A control method and system for a complete set of bamboo powder crushing processing units

Abstract

The present application relates to the field of grinder unit control, and particularly relates to a bamboo powder grinding complete unit control method and system, which comprises the following steps: collecting data in real time when the unit is processing, extracting a preset number of data sets through a sliding window, and preprocessing the data sets to form training samples; training an LSTM prediction model according to the training samples, the input being the average value of the feeding amount, the fluctuation value of the cutter disc speed, the historical production efficiency, the grinding difficulty value and the grinding distance in the current sliding window, and the output being the predicted value of the production efficiency in a future preset period; correcting the reference speed of the cutter disc based on the predicted value of the future production efficiency, inputting the corrected reference speed into a PI controller, and the PI controller calculating the speed adjustment amount and outputting a control signal to adjust the cutter disc speed. The present application can dynamically adjust the cutter disc speed, reduce the speed fluctuation caused by the change of the raw material characteristics, and ensure the consistency of the coarse grinding.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of shredder unit control. In particular, it relates to a bamboo powder shredding processing complete unit control method and system. BACKGROUND

[0002] As a key raw material in the fields of biomass materials and composite material fillers, the shredding processing of bamboo powder relies on the complete unit process of "primary coarse shredding-secondary grinding-drying-classification screening". Traditional bamboo powder shredding processing units usually use fixed parameter control. For example, in the primary coarse shredding process, the fixed gear control is used to control the speed of the cutter disc engine, so that the cutter disc cuts and breaks at a fixed speed.

[0003] This control method is suitable for scenarios with relatively stable raw material characteristics and small production scale, and cannot adapt to changes in raw material characteristics (such as hardness and moisture content). When processing different batches of raw materials, the system cannot dynamically adjust the cutter disc speed according to the actual working conditions, which easily leads to large fluctuations in production efficiency, high energy consumption, and unstable product quality. SUMMARY

[0004] To solve the technical problem that the above control method cannot dynamically adjust the cutter disc speed according to the changes in raw material characteristics, the present application provides solutions in the following aspects.

[0005] In a first aspect, a bamboo powder shredding processing complete unit control method includes:

[0006] Real-time data acquisition during unit processing and extraction of a preset number of data sets through a sliding window to form training samples after preprocessing the data sets;

[0007] Training of an LSTM prediction model according to the training samples, with the input being the average value of the feeding amount, the cutter disc speed fluctuation value, the historical production efficiency, the shredding difficulty value, and the grinding distance in the current sliding window, and the output being the predicted value of the future production efficiency in a preset period;

[0008] Correction of the reference speed of the cutter disc based on the predicted value of the future production efficiency, input of the corrected reference speed into a PI controller, calculation of the speed adjustment value by the PI controller, and output of the control signal to adjust the cutter disc speed;

[0009] The correction method of the reference speed of the cutter head comprises the following steps: calculating the weight of each training sample based on the crushing difficulty value of the single training sample and the crushing difficulty value of the current sliding window; calculating the middle rail of the Boll chart based on the weight of each training sample and the actual future production efficiency; calculating the upper rail and the lower rail based on the middle rail, the actual future production efficiency of each training sample and the grinding distance; calculating the optimization coefficient of the reference speed based on the upper rail, the middle rail, the lower rail and the production efficiency prediction value of the current sliding window; and multiplying the optimization coefficient by the reference speed of the cutter head set by the system to obtain the reference speed correction value of the cutter head.

[0010] Preferably, the method for calculating the speed adjustment amount by the PI controller comprises the following steps:

[0011] The difference between the reference speed correction value and the average speed of the cutter head in the current sliding window is multiplied by a proportional coefficient to obtain a proportional term; the integral coefficient is set, and the accumulated value of the deviation between the actual speed at each collection time and the reference speed correction value within the time range of the current sliding window is calculated to obtain an integral term; and the proportional term and the integral term are added to obtain the speed adjustment amount.

[0012] Preferably, the setting method of the proportional coefficient and the integral coefficient comprises the following steps:

[0013] The proportional coefficient is obtained by multiplying the basic proportional coefficient by the crushing difficulty value of the current sliding window after positive correlation normalization processing; and the integral coefficient is obtained by multiplying the basic integral coefficient by the crushing difficulty value of the current sliding window after negative correlation normalization processing.

[0014] Preferably, the calculation method of the crushing difficulty value of the training sample comprises the following steps:

[0015] The temperature mean value of the real-time temperature of the crushing chamber extracted from the corresponding window of the training sample is calculated, the difference between the set temperature and the temperature mean value is calculated, a concave function with the set temperature as the independent variable is constructed, and the ratio of the difference to the concave function is taken as the moisture content of the raw material; the mean value of the torque of the cutter head engine extracted from the corresponding window of the training sample is calculated, the ratio of the mean value of the torque to the rated torque of the cutter head engine is calculated, and the ratio is multiplied by the moisture content of the raw material to obtain the crushing difficulty value.

[0016] Preferably, the calculation method of the weight of the training sample comprises the following steps:

[0017] The absolute value of the difference between the crushing difficulty value of the current sliding window and the crushing difficulty value of the training sample is calculated, and the absolute value is subjected to negative correlation normalization processing to obtain the weight of the corresponding training sample.

[0018] Preferably, the calculation method of the middle rail comprises the following steps:

[0019] Based on the weight of each training sample, a weighted average of the actual future production efficiency of all training samples is calculated to obtain the middle rail.

[0020] Preferably, the calculation method of the upper rail and the lower rail comprises:

[0021] Based on the weight of each training sample, a weighted standard deviation of the actual future production efficiency of all training samples around the weighted average is calculated; the reciprocal of the sum of the grinding distance set by the system at present and 1 is calculated, the sum of the reciprocal and a preset value is multiplied by the weighted standard deviation, and the product is the upper rail, and the difference between the middle rail and the product is the lower rail.

[0022] Preferably, the calculation process of the optimization coefficient comprises:

[0023] The difference between the middle rail and the predicted value of the future production efficiency of the current sliding window is calculated as a first difference, the difference between the upper rail and the lower rail is calculated as a second difference, and the ratio of the first difference to the second difference is the optimization coefficient.

[0024] Preferably, the method for obtaining the length and the step length of the current sliding window comprises:

[0025] An initial window length is set, the comminution difficulty value corresponding to the previous window of the current sliding window is subjected to negative correlation normalization processing, the sum of the comminution difficulty value after negative correlation normalization processing and 1 is multiplied by the initial window length to obtain the length of the current sliding window, and the step length between the current sliding window and the previous window is equal to the length of the current sliding window.

[0026] By adaptively adjusting the length and the step length of the sliding window according to the characteristics of the raw material, data fluctuations can be better captured, and the system response speed can be improved.

[0027] In a second aspect, a bamboo powder comminution complete unit control system comprises a processor and a memory, and the memory stores computer program instructions.

[0028] The present application has the following effects:

[0029] 1、The present application adjusts the cutter disc speed in advance through the future production efficiency prediction value, avoiding product quality fluctuations caused by adjustment lag.

[0030] 2、The application considers the grinding distance (grinding difficulty) in the process of adjusting the rotating speed, when the grinding difficulty of the secondary crushing bin is larger, the adjustment range of the optimization coefficient is smaller, avoiding the regulation of the cutter head rotating speed being too fast, which leads to the grinding being blocked; when the grinding difficulty of the secondary crushing bin is smaller, the adjustment range of the optimization coefficient is relatively larger, that is, the regulation of the cutter head rotating speed is more flexible, so that the crushing efficiency is ensured; the cutter head rotating speed is dynamically adjusted, the rotating speed fluctuation caused by the change of the raw material characteristics is reduced, and the consistency of the coarse crushing is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 is a method flow chart of steps S1-S4 in the bamboo powder crushing complete unit control method of the embodiment of the application.

[0032] Figure 2 is a method flow chart of steps S30-S33 in the bamboo powder crushing complete unit control method of the embodiment of the application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are some but not all of the embodiments of the application.

[0034] The specific embodiments of the application will be described in detail below with reference to the drawings.

[0035] The bamboo powder crushing complete unit operates, the bamboo chip raw material is put into the primary crushing bin through the feeding device, and is coarsely crushed by the cutter head (driven by an independent motor); the coarsely crushed material enters the secondary crushing bin, the grinding distance of the crushing part and the grinding disc is set according to the target bamboo powder mesh number, the crushing part and the grinding disc finely grind the coarsely crushed material, the drying device is arranged in the crushing bin to dry the material in real time (the crushing cavity temperature is monitored by an infrared temperature measuring instrument), and finally the finished product bamboo powder is obtained through grading screening. The application considers that the same batch of raw materials may also have different breaking difficulties due to different water contents, and the breaking difficulty of different batches of raw materials is different due to different hardness and water contents. If the cutter head in the primary crushing bin is always controlled to operate at a fixed reference rotating speed, there will be problems of low grinding efficiency and blocking caused by improper coarse crushing. Therefore, a control method capable of dynamically adjusting the cutter head rotating speed according to the actual working condition is proposed.

[0036] REFERENCE Figure 1 A bamboo powder crushing complete unit control method includes steps S1-S4, and specifically as follows:

[0037] S1: Real-time data acquisition during unit processing is performed, a preset number of data sets are extracted through a sliding window, the data sets are preprocessed to form training samples.

[0038] The real-time temperature in the crushing cavity of the first crushing bin is collected by an infrared thermometer, the sampling frequency is 2 Hz, and the collected data form a temperature time sequence;

[0039] The real-time torque of the driving engine driving the cutter head is collected by a torque sensor, the sampling frequency is 2 Hz, and the collected data form a torque time sequence;

[0040] The real-time rotating speed of the cutter head is collected by a rotating speed sensor, the sampling frequency is 2 Hz, and the collected data form a rotating speed time sequence;

[0041] The real-time feeding amount of the first crushing bin and the real-time output amount of the second crushing bin are collected by a weighing sensor, the sampling frequency is 2 Hz, and the collected data form a feeding amount time sequence and an output amount time sequence;

[0042] According to the man-machine interaction system, the system set temperature for drying, the system set reference rotating speed of the cutter head, and the grinding distance between the crushing part and the grinding disc of the second crushing bin are obtained.

[0043] An adaptive sliding window mechanism is constructed to extract the above data time sequence: the initial window length is set, for example, the initial window length is The length of the subsequent sliding window is adaptively adjusted based on the raw material crushing difficulty value on the basis of the initial window length, so that the larger the crushing difficulty, the smaller the sliding window length, and the corresponding speed of the system control is improved.

[0044] The method for obtaining the length of the current sliding window is expressed by the following formula:

[0045]

[0046] In the formula, represents the length of the data extracted by the current sliding window; represents the crushing difficulty value corresponding to the previous window of the current sliding window, which is calculated by the data intercepted by the previous window; represents an exponential function with natural number as the base number, represents a negative correlation normalization processing of the crushing difficulty value of the previous window, which satisfies that the larger the crushing difficulty value, the smaller the length of the current sliding window. The step length between the previous window and the current sliding window is equal to the length of the current sliding window.

[0047] For the convenience of understanding, the calculation of the length of the sliding window is described in detail in this embodiment: when the sliding window is extracted for the first time, the length of the extracted data set is the initial window length 10s. When the sliding window is extracted for the second time, the length of the second extraction window needs to be calculated first, that is, the length of the current sliding window needs to be calculated first. First, the crushing difficulty value corresponding to the window formed by the first extraction of the sliding window (i.e. the initial window) is calculated according to the first extracted data set. Then, after the crushing difficulty value is negatively correlated and normalized, it is added to 1. The sum is multiplied by the initial window length to obtain the window length of the second extraction of the sliding window, that is, the length of the current sliding window. At the same time, the step length between the current sliding window and the initial window is obtained, that is, after the initial window stops, the data is extracted with the length of the current sliding window after the interval of the step length. In this way, a plurality of windows containing data sets are obtained.

[0048] The calculation method of the crushing difficulty value corresponding to a single window comprises: calculating the temperature mean value of the real-time temperature of the crushing chamber extracted by the sliding window, calculating the difference between the set temperature and the temperature mean value, constructing a concave function with the set temperature as the independent variable, and taking the ratio of the aforementioned difference to the concave function as the moisture content of the raw material; calculating the mean value of the torque of the cutter engine extracted by the sliding window, calculating the ratio of the mean value of the torque to the rated torque of the cutter engine, and multiplying the ratio by the moisture content of the raw material to obtain the crushing difficulty value.

[0049] The specific calculation formula of the moisture content of the raw material is as follows:

[0050]

[0051] In the formula, represents the moisture content of the raw material corresponding to the previous window of the current sliding window; represents the set temperature of the system; represents the temperature mean value corresponding to the previous window of the current sliding window, which is calculated from the real-time temperature in the crushing chamber extracted by the previous window; represents a small amount, such as , in order to avoid a zero denominator; is a concave function with the set temperature as the independent variable.

[0052] There is a clear negative correlation between the moisture content of the raw material and the temperature mean value of the real-time temperature of the crushing chamber. The deviation between the set temperature and the temperature mean value is used as the "original signal" of the moisture content of the raw material. According to the marginal decreasing effect of the influence of temperature on the moisture content of the raw material in the actual drying process, that is, when the set temperature of the system is higher, the evaporation rate of the moisture of the material is higher, which also leads to the fact that the material will be more difficult to reduce moisture due to the significant reduction of the moisture content, so the concave function is used as the denominator to simulate the marginal effect that the speed of the reduction of the moisture content of the raw material slows down with the increase of the set temperature of the system to a certain extent.

[0053] The difficulty of raw material crushing is jointly determined by "shear resistance caused by water content" and "cutting resistance caused by hardness", wherein the shear resistance is positively correlated with the water content of the raw material, and the cutting resistance is positively correlated with the average value of the real-time torque of the engine, so the specific calculation formula of the crushing difficulty value of the raw material is as follows:

[0054]

[0055] In the formula, represents the crushing difficulty value corresponding to the previous window of the current sliding window; represents the water content of the raw material corresponding to the previous window of the current sliding window; represents the average torque of the cutter engine corresponding to the previous window of the current sliding window, which is calculated from the real-time torque of the cutter engine extracted from the previous window; represents the rated torque of the cutter engine set by the system.

[0056] At this point, the difficulty crushing value as the basis for adaptive adjustment is obtained, and an adaptive sliding window mechanism for extracting real-time data is constructed. The calculation method of the crushing difficulty value corresponding to the current sliding window is the same as that of the crushing difficulty value corresponding to the previous window.

[0057] According to the real-time data sequence extracted by the adaptive sliding window mechanism, a preset number (such as 500) of windows are obtained, and each window contains a set of extracted data. The average value of the feeding amount in the window is calculated according to the feeding amount time sequence segment extracted in the window; the standard deviation of the speed of the cutter in the window, i.e. the cutter speed fluctuation value, is calculated according to the speed time sequence segment of the cutter extracted in the window; the historical production efficiency in the window is calculated according to the feeding amount time sequence segment and the output time sequence segment in the window: the ratio of the data of the corresponding sampling points in the output time sequence segment and the feeding amount time sequence segment is the production efficiency corresponding to the sampling point, and the average value of the production efficiencies of all sampling points in the window is the historical production efficiency of the window; the crushing difficulty value of the window and the grinding distance corresponding to the window set by the system are obtained; at this point, the average value of the feeding amount, the cutter speed fluctuation value, the historical production efficiency, the crushing difficulty value and the grinding distance corresponding to each window are obtained.

[0058] The feeding amount data and the output data of 20 sampling points are collected at a collection frequency of 2Hz starting from the end time of each window, the average value of the production efficiencies of the 20 sampling points is calculated as the actual future production efficiency of the future 10s corresponding to each window.

[0059] At this point, the training sample formed by each window is obtained, including the input feature set: the average value of the feeding amount, the cutter speed fluctuation value, the historical production efficiency, the crushing difficulty value and the grinding distance; the training label: the actual future production efficiency of the future 10s corresponding to each window.

[0060] This step adaptively adjusts the length and step size of the sliding window based on the characteristics of the raw materials to better capture data fluctuations and improve system response speed.

[0061] S2: Train an LSTM prediction model (Long Short-Term Memory Prediction Model) based on the training samples. The inputs are the average feed rate, the shroud speed fluctuation, the historical production efficiency, the crushing difficulty, and the grinding gap within the current sliding window. The output is the predicted production efficiency for a future preset period.

[0062] Construct the mean squared error loss function:

[0063]

[0064] In the formula, Represents the loss function; This represents the total number of training samples; Indicates the first Predicted future productivity values ​​for each training sample; Indicates the first The actual future output efficiency of each training sample.

[0065] The training samples obtained in step S1 are normalized using the Min-Max normalization method, and then the LSTM prediction model is trained. The training process is an existing technique and will not be described in detail here.

[0066] The input feature set of the current window is input into the trained LSTM prediction model to obtain the future prediction efficiency prediction value of the current sliding window.

[0067] S3: Based on the predicted value of future production efficiency, the reference speed of the cutter head is corrected. The corrected reference speed is input to the PI controller. The PI controller calculates the speed adjustment amount and outputs a control signal to control the speed of the cutter head.

[0068] Among them, see Figure 2 The method for correcting the reference speed of the cutter head includes steps S30-S33, as follows:

[0069] S30: Calculate the weight of the corresponding training sample based on the crushing difficulty value of a single training sample and the crushing difficulty value of the current window.

[0070] The method for calculating the weights of training samples includes: calculating the absolute value of the difference between the current sliding window's crushing difficulty value and the crushing difficulty value of the training sample; performing negative correlation normalization on the absolute value to obtain the weight of the corresponding training sample. This is specifically expressed by the following formula:

[0071]

[0072] In the formula, Indicates the first The weights of each training sample; This indicates the difficulty value for smashing the current window; Indicates the first The crushing difficulty value of each training sample corresponding to a window; Represented by natural numbers An exponential function with base 1. This indicates that the absolute value of the difference between two crushing difficulty values ​​is negatively correlated and normalized. The weight of the training sample is positively correlated with the similarity between the training sample and the data in the current window. That is, the closer the crushing difficulty value of the training sample is to the crushing difficulty value of the current window, the greater the weight, so as to avoid training samples with large differences in characteristics from affecting the subsequent calculation of the middle track.

[0073] S31: Calculate the middle rail of the Bollinger Bands based on the weights of each training sample and the actual future production efficiency. Calculate the upper and lower rails based on the middle rail, the actual future production efficiency of each training sample, and the grinding spacing.

[0074] The calculation method for the median efficiency includes: based on the weights of each training sample, calculating the weighted average of the actual future productivity of all training samples to obtain the median efficiency. The specific formula is as follows:

[0075]

[0076] In the formula, Indicates the middle track; This represents the total number of training samples; Indicates the first The weights of each training sample; Indicates the first The actual future production efficiency of each training sample. The middle track reflects the central trend of system operation under current conditions and provides a benchmark for the deviation of predicted efficiency. By comparing the middle track with the predicted value of future production efficiency, the deviation of the predicted value from the system average level can be evaluated, thus providing a basis for the subsequent calculation of the benchmark speed optimization coefficient.

[0077] The calculation method for the upper track includes: based on the weights of each training sample, calculating the weighted standard deviation of the actual future production efficiency of all training samples around the weighted average; taking the reciprocal of the sum of the system's currently set grinding interval and 1, multiplying the sum of the reciprocal and the preset value by the weighted standard deviation, and summing the product with the middle track as the upper track. The specific formula is as follows:

[0078]

[0079] In the formula, Indicates the upper rail; Indicates the middle track; This indicates the system-defined grinding spacing for the current window; This represents the weighted standard deviation of all training samples; These are preset values. When calculating the upper and lower bands of a Bollinger Bands chart, the multiplier used to multiply the weighted standard deviation is usually taken as... In this embodiment, the grinding spacing is included in the calculation, therefore, it is taken as... The smaller the grinding spacing set by the system (the greater the grinding difficulty), the higher the upper rail; conversely, the larger the grinding spacing, the lower the upper rail.

[0080] The formula for calculating the weighted standard deviation is as follows:

[0081]

[0082] In the formula, This represents the weighted standard deviation of all training samples; This represents the total number of training samples; Indicates the first The weights of each training sample; Indicates the first The actual future output efficiency of each training sample; This indicates the middle track.

[0083] The calculation method for the lower track includes: based on the weights of each training sample, calculating the weighted standard deviation of the actual future production efficiency of all training samples around the weighted average; taking the reciprocal of the sum of the system's currently set grinding interval and 1, multiplying the sum of the reciprocal and the preset value by the weighted standard deviation, and using the difference between the middle track and the resulting product as the lower track. The specific formula is as follows:

[0084]

[0085] In the formula, Indicates the lower rail; Indicates the middle track; This indicates the system-set grinding spacing for the current window. A smaller system-set grinding spacing (resulting in greater grinding difficulty) results in a lower bottom rail, while a larger grinding spacing results in a higher bottom rail.

[0086] S32: Calculate the optimization coefficient of the base rotation speed based on the production efficiency prediction values ​​of the upper rail, middle rail, lower rail, and the current sliding window.

[0087] The calculation process of the optimization coefficient is as follows: the difference between the middle rail and the predicted value of the future production efficiency of the current sliding window is calculated as a first difference, the difference between the upper rail and the lower rail is calculated as a second difference, and the ratio of the first difference to the second difference and the sum of 1 is taken as the optimization coefficient. The specific formula is as follows:

[0088]

[0089] In the formula, Optimization coefficient; Middle rail; Predicted value of future production efficiency of the current sliding window; Upper rail; Lower rail. When the predicted value of future production efficiency is low, the optimization coefficient is increased, so as to increase the cutter head speed; when the predicted value of future production efficiency is high, the optimization coefficient is decreased, so as to decrease the cutter head speed. The cutter head speed is adjusted in advance through the predicted value of future production efficiency, so as to avoid product quality fluctuations caused by adjustment lag.

[0090] Moreover, in the adjustment process, the grinding distance (grinding difficulty) is considered. When the grinding distance is small (the grinding difficulty is large), the upper rail is high and the lower rail is low, so the difference between the upper rail and the lower rail is large. The difference between the upper rail and the lower rail is the denominator when calculating the optimization coefficient. When the denominator is large, the adjustment range of the optimization coefficient is small. That is, the greater the grinding difficulty of the secondary crushing chamber, the slower the adjustment of the cutter head speed of the primary crushing chamber, otherwise the grinding place is easy to be blocked. When the grinding distance is large (the grinding difficulty is small), the upper rail is low and the lower rail is high, so the difference between the upper rail and the lower rail is small. The difference between the upper rail and the lower rail is the denominator when calculating the optimization coefficient. When the denominator is small, the adjustment range of the optimization coefficient is relatively large. That is, the smaller the grinding difficulty of the secondary crushing chamber, the more flexible the adjustment of the cutter head speed of the primary crushing chamber, so as to ensure the crushing efficiency, dynamically adjust the cutter head speed, reduce the speed fluctuation caused by the change of raw material characteristics, and ensure the consistency of coarse crushing.

[0091] S33: Multiply the optimization coefficient by the system-set cutter head reference speed to obtain a cutter head reference speed correction value.

[0092] The calculation formula of the cutter head reference speed correction value satisfies:

[0093]

[0094] In the formula, Reference speed correction value; The reference speed of the cutterhead set by the system, and it is necessary to point out that the reference speed set by the system is updated to the reference speed correction value calculated in the current iteration process after each iteration; The optimization coefficient is represented.

[0095] After obtaining the reference speed correction value of the cutterhead, the deviation between the average speed of the cutterhead in the current window and the reference speed correction value is used as the core adjustment basis, and the crushing difficulty value is introduced to modify the PI control coefficient according to the PI algorithm idea (the PI controller is a common closed-loop control algorithm, which is applied to various industrial and automation control systems), so as to avoid excessive or insufficient adjustment, and a model of the cutterhead speed adjustment amount is constructed, which is specifically expressed by the following formula:

[0096]

[0097] In the formula, The speed adjustment amount of the cutterhead is represented. The proportion coefficient is represented, , The basic proportion coefficient is represented, The hyperbolic tangent function is represented, The crushing difficulty value in the current window is positively normalized; The reference speed correction value is represented. The average speed of the cutterhead in the current window is represented. The integral coefficient is represented, , The basic integral coefficient is represented, The crushing difficulty value in the current window is negatively normalized; The length of the current window is represented. The real-time speed at time is represented. The speed deviation at time is represented. The integral accumulates these deviations in the time interval to obtain the total deviation amount.

[0098] The proportion coefficient increases with the increase of the crushing difficulty value in the current window. When the crushing difficulty value is large, the proportion term of the cutterhead speed adjustment amount increases, which can more quickly respond to the speed deviation, improve the cutterhead speed, and ensure the coarse crushing efficiency; and the integral coefficient decreases with the increase of the crushing difficulty value. When the crushing difficulty value is large, the integral coefficient decreases, which can avoid the integral saturation phenomenon, i.e., avoid excessive adjustment caused by long-term speed deviation accumulation, so as to maintain the stability of the system; when the crushing difficulty value is small, the proportion coefficient Decreasing, so that the proportional term of the cutterhead speed adjustment amount is reduced, avoiding excessive adjustment, while the integral coefficient Increasing, which can better handle small amplitude speed deviation, ensure long-term stability of cutterhead speed, and reduce unnecessary energy consumption.

[0099] The basic proportional coefficient and the basic integral coefficient can be set by experimental calibration, and a set of basic proportional coefficient and basic integral coefficient is found to enable the system to achieve the best overall performance.

[0100] In one embodiment, a number of sets of basic proportional coefficient and basic integral coefficient are set, and put into rated calibration working conditions (raw materials with the same batch characteristics, the same feeding amount, specified cutterhead reference speed, etc.) for operation, each set for 30 minutes, and key indicators are collected: cutterhead speed fluctuation amplitude (i.e. the difference between the maximum and minimum values of the collected cutterhead speed), speed adjustment response time (from the cutterhead real-time speed deviating from the reference speed by 10% to recovering to deviating from the reference speed by ±2%), and engine energy consumption for driving the cutterhead rotation. According to the three key indicators, the reciprocal of the product of the three key indicators is calculated as the comprehensive evaluation value, and the set of basic proportional coefficient and basic integral coefficient with the highest comprehensive evaluation value is selected.

[0101] After calculating the speed adjustment amount of the cutterhead, the PI controller outputs a control signal to adjust the cutterhead speed according to the speed adjustment amount.

[0102] The present application adjusts the cutterhead speed dynamically to adapt to the changes in raw material characteristics, thereby improving production efficiency, reducing energy consumption, and ensuring the stability of product quality.

[0103] The bamboo powder crushing complete machine set control system of the present application comprises a processor and a memory, and the memory stores computer program instructions, which realize the bamboo powder crushing complete machine set control method according to the above-mentioned embodiments of the present application when executed by the processor.

[0104] The system also includes a communication bus and a communication interface and other components familiar to those skilled in the art, the settings and functions of which are known in the art, and therefore will not be described here.

[0105] It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. A control method for a complete set of bamboo powder crushing processing units, characterized in that, The application relates to a method for dynamically adjusting the reference speed of a cutter head of a grinding mill. The method comprises the following steps: collecting data in real time during the processing of a unit and extracting a preset number of data sets through a sliding window, pre-processing the data sets to form training samples; training an LSTM prediction model according to the training samples, wherein the input is the average value of the feeding amount, the speed fluctuation value of the cutter head, the historical production efficiency, the grinding difficulty value and the grinding gap in the current sliding window, and the output is the predicted value of the future production efficiency in a preset period; correcting the reference speed of the cutter head based on the predicted value of the future production efficiency, inputting the corrected reference speed into a PI controller, calculating the speed adjustment value of the PI controller and outputting a control signal to adjust the speed of the cutter head; wherein the correction method of the reference speed of the cutter head comprises the following steps: calculating the weight of the corresponding training sample based on the grinding difficulty value of the single training sample and the grinding difficulty value of the current sliding window; calculating the middle rail of the Boll chart based on the weight of each training sample and the actual future production efficiency; calculating the upper rail and the lower rail based on the middle rail, the actual future production efficiency of each training sample and the grinding gap; calculating the optimization coefficient of the reference speed based on the upper rail, the middle rail, the lower rail and the predicted value of the production efficiency in the current sliding window; and multiplying the optimization coefficient by the reference speed of the cutter head set by the system to obtain the reference speed correction value of the cutter head.

2. The control method for a complete set of bamboo powder crushing processing unit according to claim 1, characterized in that, The method for calculating the speed adjustment value of the PI controller comprises the following steps: setting a proportional coefficient, multiplying the difference between the reference speed correction value and the average value of the speed of the cutter head in the current sliding window by the proportional coefficient to obtain a proportional term; setting an integral coefficient, calculating the accumulated value of the deviation between the actual speed and the reference speed correction value at each collection time within the time range of the current sliding window to obtain an integral term; and adding the proportional term and the integral term to obtain the speed adjustment value.

3. The control method for a complete set of bamboo powder crushing processing unit according to claim 2, characterized in that, The setting method of the proportional coefficient and the integral coefficient comprises the following steps: setting a basic proportional coefficient, positively normalizing the grinding difficulty value of the current sliding window, multiplying the basic proportional coefficient by the positively normalized grinding difficulty value to obtain the proportional coefficient; and setting a basic integral coefficient, negatively normalizing the grinding difficulty value of the current sliding window, multiplying the basic integral coefficient by the negatively normalized grinding difficulty value to obtain the integral coefficient.

4. The control method for a complete set of bamboo powder crushing processing unit according to claim 1, characterized in that, The calculation method of the grinding difficulty value of the training sample comprises the following steps: calculating the temperature average value of the real-time temperature of the grinding cavity extracted from the corresponding window of the training sample, calculating the difference between the set temperature and the temperature average value, constructing a concave function with the set temperature as the independent variable, taking the ratio of the difference and the concave function as the moisture content of the raw material; calculating the average value of the torque of the cutter head engine extracted from the corresponding window of the training sample, calculating the ratio of the average value of the torque and the rated torque of the cutter head engine, and multiplying the ratio by the moisture content of the raw material to obtain the grinding difficulty value.

5. The control method for a complete set of bamboo powder crushing processing unit according to claim 4, characterized in that, The calculation method of the weight of the training sample comprises the following steps: calculating the absolute value of the difference between the grinding difficulty value of the current sliding window and the grinding difficulty value of the training sample, negatively normalizing the absolute value to obtain the weight of the corresponding training sample.

6. The control method for a complete set of bamboo powder crushing processing unit according to claim 5, characterized in that, The calculation method of the middle rail comprises the following steps: calculating the weighted average value of the actual future production efficiency of all the training samples based on the weight of each training sample to obtain the middle rail.

7. The control method for a complete set of bamboo powder crushing processing unit according to claim 6, characterized in that, The calculation method of the upper rail and the lower rail comprises the following steps: Based on the weight of each training sample, the weighted standard deviation of the actual future production efficiency of all training samples around the weighted average is calculated; the reciprocal of the sum of the current setting of the grinding distance and 1 is multiplied by the weighted standard deviation, and the product is taken as the upper rail, and the difference between the middle rail and the product is taken as the lower rail.

8. The control method for a complete set of bamboo powder crushing processing unit according to claim 7, characterized in that, The calculation process of the optimization coefficient is as follows: The difference between the middle rail and the predicted value of the future production efficiency of the current sliding window is calculated as the first difference, the difference between the upper rail and the lower rail is calculated as the second difference, and the ratio of the first difference to the second difference is taken as the sum of 1 as the optimization coefficient.

9. The control method for a complete set of bamboo powder crushing processing unit according to claim 1, characterized in that, The length and step length of the current sliding window are obtained by the following method: Set the initial window length, and perform negative correlation normalization on the crushing difficulty value corresponding to the previous window of the current sliding window, multiply the sum of the negative correlation normalized crushing difficulty value and 1 by the initial window length to obtain the length of the current sliding window, and the step length between the current sliding window and the previous window is equal to the length of the current sliding window.

10. A control system for a complete set of bamboo powder crushing processing unit, characterized in that, It comprises: A processor and a memory, the memory stores computer program instructions, when the computer program instructions are executed by the processor, the control method of the bamboo powder crushing processing complete set according to any one of claims 1-9 is realized.

Images