CN103723292A - Control method for improving performance of automatic weighing and bagging system - Google Patents

Abstract

The invention provides a control method for improving performance of an automatic weighing and bagging system. The control method includes the steps of firstly, setting the feed speed, the sampling frequency and the number of sampling times, then, conducting function fitting on four closest sampling points to obtain forecast bagging weight reaching time, when a controller judges that the renewal stopping condition is met, if the renewal stopping condition can not be met, continuing to collect samples and conduct function fitting, and when discharge stopping time is reached, enabling the controller to control a hopper valve to stop discharging. By means of the control method, bagging can be conducted at the highest bagging speed within the range allowed by the system, and high bagging accuracy is achieved.

Description

A kind of control method that promotes automatic Weighing filling packet system performance

Technical field

The present invention relates to a kind of automatic Weighing and fill with the control method of packet system, especially a kind of automatic Weighing that can promote is filled with the filling speed packet of packet system and the control method of precision.

Background technology

Automatic Weighing is filled with packet system can fill with bag according to given mass by material, adopt it to replace Traditional Man filling, not only can significantly increase work efficiency, and avoided the high-strength working of staff at high dust, strong stimulation environment, be widely used at present the industries such as chemical industry, feed, cement, logistics.The lifting that automatic Weighing is filled with packet system performance has larger construction value and outstanding economic outlook.If an automatic Weighing is filled with packet system containing 4 groups of F800 weighing controllers, 50KG chemical fertilizer needs F800 controller to carry out filling 10 seconds, if make improvements, make the filling time be reduced to 9 seconds, every group of F800 can improve 2 tons/hour of operational capabilities, according to unit, move calculating in 22 hours every day, every automatic Weighing sack filling machine additionally filling 176 tons of chemical fertilizer every day, economic benefit is obvious.

For automatic Weighing, fill with packet system, its key technical problem that need solve is how rapidly material to be carried out to filling packing according to specified rate, need to consider its filling speed and precision.

Traditional control method is to be less than a certain threshold value when the difference of measuring weight and filling bag weight setting value, and by-pass valve control is closed, and completes and fills with bag, and after filling has been wrapped, cloth bag is sent into follow-up link by belt conveyor and sealed and piling.The frequency of supposing to weigh is f (Hz), and filling with bag weight setting value is Ws(kg), material flow is V(kg/s), system fill with absolute error that bag exists should be

between do normal distribution, error upper limit ξ is filling with bag time t is

this shows: (1), as long as material flow V is not 0, it is infinitely great that the frequency f of weighing does not tend to, and absolute error exists all the time so, be lowered only and cannot eliminate; (2) material flow is larger, fills with speed packet faster, and it is also larger to fill with the error of bag; There is problem of game in the accuracy and runtime of filling with bag.

Existing improving one's methods is that filling packet procedures is divided into two stages: thick feed stage and thin feed stage, as the DCC type automatic Weighing sack filling machine that South Sea boats and ships Instrument and Equipment Company produces, the method adopts a larger material flow V in the incipient stage of feed _slightly, and work as inventory, reach certain threshold level W _tafter, adopt less material flow V _carefully.Now absolute error in

between normal distribution, error upper limit ξ ' is

filling with bag time t' is

make V _slightly> V, V _carefully≤ V, adopts the method, as long as to V parameter _slightly, V _carefully, W _trationally arrange, just can realize the lifting of precision and speed.As order can obtain error upper limit ξ ' is

filling with bag time t' is

can find out, precision and the speed of filling with bag have all obtained certain lifting, successful.But this is improved one's methods and also has certain defect: (1) still exists larger error; (2) although accelerated to fill with the speed of bag initial period, its later stage speed is slower; (3) while switching in thick, the thin stage, can bring obstacle of vibration and extra momentum, affect controller performance.

Summary of the invention

Technical matters to be solved by this invention is to provide a kind of automatic Weighing that can promote to fill with the filling bag precision of packet system and the control method of speed ability.

In order to solve the problems of the technologies described above, the present invention proposes a kind of control method that promotes automatic Weighing filling packet system performance, by controller, realize automatic Weighing filling packet system is controlled, comprise the steps:

Step 1, controller is set delivery rate V, sampling frequency f and sampling number N, and described delivery rate V, sampling frequency f and sampling number N meet following relational expression:

$V=\left\{\begin{array}{cc}{V}_{\max },& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}>T\\ \frac{\mathrm{Ws}\·f}{N},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}\≤T\end{array}\right.$

$f=\left\{\begin{array}{cc}\frac{{\mathrm{NV}}_{\max }}{\mathrm{Ws}},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}>T\\ \frac{1}{T},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}\≤T\end{array}\right.$

Wherein, Ws fills with bag weight, V for specifying _maxfor maximum delivery rate, T is the time that controller is processed a sampled data;

Step 2, when filling with bag control, controller gathers with the corresponding bag weight y of filling with the time point x that weighs of weighting apparatus according to sampling frequency f, the sample point (x of employing support vector regression method to collection _i-3, y _i-3), (x _i-2, y _i-2), (x _i-1, y _i-1) and (x _i, y _i) carry out matching, show that the functional relation of filling with between bag weight y and acquisition time x is:

y＝f _i(x),4≤i≤N

The functional relation of matching is solved:

x＝f _i ^-1(y)

By the i time, gather the correspondence filling bag weight doping and reach time t _z(i) be:

t _z(i)＝f _i ^-1(Ws)

When the correspondence that controller matching dopes is filled with the bag weight t time of advent _z(i) meet and stop update condition:

$t_z\left(i\right)-x_i<\frac{1}{f}$

Controller stops weigh time point x and the corresponding collection of filling with bag weight y to weighting apparatus;

Step 3, when stopping blowing time t _sthe correspondence reaching while stopping upgrading is filled with the bag weight t time of advent _z(i), controller cuts out feed hopper valve, stops blowing.

Adopt the step 1 of this control method, delivery rate that can be the highest according to the performance setting of controller, and there is no dividing of thick feed and thin feed stage, can effectively improve single and fill with speed packet; Adopt this control method step 2, utilize support vector regression method in real time four of nearest collection sample points to be carried out to matching, obtain real-time acquisition time x and the corresponding bag weight y fitting function of filling with, along with the carrying out of gatherer process, the function of matching is also in real-time update, what calculate real-time update stops the blowing time, can effectively improve the precision of filling with bag.

Scheme as a further improvement on the present invention, gathers the filling bag weight that dopes for the i time and reaches the time and be in step 2:

$t_z\left(i\right)=\left\{\begin{array}{c}0.2t_z(i-1)+0.8{f_i}^{-1}\left(\mathrm{Ws}\right),5\&le;i\&le;N\\ {f_i}^{-1}\left(\mathrm{Ws}\right),i=4\end{array}\right.$

Wherein, t _z(i-1) be to gather the filling bag weight doping for the i-1 time to reach the time.Adopt and thisly by gathering the filling bag weight that dopes for the i-1 time, reach filling bag weight that time and the i time collection dope and reach the method that the time carries out iteration renewal, can prevent from causing because of the data noise of equipment sampling the sudden change of matched curve slope, effectively improve the reliability of system.

As further improvement project more of the present invention, when existence shifts to an earlier date closing control, will stop blowing time t _sdeduct shut t in advance _p, obtain with the standing time t closing in advance _spfor:

t _sp＝t _s-t _p

Shut t in advance _pfor the lowering time of material from material inlet to sack, be:

$t_p=\frac{\sqrt{v_0^2+2\mathrm{gH}}-v_0}{g}$

Wherein, g is acceleration due to gravity, v ₀for the rate of onset that material falls at material inlet place, H is the vertical distance of material inlet to sack.In the blowing time, reach with the standing time of closing in advance and just stop blowing, can effectively prevent from making to fill with the phenomenon that the final weight wrapping is greater than Ws owing to flying material in the air, reduced the waste of unnecessary material, also improved the filling bag precision of filling with packet system simultaneously.

Scheme as a further improvement on the present invention, the regression parameter of support vector regression method obtains by grid search method, and concrete steps are:

Step a, fills with bag experiment by an off-line automatic Weighing, gathers weigh time point x and the corresponding bag weight y of filling with of weighting apparatus, makes the set of three regression parameters of support vector regression method be respectively: C ∈ { 0.1,1,10,100,500,10, ε ∈ { 0.01,0.05,0.1,0.5,1,10} and σ ∈ { 0.01,0.1,0.5,1,5,10};

Step b, by 6 × 6 × 6 groups of different regression parameter collection of three regression parameter combination producings of setting, under the condition of every group of regression parameter collection, adopt support vector regression method to carry out Function Fitting to weigh time point x and the corresponding bag weight y of filling with that gather, the time point x that obtains weighing under every group of regression parameter condition wraps the functional relation between weight y with corresponding filling:

y＝f _k(x)

Wherein, which group regression parameter condition k is illustrated under is carried out Function Fitting, and 1≤k≤216;

Step c, the optimum set of parameters solving according to the functional relation of matching under every group of Parameter Conditions is:

$(C^{*},{\mathrm{\&epsiv;}}^{*},{\mathrm{\&sigma;}}^{*})=\underset{(C,\mathrm{\&epsiv;},\mathrm{\&sigma;})}{\arg \min }:\underset{k=1}{\overset{216}{\mathrm{\&Sigma;}}}{\left|\right|y-f_k\left(x\right)\left|\right|}^2.$

Adopt grid search method one by one more preferentially, tries to achieve optimum set of parameters (C ^*, ε ^*, σ ^*), when controller utilizes this optimal parameter set pair acquisition time x and corresponding filling bag weight y to carry out matching, can avoid occurring larger error of fitting, improved the precision of filling with bag.

Beneficial effect of the present invention is: (1) adopts control method of the present invention there is no the thin blowing stage, with respect to traditional control method, can effectively improve filling speed packet, and the vibration producing can avoid thick, the thin blowing stage to switch time; (2) adopt support vector regression (Support Vector Regression, SVR) method is carried out matching to acquisition time x with corresponding filling bag weight y, and fitting function is carried out to real-time update, make the filling bag weight of prediction reach the time also in real-time update, improve the real-time accuracy of filling with packet system filling bag; (3) employing iteration update method can prevent from causing because of the data noise of equipment sampling the sudden change of matched curve slope, can effectively improve the reliability of system; (4) in the blowing time, reach with the standing time of closing in advance and just stop blowing, can effectively prevent from making to fill with the final weight phenomenon bigger than normal of wrapping owing to flying material in the air, reduced unnecessary waste, also improved the precision of filling with bag simultaneously; (5) adopt grid search method regression parameter more preferentially one by one, when controller carries out actual function matching, can avoid occurring larger error of fitting, improved the precision of filling with bag.

Accompanying drawing explanation

Below in conjunction with drawings and Examples, the present invention is further described.

Fig. 1 is that common automatic Weighing is filled with packet system schematic diagram;

Fig. 2 is the basic flow sheet of institute of the present invention extracting method;

Fig. 3 is method of the present invention and orthodox method and the comparison schematic diagram of improving one's methods;

Fig. 4 is the schematic diagram of matching for the first time that the present invention adopts support vector regression method;

Fig. 5 is the schematic diagram of matching for the second time that the present invention adopts support vector regression method.

The specific embodiment

As shown in Figure 1, when automatic Weighing is filled with packet system work, first material is transported to feed hopper 1, then by controller 5, control feed hopper valve 2, open feed hopper valve 2, material falls into cloth bag 3 according to certain flow rate, in feed process, weighting apparatus 4 is weighed according to certain frequency, and surveyed data are delivered in controller 5.In traditional control method, when the difference of measuring weight and filling bag weight setting value is less than a certain threshold value, controller 5 is controlled feed hopper valve closing, completes and fills with bag, and after filling has been wrapped, cloth bag 3 is sent into follow-up link by belt conveyor 6 and sealed and piling.

As shown in Figure 2, be the diagram of circuit of method provided by the present invention, by controller, realize automatic Weighing filling packet system is controlled, comprise the steps:

Step 1, controller is set delivery rate V, sampling frequency f and sampling number N, and described delivery rate V, sampling frequency f and sampling number N meet following relational expression:

$V=\left\{\begin{array}{cc}{V}_{\max },& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}>T\\ \frac{\mathrm{Ws}\&CenterDot;f}{N},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}\&le;T\end{array}\right.$

$f=\left\{\begin{array}{cc}\frac{{\mathrm{NV}}_{\max }}{\mathrm{Ws}},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}>T\\ \frac{1}{T},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}\&le;T\end{array}\right.$

Wherein, Ws fills with bag weight, V for specifying _maxfor maximum delivery rate, T is the time that controller is processed a sampled data, and this is arranged on guarantees that algorithm has under the prerequisite of sufficient time of run and sampled data, has minimized the filling bag time.

Delivery rate V has determined to fill with the time of bag, on the basis of guaranteeing system performance, V is the bigger the better, sampling frequency f has determined sampling time interval, and f more sampling time interval is shorter, and sampling number N has determined the data volume of sampling, sampling number fill with more at most bag precision higher, normal conditions down-sampling can meet the demand of Function Fitting for 10 times substantially, and the time that controller is processed a sampled data is no more than 0.3 second, and therefore given delivery rate V and sampling frequency f are:

$V=\left\{\begin{array}{cc}{V}_{\max },& \frac{\mathrm{Ws}}{{10V}_{\max }}>0.3\\ \frac{\mathrm{Ws}\&CenterDot;f}{10},& \frac{\mathrm{Ws}}{{10V}_{\max }}\&le;0.3\end{array}\right.$

$f=\left\{\begin{array}{cc}10/\frac{\mathrm{Ws}}{V_{\max }},& \frac{\mathrm{Ws}}{{10V}_{\max }}>0.3\\ 3,& \frac{\mathrm{Ws}}{{10V}_{\max }}\&le;0.3\end{array}\right.$

Step 2, when filling with bag control, controller gathers with the corresponding bag weight y of filling with the time point x that weighs of weighting apparatus according to sampling frequency f, the sample point (x of employing support vector regression method to collection _i-3, y _i-3), (x _i-2, y _i-2), (x _i-1, y _i-1) and (x _i, y _i) carry out matching, show that the functional relation of filling with between bag weight y and acquisition time x is:

y＝f _i(x),4≤i≤N

The functional relation of matching is solved:

x＝f _i ^-1(y)

By the i time, gather the correspondence filling bag weight doping and reach time t _z(i) be:

t _z(i)＝f _i ^-1(Ws)

When i=4, to front four sample (x ₁, y ₁), (x ₂, y ₂), (x ₃, y ₃) and (x ₄, y ₄) gather, the functional relation that adopts support vector regression method to simulate between filling bag weight y and acquisition time x is:

y＝f ₄(x)＝w·x+b

Wherein, weight vector w and biasing b obtain by solving following formula in support vector regression method fit procedure,

$\mathrm{Min}:\frac{1}{2}(w\&CenterDot;w)+C\underset{i=1}{\overset{4}{\mathrm{\&Sigma;}}}L(x_i,y_i,f)$

Wherein, $L(x_i,y_i,f)={|y_i-f\left(x_i\right)|}_{\mathrm{\&epsiv;}}=\max (0,|y_i-(\underset{i=1}{\overset{4}{\mathrm{\&Sigma;}}}{w}_i\&CenterDot;K(x,x_i)+b|)-\mathrm{\&epsiv;}),$ For the error between predictor and actual value, $K(x,x_i)=e^{-\frac{{\left|\right|x-x_i\left|\right|}^2}{{2\mathrm{\&sigma;}}^2},}$ σ, C and ε are regression parameter,

The filling bag weight that the 4th collection calculating according to fitting function relational expression dopes reaches the time and is:

t _z(4)＝f ₄ ^-1(Ws)

Along with the continuation of controller gatherer process, the filling bag weight doping reaches time meeting real-time update, when the correspondence that controller matching dopes is filled with the bag weight t time of advent _z(i) meet and stop update condition:

$t_z\left(i\right)-x_i<\frac{1}{f}$

Controller stops weigh time point x and the corresponding collection of filling with bag weight y to weighting apparatus;

Step 3, when stopping blowing time t _sthe correspondence reaching while stopping upgrading is filled with the bag weight t time of advent _z(i), controller cuts out feed hopper valve, stops blowing.

Adopt the step 1 of the method, delivery rate that can be the highest according to the performance setting of controller, and there is no dividing of thick feed and thin feed stage, can effectively improve single and fill with speed packet; Adopt the method step 2, utilize support vector regression method to carry out matching to four of nearest collection sample points, obtain real-time acquisition time x and the matched curve of corresponding filling bag weight y, that calculates real-time update stops the blowing time, can effectively improve the precision of filling with bag.

For the further precision of filling with bag that improves, when existence shifts to an earlier date closing control, stop blowing time t _sdeduct shut t in advance _p, obtain with the standing time t closing in advance _spfor:

t _sp＝t _s-t _p

Shut t in advance _pfor the lowering time of material from material inlet to sack, shift to an earlier date shut t _pfor:

$t_p=\frac{\sqrt{v_0^2+2\mathrm{gH}}-v_0}{g}$

Wherein, g is acceleration due to gravity, v ₀for the rate of onset that material falls at material inlet place, H is the vertical distance of material inlet to sack.Employing reached with the standing time of closing in advance and just stops blowing in the blowing time, can effectively prevent from making to fill with the phenomenon that the final weight wrapping is greater than Ws owing to flying material in the air, had reduced unnecessary waste, had also improved the precision of filling with bag simultaneously.

In order further to improve the fitting effect of support vector regression method, improve the precision of filling with bag, the regression parameter of support vector regression method obtains by grid search method, and concrete steps are:

Step a, fills with bag experiment by an off-line automatic Weighing, gathers weigh time point x and the corresponding bag weight y of filling with of weighting apparatus, makes the set of three regression parameters of support vector regression method be respectively: C ∈ { 0.1,1,10,100,500,1000}, ε ∈ { 0.01,0.05,0.1,0.5,1,10} and σ ∈ { 0.01,0.1,0.5,1,5,10};

Step b, by 6 × 6 × 6 groups of different regression parameter collection of three regression parameter combination producings of setting, under the condition of every group of regression parameter collection, adopt support vector regression method to carry out Function Fitting to weigh time point x and the corresponding bag weight y of filling with that gather, the time point x that obtains weighing under every group of regression parameter condition wraps the functional relation between weight y with corresponding filling:

y＝f _k(x)

Wherein, which group regression parameter condition k is illustrated under is carried out Function Fitting, and 1≤k≤216;

Step c, the optimum set of parameters solving according to the functional relation of matching under every group of Parameter Conditions is:

$(C^{*},{\mathrm{\&epsiv;}}^{*},{\mathrm{\&sigma;}}^{*})=\underset{(C,\mathrm{\&epsiv;},\mathrm{\&sigma;})}{\arg \min }:\underset{k=1}{\overset{216}{\mathrm{\&Sigma;}}}{\left|\right|y-f_k\left(x\right)\left|\right|}^2.$

Adopt grid search method one by one more preferentially, tries to achieve optimum set of parameters (C ^*, ε ^*, σ ^*), when controller utilizes this optimal parameter set pair acquisition time x and corresponding filling bag weight y to carry out matching, can avoid occurring larger error of fitting, improved the precision of filling with bag.

As shown in Figure 3, the filling bag control system of traditional control method adopts uniform delivery rate to carry out blowing, when sampling frequency one timing, delivery rate is faster, the filling bag weight W s that final filling bag weight departs from setting is more, and delivery rate is slower, and it is longer that single is filled with the bag deadline, all the time accurate filling bag cannot be reached, and the final filling bag weight W s that bag weight departs from setting that fills with cannot be avoided; Two-stage fills with bag method and adopts extensive material and thin two such stages of blowing to fill with bag, the delivery rate in extensive material stage is very fast, the delivery rate in thin blowing stage is slower, can reduce so certain filling bag time, but still cannot avoid the final filling bag weight W s that bag weight departs from setting that fills with; The filling bag control system of method of the present invention adopts support vector regression method to carry out matching to the sample gathering, and the function of matching is carried out to real-time update, and adopt under higher delivery rate and complete and fill with bag in the scope that can allow in system, so it is the shortest that single is filled with the bag deadline, and the final bag weight ratio of filling with approaches the filling bag weight W s setting.

As shown in Figures 4 and 5, for the further reliability of raising system, adopt support vector regression method to carry out matching for the first time to front four sampling points, according to the iteration update method in improvement project, show that the 4th collection predicts that the filling bag weight obtaining reaches time t _z(4), adopt support vector regression method to carry out matching for the second time to the the 2nd, 3,4 and the 5th sampling point, can obtain the filling bag weight that matching for the second time obtains and reach time t' _z(5)=f ₅ ^-1(Ws), according to the iteration update method in improvement project, calculate the 5th collection and predict that the filling bag weight obtaining reaches the time and is:

t _z(5)＝0.2t _z(4)+0.8t _z'(5)

Because the mode that has adopted iteration to upgrade can prevent from, because the data noise of equipment sampling causes the sudden change of matched curve slope, although likely reduced filling speed packet, but can effectively improving the reliability of system.

Claims (4)

1. promote automatic Weighing and fill with a control method for packet system performance, by controller, realize automatic Weighing filling packet system is controlled, it is characterized in that, comprise the steps:

Step 1, controller is set delivery rate V, sampling frequency f and sampling number N, and described delivery rate V, sampling frequency f and sampling number N meet following relational expression:

$V=\left\{\begin{array}{cc}{V}_{\max },& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}>T\\ \frac{\mathrm{Ws}\&CenterDot;f}{N},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}\&le;T\end{array}\right.$

$f=\left\{\begin{array}{cc}\frac{{\mathrm{NV}}_{\max }}{\mathrm{Ws}},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}>T\\ \frac{1}{T},& \frac{\mathrm{Ws}}{{\mathrm{NV}}_{\max }}\&le;T\end{array}\right.$

Wherein, Ws fills with bag weight, V for specifying _maxfor maximum delivery rate, T is the time that controller is processed a sampled data;

Step 2, when filling with bag control, controller gathers with the corresponding bag weight y of filling with the time point x that weighs of weighting apparatus according to sampling frequency f, the sample point (x of employing support vector regression method to collection _i-3, y _i-3), (x _i-2, y _i-2), (x _i-1, y _i-1) and (x _i, y _i) carry out matching, show that the functional relation of filling with between bag weight y and acquisition time x is:

y＝f _i(x),4≤i≤N

The functional relation of matching is solved:

x＝f _i ^-1(y)

By the i time, gather the correspondence filling bag weight doping and reach time t _z(i) be:

t _z(i)＝f _i ^-1(Ws)

When the correspondence that controller matching dopes is filled with the bag weight t time of advent _z(i) meet and stop update condition:

$t_z\left(i\right)-x_i<\frac{1}{f}$

Controller stops weigh time point x and the corresponding collection of filling with bag weight y to weighting apparatus;

Step 3, when stopping blowing time t _sthe correspondence reaching while stopping upgrading is filled with the bag weight t time of advent _z(i), controller cuts out feed hopper valve, stops blowing.

2. promote automatic Weighing and fill with a control method for packet system performance, it is characterized in that, in described step 2, gather the correspondence doping for the i time and fill with and wrap weight and reach the time and be:

$t_z\left(i\right)=\left\{\begin{array}{c}0.2t_z(i-1)+0.8{f_i}^{-1}\left(\mathrm{Ws}\right),5\&le;i\&le;N\\ {f_i}^{-1}\left(\mathrm{Ws}\right),i=4\end{array}\right.$

Wherein, t _z(i-1) be to gather the filling bag weight doping for the i-1 time to reach the time.

3. lifting automatic Weighing according to claim 1 and 2 is filled with the control method of packet system performance, it is characterized in that, when existence shifts to an earlier date closing control, will stop blowing time t _sdeduct shut t in advance _p, obtain with the standing time t closing in advance _spfor:

t _sp＝t _s-t _p

Described shut t in advance _pfor the lowering time of material from material inlet to sack, be:

$t_p=\frac{\sqrt{v_0^2+2\mathrm{gH}}-v_0}{g}$

Wherein, g is acceleration due to gravity, v ₀for the rate of onset that material falls at material inlet place, H is the vertical distance of material inlet to sack.

4. lifting automatic Weighing according to claim 1 and 2 is filled with the control method of packet system performance, it is characterized in that, the regression parameter of described support vector regression method obtains by grid search method, and concrete steps are:

Step a, fills with bag experiment by an off-line automatic Weighing, gathers weigh time point x and the corresponding bag weight y of filling with of weighting apparatus, makes the set of three regression parameters of support vector regression method be respectively: C ∈ { 0.1,1,10,100,500,10, ε ∈ { 0.01,0.05,0.1,0.5,1,10} and σ ∈ { 0.01,0.1,0.5,1,5,10};

Step b, by 6 × 6 × 6 groups of different regression parameter collection of three regression parameter combination producings of setting, under the condition of every group of regression parameter collection, adopt support vector regression method to carry out Function Fitting to weigh time point x and the corresponding bag weight y of filling with that gather, the time point x that obtains weighing under every group of regression parameter condition wraps the functional relation between weight y with corresponding filling:

y＝f _k(x)

Wherein, which group regression parameter condition k is illustrated under is carried out Function Fitting, and 1≤k≤216;

Step c, the optimum set of parameters solving according to the functional relation of matching under every group of Parameter Conditions is:

$(C^{*},{\mathrm{\&epsiv;}}^{*},{\mathrm{\&sigma;}}^{*})=\underset{(C,\mathrm{\&epsiv;},\mathrm{\&sigma;})}{\arg \min }:\underset{k=1}{\overset{216}{\mathrm{\&Sigma;}}}{\left|\right|y-f_k\left(x\right)\left|\right|}^2.$

Images