CN117407822A - Full-automatic bud seedling machine and control method thereof - Google Patents
Full-automatic bud seedling machine and control method thereof Download PDFInfo
- Publication number
- CN117407822A CN117407822A CN202311695198.5A CN202311695198A CN117407822A CN 117407822 A CN117407822 A CN 117407822A CN 202311695198 A CN202311695198 A CN 202311695198A CN 117407822 A CN117407822 A CN 117407822A
- Authority
- CN
- China
- Prior art keywords
- current
- abnormality
- isolated
- dimension
- full
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 69
- 230000002159 abnormal effect Effects 0.000 claims abstract description 60
- 238000005520 cutting process Methods 0.000 claims abstract description 36
- 238000001514 detection method Methods 0.000 claims abstract description 26
- 238000007405 data analysis Methods 0.000 claims abstract description 25
- 238000012544 monitoring process Methods 0.000 claims abstract description 16
- 230000005856 abnormality Effects 0.000 claims description 112
- 238000009826 distribution Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 10
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 238000012216 screening Methods 0.000 claims description 4
- 230000000737 periodic effect Effects 0.000 claims description 3
- 238000010899 nucleation Methods 0.000 claims 3
- 238000010276 construction Methods 0.000 abstract description 4
- 238000004458 analytical method Methods 0.000 description 7
- 241000196324 Embryophyta Species 0.000 description 4
- 238000002955 isolation Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 240000000111 Saccharum officinarum Species 0.000 description 2
- 235000007201 Saccharum officinarum Nutrition 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R19/00—Arrangements for measuring currents or voltages or for indicating presence or sign thereof
- G01R19/0092—Arrangements for measuring currents or voltages or for indicating presence or sign thereof measuring current only
-
- A—HUMAN NECESSITIES
- A01—AGRICULTURE; FORESTRY; ANIMAL HUSBANDRY; HUNTING; TRAPPING; FISHING
- A01C—PLANTING; SOWING; FERTILISING
- A01C1/00—Apparatus, or methods of use thereof, for testing or treating seed, roots, or the like, prior to sowing or planting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
Abstract
The invention relates to the technical field of agricultural cutting machinery, in particular to a full-automatic bud seedling machine and a control method thereof. The automatic bud seedling cutting machine comprises a full-automatic bud seedling cutting machine body and a current monitoring and analyzing module, wherein the current monitoring and analyzing module comprises a current sensor, a detection wire, a current amplifier, a fixed base and a current data analysis control chip. The current data analysis control chip analyzes the references of the current signals in each dimension through the construction of the twice isolated forest, accurately identifies the abnormal state of the current signals at real time, and further timely and effectively feeds back control signals. The invention improves the working efficiency and service life of the full-automatic bud seedling machine by improving the accuracy of detecting the abnormal running state of the full-automatic bud seedling machine.
Description
Technical Field
The invention relates to the technical field of agricultural cutting machinery, in particular to a full-automatic bud seedling machine and a control method thereof.
Background
The full-automatic bud seedling cutter can cut the stems of complete plants of gramineous plants such as sugarcane into double sections with proper lengths, and the double sections can be used as seedlings for subsequent planting. Full-automatic bud seedling cutter can improve the efficiency and the quality of seedling preparation, reduces the human cost.
In the actual working process of the full-automatic bud seedling cutting machine, because the quality of agricultural plants is unstable, for example, a certain batch of plants have more fibers, stems are harder and the like, and a lot of accidents such as motor overload, power system short circuit and the like exist in the machine operation process, the current information of the bud seedling cutting machine needs to be monitored in real time, and the abnormal situation can be timely detected. The existing abnormal running state detection function is often realized through a sensor or a safety device arranged on the machine, but the detection means is only suitable for the condition that the abnormal current state is more violent and obvious, and can not accurately detect and identify some small abnormalities or abnormal trends before the occurrence of the abnormalities, thereby influencing the working efficiency and the service life of the bud seedling cutting machine.
Disclosure of Invention
In order to solve the technical problem that the working efficiency and the service life are affected due to the fact that the current abnormal state cannot be accurately identified by the existing full-automatic bud seedling machine, the invention aims to provide the full-automatic bud seedling machine and the control method thereof, and the adopted technical scheme is as follows:
the invention provides a full-automatic bud seedling machine control method, which comprises the following steps:
acquiring current information of a full-automatic bud seedling machine cutting machine body; acquiring signal characteristics of each data point on the current signal in multiple dimensions;
obtaining an isolated forest formed by all dimension signal characteristics of all data points; the isolated trees belonging to the same dimension in the isolated forest are dimension isolated trees; the number of the isolated trees in the isolated forest is a preset first number;
obtaining initial abnormality of each data point according to abnormality indexes of each data point in the isolated forest and distribution information of each data point in isolated trees with different dimensions; screening out characteristic data points according to the initial abnormality; obtaining a dimension abnormality sequence of each dimension according to the fluctuation characteristics of the abnormality indexes of the characteristic data points on the isolated tree in each dimension isolated tree; obtaining the referential property of each dimension according to the data distribution in the dimension abnormality sequence;
reconstructing an isolated forest based on the references to obtain an optimized isolated forest; the number of the isolated trees in the optimized isolated forest is larger than the preset first number; taking the initial abnormality in the optimized isolated forest as the optimized abnormality of each data point; judging whether abnormality occurs at real time according to the optimized abnormality, and if so, feeding back a control signal to the full-automatic bud seedling cutting machine body.
Further, the signal features include current data statistics, periodicity, and waveform features;
obtaining the period length of the current signal through a time-frequency conversion method, segmenting the current signal according to the period length, and obtaining a plurality of segmented signals;
taking the mean value, variance, maximum value and minimum value of each segmented signal as the current data statistical characteristics of each data point in the segmented signal;
performing EMD (empirical mode decomposition) on the current signals to obtain IMF components and frequencies and phases on each IMF component, and taking the maximum frequency and the phase corresponding to the maximum frequency in each segmented signal as the periodic characteristic of each data point in the segmented signal;
obtaining peak values and waveform shapes on the segmented signal, and taking arithmetic coding values of the peak values and the waveform shapes as the waveform characteristics of each data point on the segmented signal.
Further, the method for acquiring the initial abnormality comprises the following steps:
obtaining the first quantity of the isolated trees in each dimension isolated tree, and taking the ratio of the first quantity of the isolated trees in each dimension isolated tree to the maximum first quantity of the isolated trees as the credibility of the corresponding dimension;
obtaining the abnormality index of the signal characteristic of each data point in each dimension according to an isolated forest algorithm, and carrying out weighted summation on the abnormality index and the credibility of the corresponding dimension to obtain a weighted abnormality index;
obtaining the second number of the isolated trees of the isolated tree where each data point is located, and taking the ratio of the second number of the isolated trees to the total number of the dimension isolated trees in the isolated forest as the number credibility of the corresponding data points;
taking the product of the number confidence and the weighted anomaly index as the initial anomaly.
Further, the method for acquiring the characteristic data points comprises the following steps:
and taking the data points with the initial abnormality greater than a preset initial abnormality threshold as the characteristic data points.
Further, the method for acquiring the dimension abnormality sequence comprises the following steps:
under one dimension, obtaining an abnormal index mean value of all characteristic data points on each isolated tree, wherein the abnormal index mean value of all the isolated trees forms an abnormal index sequence; obtaining a coefficient of variation of the abnormality index sequence; and calculating the product of the variation coefficient and each element in the abnormality index sequence under the corresponding dimension to obtain a dimension abnormality sequence.
Further, the reference acquisition method includes:
processing the dimension abnormality sequence by using an Ojin threshold algorithm to obtain two sections of subsequences; obtaining an inter-class variance between two sections of the subsequences; obtaining element quantity ratio between two sub-sequences, wherein the element quantity ratio is less than or equal to 1;
obtaining the referenceusing a referenceformula comprising:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>For the reference, +_>For the inter-class variance +_>Is the element number ratio.
Further, the method for acquiring the optimized isolated forest comprises the following steps:
and taking the referential of each dimension as a random probability value in the process of constructing the optimized isolated forest, and obtaining the optimized isolated forest by utilizing an isolated forest algorithm.
Further, the method for judging whether the abnormality occurs at the real-time moment according to the optimized abnormality comprises the following steps:
taking the data points with the optimized abnormality greater than a preset optimized abnormality threshold as abnormal data points; if the data points at the real-time moment are abnormal data points and the data points with continuous preset number are the abnormal data points, judging that the real-time moment is abnormal.
Further, the full-automatic bud seedling cutting machine body comprises a pneumatic accessory;
if the real-time moment is abnormal, the current speed of the pneumatic accessory is regarded as an initial speed, the adjustment speed of the initial speed is obtained according to the preset speed gain, the current speed of the pneumatic accessory is reduced until the real-time moment is judged to be abnormal, then the speed of the pneumatic accessory is increased to the adjustment speed, and if the abnormality can still be detected, the bud seedling cutting machine is stopped.
The invention also provides a full-automatic bud seedling machine, which comprises a full-automatic bud seedling cutting machine body, and further comprises a current monitoring and analyzing module, wherein the current monitoring and analyzing module is arranged on the full-automatic bud seedling cutting machine body and comprises a current sensor 1, a detection lead 2, a current amplifier 3, a fixed base 4 and a current data analysis control chip 5; the current sensor 1, the detection lead 2, the current amplifier 3 and the current data analysis control chip 5 are fixedly connected through the fixed base 4; the current information of the full-automatic bud seedling cutting machine body is transmitted to the current amplifier 3 through the detection lead 2, and the current sensor 1 obtains a current signal after the current signal is processed by the current amplifier 3; the current data analysis control chip 5 receives the current signal and performs data processing; the current data analysis control chip realizes the data processing step in the full-automatic bud seedling machine control method.
The invention has the following beneficial effects:
according to the invention, the current monitoring and analyzing module is arranged on the full-automatic bud seedling machine cutting machine body, wherein the current information is amplified through the current amplifier, so that the current sensor can detect clearly available current signals, and the current data analysis control chip is further utilized to analyze the current signals. The problem that the abnormality of the data points identified by the existing data abnormality algorithm is weak in referential property is considered; in order to realize timely and accurate current abnormality detection of the full-automatic bud seedling machine, the current signals involved in data analysis have the characteristics under a plurality of dimensions, so that the current data analysis control chip is improved on the existing isolated forest abnormality detection algorithm, an isolated forest with fewer isolated trees is firstly constructed, the characteristics of the isolated trees under each dimension are analyzed to obtain initial abnormality, and the abnormality information of each data point is primarily evaluated by utilizing the initial abnormality. And further obtaining the referential property of each dimension by utilizing the abnormal index distribution of the characteristic data points in the isolated forest. Because the signal characteristic dimension of the current information is more, the isolated forest is rebuilt by obtaining the referential of each dimension, so that the dimension with the larger referential has stronger referential information in the isolated forest, namely, the accurate referential information of each dimension is determined by the isolated forest with the smaller number of the isolated trees, and then the optimized isolated forest is rebuilt, so that more accurate abnormal information can be detected in the abnormal detection process. By accurately identifying the abnormal state at the real time and feeding back the control signal, the abnormal state detection and control of the full-automatic bud seedling machine are realized, and the working efficiency and the service life of the full-automatic bud seedling machine are improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a block diagram of a current monitoring and analyzing module of a full-automatic bud seedling machine according to an embodiment of the present invention;
fig. 2 is a flowchart of a control method of a full-automatic bud seedling machine according to an embodiment of the present invention.
The reference numerals in the figures are: 1. a current sensor; 2. detecting a wire; 3. a current amplifier; 4. a fixed base; 5. the current data analysis control chip; s1, a step S1; s2, a step S2; s3, a step S3; s4, step S4.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a detailed description of a full-automatic bud seedling machine and a control method thereof according to the invention, wherein the detailed description of the specific implementation, structure, characteristics and effects thereof is as follows, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The full-automatic bud seedling machine provided by the embodiment of the invention comprises a full-automatic bud seedling cutting machine body, wherein a current monitoring and analyzing module is further arranged on the full-automatic bud seedling cutting machine body and is used for monitoring and analyzing current operation data of the full-automatic bud seedling cutting machine body, so that the current monitoring and analyzing module comprises a current sensor, a detection wire, a current amplifier, a fixed base and a current data analysis control chip.
The following specifically describes a specific scheme of a full-automatic bud seedling machine and a control method thereof.
Referring to fig. 1, a block diagram of a current monitoring and analyzing module of a full-automatic bud seedling machine according to an embodiment of the present invention is shown, where the current monitoring and analyzing module includes: the device comprises a current sensor (1), a detection lead (2), a current amplifier (3), a fixed base (4) and a current data analysis control chip (5). The current sensor (1), the detection lead (2), the current amplifier (3) and the current data analysis control chip (5) are fixedly connected through the fixing base (4), and all components related to the current monitoring analysis module are integrated and mounted on the full-automatic bud seedling cutting machine body through the fixing base (4). The detection lead (2) is used for transmitting current information, the full-automatic bud seedling cutting machine body transmits the current information to the current amplifier (3) through the detection lead (2), and the current amplifier (3) amplifies the current information to ensure that the current sensor (1) obtains clear and effective current signals. The current sensor (1) converts the current information into a current signal and transmits the current signal to the current data analysis control chip (5), and the current data analysis control chip (5) receives the current signal and performs data processing. The current data analysis control chip (5) can realize the data processing step in a full-automatic bud seedling machine control method in the following embodiment.
Referring to fig. 2, a flowchart of a control method of a full-automatic bud seedling machine according to an embodiment of the invention is shown, where the method includes:
step S1: and acquiring signal characteristics of each data point on the current signal in multiple dimensions.
When the full-automatic bud seedling cutting machine body runs stably, the reflected current signals are sinusoidal alternating current as a whole, however, local fluctuation exists among each data point, and the sinusoidal alternating current is not smooth, so that signal characteristics under multiple dimensions, such as amplitude, frequency, period and the like, exist in the current signals. In order to realize accurate anomaly detection, the characteristics of the current signals need to be analyzed in multiple dimensions, and missing detection caused by analyzing the characteristics of signals in one dimension or a small number of dimensions is avoided. It is therefore first necessary to obtain signal characteristics in multiple dimensions for each data point on the current signal. It should be noted that, the original current signal is a series of time sequence signals, and the current data at the real-time moment and the previous historical moment in the preset time period form the current signal, that is, each data point in the current signal corresponds to one moment. In one embodiment of the present invention, the current collection frequency is set to 0.01 seconds, i.e., current data is collected every 0.01 seconds, and the preset time period is set to 10 minutes.
Preferably, in one embodiment of the present invention, the signal features include a current data statistics feature, a period feature and a waveform feature, and the specific acquisition method includes:
the cycle length of the current signal is obtained through a time-frequency conversion method, the current signal is segmented according to the cycle length, and a plurality of segmented signals are obtained. And taking the mean value, the variance, the maximum value and the minimum value of each segmented signal as the current data statistical characteristics of each data point in the corresponding segmented signal. In the embodiment of the invention, the Fourier transform is adopted to realize the time-frequency conversion.
The current signal is subjected to EMD decomposition to obtain IMF components and the frequency and phase at each IMF component, it being noted that since the frequency and phase of the IMF component of the current signal are obtained and the segmented signal is part of the current signal, the segmented signal also has frequency and phase information of the IMF component at the same time. And taking the maximum frequency in each segmented signal and the phase corresponding to the maximum frequency as the periodic characteristic of each data point in the segmented signal. It should be noted that the EMD decomposition is a technical means well known to those skilled in the art, and will not be described herein.
The peak value and the waveform shape on the segment signal are obtained, and the arithmetic coding value of the peak value and the waveform shape is taken as the waveform characteristic of each data point on the segment signal. After the peak value is obtained in one embodiment of the invention, the peak value can be converted into decimal system to obtain the arithmetic coding value; the description of the waveform shape, such as "smooth", "saw tooth", "step" and the like, is encoded and converted into decimal system to obtain its arithmetic coding value. Other arithmetic coding schemes may be selected in other embodiments of the invention. It should be noted that, the method for acquiring the peak value may be a multi-scale peak value query method, the method for acquiring the shape description of the signal waveform may be a method for judging which waveform shape belongs to by analyzing the distribution of the data point signal values, the specific method is a technical means well known to those skilled in the art, and in other embodiments of the present invention, other existing technologies may be used to extract the corresponding features, which is not limited and described herein.
Step S2: obtaining an isolated forest formed by all dimension signal characteristics of all data points; the isolated trees belonging to the same dimension in the isolated forest are dimension isolated trees; the number of the isolated trees in the isolated forest is a preset first number.
Because there are many data points on the current signal and each data point has a multidimensional signal characteristic, if the construction of an isolated forest cannot accurately represent the data characteristic of the current signal, the accuracy of the finally executed abnormality analysis result will also be affected. Therefore, the embodiment of the invention firstly builds the isolated forest with fewer isolated trees, analyzes the information of each dimension in the isolated forest, and then rebuilds the isolated forest to obtain the optimized isolated forest, so that the final abnormality analysis result is more accurate. Therefore, in the embodiment of the invention, the first quantity is set, and the isolated forest is constructed according to the first quantity and the multidimensional signal characteristics of each data, namely, fewer isolated trees are included in the isolated forest, and the fewer isolated trees are analyzed in the subsequent process, and the analysis result is used as a reference when reconstructing the isolated trees. It should be noted that, in the isolated forest, each isolated tree corresponds to a dimension, one dimension corresponds to a plurality of isolated trees, and the isolated tree belonging to the unified dimension is a dimension isolated tree.
In one embodiment of the present invention, the first number is set to 50, and each orphan tree is randomly selected and constructedThe nearest positive integer n is taken as the number of data points, wherein +.>Is the number of all data points. In other embodiments, other number rules may be set to construct an isolated forest.
Step S3: obtaining initial abnormality of each data point according to abnormality indexes of each data point in the isolated forest and distribution information of each data point in the isolated tree with different dimensions; screening out characteristic data points according to the initial abnormality; obtaining a dimension abnormality sequence of each dimension according to the fluctuation characteristics of the abnormality indexes of the feature data points in the isolated tree of each dimension; and obtaining the referential property of each dimension according to the data distribution in the dimension abnormality sequence.
In the isolated forest anomaly detection algorithm, the anomaly index of each node can be obtained by a method of calculating a node path, because the current signal of the full-automatic bud seedling machine has signal characteristics under multiple dimensions, and joint analysis is required under the multiple dimensions in order to accurately judge the anomaly of the data point. Because the isolated forest algorithm is constructed by randomly selecting data, when a certain data point is constructed, the signal characteristic of the data point in a certain dimension is not selected, and the reference degree of the data point is weaker, so that the abnormality index in an isolated tree is considered when the data point is analyzed, and the distribution information of the data point in the isolated tree in different dimensions is analyzed to obtain the initial abnormality of each data point. By combining the abnormality index and the distribution information, the obtained initial abnormality reference degree is stronger, and the subsequent algorithm processing is convenient.
Preferably, in one embodiment of the present invention, the method for acquiring initial abnormality includes:
obtaining the first number of the isolated trees in each dimension, and taking the ratio of the first number of the isolated trees in each dimension to the first number of the largest isolated tree as the credibility of the corresponding dimension. The higher the credibility is, the more signal characteristic data of the corresponding dimension is selected in the construction process of the isolated forest algorithm, and the higher the credibility of the abnormal index in the dimension isolated tree corresponding to the dimension is.
And obtaining an abnormality index of the signal characteristic of each data point in each dimension according to an isolated forest algorithm, wherein the reliability represents the reliability degree of the dimension in the current isolated forest, so that the reliability is combined with the obtained abnormality index, namely the abnormality index and the reliability of the corresponding dimension are subjected to weighted summation, and the weighted abnormality index is obtained. The weighted abnormality index can initially correct the abnormality index to include dimensional information features in the current isolated forest.
Further obtaining the second number of the isolated trees of the isolated tree where each data point is located, and taking the ratio of the second number of the isolated trees to the total number of the dimension isolated trees in the isolated forest as the number credibility of the corresponding data points. Because the data is randomly selected in the process of constructing the isolated forest, a preset first number of isolated trees is finally constructed, and therefore, signal characteristic data of certain dimensions of a certain data point are not selected, and the larger the second number of the isolated trees is, the more data of the data point is selected, the more accurate the information reflected by the data point is, namely, the more the number reliability is, the more accurate the abnormal information reflected by the corresponding data point is. The product of the number confidence and the weighted anomaly index is therefore taken as the initial anomaly.
In one embodiment of the invention, the initial anomaly is formulated as:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>Is->Initial abnormality of data points ++>For a second number of the orphan trees,for the total number of dimension isolation trees in an isolation forest, i.e.>The number of dimensions that can be regarded as signal features, +.>Is->Reliability of individual dimension, +.>Is->Data points at->Abnormality indexes in a dimension isolation tree of individual dimensions.
It should be noted that, the method for obtaining the anomaly index is the prior art in the isolated forest algorithm, that is, the shortest path between the corresponding node and the root node of the data point in the isolated tree is calculated, the maximum value of the shortest path of all the data points in all the isolated forests is further calculated, and the ratio of the shortest path to the maximum value is used as the anomaly index of the corresponding data. The specific implementation method is a technical means well known to those skilled in the art, and will not be described herein.
Because the data of the isolated forest algorithm when constructing the isolated tree is randomly selected, and the first quantity defined when constructing the isolated forest for the first time is smaller in the embodiment of the invention, data points in the isolated forest do not all have reference degrees, and data points with larger reference degrees need to be screened out to analyze the characteristics in each dimension of the current isolated forest. Because the initial abnormality includes both the abnormality information of the data point and the credibility information of the data point, the initial abnormality can be used as an index for screening the characteristic data point, and the characteristic data point can be screened according to the initial abnormality.
Preferably, in one embodiment of the present invention, data points with initial anomalies greater than a preset initial anomaly threshold are taken as characteristic data points. In one embodiment of the invention, the value range of the initial abnormality is limited between 0 and 1 after the initial abnormality is normalized, and the initial abnormality threshold is set to be 0.7.
If the feature data points have larger abnormal information in certain dimensions, the information is less in certain dimensions, and a plurality of feature data points exist in the dimension with larger abnormal information and have the same larger abnormal information, the fact that the dimension with larger abnormal information contains more abnormal point information compared with other dimensions is indicated, namely the dimension can better represent the abnormal information of the abnormal points, and the abnormal information is more obvious in the dimension, so that fluctuation features of the abnormal information of the feature data points on the isolated tree in each dimension can be analyzed, a dimension abnormality sequence corresponding to each dimension is obtained, and each element in the dimension abnormality sequence corresponds to one isolated tree in the dimension isolated tree. The data distribution in the dimension abnormality sequence can show the fluctuation characteristics of the initial abnormality of the characteristic data points in the current dimension isolation tree, and if the data distribution of the elements in the sequence has clear category boundaries, the data distribution of the elements in the sequence shows that the data distribution has obvious abnormality information distribution information in the current dimension, namely the bigger the reference corresponding to the dimension.
Preferably, in one embodiment of the present invention, the method for acquiring the dimension abnormality sequence includes:
under one dimension, obtaining an abnormal index mean value of all characteristic data points on each isolated tree, wherein the abnormal index mean value of all the isolated trees forms an abnormal index sequence, and obtaining a variation coefficient of the abnormal index sequence. And calculating the product of the variation coefficient and each element in the abnormal index sequence under the corresponding dimension to obtain a dimension abnormal sequence, namely weighting the data in the abnormal index sequence through the variation coefficient, and amplifying the information in the abnormal index sequence to obtain the dimension abnormal sequence.
Preferably, in one embodiment of the present invention, the method for obtaining the reference includes:
processing the dimension abnormality sequence by using an Ojin threshold algorithm to obtain two sections of subsequences; obtaining an inter-class variance between two sub-sequences; and obtaining the element number ratio between the two sub-sequences, wherein the element number ratio is less than or equal to 1. It should be noted that, the oxford threshold algorithm is a technical means well known to those skilled in the art, and will not be described herein.
Obtaining the reference using a reference formula, the reference formula comprising:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>For reference, ->Is inter-class variance>Is the element number ratio. That is, the larger the inter-class variance, the more obvious the distinction between two classes in the sequence; the closer the element number ratio is to 1, the more uniform the number between the two categories is, the +.>The smaller the description sequence, the less obvious the distinction between the two categories; that is, the larger the inter-class variance, the less uniform the number between the two classes, the more obvious the distinction between the two classes in the specification sequence, and the greater the references to the corresponding dimensions.
Step S4: reconstructing an isolated forest based on the referential property to obtain an optimized isolated forest; the number of the isolated trees in the optimized isolated forest is larger than a preset first number; taking the initial abnormality in the optimized isolated forest as the optimized abnormality of each data point; judging whether abnormality occurs at real time according to the optimized abnormality, and if so, feeding back a control signal to the body of the full-automatic bud seedling cutting machine.
The referential property of each dimension is obtained in the step S3, and because the referential property is obtained according to the change and the distribution of the abnormal information and is irrelevant to the quantity and the size of the abnormal information, the obtained referential property can still represent the signal characteristic information of the current signal of the current full-automatic bud seedling machine in each dimension even though the isolated forest in the step S2 is constructed by adopting a small first quantity. Therefore, the isolated forest can be reconstructed based on the referential property of each dimension to obtain an optimized isolated forest, wherein the number of the isolated trees in the optimized isolated forest is larger than the first number, namely, the referential property of each dimension is introduced in the process of constructing the isolated forest, and the number of the isolated trees is increased, so that the abnormality shown in the optimized isolated forest is more accurate and effective, the initial abnormality shown by each data point in the optimized isolated forest is used as the optimized abnormality of each data point, whether the abnormality occurs at real time or not can be accurately judged according to the optimized abnormality, and a control signal is fed back to the body of the full-automatic bud seedling cutter according to the judgment result.
Preferably, in one embodiment of the present invention, the method for optimizing the acquisition of an isolated forest includes:
the referential of each dimension is used as a random probability value in the process of constructing and optimizing the isolated forest, namely, the larger the referential is, the larger the probability is when randomly selected, and the effect of showing more accurate abnormal information by using fewer isolated trees can be achieved. And obtaining the optimized isolated forest by utilizing an isolated forest algorithm.
In one embodiment of the invention, the number of orphan trees in the optimized orphan forest is set to 500.
Preferably, the method for judging whether the abnormality occurs at the real-time moment according to the optimized abnormality comprises the following steps:
taking the data points with the optimized abnormality greater than a preset optimized abnormality threshold as abnormal data points; if the data points at the real-time moment are abnormal data points and the data points with continuous preset number are the abnormal data points, judging that the real-time moment is abnormal. In one embodiment of the present invention, after normalizing the optimized anomalies, the optimized anomalies threshold is set to 0.7 and the preset number is set to 20.
In one embodiment of the invention, the current data analysis control chip (5) adopts a field programmable gate array (Field Programmable Gate Array, FPGA) chip, and the FPGA chip is used as a semi-custom circuit in the field of application-specific integrated circuits, so that the defect of custom circuits can be overcome, and the defect of limited gate circuits of the original programmable devices can be overcome. The programmable function enables the practitioner to set the parameters of the on-chip storage algorithm, such as the first number, control signals, etc., in the embodiments of the present invention.
Preferably, in one embodiment of the present invention, the full-automatic bud seedling cutter body comprises a pneumatic fitting, and the current detection and analysis module is mounted on the pneumatic fitting. The dual-tooth sugarcane seedling cutting machine as set forth in patent CN210551503U is a full-automatic seedling machine, and the main body comprises a waste frame, a stand, a finished product frame, a control cabinet, an electric cutting machine, a cutting sensor, a counter and pneumatic accessories. Because the pneumatic accessory mainly realizes the functions of operation, stop and speed regulation of mechanical equipment, the current detection and analysis module is arranged on the pneumatic accessory, so that the distortion and noise in the transmission process of control signals can be reduced, and the full-automatic bud seedling cutting machine body can be accurate and efficient when realizing the control command of the control signals. Therefore, under the embodiment of the invention, the control signal fed back by the current data analysis control chip (5) comprises:
if the real-time abnormality occurs, the speed of the current pneumatic accessory is considered as an initial speed, the adjusting speed of the initial speed is obtained according to the preset speed gain, the speed of the current pneumatic accessory is reduced until the real-time abnormality is judged, then the speed of the pneumatic accessory is increased to the adjusting speed, and if the abnormality can still be detected, the bud seedling cutting machine is stopped. Through the verification type control speed regulation, whether the full-automatic bud seedling machine has abnormal operation or not is further determined, and false detection caused by data errors and the like is prevented.
It should be noted that, the control signal fed back by the current data analysis control chip (5) can be transmitted wirelessly, and also can be transmitted to the full-automatic bud seedling cutting machine body through the detection wire (2).
In summary, the embodiment of the invention provides a full-automatic bud seedling machine and a control method thereof, wherein the full-automatic bud seedling machine comprises a full-automatic bud seedling cutting machine body and a current monitoring and analyzing module, and the current monitoring and analyzing module comprises a current sensor, a detection wire, a current amplifier, a fixed base and a current data analysis and control chip. The current data analysis control chip accurately identifies the abnormal state of the current signal at real time through the construction of the twice isolated forest, and further timely and effectively feeds back the control signal. According to the embodiment of the invention, the working efficiency and the service life of the full-automatic bud seedling machine are improved by improving the accuracy of detecting the abnormal running state of the full-automatic bud seedling machine.
It should be noted that: the sequence of the embodiments of the present invention is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments.
Claims (10)
1. The control method of the full-automatic bud seedling machine is characterized by comprising the following steps:
acquiring current information of a full-automatic bud seedling machine cutting machine body; acquiring signal characteristics of each data point on the current signal in multiple dimensions;
obtaining an isolated forest formed by all dimension signal characteristics of all data points; the isolated trees belonging to the same dimension in the isolated forest are dimension isolated trees; the number of the isolated trees in the isolated forest is a preset first number;
obtaining initial abnormality of each data point according to abnormality indexes of each data point in the isolated forest and distribution information of each data point in isolated trees with different dimensions; screening out characteristic data points according to the initial abnormality; obtaining a dimension abnormality sequence of each dimension according to the fluctuation characteristics of the abnormality indexes of the characteristic data points on the isolated tree in each dimension isolated tree; obtaining the referential property of each dimension according to the data distribution in the dimension abnormality sequence;
reconstructing an isolated forest based on the references to obtain an optimized isolated forest; the number of the isolated trees in the optimized isolated forest is larger than the preset first number; taking the initial abnormality in the optimized isolated forest as the optimized abnormality of each data point; judging whether abnormality occurs at real time according to the optimized abnormality, and if so, feeding back a control signal to the full-automatic bud seedling cutting machine body.
2. The method according to claim 1, wherein the signal characteristics include current data statistics, period characteristics and waveform characteristics;
obtaining the period length of the current signal through a time-frequency conversion method, segmenting the current signal according to the period length, and obtaining a plurality of segmented signals;
taking the mean value, variance, maximum value and minimum value of each segmented signal as the current data statistical characteristics of each data point in the segmented signal;
performing EMD (empirical mode decomposition) on the current signals to obtain IMF components and frequencies and phases on each IMF component, and taking the maximum frequency and the phase corresponding to the maximum frequency in each segmented signal as the periodic characteristic of each data point in the segmented signal;
obtaining peak values and waveform shapes on the segmented signal, and taking arithmetic coding values of the peak values and the waveform shapes as the waveform characteristics of each data point on the segmented signal.
3. The control method of a fully automatic bud seeding machine according to claim 1, wherein the initial abnormality acquisition method comprises:
obtaining the first quantity of the isolated trees in each dimension isolated tree, and taking the ratio of the first quantity of the isolated trees in each dimension isolated tree to the maximum first quantity of the isolated trees as the credibility of the corresponding dimension;
obtaining the abnormality index of the signal characteristic of each data point in each dimension according to an isolated forest algorithm, and carrying out weighted summation on the abnormality index and the credibility of the corresponding dimension to obtain a weighted abnormality index;
obtaining the second number of the isolated trees of the isolated tree where each data point is located, and taking the ratio of the second number of the isolated trees to the total number of the dimension isolated trees in the isolated forest as the number credibility of the corresponding data points;
taking the product of the number confidence and the weighted anomaly index as the initial anomaly.
4. The method for controlling a fully automatic bud seeding machine according to claim 1, wherein the method for acquiring the characteristic data points comprises:
and taking the data points with the initial abnormality greater than a preset initial abnormality threshold as the characteristic data points.
5. The method for controlling a fully automatic bud seeding machine according to claim 1, wherein the method for acquiring the dimension abnormality sequence comprises:
under one dimension, obtaining an abnormal index mean value of all characteristic data points on each isolated tree, wherein the abnormal index mean value of all the isolated trees forms an abnormal index sequence; obtaining a coefficient of variation of the abnormality index sequence; and calculating the product of the variation coefficient and each element in the abnormality index sequence under the corresponding dimension to obtain a dimension abnormality sequence.
6. The full-automatic bud seedling machine and the control method thereof according to claim 1, wherein the reference acquisition method comprises:
processing the dimension abnormality sequence by using an Ojin threshold algorithm to obtain two sections of subsequences; obtaining an inter-class variance between two sections of the subsequences; obtaining element quantity ratio between two sub-sequences, wherein the element quantity ratio is less than or equal to 1;
obtaining the referenceusing a referenceformula comprising:
the method comprises the steps of carrying out a first treatment on the surface of the Wherein (1)>For the reference, +_>For the inter-class variance +_>Is the element number ratio.
7. The full-automatic bud seedling machine and the control method thereof according to claim 1, wherein the method for acquiring the optimized isolated forest comprises the following steps:
and taking the referential of each dimension as a random probability value in the process of constructing the optimized isolated forest, and obtaining the optimized isolated forest by utilizing an isolated forest algorithm.
8. The full-automatic bud seedling machine and the control method thereof according to claim 1, wherein the method for judging whether abnormality occurs at real time according to the optimized abnormality comprises:
taking the data points with the optimized abnormality greater than a preset optimized abnormality threshold as abnormal data points; if the data points at the real-time moment are abnormal data points and the data points with continuous preset number are the abnormal data points, judging that the real-time moment is abnormal.
9. The full-automatic bud seedling machine and the control method thereof according to claim 1, wherein the full-automatic bud seedling cutting machine body comprises a pneumatic fitting;
if the real-time moment is abnormal, the current speed of the pneumatic accessory is regarded as an initial speed, the adjustment speed of the initial speed is obtained according to the preset speed gain, the current speed of the pneumatic accessory is reduced until the real-time moment is judged to be abnormal, then the speed of the pneumatic accessory is increased to the adjustment speed, and if the abnormality can still be detected, the bud seedling cutting machine is stopped.
10. The full-automatic bud seedling machine comprises a full-automatic bud seedling cutting machine body, and is characterized by further comprising a current monitoring and analyzing module, wherein the current monitoring and analyzing module is installed on the full-automatic bud seedling cutting machine body and comprises a current sensor (1), a detection lead (2), a current amplifier (3), a fixed base (4) and a current data analysis control chip (5); the current sensor (1), the detection lead (2), the current amplifier (3) and the current data analysis control chip (5) are fixedly connected through the fixed base (4); the current information of the full-automatic bud seedling cutting machine body is transmitted to the current amplifier (3) through the detection lead (2), and the current sensor (1) obtains a current signal after the current amplifier (3) is processed; the current data analysis control chip (5) receives the current signals and processes data; the current data analysis control chip realizes the data processing step in the full-automatic bud seedling machine control method according to any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311695198.5A CN117407822B (en) | 2023-12-12 | 2023-12-12 | Full-automatic bud seedling machine and control method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311695198.5A CN117407822B (en) | 2023-12-12 | 2023-12-12 | Full-automatic bud seedling machine and control method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117407822A true CN117407822A (en) | 2024-01-16 |
| CN117407822B CN117407822B (en) | 2024-02-20 |
Family
ID=89491126
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311695198.5A Active CN117407822B (en) | 2023-12-12 | 2023-12-12 | Full-automatic bud seedling machine and control method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117407822B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113284004A (en) * | 2021-05-10 | 2021-08-20 | 广州汇通国信科技有限公司 | Power data diagnosis treatment method based on isolated forest algorithm |
| CN113420073A (en) * | 2021-08-23 | 2021-09-21 | 平安科技（深圳）有限公司 | Abnormal sample detection method based on improved isolated forest and related equipment |
| CN113420652A (en) * | 2021-06-22 | 2021-09-21 | 中冶赛迪重庆信息技术有限公司 | Method, system, medium and terminal for recognizing abnormity of time sequence signal fragment |
| CN116502043A (en) * | 2023-04-27 | 2023-07-28 | 北京中鑫联物联科技有限公司 | Finish rolling motor state analysis method based on isolated forest algorithm |
| WO2023169098A1 (en) * | 2022-03-10 | 2023-09-14 | 东南大学 | Isolation forest-based method for diagnosing open-circuit fault of modular multilevel converter |
-
2023
- 2023-12-12 CN CN202311695198.5A patent/CN117407822B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113284004A (en) * | 2021-05-10 | 2021-08-20 | 广州汇通国信科技有限公司 | Power data diagnosis treatment method based on isolated forest algorithm |
| CN113420652A (en) * | 2021-06-22 | 2021-09-21 | 中冶赛迪重庆信息技术有限公司 | Method, system, medium and terminal for recognizing abnormity of time sequence signal fragment |
| CN113420073A (en) * | 2021-08-23 | 2021-09-21 | 平安科技（深圳）有限公司 | Abnormal sample detection method based on improved isolated forest and related equipment |
| WO2023169098A1 (en) * | 2022-03-10 | 2023-09-14 | 东南大学 | Isolation forest-based method for diagnosing open-circuit fault of modular multilevel converter |
| CN116502043A (en) * | 2023-04-27 | 2023-07-28 | 北京中鑫联物联科技有限公司 | Finish rolling motor state analysis method based on isolated forest algorithm |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117407822B (en) | 2024-02-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20060210141A1 (en) | Inspection method and inspection apparatus | |
| CN108038553B (en) | Rolling mill equipment state on-line monitoring and diagnosing system and monitoring and diagnosing method | |
| CA2634328C (en) | Method and system for trend detection and analysis | |
| CN111259730B (en) | State monitoring method and system based on multivariate state estimation | |
| CN105184319A (en) | Fiber perimeter intrusion signal identification method and apparatus, and perimeter intrusion alarm system | |
| US10311703B1 (en) | Detection of spikes and faults in vibration trend data | |
| CN105651376A (en) | Mechanical equipment offline detection system vibration signal spectrum analysis and alarm method | |
| CN117407822B (en) | Full-automatic bud seedling machine and control method thereof | |
| CN116070163A (en) | Indoor harmful gas concentration anomaly monitoring data processing method | |
| CN109612757B (en) | Method for diagnosing equipment based on sound characteristic and temperature characteristic | |
| CN117708748B (en) | Operation monitoring system and method for centrifugal fan | |
| CN117235652B (en) | Steel wire processing environment supervision method and system based on big data | |
| CN115114565B (en) | Linear motor fault detection method and system based on big data analysis | |
| CN116502141A (en) | Data-driven-based electric screw press fault prediction method and system | |
| Zhang et al. | Gearbox fault diagnosis based on multifractal detrended fluctuation analysis and improved k means clustering | |
| CN115041544A (en) | Method and device for detecting abnormality of stamping part and storage medium | |
| CN115326393A (en) | Wind turbine generator bearing pair fault diagnosis method based on temperature information | |
| Wang et al. | Research on a method for diagnosing clogging faults and longitudinal axial flow in the threshing cylinders of drum harvesters | |
| CN111980900B (en) | Water pump fault diagnosis method based on multi-source data fusion analysis | |
| CN115854490B (en) | Analysis method and system for dehumidifier state monitoring | |
| CN112710488B (en) | Stress wave signal acquisition and analysis method for mechanical power equipment | |
| CN117574307B (en) | Security management system and method for data of Internet of things | |
| CN117273547B (en) | Production equipment operation data processing method based on edge calculation | |
| US20240094094A1 (en) | Pump health analysis method and pump health analysis device using the same | |
| CN117649249B (en) | Food production traceability management method and system based on block chain |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |