k), g (k) are called low-pass filter coefficients and Hi-pass filter coefficient, and meet:

g(k)＝(-1) ^kh(1-k) (8)

Initial scale coefficient c _jk () can be obtained by signal x (t) Direct Sampling; If the sample frequency of signal is greater than nyquist frequency, so c _jk () just can well approximation signal x (t), that is, and not Required coefficient d completely under this yardstick _j(k);

S3.4: by analyzing above and deriving, obtains the coefficient of each frequency content of signal decomposition, can be expressed as the form of matrix, shown in (9):

C _signal＝[c ₀,d ₀,d ₁,…,d _J-1] (9)

In formula, J represents Decomposition order;

According to the feature of measurement noises, arrange the threshold value of each coefficient, the part higher than setting threshold value retains, and the part lower than setting threshold value is set to 0, obtains the coefficient of reconstruct

C' _signal＝[c' ₀,d' ₀,d' ₁,…,d' _J-1] (11)

S3.5: formula (11) is substituted into reconstruct equation

Get final product the multinomial noise in filtering wind speed original signal;

S4: transmission line forest fire early warning;

Maximum flame spread rates is formula (14):

v＝v ₀(1+ψ _V) (14)

As long as by wind speed x'(t) formula expression (12) substitute into formula (14), just can calculate mountain fire propagation and the speed that spreads, accurate early warning is carried out to transmission line forest fire.