CN103447513B - A kind of medium-frequency induction furnace automatic casting control system - Google Patents

Abstract

The invention discloses a kind of medium-frequency induction furnace Auto-pouring System, comprise hydraulic cylinder, hydraulic cylinder is overturn by leverage and body of heater and is in transmission connection; Hydraulic cylinder oil inlet is provided with flow control valve, also comprises feedback control system; Flow control valve is the electric proportional-regulation valve comprising proportional plus integral control module, and controller major loop comprises comparator module; Backfeed loop comprises support vector regression algoritic module, and sensor comprises the current sensor of electric proportional-regulation valve, the angular transducer at electric furnace inclination angle, the angular-rate sensor of electric furnace upset; The output of each sensor connects with the corresponding input signal of telecommunication of support vector regression algoritic module, the output of support vector regression algoritic module is connected with the input signal of telecommunication of comparator module, and the output of comparator module is connected with the input signal of telecommunication of electric proportional-regulation valve.High-accuracy intelligent automatic casting can be realized, adjustable parameters on medium-frequency induction furnace, portable strong.

Description

A kind of medium-frequency induction furnace automatic casting control system

Technical field

The invention belongs to automatic control system technical field, relate to pouring furnace automatic casting control system, particularly common medium-frequency induction furnace controls the automatic control system of casting metal flow quantity.

Background technology

Traditional electric furnace casting and pouring adopts medium-frequency induction furnace usually, in its casting and pouring process, electric furnace fluid is normally poured in casting mold or crucible by hand-guided arm bar rotary device by molten metal flow-control, hand dropping controls to need operative employee to operate under severe working environment, depend on operative employee's experience to a great extent, very high to the requirement of operative employee's quality, cast defect rate is higher, and production efficiency is lower; Due to the complex control object that running gate system is close coupling between non-linear serious, time lag large, an each parameter, the structure and parameter of Auto-pouring System controller must be determined by rule of thumb, control to be difficult to obtain desirable control effects as undertaken by the PID control system of routine, therefore, common medium-frequency induction furnace is difficult to realize conventional closed loop feedback automatic casting control.As needed automated production, precisely controlling casting metal flow quantity, improving quality and the production efficiency of casting, just need to adopt expensive vapour-pressure type insulation automatic pouring furnace.

The common medium-frequency induction furnace casting of molten metal flow-control of prior art, adopts hand-guided exactly, and cast defect rate is higher, production efficiency is lower, uses effect desirable not enough.

Summary of the invention

For overcoming the deficiencies in the prior art, the invention provides a kind of intelligence degree that can realize automatic casting on common medium-frequency induction furnace high, the control system that control effects is good.

The present invention for the technical scheme reaching above-mentioned technical purpose and adopt is: a kind of medium-frequency induction furnace Auto-pouring System, and comprise two and be arranged on the other hydraulic cylinders of electric furnace, the take-off lever of hydraulic cylinder is overturn by leverage and body of heater and is in transmission connection; The oil-in of hydraulic cylinder is provided with flow control valve, also comprises feedback control system; Flow control valve is the electric proportional-regulation valve comprising proportional plus integral control module, and controller major loop comprises comparator module; Backfeed loop comprises support vector regression algoritic module, and sensor comprises the current sensor of electric proportional-regulation valve, the angular transducer at electric furnace inclination angle, the angular-rate sensor of electric furnace upset; The output of each sensor connects with the corresponding input signal of telecommunication of support vector regression algoritic module, the output of support vector regression algoritic module is connected with the input signal of telecommunication of comparator module, and the output of comparator module is connected with the input signal of telecommunication of electric proportional-regulation valve.

The corresponding input of described support vector regression algoritic module is also connected with time input signal.

Described support vector regression algoritic module and comparator module are computer control module.

The invention has the beneficial effects as follows: owing to being provided with feedback control system, sensor; Flow control valve is electric proportional-regulation valve, and controller backfeed loop comprises support vector regression algoritic module, comparator module; Sensor comprises the current sensor of electric proportional-regulation valve, the angular transducer at electric furnace inclination angle, the angular-rate sensor of electric furnace upset; The output of each sensor connects with the corresponding input signal of telecommunication of support vector regression algoritic module, the output of support vector regression algoritic module is connected with the input signal of telecommunication of comparator module, and the output of comparator module is connected with the feedback input end signal of telecommunication of electric proportional-regulation valve.High Precision Automatic cast can be realized, and can for different cast workpiece adjustment parameter, portability is strong.

Accompanying drawing explanation

Below in conjunction with drawings and Examples, the invention will be further described.Wherein:

Running gate system sketch when Fig. 1 is smelting furnace non-heeling condition;

The running gate system sketch that Fig. 2 is smelting furnace angle of inclination when being 0≤θ≤∠ ADB;

The running gate system sketch that Fig. 3 is smelting furnace angle of inclination when being ∠ ADB≤θ≤90 °;

When Fig. 4 is smelting furnace inclination, the relation sketch between converter nose flow and hydraulic cylinder output device stroke;

Fig. 5 is closed-loop control system block diagram of the present invention.

Figure notation numbering in accompanying drawing is described as follows: comparator module 1, electric proportional-regulation valve 2, support vector regression algoritic module 3

In accompanying drawing and various in and in reckoning process relevant letter character be defined as follows table:

w	The angular speed of smelting furnace banking motion
		θ	The angle that smelting furnace tilts
I	The input current of ratio adjusting valve
		Q	Cast flow
c	Punishment parameter
		g	Nuclear parameter
ε	Slack variable
		r	Smelting furnace radius
2r	Smelting furnace diameter
		A	First summit of the maximum longitudinal section of smelting furnace
B	Second summit of the maximum longitudinal section of smelting furnace
		C	3rd summit of the maximum longitudinal section of smelting furnace
D	4th summit of the maximum longitudinal section of smelting furnace
		E	The peak of metal bath surface when certain moment smelting furnace tilts

M	Be in transmission connection a little
		N	The tie point of bar and hydraulic cylinder
O	The rotary middle point of smelting furnace
		h	The height of smelting furnace
h1	The minimum point of metal bath surface and the distance of A point when certain moment smelting furnace tilts
		L1	OM length
L2	MN length
		α	The angle of certain moment OM and ON
β	The angle of certain moment bar and ON
		t	Moment
T	Time period
		V	The molten iron volume poured out
S	The cross section of smelting furnace molten iron
		X	The cross section of smelting furnace molten iron and the distance of converter nose
v	The movement velocity of hydraulic cylinder
		S	The travel distance of hydraulic cylinder

Detailed description of the invention

Embodiments of the invention, as shown in Figure 1, Figure 2, shown in Fig. 3, Fig. 4, Fig. 5, a kind of medium-frequency induction furnace Auto-pouring System, comprise two and be arranged on the other hydraulic cylinders of electric furnace, the take-off lever of hydraulic cylinder is overturn by leverage and body of heater and is in transmission connection; The oil-in of hydraulic cylinder is provided with flow control valve, also comprises feedback control system; Flow control valve is the electric proportional-regulation valve 2 comprising proportional plus integral control module, and controller major loop comprises comparator module 1; Backfeed loop comprises support vector regression algoritic module 3, and sensor comprises the current sensor of electric proportional-regulation valve 2, the angular transducer at electric furnace inclination angle, the angular-rate sensor of electric furnace upset; The output of each sensor connects with the corresponding input signal of telecommunication of support vector regression algoritic module 3, the output of support vector regression algoritic module 3 is connected with the input signal of telecommunication of comparator module 1, and the output of comparator module 1 is connected with the input signal of telecommunication of electric proportional-regulation valve 2.

The corresponding input of described support vector regression algoritic module 3 is also connected with time input signal.

Described support vector regression algoritic module 3 and comparator module 1 are computer control module.

Principle of the present invention is: first according to the structural parameters of smelting furnace, angular speed w, the angle θ of smelting furnace inclination of smelting furnace banking motion and the input current I of ratio adjusting valve, set up the Mathematical Modeling of cast flow Q.Because in reality, the flow value in each moment is difficult to record, support vector regression algorithm is therefore utilized to seek optimum penalty parameter c, nuclear parameter g and slack variable ε, according to the flow value of each moment running gate system of the parameter prediction sought.The flow value of prediction compares with desired theoretical value, and the speed controlling body of heater both sides hydraulic cylinder in the next moment compensates with stroke, reaches the effect of automatic control.

One, angle θ, the angular speed w of smelting furnace banking motion, the relation derivation between smelting furnace radius r and the high h of smelting furnace that flow Q and smelting furnace tilt is poured into a mould:

As shown in Figure 1, suppose at a time t, the input current of ratio adjusting valve is I to running gate system sketch during the non-heeling condition of smelting furnace, and the angle that smelting furnace tilts is θ, and the angular speed of smelting furnace banking motion is w, and cast flow is Q, and the molten iron volume poured out is V.

(1) when smelting furnace angle of inclination is 0≤θ≤∠ ADB: running gate system sketch as shown in Figure 2;

h ₁=2rtgθ

The molten iron volume that t is poured out is:

$V_1=\frac{1}{2}{\mathrm{\πr}}^2{h}_1={\mathrm{\πr}}^3\mathrm{tg\θ}$

Both sides obtain flow when smelting furnace angle of inclination is 0≤θ≤∠ ADB to time t differentiate simultaneously:

$Q_1=\frac{\mathrm{dV}}{\mathrm{dt}}={\mathrm{\πr}}^{3}w{\sec }^2\mathrm{\θ}$

(2) when stove angle of inclination is ∠ ADB≤θ≤90 °, running gate system sketch as shown in Figure 3;

Go to cut a molten iron with a cross section, the section S intercepting smelting furnace molten iron is parallel with furnace bottom face, and the section S of smelting furnace molten iron and the distance of converter nose are x, if δ=90-is θ, then can calculates and intercept molten iron area and be:

$S\left(x\right)=r^2\arccos (1-\frac{\mathrm{xtg\δ}}{r})+\mathrm{tga}(\mathrm{xtg\δ}-r)\sqrt{-x^2+2\mathrm{xrctg\δ}}$

Utilize S (x) to quadrature on interval [0, h], remaining molten iron volume in t stove can be obtained:

$V_2=\underset{0}{\overset{h}{\∫}}S\left(x\right)\mathrm{dx}=\underset{0}{\overset{h}{\∫}}[r^2\arccos (1-\frac{\mathrm{xtg\δ}}{r})+\mathrm{tg\δ}(\mathrm{xtg\δ}-r)\sqrt{-x^2+2\mathrm{xrctg\δ}}]\mathrm{dx}$

$=r^3\mathrm{ctg\δ}[\sqrt{1-\frac{h^2{\mathrm{tg}}^2\mathrm{\δ}}{r^2}}-(1-\frac{\mathrm{htg\δ}}{r})\arccos (1-\frac{\mathrm{htg\δ}}{r})]-\frac{1}{3}{\mathrm{tg}}^2\mathrm{\δ}\·\sqrt[3]{2\mathrm{hrctg\δ}-h^2}$

Because δ=90-is θ, then

$V_2=r^3\mathrm{tg\θ}[\sqrt{1-\frac{h^2\mathrm{ct}{g}^2\mathrm{\θ}}{r^2}}-(1-\frac{\mathrm{hctg\θ}}{r})\arccos (1-\frac{\mathrm{hctg\θ}}{r})]-\frac{1}{3}{\mathrm{ctg}}^2\mathrm{\θ}\·\sqrt[3]{2\mathrm{hrtg\θ}-h^2}$

The molten iron volume poured out when smelting furnace angle of inclination is ∠ ADB≤θ≤90 ° is:

V=πr ²h-V ₂

Then both sides are simultaneously to t differentiate, can obtain flow Q when smelting furnace angle of inclination is ∠ ADB≤θ≤90 ° ₂: $Q_2=\frac{\mathrm{dV}}{\mathrm{dt}}=r^{3}w[{\sec }^2\mathrm{\θ}\sqrt{1-\frac{h^2{\mathrm{ctg}}^2\mathrm{\θ}}{r^2}}-\frac{2h^2{\csc }^2\mathrm{\θ}}{r^2\sqrt{1-\frac{h^2{\mathrm{tg}}^2\mathrm{\θ}}{r}}}-{\sec }^2\mathrm{\θarccoc}(1-\frac{\mathrm{gctg\θ}}{r})$ $-(r^3\mathrm{tg\θ}-r^{2}h)\frac{\mathrm{hw}{\csc }^2\mathrm{\θ}}{\sqrt{2\mathrm{rhctg\θ}-h^2{\mathrm{ctg}}^2\mathrm{\θ}}}+\frac{w{\csc }^2\mathrm{\θ}}{3}\sqrt[3]{2\mathrm{hrtg\θ}-h^2}+\mathrm{hrwctg\θ}{\sec }^2\mathrm{\θ}\sqrt{2\mathrm{hrtg\θ}-h^2}$

Therefore can obtain pouring into a mould flow Q and θ, Mathematical Modeling between w, r, h: (formula 1)

Two, the relation derivation between flow Q and the input current I of ratio adjusting valve is poured into a mould:

When smelting furnace cant angle theta angle, the relation sketch between converter nose flow and hydraulic cylinder output device stroke as shown in Figure 4.Wherein OM=L ₁, MN=L ₂, the movement velocity of hydraulic cylinder is v, and the travel distance of hydraulic cylinder is s.

Then can obtain according to geometrical relationship:

S=L ₁+L ₂-L ₁cosα-l ₂cosβ

=L ₁(1-cos α)+L ₂(1-cos β) (formula 2)

Due to: L ₂sin β=L ₁sin α, namely $\sin \mathrm{\β}=\frac{L_1}{L_2}\sin \mathrm{\α}=k\sin \mathrm{\α},(\frac{L_1}{L_2}=k),$ Can obtain:

$\cos \mathrm{\β}=\sqrt{1-{\sin }^2\mathrm{\β}}=\sqrt{1-k^2{\sin }^2\mathrm{\α}}\≈1-\frac{1}{2}{k}^2{\sin }^2\mathrm{\α}$

Again because: ${\sin }^2\mathrm{\α}=\frac{1}{2}(1-\cos 2\mathrm{\α}),$ So $\cos \mathrm{\β}=1-\frac{1}{4}{k}^2(1-\cos 2\mathrm{\α}),$ (formula 2) can be deformed into: $S=L_1(1-\cos )+L_2\frac{1}{4}{k}^2(1-\cos 2\mathrm{\α})$ $=L_1[(1-\cos \mathrm{\α})+\frac{L_2}{L_1}k\frac{1}{4}k(1-\cos 2\mathrm{\α})]$ (formula 3) $=L_1[(1-\cos \mathrm{\α})\frac{1}{4}k(1-\cos 2\mathrm{\α})]$

Because:

$v=\frac{\mathrm{dS}}{\mathrm{dt}}=\frac{\mathrm{dS}}{\mathrm{d\α}}\·\frac{\mathrm{d\α}}{\mathrm{dt}}=\mathrm{\ω}{L}_1[\sin \mathrm{\α}+\frac{1}{4}k\·2\sin 2\mathrm{\α}]$

$=w{L}_1[\sin \mathrm{\α}+\frac{1}{2}k\sin 2\mathrm{\α}]$

So the angular speed w that can obtain smelting furnace banking motion is:

$w=\frac{v}{L_1[\sin \mathrm{\α}+\frac{1}{2}k\sin 2\mathrm{\α}]}$ (formula 4)

The movement velocity v of hydraulic cylinder can the input current I of passing ratio control valve control, i.e. v=f (I), and θ=α, then w can be deformed into:

$w=\frac{f\left(I\right)}{L_1[\sin \mathrm{\θ}+\frac{1}{2}k\sin 2\mathrm{\θ}]}$ (formula 5)

(formula 5) is substituted into (formula 1) can obtain

$Q_1=\frac{\mathrm{dV}}{\mathrm{dt}}=\frac{f\left(I\right){\mathrm{\πr}}^3{\sec }^2\mathrm{\θ}}{L_1[\sin \mathrm{\θ}+\frac{1}{2}k\sin 2\mathrm{\θ}]}$

$Q_2=\frac{\mathrm{dV}}{\mathrm{dt}}=r^3\frac{f\left(I\right)}{L_1[\sin \mathrm{\θ}+\frac{1}{2}k\sin 2\mathrm{\θ}]}[{\sec }^2\mathrm{\θ}\sqrt{1-\frac{h^2{\mathrm{ctg}}^2\mathrm{\θ}}{r^2}}-\frac{2h^2{\csc }^2\mathrm{\θ}}{r^2\sqrt{1-\frac{h^2{\mathrm{tg}}^2\mathrm{\θ}}{r}}}-{\sec }^2\mathrm{\θarccoc}(1-\frac{\mathrm{gctg\θ}}{r})]$

$-\frac{f\left(I\right)}{L_1[\sin \mathrm{\θ}+\frac{1}{2}k\sin 2\mathrm{\θ}]}[\frac{(r^3\mathrm{tg\θ}-r^{2}h)h{\csc }^2\mathrm{\θ}}{\sqrt{2\mathrm{rhctg\θ}-h^2{\mathrm{ctg}}^2\mathrm{\θ}}}-\frac{{\csc }^2\mathrm{\θ}}{3}\sqrt[3]{2\mathrm{hrth\θ}-h^2}-\mathrm{hrctg\θ}{\sec }^2\mathrm{\θ}\sqrt{2\mathrm{hrtg\θ}-h^2}]$

That is:

(formula 6)

Therefore cast flow Q can be controlled by the input current I of control ratio control valve.

Three, the realization of closed-loop control

During for certain workpiece of cast, the weight of required molten iron and volume are certain.That is:

$V=\underset{0}{\overset{T}{\∫}}Q\left(t\right)\mathrm{dt}=\underset{0}{\overset{T}{\∫}}v\left(t\right)s\left(t\right)\mathrm{dt}$ (formula 7)

Wherein: V is the molten iron volume (i.e. the volume of required molten iron) poured out;

Q (t) is the flow of a certain moment molten iron one by one, is the function about time t;

V (t) is the speed that a certain moment molten iron flows out converter nose one by one;

The area of shared converter nose when s (t) is a certain moment molten iron outflow converter nose one by one;

(formula 1) is out of shape to (formula 7):

(formula 8)

Can obtain in during 0-T, the average discharge Q of molten iron _{on average}if, can control two other hydraulic cylinders of smelting furnace make running gate system during 0-T in, flow Q (t) of each moment molten iron equals Q _{on average}, just can complete automatic casting.

The input current I of from (formula 1) to (formula 6) known Q (t) and ratio adjusting valve, the angular speed w of smelting furnace banking motion is relevant with the angle θ that smelting furnace tilts, moment t.Because also may error be there is in control procedure, the error to producing is needed to compensate (i.e. closed-loop control).

Due to running gate system be one non-linear serious, time lag large, therefore, the flow value that will record each moment is in practice very difficult, or even impossible.Therefore can be predicted the flow value in each moment by theoretical algorithm, conventional nerual network technique and the advantage of fuzzy logic control are the mathematical descriptions do not needed between constrained input, can well solve the above problems.But because the determination of neural network structure and the extraction of fuzzy rule are important problems, depend in the utilization process of reality " skill " of designer, the theoretical foundation mainly statistics of simultaneously above machine learning method.Be be tending towards infinitely-great progressive theory to number of samples at traditional statistical research, existing many machine learning methods also have this to suppose.Which results in many learning methods more outstanding in theory and but do not reach desirable effect in actual applications, but SVMs is a kind of new machine learning algorithm, its principle is that the VC being based upon Statistical Learning Theory ties up on theoretical and structural risk minimization basis.Support vector determines its topological structure, this just well compensate for and adopts traditional neural network to the undue dependence of the selection of topological structure to " skill " of designer, can reasonable solution small sample, non-linear, the problem such as seniority top digit and local minimum point, there is relatively strong generalization ability.Therefore, native system adopts algorithm of support vector machine to predict flow Q (t) of t molten iron.

For regression problem, as given training set

S={ (x ₁, Q ₁), (x ₂, Q ₂) .., (x ₁, Q _l) ∈ (R ⁿ, Q) ^l(formula 9)

X in (formula 9) _i=(I _i, w _i, θ _i, t _i) be input model, Q _i∈ Q=R is output model, wherein i=1,2 ..., l and R is set of real numbers.The problem that regression problem finally will solve is, according to training set S at R ⁿupper searching real-valued function Q (x), then a given new input model x, can obtain corresponding Q value.

Therefore, introduce kernel function and can obtain normally used ε-support vector regression, algorithm steps is as follows:

Step1: given training set S={ (x ₁, y ₁), (x ₂, y ₂) .., (x _i, y _i), wherein, y _i∈ R, i=1,2 ..., l, x _i∈ R ⁿ, S ∈ (R ⁿ× R);

Step2: for non-linear selection Radial basis kernel function K (x, x '); Gradient algorithm is utilized to seek optimum parameter ε, C, g;

Step3: construct and solve convex quadratic programming problem

$\underset{{\mathrm{\α}}^{(*)}\∈R^{2L}}{\max }-\frac{1}{2}\underset{i,j=1}{\overset{l}{\mathrm{\Σ}}}({\mathrm{\α}}_i^{*}-{\mathrm{\α}}_i)({\mathrm{\α}}_j^{*}-{\mathrm{\α}}_j)K(x,x^{\′})-\mathrm{\ϵ}\underset{i=1}{\overset{l}{\mathrm{\Σ}}}({\mathrm{\α}}_i^{*}+{\mathrm{\α}}_i)+\underset{i=1}{\overset{l}{\mathrm{\Σ}}}{y}_i({\mathrm{\α}}_i^{*}-{\mathrm{\α}}_i)$ (formula 10)

Constraints

$\underset{i=1}{\overset{l}{\mathrm{\Σ}}}({\mathrm{\α}}_i^{*}-{\mathrm{\α}}_i)=0$ (formula 11)

$0\≤{\mathrm{\α}}_i^{*}\≤C,i=\mathrm{1,2},...l$

Must separate ${\mathrm{\α}}^{(*)}={({\mathrm{\α}}_i,{\mathrm{\α}}_i^{*},...,{\mathrm{\α}}_l,{\mathrm{\α}}_l^{*})}^T$ ；

Step4: solve : choose and be positioned in open interval (0, C) component or if choosing then:

$\stackrel{\‾}{b}=y_j-\underset{i=1}{\overset{l}{\mathrm{\Σ}}}({\stackrel{\‾}{\mathrm{\α}}}_i^{*}-{\stackrel{\‾}{\mathrm{\α}}}_i)K(x_i,x_j)+\mathrm{\ϵ}$ (formula 12)

If choosing then:

$\stackrel{\‾}{b}=y_k-\underset{i=1}{\overset{l}{\mathrm{\Σ}}}({\stackrel{\‾}{\mathrm{\α}}}_i^{*}-{\stackrel{\‾}{\mathrm{\α}}}_i)K(x_i,x_j)-\mathrm{\ϵ}$ (formula 13)

Step5: structure decision function

$Q=\underset{i=1}{\overset{l}{\mathrm{\Σ}}}({\stackrel{\‾}{\mathrm{\α}}}_i^{*}-{\stackrel{\‾}{\mathrm{\α}}}_i)K(x_i,x)+\stackrel{\‾}{b}$ (formula 14)

(formula 1) to (formula 14) Mathematical Modeling for utilizing support vector regression to predict each moment flow.

Closed-loop control system block diagram of the present invention as shown in Figure 5, utilizes support vector regression algorithm to solve the relational expression of Q, w, θ, t, prediction Q ' (t), by flow value Q ' (t) of prediction and Q _{on average}do difference, difference compensates at subsequent time by the size according to the input current I of difference control ratio control valve, improves the precision of automatic casting.

Automatic control running gate system of the present invention, can not only realize the high automatic casting of intelligence degree on common medium-frequency induction furnace, and parameter that can be different for different workpiece setting, has extraordinary portability.

Claims (3)

1. a medium-frequency induction furnace Auto-pouring System, comprise two and be arranged on the other hydraulic cylinder of electric furnace, the take-off lever of hydraulic cylinder is overturn by leverage and body of heater and is in transmission connection; The oil-in of hydraulic cylinder is provided with flow control valve, it is characterized in that: also comprise feedback control system; Flow control valve is the electric proportional-regulation valve comprising proportional plus integral control module, and controller major loop comprises comparator module; Backfeed loop comprises support vector regression algoritic module, and sensor comprises current sensor, the angular transducer at electric furnace inclination angle, the angular-rate sensor of electric furnace upset of electric proportional-regulation valve; The output of each sensor connects with the corresponding input signal of telecommunication of support vector regression algoritic module, the output of support vector regression algoritic module is connected with the input signal of telecommunication of comparator module, and the output of comparator module is connected with the input signal of telecommunication of electric proportional-regulation valve.

2. a kind of medium-frequency induction furnace automatic casting control system according to claim 1, is characterized in that: the corresponding input of described support vector regression algoritic module is also connected with time input signal.

3. a kind of medium-frequency induction furnace automatic casting control system according to claim 1 and 2, is characterized in that: described support vector regression algoritic module and comparator module are computer control module.