A steady flow bin weight fuzzy control method based on fuzzy subtraction clustering

Abstract

The application discloses a steady flow bin weight fuzzy control method based on fuzzy subtraction clustering, which comprises the following steps: inputting the real-time deviation GE t and the deviation change rate GEC t of the bin weight at time t into a fuzzy controller to obtain a feeding amount action amplitude QE t ; performing inverse normalization on the feeding amount action amplitude QE t , and then adding the feeding amount QE t‑1 at time (t-1) to obtain the feeding amount Q t at time t, so as to control the steady flow bin weight in real time; the steady flow bin weight fuzzy control method can adaptively acquire expert experience and obtain a fuzzy controller, thereby realizing real-time control of the steady flow bin weight.

Description

Technical Field

[0001] This invention relates to the field of combined grinding technology, and in particular to a fuzzy control method for steady-flow bin weight based on fuzzy subtractive clustering. Background Technology

[0002] The combined grinding system consists of a stabilizing chamber, roller press, ball mill, and classifier. The stabilizing chamber holds the material to be pre-ground, buffering it and ensuring uniform mixing. The clinker enters the stabilizing chamber after passing through a belt scale. In recent years, intelligent control schemes combining fuzzy control with industrial processes have seen increased application. However, the fuzzification and fuzzy rules in fuzzy control rely heavily on expert experience; too many rules can easily lead to a "dimensionality explosion," resulting in poor control performance. This limits the application of fuzzy control. Summary of the Invention

[0003] Based on the technical problems existing in the background technology, this invention proposes a fuzzy control method for steady flow silo weight based on fuzzy subtraction clustering, in order to adaptively acquire expert experience and obtain a fuzzy controller, thereby realizing real-time control of steady flow silo weight.

[0004] The present invention proposes a fuzzy control method for steady-flow warehouse weight based on fuzzy subtraction clustering, comprising the following steps:

[0005] The real-time deviation of the bulk weight at time t is GE t and the rate of change of deviation (GEC) t The data is input into the fuzzy controller to obtain the feed rate action amplitude QE. t ;

[0006] Feeding volume action amplitude QE t Perform inverse normalization, then compare with the feed rate QE at time (t-1). t-1 The summation yields the feed amount Q at time t. t This is to enable real-time control of the weight of the steady flow bin;

[0007] The fuzzy controller is constructed as follows:

[0008] (a1) Obtain (t1~t) d The feed amount and bin weight for a given time period are set into a set D, where D = {X1, ..., X2}, in chronological order. i , ..., X d}, where X i ={Q i G i}, i = 1, ..., d, Q i Represented as t i The feed rate of the constant flow silo, G i Indicates t iThe weight of time;

[0009] (a2) Perform outlier removal on set D to obtain set D2. in, Where p = 1, ..., q, Indicates t i The feed rate in the constant flow bin conforms to a normal distribution. Indicates t i The weight of the constant-flow storage bin conforms to a normal distribution.

[0010] (a3) Based on set D2, t p Feed rate of the constant flow silo and t p The weight of time We obtain the set D3, which is a combination of time elements, where D3 = {E1, ..., E...} p , ..., E q-1}, where E p ={GE p GEC p QE p}, p = 1, ..., q-1, GE p Indicates t p Time-based weight deviation, GEC p Indicates t p The rate of change of deviation at time, QE p Indicates t p The amount of feed and the range of motion at any given moment;

[0011] (a4) Normalize set D3 to obtain set D4, calculate the density index of set D4, and obtain the cluster center and density radius of set D4.

[0012] (a5) The fuzzy membership function of the fuzzy control is initialized based on set D3, cluster center and density radius, so as to construct the fuzzy controller.

[0013] Furthermore, in (a2), outlier removal is performed on set D to obtain set D2, which specifically includes:

[0014] Using DBSCAN to cluster the set D, we obtain the set {C1, ..., C...} j C v}, j = 1, ..., v, C j Let j be the set of data contained in all clusters.

[0015] Based on set D and set D′, we obtain a data set D″=DD′∩D that is not in any cluster;

[0016] Define the data in set D″ as outlier data and remove it;

[0017] Define the data in set D′ as a normal dataset, D′={X1,…,X r , ..., X n}, X r ={Q r G r}, r = 1, ..., n, where Q r Indicates t r Normal dataset of constant flow silo feed rate, G r Indicates t r A normal dataset with a constant flow and heavy load;

[0018] The set D′ is filtered using the Laida-means algorithm to obtain set D2.

[0019] Furthermore, the step of using the Laida-means algorithm to filter set D′ to obtain set D2 specifically includes:

[0020] Applying a sliding filter to set D′ yields the filtered set. in, Indicates t i The feed rate after filtering in the constant flow chamber. Indicates t i The weight of the silo after filtering in the constant flow silo;

[0021] Determine whether the data in data D1 conforms to a normal distribution according to the Laida criterion, and remove the data that does not conform to a normal distribution, resulting in set D2;

[0022] The specific formula for sliding filtering on set D′ is as follows:

[0023]

[0024] Where N represents the filter window length, Q m and G m This indicates the feed rate and bin weight within the filter window. i represents time t i The corresponding subscript value, m, represents the intermediate parameter of the summation function during the sliding filter process;

[0025] The specific formula for determining whether the data in data D1 conforms to a normal distribution is as follows:

[0026]

[0027]

[0028]

[0029] Where q represents the amount of data in a normal dataset that conforms to a normal distribution, μ and σ represent the parameters of the normal distribution, and ρ(·) represents the normal distribution.

[0030] Furthermore, in (a4), it specifically includes:

[0031] (a4-1) Normalize set D3 to obtain set in, Indicates t p Time-normalized warehouse weight deviation Indicates t p The rate of change of deviation after time-normalization. Indicates t p Feeding volume and movement amplitude after time-normalization;

[0032] (a4-2) Calculate the sample point E in set D4 p The density index is used to select the sample point E with the highest density index. cp As the first cluster center;

[0033] (a4-3) Calculate the sample point E after removing the sample point. cp For the density indices of other sample points besides E, select the sample point E with the highest density index among the other sample points. cp+1 As another cluster center;

[0034] (a4-4) Determine sample point E cp+1 With sample point E cp If the density ratio between the two is less than the preset value δ, proceed to (a4-5); otherwise, proceed to (a4-2) to select another sample point with the largest density index as the first cluster center.

[0035] (a4-5) Determine the cluster centers and density radii of set D4, and define the cluster centers {c} for the weight deviation. GE1 c GEg c GEf}, Cluster centers of the rate of change of deviation {c GEC1 c GECg c GECf} and the cluster centers {c} of the feeding amount movement amplitude QE1 c QEg c QEf}, density radius of the bulk weight deviation {σ GE1 , …, σ GEg , …, σ GEf}, Density radius of the rate of change of deviation {σ GEC1 , …, σ GECg , …, σGECf}, Density radius of the feeding amount movement amplitude {σ QE1 , …, σ QEg , …, σ QEf}, where c GEg σ represents the g-th cluster center of the weight deviation. GEg C represents the g-th density radius representing the weight deviation. GECg σ represents the g-th cluster center of the rate of change of warehouse weight deviation. GECg c represents the g-th density radius representing the rate of change of bulk weight deviation. QEg σ represents the g-th cluster center of the feeding action amplitude. QEg Let f be the g-th density radius representing the amplitude of the feeding action, where g = 1, ..., f.

[0036] Furthermore, in (a4-2), the sample point E in set D4 is calculated. p The density index is calculated using the following formula:

[0037]

[0038]

[0039] Among them, M p For sample point E p The density index, a is E p The neighborhood radius is given by q, which represents the amount of data in a normal dataset that follows a normal distribution, and w represents an intermediate parameter of the summation function.

[0040] Calculate the sample point E in (a4-3). cp The density indices for other sample points are calculated using the following formulas:

[0041]

[0042] Among them, M cp Indicates sample point E cp The density index, M pp Indicates removing E cp The density index of other sample points is b = 1.5a, where b represents the neighborhood radius of other sample points.

[0043] Furthermore, in (a5), the fuzzy membership function of the fuzzy control is initialized based on set D3, cluster center, and density radius to construct the fuzzy controller. The specific formula for the fuzzy membership function is as follows:

[0044]

[0045]

[0046]

[0047] Where, μ Gng μ represents the membership degree corresponding to the g-th cluster center of the weight deviation. GECg μ represents the membership degree of the g-th cluster center corresponding to the rate of change of warehouse weight deviation. QEg This represents the membership degree of the g-th cluster center corresponding to the feeding amount movement amplitude.

[0048] Furthermore, the real-time deviation of the warehouse weight at time t, GE t and the rate of change of deviation (GEC) t The input is sent to the fuzzy controller as follows:

[0049] Set the target value G of the weighing bin goal ;

[0050] The weight G of the steady-flow chamber at time t is collected using a weighing sensor. t , warehouse weight G t With the set target value G goal The real-time deviation GE was obtained after comparison. t and the rate of change of deviation (GEC) t ;

[0051] Real-time deviation GE t and the rate of change of deviation (GEC) t The fuzzy controller processes the data to obtain the feed rate amplitude QE. t .

[0052] The advantages of the fuzzy control method for steady-flow silo weight based on fuzzy subtractive clustering provided by this invention are as follows: This invention provides a fuzzy controller that uses existing steady-flow silo weight data and feed rate data, employs subtractive clustering to obtain fuzzy controller parameters, converts them into fuzzy rules, and feeds the feed rate Q at time t back to the belt scale. t To achieve real-time control of the silo weight of the steady flow bin, since the feed rate is a key data point for the silo weight; without relying too much on expert experience, a fuzzy controller is obtained in order to adaptively acquire expert experience, thereby achieving real-time control of the silo weight of the steady flow bin. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the structure of the present invention;

[0054] Figure 2 A graph showing the relationship between the warehouse weight setpoint and the system's predicted warehouse weight.

[0055] Figure 3This is a graph showing the error between the predicted feed amount and the bin weight. Detailed Implementation

[0056] The technical solution of the present invention will now be described in detail through specific embodiments. Many specific details are set forth in the following description to provide a thorough understanding of the invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0057] like Figures 1 to 3 As shown, the present invention proposes a fuzzy control method for steady-flow warehouse weight based on fuzzy subtraction clustering, which includes the following steps:

[0058] S1: The real-time deviation of the bulk weight at time t (GE) t and the rate of change of deviation (GEC) t The data is input into the fuzzy controller to obtain the feed rate action amplitude QE. t ;

[0059] Set the target value G of the weighing bin goal The weight G of the steady-flow chamber at time t is collected using a weighing sensor. t , warehouse weight G t With the set target value G goal The real-time deviation GE was obtained after comparison. t and the rate of change of deviation (GEC) t GE real-time deviation t and the rate of change of deviation (GEC) t The fuzzy controller processes the data to obtain the feed rate amplitude QE. t .

[0060] S2: Feeding amount action amplitude QE t Perform inverse normalization, then compare with the feed rate QE at time (t-1). t-1 The summation yields the feed amount Q at time t. t The data is then transmitted to the belt scale for real-time control of the stabilizing bin weight.

[0061] Through steps S1 to S2, suitable fuzzy rules can be obtained without expert experience. Furthermore, the resulting fuzzy controller has a good effect on the control of the weight of the steady flow bin.

[0062] The fuzzy control method for the steady flow bin weight is applied to a device consisting of a steady flow bin, a weighing sensor, a belt scale, and a controller. The construction process of the fuzzy controller is as follows.

[0063] (b1) The feed rate of the steady flow bin is collected in real time using a belt scale to obtain the time period t1~t dFeeding quantity information {Q1, ..., Q} i Q d}, where Q i For t i The feed rate of the constant flow silo;

[0064] Weighing sensors are used to collect the weight data of the uniformly mixed material in the steady flow bin in real time, thereby obtaining the weight data for the time period t1~t2. d Warehouse weight data {G1, ..., G i , ..., G d}; where G i For t i The weight of the warehouse at each moment, i = 1, ..., d;

[0065] All Q i and G i They are grouped into a set D according to their time correspondence, D = {X1, ..., X2} i , ..., X d}, where X i ={Q i G i} corresponds to (a1) in the invention description.

[0066] In this embodiment, nearly 8 hours of data from a cement plant were obtained, with values ​​taken at 2-second intervals, resulting in a total of 14,000 data entries.

[0067] (b2) Given the neighborhood range ε, the number of core sample points MinPts;

[0068] In this embodiment, the given neighborhood range ε ​​= 0.5 and the core sample point number MinPts = 500.

[0069] (b3) Perform abnormal data removal processing on set D to obtain set D′, which corresponds to (a2) in the invention content;

[0070] Using DBSCAN and a given neighborhood range ε, and the number of core sample points MinPts, the cluster set D is used to obtain the set {C1, ..., C2}. j C v}, j = 1, ..., v, C j Let j be the set of data contained in all clusters.

[0071] Based on set D and set D′, we obtain a data set D″=DD′∩D that is not in any cluster;

[0072] Define the data in set D″ as outlier data and remove it;

[0073] Define the data in set D′ as a normal dataset, D′={X1,…,X r, ..., X n}, X r ={Q r G r}, r = 1, ..., n, where Q r Indicates t r Normal dataset of constant flow silo feed rate, G r Indicates t r A normal dataset with a constant flow of data in the warehouse.

[0074] (b4) The set D′ is further processed using the Laida-means processing algorithm, corresponding to (a2) in the invention content;

[0075] Using formula (1), a sliding filter is applied to set D′ to obtain the filtered set. in, Indicates t i The feed rate after filtering in the constant flow chamber. Indicates t i The weight of the silo after filtering in the constant flow silo;

[0076]

[0077] In this embodiment, N = 5;

[0078] According to the Raida criterion, based on formula (2), determine whether the data in data D1 conforms to a normal distribution. Remove the data that does not conform to a normal distribution, and set D2. in, Where p = 1, ..., q, Indicates t i The feed rate in the constant flow bin conforms to a normal distribution. Indicates t i The weight of the constant-flow storage bin conforms to a normal distribution.

[0079]

[0080]

[0081]

[0082] Where q represents the amount of data in a normal dataset that conforms to a normal distribution, μ and σ represent the parameters of the normal distribution, and ρ(·) represents the normal distribution.

[0083] (b5) Based on set D2, t p Feed rate of the constant flow silo and t p The weight of time We obtain the set D3, which is a combination of time elements, where D3 = {E1, ..., E...} p , ..., E q-1}, where E p ={GE p GEC p QE p}, p = 1, ..., q-1, GE p Indicates t p Time-based weight deviation, GEC p Indicates t p The rate of change of deviation at time, QE p Indicates t p The feeding amount and movement amplitude at any given time correspond to (a3) ​​in the invention description;

[0084] That is, by combining the feed amount and bin weight data within set D2, the corresponding bin weight deviation {GE1, ..., GE1} is obtained. p , ..., GE q-1}, Deviation change rate {GEC1, ..., GEC} p , ..., GEC q-1}, Feeding amount movement range {QE1, ..., QE p QE q-1}

[0085] (b6) Normalize set D3 to eliminate the influence of different dimensions, making the calculation more accurate, and obtain... in, Indicates t p Time-normalized warehouse weight deviation Indicates t p The rate of change of deviation after time-normalization. Indicates t p The normalized feeding amount movement amplitude corresponds to (a4) in the invention.

[0086] (b7) Calculate the sample point E in set D4 using formula (3). p The density index corresponds to (a4) in the invention content;

[0087]

[0088]

[0089] M p For sample point E p The density index, a is E p The neighborhood radius is given by q, which represents the amount of data in a normal dataset that follows a normal distribution, and w represents an intermediate parameter of the summation function.

[0090] (b8) Select the sample point E with the highest density index. cp As the first cluster center, E cp Let M be a value in set D4, and denote its density index as M. cp This corresponds to (a4) in the invention description.

[0091] (b9) Calculate the sample point E according to formula (4). cp Then, the density index of other sample points corresponds to (a4) in the invention content;

[0092]

[0093] Among them, M cp Indicates sample point E cp The density index, M pp Indicates removing E cp The density index of other sample points is b = 1.5a, where b represents the neighborhood radius of other sample points.

[0094] Each time a cluster center is found, the density index of other sample points is calculated;

[0095] (b10) Determine the sample point E cp+1 With sample point E cp If the density ratio between the two is less than the preset value δ, proceed to step (b7); otherwise proceed to step (b11), which corresponds to (a4) in the invention content.

[0096] δ is a value given by the user; in this embodiment, δ = 0.25 is given.

[0097] (b11) Obtain the cluster centers and density radii of set D4, and define the cluster centers {C} for the weight deviation. GE1 C GEg C GEf}, Cluster centers of the rate of change of deviation {c GEC1 C GECg c GECf} and the cluster centers {c} of the feeding amount movement amplitude QE1 c QEg c QEf}, density radius of the bulk weight deviation {σ GE1 , …, σ GEg , …, σ GEf}, Density radius of the rate of change of deviation {σ GEC1 , …, σ GECg , …, σ GECf}, Density radius of the feeding amount movement amplitude {σ QE1 , …, σ QEg, …, σ QEf}, where C GEg σ represents the g-th cluster center of the weight deviation. GEg c represents the g-th density radius representing the weight deviation. GECg σ represents the g-th cluster center of the rate of change of warehouse weight deviation. GECg C represents the g-th density radius representing the rate of change of bulk weight deviation. QEg The first movement representing the range of the feeding action g Cluster centers, σ QEg The g-th density radius represents the range of the feeding action, where g = 1, ..., f, corresponding to (a4) in the invention.

[0098] (b12) Initialize the fuzzy membership function for fuzzy control, corresponding to (a5) in the invention description, with the following formula:

[0099]

[0100]

[0101] Where, μ GEg μ represents the membership degree corresponding to the g-th cluster center of the weight deviation. GECg μ represents the membership degree of the g-th cluster center corresponding to the rate of change of warehouse weight deviation. QEg This represents the membership degree of the g-th cluster center corresponding to the feeding amount movement amplitude.

[0102] (b13) Establish fuzzy rules: If GE p The membership degree is μ GEg GEC p The membership degree is μ GECg QE p The membership degree is μ QEg .

[0103] A fuzzy controller is constructed through steps (b1) to (b13). This fuzzy controller obtains fuzzy controller parameters by using subtractive clustering based on existing steady-flow bin weight data and feed rate data, and converts them into fuzzy rules. The feed rate Q at time t is fed back to the belt scale. t This is to achieve real-time control of the slack weight in the slack, because the feed rate is a key data point for the slack weight.

[0104] To verify the effectiveness and superiority of the fuzzy control method for steady-flow warehouse based on fuzzy subtraction clustering, MATLAB simulation was used for the experiment.

[0105] Using data obtained from the field, after processing (b3) of S1 to S2, the number of clusters obtained is 3, which corresponds to C in (b3) of S2. vThe subscripts, i.e., j = 1, 2, 3, indicate three different operating conditions. Data not falling into these three conditions is labeled as -1, representing abnormal operating condition data.

[0106] Then, given δ = 0.25, the fuzzy controller parameters are obtained through subtractive clustering and converted into fuzzy rules.

[0107] Since the control quantity, namely the feed rate, is required, and the output of the steady flow silo weight fuzzy controller based on fuzzy subtraction clustering is the feed rate action amplitude, the stable feed rate value is set to 230t / h.

[0108] pass Figure 2 and 3 As can be seen, during the control process, when the set value is 30, the controller can control the feeding amount well, so that the model output follows the set value, and the error between the set value and the actual value of the bin weight is only 0.17.

[0109] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A fuzzy control method for steady-flow warehouse weight based on fuzzy subtractive clustering, characterized in that, Includes the following steps: Will Real-time deviation of warehouse weight and the rate of change of deviation The data is input into the fuzzy controller to obtain the feeding amount and movement amplitude. ; Feeding volume movement range Perform inverse normalization, then with Feeding amount at any time After adding, we get Feeding amount at any time This is to enable real-time control of the weight of the steady flow bin; The fuzzy controller is constructed as follows: (a1) Obtain Feeding volume and bin weight for a given time period, arranged in chronological order. , ,in, , , Represented as The feed rate of the constant flow silo is maintained at all times. express The weight of time; (a2) For the set Perform outlier removal to obtain a set. , ,in, ,in ， express The feed rate in the constant flow bin conforms to a normal distribution. express The weight of the constant-flow storage bin conforms to a normal distribution. (a3) Based on sets Inside Feed rate of the constant flow silo and The weight of time This yields a set of combinations based on time. , ,in, , , express Weight deviation at any given time express Rate of change of deviation at time, express The amount of feed and the range of motion at any given moment; (a4) For sets Normalization is performed to obtain the set Calculate the set The density index is used to obtain the set. Cluster centers and density radii; (a5) Based on sets The fuzzy membership function of the fuzzy control is initialized by the cluster center and density radius, thereby constructing the fuzzy controller; In (a4), specifically including: (a4-1) For sets Normalization is performed to obtain the set ,in, , express Time-normalized warehouse weight deviation express The rate of change of deviation after time-normalization. express Feeding volume and movement amplitude after time-normalization; (a4-2) Calculate the set medium sample points The density index is used to select the sample points with the highest density index. As the first cluster center; (a4-3) Calculate the number of sample points removed. For the density index of other sample points, select the sample point with the highest density index among the other sample points. As another cluster center; (a4-4) Determine the sample points With sample points Is the density ratio between them less than the preset value? If yes, proceed to (a4-5); otherwise, proceed to (a4-2) to select another sample point with the highest density index as the first cluster center. (a4-5) sets The cluster centers and density radii are defined, and the cluster centers for the weight deviation are defined. Cluster centers of deviation change rate and cluster centers of feeding amount movement amplitude Density radius of silo weight deviation Density radius of deviation change rate Density radius of feeding amount movement amplitude ,in, The first indicates the weight deviation. Cluster centers, The first indicates the weight deviation. Density radius, The first digit representing the rate of change of warehouse weight deviation Cluster centers, The first digit representing the rate of change of warehouse weight deviation Density radius, The first movement representing the range of the feeding action Cluster centers, The first movement representing the range of the feeding action Density radius, .

2. The fuzzy control method for steady-flow warehouse weight based on fuzzy subtraction clustering according to claim 1, characterized in that, In (a2) pairs of sets Perform outlier removal to obtain a set. Specifically, this includes: Clustering sets using DBSCAN Get the set , , Indicates the first j Each cluster contains a set of data. ; Based on sets and set This yields a dataset that is not located in any cluster. ; set Data within this range is defined as outlier and removed. set The data within is defined as a normal dataset. , , ,in express Normal dataset of constant flow silo feed rate express A normal dataset with a constant flow and heavy load; Using the Laida-means algorithm to process the set Filtering yields a set .

3. The fuzzy control method for steady-flow warehouse weight based on fuzzy subtraction clustering according to claim 2, characterized in that, The Laida-means algorithm is used to process the set. Filtering yields a set Specifically, this includes: For sets Perform sliding filtering to obtain the filtered set. ,in, , express The feed rate after filtering in the constant flow chamber. express The weight of the silo after filtering in the constant flow silo; Data is judged according to the Raida Criterion. If the data within the set conforms to a normal distribution, remove data that does not conform to a normal distribution. ; gather The specific formula for performing sliding filtering is as follows: in, Indicates the length of the filtering window. and This indicates the feed rate and bin weight within the filter window. , Indicates time The corresponding index value, This represents the intermediate parameters of the summation function during the sliding filter process; Judge data The formula for determining whether the internal data conforms to a normal distribution is as follows: in, This represents the amount of data in a normal dataset that conforms to a normal distribution. and The parameters representing the normal distribution, This indicates a normal distribution.

4. The fuzzy control method for steady-flow warehouse weight based on fuzzy subtraction clustering according to claim 1, characterized in that, Calculate the set in (a4-2) medium sample points The density index is calculated using the following formula: in, For sample points Density index, for neighborhood radius, This represents the amount of data in a normal dataset that conforms to a normal distribution. This represents the intermediate parameter of the summation function. ; Calculate the sample points removed in (a4-3) The density indices for other sample points are calculated using the following formulas: in, Represents sample points Density index, Indicates removal Density indices of other sample points , This represents the neighborhood radius of other sample points.

5. The fuzzy control method for steady-flow warehouse weight based on fuzzy subtraction clustering according to claim 1, characterized in that, In (a5) based on sets The fuzzy membership function of the fuzzy controller is initialized with the cluster center and density radius, thus constructing the fuzzy controller. The specific formula for the fuzzy membership function is as follows: in, The first indicates the weight deviation. The membership degree corresponding to each cluster center The first digit representing the rate of change of warehouse weight deviation The membership degree corresponding to each cluster center The first movement representing the range of the feeding action The membership degree corresponding to each cluster center.

6. The fuzzy control method for steady-flow warehouse weight based on fuzzy subtraction clustering according to claim 1, characterized in that, In Real-time deviation of warehouse weight and the rate of change of deviation The input is sent to the fuzzy controller as follows: Set the target value for the weighing bin ; Data collected using weighing sensors The weight of the constant flow warehouse , warehouse weight With the set target value The real-time deviation was obtained after comparison. and the rate of change of deviation ; Real-time deviation and the rate of change of deviation The fuzzy controller processes the data to obtain the feeding amount and amplitude. .

Images