Method and apparatus for manufacturing a plant fiber board

Abstract

The application relates to the technical field of fiberboard preparation, and discloses a preparation method and device of a plant fiberboard, which comprises the following steps: sequentially performing cleaning, drying, crushing and screening treatment on crop straws to obtain fiberboard preparation raw materials; feeding the fiberboard preparation raw materials into a preliminary cooking cylinder for preliminary cooking softening treatment, and then feeding the fiberboard preparation raw materials into a cooking pot for pressure cooking treatment; obtaining optimal cooking pressure during the pressure cooking treatment according to the fluctuation of the crop straws in the discharging area of the pressure cooking pot, and obtaining the crop straws after the pressure cooking treatment; feeding the crop straws after the pressure cooking treatment into a hot mill for fiber separation to obtain plant fibers, and then mixing the plant fibers to obtain magnesium plant fiberboard slurry; feeding the magnesium plant fiberboard slurry into a press for cold pressing forming, and then performing maintenance, cutting and polishing processes to obtain the plant fiberboard. The application improves the accuracy of the optimal cooking pressure and the quality of the plant fiberboard product.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of fiberboard preparation, in particular to a preparation method and device of plant fiberboard. BACKGROUND

[0002] Crop straw is a kind of green renewable resource which is easy to obtain, and contains a large amount of plant fiber. The plant fiberboard is a kind of artificial board made of wood fiber or other plant fiber. Since the plant fiber can not only improve the mechanical properties of cement-based materials, but also improve the physical properties such as thermal insulation and sound insulation, the plant fiberboard is widely used in the fields of building and fire protection.

[0003] The crop straw usually contains a large amount of lignin, and the lignin is a relatively hard biopolymer. When the lignin exists, the combination between the plant fibers is relatively strong, which may make the plant fibers difficult to separate and process, and affect the uniformity and mechanical properties of the plant fiberboard. Therefore, it is often necessary to first cook the crop straw to remove part of the lignin, improve the processability and plasticity of the plant fiber, and help to optimize the production process and performance of the plant fiberboard. The cooking pressure in the cooking process has a great influence on the performance of the plant fiberboard. However, due to the complex process environment and the large fluctuation of the composition of the crop straw, the cooking pressure cannot be self-adaptively adjusted in the preparation process of the plant fiberboard, which affects the quality of the plant fiberboard product. SUMMARY

[0004] The application provides a preparation method and device of plant fiberboard to solve the problem that the cooking pressure cannot be self-adaptively adjusted. The technical solutions adopted are as follows:

[0005] In a first aspect, one embodiment of the application provides a preparation method of plant fiberboard, which comprises the following steps:

[0006] Crop straw is selected and washed, and 10% moisture mass fraction of the crop straw is obtained after drying. Then, the crop straw is crushed and sieved to obtain plant fiberboard raw material. The plant fiberboard raw material is cooked and softened at a cooking temperature of 80 DEG C and a cooking pressure of 0.9 MPa to obtain treated plant fiberboard raw material.

[0007] The treated plant fiberboard raw material is sent into a pressure cooking tank for pressure cooking treatment. The optimal cooking pressure during the pressure cooking treatment is obtained according to the information fluctuation of the plant fiberboard raw material in the discharge area of the pressure cooking tank, and the plant fiberboard raw material after the pressure cooking treatment is obtained.

[0008] The plant fiber board raw material after pressure cooking treatment is subjected to fiber separation to obtain plant fiber; the plant fiber is mixed with an additive, a magnesium sulfate aqueous solution and 65% active light-burned magnesium oxide to obtain a magnesium plant fiber board slurry; the magnesium plant fiber board slurry is subjected to cold-press forming treatment at a board surface pressure of 2.5 MPa to obtain a plant fiber board rough product, which is then cured at a temperature of 15-30℃ and an air humidity of 30-50% rh for 2 days, and is subjected to cutting and polishing to obtain a plant fiber board finished product.

[0009] Preferably, the mesh size of the screen is 40 mesh.

[0010] Preferably, the cooking temperature in the pressure cooking tank is 170℃, and the material height is 3.5 m.

[0011] Preferably, the gap between the grinding plates in the thermal grinder is 0.1-0.2 mm.

[0012] Preferably, the mass ratio of silica ash, building gypsum and cellulose ether in the additive is 1:5:0.5.

[0013] Preferably, the method for obtaining the optimal cooking pressure is as follows:

[0014] The fibers in each discharge port area are sampled, and each fiber sample after sampling is evenly divided into first preset parameter sub-samples, and the fiber length, fiber width, fiber circularity and fiber distribution density of each sub-sample are collected based on a fiber image analyzer and fiber measurement software.

[0015] The sample fiber parameter steady state vector of each sampling in each discharge port area is obtained according to the data collection results of each sampling in each discharge port area.

[0016] The clustering results of the sample fiber parameter steady state vectors of all samplings in each discharge port area are obtained according to the sample fiber parameter steady state vectors of all samplings in each discharge port area.

[0017] The cooking pressure insufficiency index of each discharge port area is obtained according to the clustering results of the sample fiber parameter steady state vectors of all samplings in each discharge port area.

[0018] A sequence composed of the cooking pressure insufficiency indexes of all discharge port areas is taken as a cooking pressure insufficiency sequence, and the cooking pressure insufficiency sequence and the cooking pressure during cooking are taken as inputs of a CNN convolutional neural network, and the optimal cooking pressure during the preparation of a plant fiber board is obtained by using the CNN convolutional neural network.

[0019] Preferably, the method for obtaining the sample fiber parameter steady state vector is as follows:

[0020] Step S1: for each discharge port area, a sequence consisting of the fiber lengths of all subsamples after each sampling in ascending numerical order is used as a subsample fiber length sequence; each data in the subsample fiber length sequence is used as a target data, a second preset parameter number of data is selected from the serial number position of the target data in the subsample fiber length sequence to the left, and a sequence consisting of the target data and the second preset parameter number of data in ascending numerical order is used as a secondary fiber length sequence of the target data;

[0021] Step S2, taking the difference between the last data in the secondary fiber length sequence of the target data and any data in the secondary fiber length sequence as the numerator; taking the mapping result with a natural constant as the base and the difference between the last data sequence number in the secondary fiber length sequence of the target data and any data sequence number as the exponent as the denominator; and taking the cumulative sum of the ratio of the numerator to the denominator in the secondary fiber length sequence as the increasing steepness coefficient of the target data;

[0022] Step S3: taking the sequence consisting of the increasing steepness coefficients of all data in the sub-sample fiber length sequence as the increasing steepness sequence, clustering all data elements in the increasing steepness sequence; calculating the mean of the data elements in each cluster, and taking each data element in the cluster corresponding to the maximum value of the mean of the data elements as a steep data point in the sub-sample fiber length sequence;

[0023] Step S4, calculating the absolute value of the difference between the frequency of each fiber length value in the sub-sample fiber length sequence and the average frequency of all fiber length values, and taking the average of the cumulative sum of the negative mapping results with the natural constant as the base and the absolute value as the exponent on the sub-sample fiber length sequence as the distribution balance coefficient for each sampling;

[0024] Calculating the difference between the maximum value data and the minimum value data in the subsample fiber length sequence, and using the ratio of the distribution balance coefficient to the difference as the first component factor; calculating the reciprocal of the difference between the sequence number of each steep data point and the sequence number of the previous steep data point in the subsample fiber length sequence, and using the cumulative sum of the reciprocals in the subsample fiber length sequence as the second component factor; and using the product of the first component factor and the second component factor as the sample fiber length steady-state index for each sampling;

[0025] Step S5, replacing the fiber length in step S1 with fiber width, fiber circularity, and fiber distribution density in sequence, and repeatedly calculating steps S1, S2, S3, and S4 to obtain the steady-state index of sample fiber width, the steady-state index of sample fiber circularity, and the steady-state index of sample fiber distribution density for each sampling;

[0026] Step S6: taking the vector consisting of the steady-state index of the sample fiber length, the steady-state index of the sample fiber width, the steady-state index of the sample fiber circularity, and the steady-state index of the sample fiber distribution density of each sampling as the steady-state vector of the sample fiber parameters of each sampling.

[0027] Preferably, the method for obtaining the clustering results of the steady-state vectors of the fiber parameters of all samples taken in each discharge port area is:

[0028] For each discharge port area, a data set consisting of a steady-state index of sample fiber length, a steady-state index of sample fiber width, a steady-state index of sample fiber roundness, and a steady-state index of sample fiber distribution density in the steady-state vector of all sampled sample fiber parameters is taken as a fiber steady-state data set; a data set consisting of a steady-state index of each sample fiber parameter in the steady-state vector of all sampled sample fiber parameters is taken as a data set of each fiber parameter, wherein the fiber parameters include fiber length, fiber width, fiber roundness, and fiber distribution density;

[0029] The information gain of each fiber parameter is calculated based on the fiber steady-state data set and the data set of each fiber parameter, and the information gain of each fiber parameter is used as the numerator; the sum of the information gains of all fiber parameters is used as the denominator; and the ratio of the numerator to the denominator is used as the cooking pressure characteristic weight of each fiber parameter;

[0030] For any two sample fiber parameter steady-state vectors, calculating the product of the Euclidean distance between the two sample fiber parameter steady-state vectors in each dimension and the cooking pressure characteristic weight, and taking the cumulative sum of the products on the fiber parameters as the metric distance between the two sample fiber parameter steady-state vectors;

[0031] A clustering algorithm is used to obtain clustering results of steady-state vectors of fiber parameters of all samples taken from the discharge port area based on the metric distance.

[0032] Preferably, the method for obtaining the cooking pressure insufficiency index is:

[0033] For each cluster in the clustering results of all sample fiber parameter steady-state vectors of each discharge port area, each element in the cluster represents a sample fiber parameter steady-state vector, and a sequence consisting of the sampling moments of all elements in the cluster in ascending time order is used as the intra-cluster sampling sequence;

[0034] The sum of the differences between each element and the previous element in the cluster sampling sequence is used as the numerator; the difference between the first element and the last element in the cluster sampling sequence is calculated, and the mapping result with the natural constant as the base and the difference as the exponent is used as the denominator; the ratio of the numerator to the denominator is used as the sampling time disorder index of the cluster;

[0035] The product of multiplication of the sample fiber parameter steady state index of each fiber parameter in the sample fiber parameter steady state vector of each element in the intra-cluster sampling sequence and the cooking pressure characteristic weight of each fiber parameter is accumulated and summed on the intra-cluster sampling sequence, and the mean of the accumulated sum of the intra-cluster sampling sequence is taken as the overall steady state upward trend coefficient of the cluster.

[0036] The ratio of the sampling timing disorder index and the overall steady state upward trend coefficient is accumulated and summed on all the cluster, and the accumulated sum is taken as the cooking pressure insufficiency index of the discharge area.

[0037] In a second aspect, an embodiment of the present application provides a plant fiber board preparation device, which is used in any one of the plant fiber board preparation methods.

[0038] The beneficial effects of the present application are: based on the characteristics of the fiber parameters between the crop straw sampling samples each time, combined with information gain and clustering algorithm, the clustering results of each sampling sample in the discharge area are obtained, which has the beneficial effect of considering that different fiber parameters may have different pressure cooking characteristic weights, avoiding the problem that the accuracy of the clustering results is low due to the fact that the collected and obtained certain fiber parameter is less affected by the pressure cooking; according to the timing characteristics between the samples in the discharge area and the steady state characteristics between the sub-samples, the cooking pressure insufficiency index of different discharge areas is comprehensively constructed, the cooking pressure insufficiency sequence composed of the cooking pressure insufficiency index is taken as the input of the CNN convolutional neural network, and the optimal cooking pressure for the preparation of the plant fiber board is obtained. The beneficial effects are: the timing characteristics of the sampling are comprehensively considered, the situation that the optimal cooking pressure output by the CNN convolutional neural network is large when the accuracy of the fiber parameter data obtained by single sampling is low is avoided, the accuracy of the optimal cooking pressure is improved, and the product quality of the plant fiber board preparation is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0040] Figure 1 A flowchart of a plant fiber board preparation method provided by an embodiment of the present application;

[0041] Figure 2 An implementation flowchart of a plant fiber board preparation method provided by an embodiment of the present application. DETAILED DESCRIPTION

[0042] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of the present application.

[0043] Please refer to Figure 1 which shows a flow chart of a method for preparing a plant fiber board according to an embodiment of the present application, the method comprising the following steps:

[0044] In step S001, fiber parameter data during cooking treatment is obtained.

[0045] In the present application, the preparation of the plant fiber board is jointly completed through the steps of raw material treatment, pre-cooking treatment, pressure cooking treatment, hot grinding treatment, mixing treatment, cold pressing forming treatment, curing treatment, etc., to prepare a magnesium plant fiber board.

[0046] Raw material treatment: crop straw is input into a washing machine through a conveyor for cleaning to remove impurities such as mud, stones and the like attached to the crop straw, and then the cleaned crop straw is sent into an electric heat air drying oven for drying, wherein the moisture mass fraction of the crop straw is controlled at 10%, further, a high-speed pulverizer is used to pulverize the dried crop straw, and at the same time, the crop straw passing through a 40-mesh screen is taken as the raw material for preparing the plant fiber board.

[0047] Pre-cooking: the crop straw raw material after the raw material treatment is sent into a pre-cooking cylinder for preliminary cooking softening treatment, wherein the preliminary cooking temperature is set at 80°C and the cooking pressure is set at 0.9 MPa.

[0048] Pressure cooking treatment: the crop straw after pre-cooking is continuously and uniformly sent into a cooking pot in a spiral motion through a spiral feeding device by using the friction between the crop straw after pre-cooking and the screw, wherein the cooking temperature in the pressure cooking pot is set at 170°C and the material height in the pressure cooking pot is set at 3.5 meters.

[0049] Due to the complex process environment and the large fluctuation of the composition of crop straw, the cooking pressure cannot be self-adaptively adjusted during the pressure cooking treatment, the discharge port in the pressure cooking pot is evenly divided into M discharge regions, and every time interval t, the crop straw is sampled from each discharge region, and the total sampling time length is T, in the present application, the number M of the discharge regions is 100, the sampling time interval t is set at 1 minute, and the total sampling time length T is 60 minutes, and the implementer can self-value according to the actual situation. Since the crop straw contains a large amount of fiber, it is not conducive to the measurement of fiber parameter data, and further division is required.

[0050] Further, for each fiber sample after each sampling of each discharge port area, the fiber sample is uniformly divided into N sub-samples according to the mass size, and N is empirically valued as 50. The fiber length, fiber width, fiber circularity, and fiber distribution density of each sub-sample are measured by a fiber image analyzer and fiber measurement software (such as FiberMetric). When multiple fibers appear in the sub-sample, the average of all fiber parameters is taken as the fiber parameter measurement value of the sub-sample, and the fiber parameters include fiber length, fiber width, fiber circularity, and fiber distribution density. Since there may be missing data in the measurement data during transmission, and the dimensions of different types of measurement data are different, the application first fills the data by using the Lagrange filling method, and then eliminates the dimension by using the Z-Score standardization. The Lagrange filling method and the Z-Score standardization are both known technologies, and the specific process will not be described again.

[0051] At this point, the fiber parameter data during the cooking process is obtained, including fiber length, fiber width, fiber circularity, and fiber distribution density.

[0052] Step S002, obtaining the steepness coefficient of the increase according to the fiber parameter data during the cooking process, obtaining the distribution balance coefficient and the sample fiber parameter steady-state index according to the steepness coefficient of the increase, and obtaining the sample fiber parameter steady-state vector according to the sample fiber parameter steady-state index.

[0053] Since plant fibers not only improve the mechanical properties of cement-based materials, but also improve the physical properties such as thermal insulation and sound insulation, magnesium plant fiber boards are widely used in the fields of construction, fire protection, etc. In the process of preparing plant fiber boards, crop straw is often subjected to pressure cooking treatment. However, due to the large fluctuation of the composition of crop straw, the optimal cooking pressure required is also uncertain. In the plant fiber board preparation process parameters, if the cooking pressure is too high, it may cause excessive thermal decomposition of the crop straw, low fiber yield, and reduce the quality of the plant fiber and the plant fiber board. If the cooking pressure is too low, the pressure cooking is insufficient, and the fiber parameters of the cooked crop straw may be quite different, resulting in low production efficiency and poor quality of the plant fiber board. Therefore, the purpose of the application is to adaptively adjust the cooking pressure in the pressure cooking process according to the fiber parameters of the cooked crop straw.

[0054] Specifically, the present application takes the fiber length in the a-th sampling in the m-th discharge area as an example, and sorts the fiber length measurement values ​​collected in each subsample in ascending numerical order as a subsample fiber length sequence. For any data element in the subsample fiber length sequence, 4 data are taken to the left, and combined with themselves to construct a secondary fiber length sequence. For example, taking the i-th fiber length data in the subsample fiber length sequence as an example, the i-1, i-2, i-3, and i-4 fiber length data are taken to the left, and combined with the i-th fiber length data, they are sorted in ascending numerical order to form a secondary fiber length sequence of the i-th fiber length data. At the same time, in order to avoid endpoint problems, the missing values ​​are filled with the existing minimum values ​​in the secondary fiber length sequence.

[0055] Based on the above analysis, for the sub-sample fiber length sequence of the a-th sampling in the m-th discharge area, calculate the steepness coefficient s of the i-th fiber length data in the sub-sample fiber length sequence i : Where s i is the steepness coefficient of the i-th fiber length data in the subsample fiber length sequence, J is the total number of data elements in the secondary fiber length sequence, and in this application, J takes the empirical value of 5, L i,J , L i,j are respectively the fiber length values ​​of the J-th and j-th data elements in the secondary fiber length sequence of the i-th fiber length data in the sub-sample fiber length sequence.

[0056] When the difference between the fiber length values ​​of the Jth and jth data elements is greater, that is, L i,J -L i,j The larger the value, the more the i-th fiber length data tends to increase compared with the left data in the subsample fiber length sequence, and the greater the degree of increase. At the same time, when the J-th and j-th data elements are closer in the secondary fiber length sequence, that is, e J-j The smaller it is, the greater the weight of the jth data element in the calculation, and the more it can avoid the problem that the farther the distance in the secondary fiber length sequence is, the greater the change in the i-th fiber length data in the sub-sample fiber length sequence tends to increase, the steeper the increase trend, and the steeper the increase coefficient s. i The bigger.

[0057] Further, according to the above method, the steepness coefficient of any data element in the sub-sample fiber length sequence of the a-th sampling in the m-th discharge area is obtained. The sequence composed of all the steepness coefficients is taken as a steepness sequence, and all the elements in the steepness sequence are taken as inputs of a k-means clustering algorithm, the number of clustering clusters is 2, and the Euclidean distance is taken as a metric distance. The output of the k-means clustering algorithm is taken as a clustering result of all the elements in the steepness sequence, and the clustering result includes two clustering clusters. The k-means clustering algorithm is a known technology, and the specific process will not be described again. The average value of the steepness coefficients of all the data elements in each clustering cluster is calculated, and each data element in the clustering cluster corresponding to the maximum average value of the steepness coefficients is recorded as a steep data point in the sub-sample fiber length sequence. It should be noted that each steep data point represents the steepness coefficient of a data element in the sub-sample fiber length sequence, that is, each steep data point corresponds to a data element in the sub-sample fiber length sequence.

[0058] Based on the above analysis, the sample fiber length stability index of the fiber length parameter of the a-th sampling in the m-th discharge area is calculated wherein, wherein, is the distribution balance coefficient of the fiber length parameter of the a-th sampling in the m-th discharge area, V z is the frequency of the z-th fiber length value in the sub-sample fiber length sequence of the a-th sampling in the m-th discharge area, is the average frequency of different fiber length values in the sample fiber length sequence of the a-th sampling in the m-th discharge area, is the number of different fiber length values in the sub-sample fiber length sequence of the a-th sampling in the m-th discharge area. The calculation of the frequency and the average frequency is a known technology, and the specific process will not be described again.

[0059] is the sample fiber length stability index of the fiber length parameter of the a-th sampling in the m-th discharge area, L max , L min are the maximum and minimum fiber length values in the sub-sample fiber length sequence of the a-th sampling in the m-th discharge area, O h , O h-1 are the serial numbers of the corresponding data elements in the sub-sample fiber length sequence of the h-th and h-1-th steep data points in the steep data sequence of the a-th sampling in the m-th discharge area, is the total number of all the steep data points in the steep data sequence of the a-th sampling in the m-th discharge area.

[0060] The smaller the difference between the frequency and the average frequency of the z-th fiber length value, that is, The greater, the more consistent the fiber length of each sub-sample, the more balanced the distribution, and the distribution balance coefficient The greater.

[0061] The more consistent the fiber length of each sub-sample, the more balanced the distribution, and the distribution balance coefficient The greater, the more balanced the fiber length distribution of the sample obtained by sampling, and the smaller the range of the fiber length sequence of the sub-sample, that is, L max -L min The smaller, the smaller the range of the fiber length of the sample obtained by the a-th sampling, and the closer the two steep data points are in the fiber length sequence of the sub-sample, that is, O h -O h-1 The smaller, the less likely the steep increase in the fiber length of the sample obtained by the a-th sampling, and the more stable the fiber length parameter of the a-th sampling, the more stable state, and the sample fiber length stability index The greater.

[0062] Based on the above method, the sample fiber length stability index of the a-th sampling in the m-th discharge area can be obtained. In actual cooking process, cooking pressure is an important factor affecting the cooking effect of crop straw. When the cooking pressure is too small, the pressure cooking is insufficient, the fiber parameters of the cooked crop straw are greatly different, and the sample fiber length stability index calculated may be low.

[0063] Further, by the same method, the sample fiber width stability index, sample fiber circularity stability index, and sample fiber distribution density stability index of the a-th sampling in the m-th discharge area can be obtained, which are respectively denoted as The sample fiber length stability index The sample fiber parameter stability index belongs to different types of sample fiber parameters, and together constitutes the sample fiber parameter stability vector of the a-th sampling in the m-th discharge area

[0064] Thus, the sample fiber parameter stability vector of each sampling in each discharge area is obtained.

[0065] Step S003, obtaining the cooking pressure feature weight according to the sample fiber parameter stability vector, obtaining the clustering result of the sample fiber parameter stability vector by using the cooking pressure feature weight, and obtaining the cooking pressure insufficiency index according to the clustering result of the sample fiber parameter stability vector.

[0066] In actual plant fiber board preparation process, various fiber parameters may reflect the effect of pressure cooking, but the feature weights of fiber parameters are not consistent. For example, in the pressure cooking process, the fiber structure of crop straw will be damaged to a certain extent, causing fiber breakage or shortening. At this time, the fiber length may better reflect the effect of pressure cooking.

[0067] Specifically, for each discharge port area, a data set consisting of the steady-state index of sample fiber length, the steady-state index of sample fiber width, the steady-state index of sample fiber roundness, and the steady-state index of sample fiber distribution density in the steady-state vector of all sampled sample fiber parameters is taken as a fiber steady-state data set F, and a data set consisting of the steady-state index of each sample fiber parameter in the steady-state vector of all sampled sample fiber parameters is taken as a data set of each fiber parameter, and the fiber parameters include fiber length, fiber width, fiber roundness, and fiber distribution density.

[0068] Based on the above analysis, for each outlet area, the cooking pressure characteristic weight α of the x-th fiber parameter is calculated: x : Where, α x is the cooking pressure characteristic weight of the x-th fiber parameter, F is the fiber steady-state data set, u x is the data set of the x-th fiber parameter in the fiber steady-state data set, Ig(F,u x ) is the information gain of the x-th fiber parameter, X is the total number of fiber parameters, and in this application, the empirical value is 4. The calculation of information gain is a well-known technology, and the specific process will not be repeated here.

[0069] When the information gain of the x-th fiber parameter is larger, that is, Ig(F,u x ) is larger, indicating that the x-th fiber parameter is more likely to divide the fiber steady-state data set. In the process of pressure cooking, the more it can reflect the effect of pressure cooking, the more likely it is to have a higher feature weight. The cooking pressure feature weight α x The bigger.

[0070] Taking the mth discharge area as an example, crop straw sampled at different times may have similar fiber characteristics. The data points of the steady-state vectors of all sample fiber parameters in the multidimensional feature space are used as the input of the K-mediods clustering algorithm, where the metric distance D between the steady-state vectors of the yth and y′th sample fiber parameters is y,y′ for: Where D y,y′ is the metric distance between the steady-state vectors of the fiber parameters of the yth and y′th samples, α x is the cooking pressure characteristic weight of the x-th fiber parameter, is the Euclidean distance between the steady-state vectors of the fiber parameters of the y-th and y′-th samples in the multidimensional feature space in the x-th dimension, where X represents the total number of fiber parameters.

[0071] Further, the number of clustering clusters of the K-mediods clustering algorithm is determined by the elbow method, the above-mentioned distance is taken as the distance of the K-mediods clustering algorithm, and the output of the K-mediods clustering algorithm is taken as a clustering result composed of K clustering clusters of the sample fiber parameter steady-state vector, wherein the K-mediods clustering algorithm and the elbow method are both known technologies, and the specific process will not be described in detail.

[0072] Further, in the preparation process of the plant fiber board, the closer the two sampling times are, the more likely the crop straws of the same production batch are, and under the condition that the cooking pressure is appropriate, the more similar the fiber parameters of the two sampling samples are, and the more likely they are divided into the same clustering cluster. However, when the cooking pressure is small, the pressure cooking is insufficient, and the crop straws after cooking sampling may have great differences, that is, the two sampling samples are close in sampling time, and may be divided into different clustering clusters.

[0073] Specifically, taking the kth clustering cluster as an example, a sequence composed of sampling times of all elements in the clustering cluster in ascending order of time is taken as an intra-cluster sampling sequence, and there are data elements in the sequence.

[0074] Based on the above analysis, the cooking pressure insufficiency index W m of the mth discharge area is calculated. Wherein, In the formula, is the sampling time disorder index of the kth clustering cluster in the mth discharge area, B z is the difference between the zth and z-1th data elements in the intra-cluster sampling sequence, is the total number of data elements in the kth intra-cluster sampling sequence, R1、 are the 1st and data elements in the intra-cluster sampling sequence, respectively.

[0075] is the overall steady-state increasing coefficient of the kth clustering cluster in the mth discharge area, is the sample fiber parameter steady-state index of the xth fiber parameter in the sample fiber parameter steady-state vector corresponding to the λth data element in the intra-cluster sampling sequence, α x is the cooking pressure characteristic weight of the xth fiber parameter.

[0076] W m is the cooking pressure insufficiency index of the mth discharge area, and K is the number of all clustering clusters of the mth discharge area.

[0077] It should be noted that the sampling time of the present application refers to the sampling order, such as the sampling time of the first sampling is 1. When the distance between the sampling times of each data element in the cluster is farther, that is, B z is larger, it means that the sampling time distribution of the cluster is more discrete, and the sampling of adjacent time is not divided into a cluster. At the same time, when the distance between the sampling times represented by the first and last data elements in the sampling sequence in the cluster is farther, that is is smaller, it means that the sampling time distribution range of the cluster is larger, and the data elements in the cluster are more disordered in the sampling time sequence. The sampling time sequence disorder index is larger.

[0078] When the stability of the fiber parameters of the data elements in the sampling sequence in the cluster is higher, and the characteristic weight of the fiber parameters is larger, that is is larger, it means that the fiber parameter stability of the data elements in the cluster is higher as a whole, and the overall stability increasing coefficient is larger.

[0079] When the data elements in the cluster are more disordered in the sampling time sequence, that is is larger, it means that there may be a large difference in the crop straw after cooking in the mth discharge area, resulting in that the sample data elements with close sampling times cannot be divided into a cluster. At the same time, when the overall fiber parameter stability of the cluster is lower, that is is smaller, it means that there may be a large difference between the fiber parameters of the sub-samples of the sampling samples, and it is more likely that the cooking pressure is too small during the pressure cooking process, resulting in that the crop straw is not fully cooked, resulting in that the fiber parameters have a large difference, that is, the cooking pressure insufficient index W m is larger.

[0080] At this point, the cooking pressure insufficient index of each discharge port area is obtained.

[0081] Step S004, obtaining a cooking pressure insufficient sequence according to the cooking pressure insufficient index, and obtaining an optimal cooking pressure based on the cooking pressure insufficient sequence by using a CNN convolutional neural network.

[0082] Based on the above steps, the cooking pressure insufficiency index of each discharge area is obtained, and the sequence of the cooking pressure insufficiency indexes of all discharge areas is taken as a cooking pressure insufficiency sequence. The cooking pressure insufficiency sequence and the cooking pressure during the cooking process are respectively taken as inputs of a CNN (Convolutional Neural Networks) convolutional neural network. The optimizer of the CNN convolutional neural network is Adam, the mean square error loss function (MSE) is taken as a loss function, and the output of the neural network is the optimal cooking pressure in the preparation process of the plant fiber board. The CNN convolutional neural network and the training process are both known technologies, and the specific process will not be described here. The implementation flowchart of the present application is shown in Figure 2

[0083] The second aspect embodiment of the present application provides a plant fiber board preparation device, which is used in the plant fiber board preparation method.

[0084] Further, the crop straw after cooking is subjected to subsequent treatment by using the following plant fiber board preparation process, so as to obtain a magnesium plant fiber board:

[0085] Hot grinding treatment: the crop straw after pressure cooking is conveyed to the grinding chamber of the hot grinder by the discharge screw to separate the fibers. In the first embodiment of the present application, the gap between the grinding plates of the hot grinder is 0.1 mm, so that the lignin in the cell wall and cell wall is softened or partially dissolved, and then separated into fibers by mechanical external force;

[0086] In the second embodiment of the present application, the gap between the grinding plates of the hot grinder during hot grinding treatment is 0.15 mm.

[0087] In the third embodiment of the present application, the gap between the grinding plates of the hot grinder during hot grinding treatment is 0.2 mm.

[0088] Mixing treatment: the active light burned magnesium oxide with an activity of 65% is fully mixed with the plant fiber obtained above, and then the reinforcing agent is added and mixed, wherein the reinforcing agent accounts for 7.5% of the weight of the light burned magnesium oxide, and the silicon ash, building gypsum and cellulose ether are mixed according to the ratio of 1:5:0.5 to form. Then, the magnesium sulfate aqueous solution is added, wherein the inorganic modifier needs to be added in the magnesium sulfate aqueous solution, and the addition amount of the inorganic modifier is 1.5% of the mass of the magnesium sulfate. The inorganic modifier is a mixture of copper sulfate, ferrous sulfate, citric acid, boric acid, zinc phosphate and fly ash, and the weight ratio is 1.5:0.65:0.25:0.75:0.25:10, to prepare the magnesium plant fiber board slurry.

[0089] ​Cold-press forming treatment: the magnesium plant fiber board slurry obtained above is laid in a mold to become a blank, the blank is sent into a press together with the mold, the mold is locked after being pressed, and cold-press (i.e. normal temperature) forming is performed, wherein the plate surface pressure is 2.5 MPa, and demolding is performed after the process demolding strength is reached.

[0090] Curing treatment: in the first embodiment of the present application, curing is performed for 2 days in an environment with a temperature of 15°C and an air humidity of 50% rh, and then cutting and polishing processes are performed, thereby obtaining a magnesium plant fiber board.

[0091] In the second embodiment of the present application, curing is performed for 2 days in an environment with a temperature of 22.5°C and an air humidity of 30% rh.

[0092] In the third embodiment of the present application, curing is performed for 2 days in an environment with a temperature of 30°C and an air humidity of 40% rh.

[0093] Thus, a preparation method and device of a plant fiber board are completed.

[0094] Each of the embodiments in the specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other. Each embodiment mainly describes the difference from other embodiments. The above only describes the preferred embodiments of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. within the principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for preparing a plant fiberboard, characterized in that: The method comprises the following steps: Crop straw was selected and cleaned, dried to obtain crop straw with a moisture content of 10%, and then crushed and sieved to obtain plant fiber board raw materials; the plant fiber board raw materials were sieved at 80 The raw material of the plant fiberboard is steamed and softened at a steaming temperature of 0.5 MPa and a steaming pressure of 0.9 MPa to obtain the processed raw material of the plant fiberboard; The processed plant fiberboard raw material is fed into a cooking tank for pressure cooking, and the optimal cooking pressure during the pressure cooking process is obtained according to the information fluctuation of the plant fiberboard raw material in the discharge area of ​​the pressure cooking tank, thereby obtaining the plant fiberboard raw material after pressure cooking; The plant fiberboard raw material after pressure cooking is subjected to fiber separation to obtain plant fiber; the plant fiber is mixed with an additive, a magnesium sulfate aqueous solution, and 65% active light-burned magnesium oxide to obtain a magnesium plant fiberboard slurry; the magnesium plant fiberboard slurry is cold-pressed at a board surface pressure of 2.5 MPa to obtain a plant fiberboard crude product, and then the raw material is heated for 15-30 minutes. The finished plant fiberboard is cured for 2 days in an environment of 30-50% RH and 50% RH, and then cut and polished to obtain the finished plant fiberboard. The method for obtaining the optimal cooking pressure is: Sampling the fibers in each discharge port area, evenly dividing the fiber samples after each sampling into a first preset parameter number of subsamples according to mass, and collecting the fiber length, fiber width, fiber circularity, and fiber distribution density of each subsample based on a fiber image analyzer and fiber measurement software; Obtain the steady-state vector of the sample fiber parameters of each sampling in each discharge port area according to the data collection results after each sampling in each discharge port area; Obtaining clustering results of all sample fiber parameter steady-state vectors sampled in each discharge port area according to all sample fiber parameter steady-state vectors sampled in each discharge port area; Obtaining the insufficient cooking pressure index of each discharge port area according to the clustering results of the steady-state vectors of the fiber parameters of all samples taken in each discharge port area; The sequence composed of the insufficient cooking pressure indexes of all discharge port areas is used as the insufficient cooking pressure sequence. The insufficient cooking pressure sequence and the cooking pressure during cooking treatment are used as the input of the CNN convolutional neural network, respectively. The CNN convolutional neural network is used to obtain the optimal cooking pressure for the preparation of plant fiberboard.

2. The method for preparing a plant fiberboard according to claim 1, characterized in that: The mesh number of the sieve for screening is 40 mesh.

3. The method for preparing a plant fiberboard according to claim 1, characterized in that: The cooking temperature in the pressure cooking tank is 170 , the material height is 3.5m.

4. The method for preparing a plant fiberboard according to claim 1, characterized in that: Fiber separation is achieved through thermal grinding, and the gap between the grinding plates in the thermal grinding machine is 0.1-0.2mm.

5. The method for preparing a plant fiberboard according to claim 1, characterized in that: The mass ratio of silica fume, building gypsum and cellulose ether in the additive is 1:5:0.

5.

6. The method for preparing a plant fiberboard according to claim 1, characterized in that: The method for obtaining the steady-state vector of the sample fiber parameters is: Step S1: for each discharge port area, a sequence consisting of the fiber lengths of all subsamples after each sampling in ascending numerical order is used as a subsample fiber length sequence; each data in the subsample fiber length sequence is used as a target data, a second preset parameter number of data is selected from the serial number position of the target data in the subsample fiber length sequence to the left, and a sequence consisting of the target data and the second preset parameter number of data in ascending numerical order is used as a secondary fiber length sequence of the target data; Step S2, taking the difference between the last data in the secondary fiber length sequence of the target data and any data in the secondary fiber length sequence as the numerator; taking the mapping result with a natural constant as the base and the difference between the last data sequence number in the secondary fiber length sequence of the target data and any data sequence number as the exponent as the denominator; and taking the cumulative sum of the ratio of the numerator to the denominator in the secondary fiber length sequence as the increasing steepness coefficient of the target data; Step S3: taking the sequence consisting of the increasing steepness coefficients of all data in the sub-sample fiber length sequence as the increasing steepness sequence, clustering all data elements in the increasing steepness sequence; calculating the mean of the data elements in each cluster, and taking each data element in the cluster corresponding to the maximum value of the mean of the data elements as a steep data point in the sub-sample fiber length sequence; Step S4, calculating the absolute value of the difference between the frequency of each fiber length value in the sub-sample fiber length sequence and the average frequency of all fiber length values, and taking the average of the cumulative sum of the negative mapping results with the natural constant as the base and the absolute value as the exponent on the sub-sample fiber length sequence as the distribution balance coefficient for each sampling; Calculating the difference between the maximum value data and the minimum value data in the sub-sample fiber length sequence, and using the ratio of the distribution balance coefficient to the difference as the first component factor; Calculating the reciprocal of the difference between the serial number of each steep data point and the serial number of the previous steep data point in the subsample fiber length sequence, and taking the cumulative sum of the reciprocals in the subsample fiber length sequence as the second component factor; taking the product of the first component factor and the second component factor as the sample fiber length steady-state index of each sampling; Step S5, replacing the fiber length in step S1 with fiber width, fiber circularity, and fiber distribution density in sequence, and repeatedly calculating steps S1, S2, S3, and S4 to obtain the steady-state index of sample fiber width, the steady-state index of sample fiber circularity, and the steady-state index of sample fiber distribution density for each sampling; Step S6: taking the vector consisting of the steady-state index of the sample fiber length, the steady-state index of the sample fiber width, the steady-state index of the sample fiber circularity, and the steady-state index of the sample fiber distribution density of each sampling as the steady-state vector of the sample fiber parameters of each sampling.

7. The method for preparing a plant fiberboard according to claim 1, characterized in that: The method for obtaining the clustering results of the steady-state vectors of the sample fiber parameters of all samples in each discharge port area is as follows: For each discharge port area, a data set consisting of a steady-state index of sample fiber length, a steady-state index of sample fiber width, a steady-state index of sample fiber roundness, and a steady-state index of sample fiber distribution density in the steady-state vector of all sampled sample fiber parameters is taken as a fiber steady-state data set; a data set consisting of a steady-state index of each sample fiber parameter in the steady-state vector of all sampled sample fiber parameters is taken as a data set of each fiber parameter, wherein the fiber parameters include fiber length, fiber width, fiber roundness, and fiber distribution density; The information gain of each fiber parameter is calculated based on the fiber steady-state data set and the data set of each fiber parameter, and the information gain of each fiber parameter is used as the numerator; the sum of the information gains of all fiber parameters is used as the denominator; and the ratio of the numerator to the denominator is used as the cooking pressure characteristic weight of each fiber parameter; For any two sample fiber parameter steady-state vectors, calculating the product of the Euclidean distance between the two sample fiber parameter steady-state vectors in each dimension and the cooking pressure characteristic weight, and taking the cumulative sum of the products on the fiber parameters as the metric distance between the two sample fiber parameter steady-state vectors; A clustering algorithm is used to obtain clustering results of steady-state vectors of fiber parameters of all samples taken from the discharge port area based on the metric distance.

8. The method for preparing a plant fiberboard according to claim 7, characterized in that: The method for obtaining the cooking pressure insufficiency index is: For each cluster in the clustering results of all sample fiber parameter steady-state vectors of each discharge port area, each element in the cluster represents a sample fiber parameter steady-state vector, and a sequence consisting of the sampling moments of all elements in the cluster in ascending time order is used as the intra-cluster sampling sequence; The sum of the differences between each element and the previous element in the cluster sampling sequence is used as the numerator; the difference between the first element and the last element in the cluster sampling sequence is calculated, and the mapping result with the natural constant as the base and the difference as the exponent is used as the denominator; the ratio of the numerator to the denominator is used as the sampling time disorder index of the cluster; Calculating the product of the sample fiber parameter steady-state index of each fiber parameter in the sample fiber parameter steady-state vector of each element in the cluster sampling sequence and the cooking pressure characteristic weight of each fiber parameter, and taking the mean of the quadratic cumulative sum of the product on the cluster sampling sequence as the overall steady-state trend coefficient of the cluster; The cumulative sum of the ratio of the sampling timing disorder index to the overall steady-state trend coefficient in all clusters is used as the cooking pressure insufficiency index of the discharge port area.

Images