Intelligent automatic coal feeding system based on multi-plc cooperative control and control method

Abstract

This invention relates to the field of automatic coal feeding control technology, and discloses an intelligent automatic coal feeding system and control method based on multi-PLC collaborative control. The method first collects state parameter signals from the collaborative operation of multiple PLCs to form a state sequence, divides the collaborative stages, and clusters them to obtain control sub-sequences. Then, based on the state change trend fluctuations and state differences of the control sub-sequences, a collaborative fluctuation index is obtained. Combined with the differences in command change trends and time sequence change trends, command descent differences and interval differences are obtained. Next, based on the above differences, the average of the collaborative fluctuation index, and the average of the sub-sequence similarity, a control disorder coefficient is obtained. Simultaneously, a feature matrix is ​​constructed and decomposed to obtain the proportion of key components and the difference index. Finally, combining the correlation of the feature matrix and the control disorder coefficient, a system stability index is obtained, which is used to control the automatic coal feeding system. This method can accurately grasp the system's operating state, improve the effect of multi-PLC collaborative control, and ensure the stable operation of the automatic coal feeding system.

Description

Technical Field

[0001] This invention relates to the field of automatic coal feeding control technology, specifically to an intelligent automatic coal feeding system and control method based on multi-PLC collaborative control. Background Technology

[0002] In industrial production, automated coal feeding systems are a crucial component of energy supply, and their operational efficiency and stability directly impact the smoothness of the overall production process. With the continuous improvement of industrial automation, single-PLC controlled automated coal feeding systems are increasingly unable to meet the demands of large-scale, highly complex production. To address this challenge, a multi-PLC collaborative control model has been introduced into automated coal feeding systems. Through the coordinated work of multiple PLCs, joint control of multiple devices such as coal feeders, conveyor belts, and coal storage bins can be achieved. Intelligent automatic coal feeding systems controlled by multiple PLCs face numerous problems in actual operation. When multiple PLCs work collaboratively, their operating parameters often exhibit complex trends due to the different control tasks they undertake. Delays or discrepancies in instruction transmission between different PLCs can easily lead to asynchronous actions of various devices, resulting in fluctuations in coal feeding and equipment operation conflicts. For example, if one PLC issues a stop coal feeding command but other PLCs responsible for transmission fail to respond promptly, coal accumulation may occur, affecting the normal operation of subsequent processes.

[0003] In the collaborative operation of multiple PLCs, the acquisition and processing of status parameter signals lack an effective analysis mechanism. Traditional control methods often focus only on the operating status of a single device, ignoring the collaborative relationship between multiple PLCs, making it difficult to accurately determine the overall operational stability of the system. When local state changes occur in the system, it is impossible to quickly identify their impact on the overall collaborative control, leading to a lag in the adjustment of control strategies and consequently affecting the system's operating efficiency. Existing technologies lack quantitative evaluation indicators for the degree of chaos in multi-PLC collaborative control, making it difficult to make targeted adjustments based on the actual operating status of the system. When faced with complex changes in operating conditions, such as fluctuations in coal quality and changes in equipment load, the issuance of commands from multiple PLCs is prone to confusion, resulting in poor system flexibility in adapting to external changes. Moreover, traditional control methods do not conduct in-depth analysis of eigenvalues ​​when constructing the characteristic matrix of the state sequence, failing to accurately extract key factors affecting system stability, leading to a lack of scientific basis for the formulation of control strategies. With the continuous expansion of production scale, the structure of automatic coal feeding systems is becoming increasingly complex, placing higher demands on the accuracy and response speed of multi-PLC collaborative control. Existing control methods, due to the aforementioned shortcomings, frequently experience problems such as insufficient control accuracy and poor stability during the operation of intelligent automatic coal feeding systems with multi-PLC collaborative control. This not only increases the equipment failure rate but may also lead to energy waste and increased production costs. Therefore, developing an intelligent automatic coal feeding control method that can effectively improve the performance of multi-PLC collaborative control has become an important issue that needs to be addressed in the current development of automatic coal feeding systems. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent automatic coal feeding system and control method based on multi-PLC collaborative control, so as to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an intelligent automatic coal feeding control method based on multi-PLC collaborative control, the method comprising: The state sequence is composed of state parameter signals collected during the coordinated operation of multiple PLCs in the coal feeding system. The collaborative phases are divided based on the distribution of nodes in the state sequence, and the collaborative phases in the state sequence are clustered according to the distribution of collaborative phases to obtain each control subsequence; Based on the fluctuation of local state change trends in the control subsequence and the differences between local states, obtain the cooperative fluctuation index of each control subsequence; Based on the differences in the changing trends of instructions in the control subsequence and the differences in the timing of instructions in the control subsequence, the differences in instruction descent and instruction interval are obtained. The control disorder coefficient of the state sequence is obtained based on the difference in instruction descent, the difference in instruction interval, the average of the cooperative fluctuation index of all control subsequences, and the average of the similarity between control subsequences. Construct a feature matrix of the state sequence, perform feature decomposition on the feature matrix to obtain each feature value, and obtain the key component proportion and difference index of the feature value sequence of the state sequence based on the distribution, average and difference of the feature values. The system stability index of the state sequence is obtained based on the correlation between data in each row of the feature matrix, the control disorder coefficient, the proportion of key components, and the difference index; the intelligent automatic coal feeding system with multi-PLC collaborative control is controlled based on the system stability index.

[0006] Preferably, the method for obtaining the control subsequence is as follows: The state node detection algorithm is used to obtain all instructions and nodes in the state sequence. The state sequence is divided into stage sequences from each node. The feature values ​​of the elements in each stage sequence are calculated. The feature values ​​of all stage sequences are used as input to the clustering algorithm, and the output is each cluster. Calculate the mean of internal elements in each cluster, select the cluster with the largest mean of internal elements as the control cluster, and select the state subsequence corresponding to the internal elements of the control cluster as the control subsequence.

[0007] Preferably, the method for obtaining the coordinated volatility index is as follows: Based on the fluctuations in the local state change trends and the differences between local states in the control subsequence, the bidirectional response coefficient and bidirectional adjustment coefficient of each control subsequence are obtained. The coordinated fluctuation index of each control subsequence is obtained by combining the bidirectional response coefficient and bidirectional adjustment coefficient of each control subsequence.

[0008] Preferably, the method for obtaining the bidirectional adjustment coefficient is as follows: For each control subsequence, obtain the maximum value in each control subsequence, take the subsequence consisting of the maximum value and all elements preceding the maximum value as the predecessor subsequence of each control subsequence, and take the subsequence consisting of all elements following the maximum value as the successor subsequence of each control subsequence. Obtain the first-order difference sequence of the predecessor subsequence of each control subsequence, and use the response function to process all elements in the first-order difference sequence to obtain the difference response sequence; The absolute value of the difference between the sum of elements in the differential response sequence and the length of the first-order difference sequence of the predecessor subsequence is calculated and used as the fluctuation index of the predecessor subsequence of each control subsequence. Using the same method as for the predecessor subsequence, the volatility index of the successor subsequence of each control subsequence is obtained; the product of the absolute value of the difference between the volatility indices of the predecessor subsequence and the successor subsequence and the mean value is calculated as the bidirectional response coefficient of each control subsequence. Calculate the variance of the elements in the predecessor subsequence and the variance of the elements in the successor subsequence for each control subsequence, and use the mean of the variances of the predecessor and successor subsequences as the bidirectional adjustment coefficient for each control subsequence.

[0009] Preferably, the method for obtaining the instruction interval difference is as follows: Sort all instructions in all control subsequences according to their timing position in the coal supply system to construct an instruction sequence; sort the timing of all instructions in the control subsequences according to their size to construct an instruction timing sequence. Obtain the first-order difference sequences of the instruction sequence and the instruction timing sequence. Use the variance of all elements in the first-order difference sequence of the instruction sequence as the instruction descent difference; use the variance of all elements in the first-order difference sequence of the instruction timing sequence as the instruction interval difference.

[0010] Preferably, the method for obtaining the control disorder coefficient is as follows: The mean of the co-variance indices of all control subsequences is calculated as the average volatility index. The mean of the similarity coefficients between each control subsequence and all other control subsequences is taken as the sequence similarity of each control subsequence; The control disorder coefficient is calculated by combining the mean of sequence similarity of all control subsequences with the instruction descent difference, instruction interval difference, and average volatility index.

[0011] Preferably, the method for obtaining the proportion of key components and the difference index is as follows: Sort all feature values ​​in descending order to construct a feature value sequence; The feature value sequence is used as the input of the threshold segmentation method, and the output of the threshold segmentation method is the threshold. All feature values ​​greater than or equal to the threshold are used as key feature values, and all feature values ​​less than the threshold are used as auxiliary feature values. The ratio of the number of key feature values ​​to the total number of all feature values ​​is used as the proportion of key components in the feature value sequence. Calculate the mean and variance of all key eigenvalues ​​in the eigenvalue sequence, and use the product of the mean and variance as the volatility index of the key eigenvalues; use the same method as the volatility index of the key eigenvalues ​​to calculate the volatility index of the auxiliary eigenvalues, and use the absolute value of the difference between the volatility index of the key eigenvalues ​​and the volatility index of the auxiliary eigenvalues ​​as the difference index of the eigenvalue sequence.

[0012] Preferably, the method for obtaining the system stability index is as follows: Calculate the absolute value of the similarity coefficient between each row of data in the feature matrix and all other rows of data, and take the mean of the absolute values ​​of all the similarity coefficients of each row of data as the co-coefficient of each row of data; take the mean of the co-coefficients of all rows of data in the feature matrix as the co-coefficient of the feature matrix itself. The system stability index is calculated by combining its own synergy coefficient with the proportion of key components, the difference index, and the control disorder coefficient.

[0013] Preferably, the control of the intelligent automatic coal feeding system based on the system stability index and the collaborative control of multiple PLCs includes: The default threshold for control parameters is obtained from the system stability index based on the state sequence; The parameters for multi-PLC collaborative operation are optimized based on the default thresholds of the control parameters, and the coal feeding control parameters in the coal feeding system are calculated using the collaborative control method.

[0014] Preferably, the present invention also includes an intelligent automatic coal feeding system based on multi-PLC collaborative control, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the intelligent automatic coal feeding control method based on multi-PLC collaborative control described above.

[0015] Compared with the prior art, the beneficial effects of the present invention are: By collecting state parameter signals from multiple PLCs operating collaboratively in the coal supply system to form a state sequence, and then performing multi-dimensional analysis and processing, we can deeply uncover the potential patterns and characteristics of the system's operation. Dividing the system into collaborative stages and clustering them to obtain control sub-sequences helps to refine the study of different operating stages, making subsequent analysis more targeted. Based on the fluctuation trends of the control sub-sequences and the differences between states, a collaborative fluctuation index is obtained, which accurately reflects the collaborative operating state within each sub-sequence, allowing operators to clearly understand the collaborative performance of each local component. By considering the differences in instruction change trends and temporal change trends among control subsequences, and obtaining differences in instruction descent and instruction intervals, we can comprehensively capture the changing characteristics of instructions during transmission and execution, providing an effective way to judge the consistency of instruction execution. Combining instruction correlation differences, the average of the cooperative fluctuation index, and the average of the similarity among control subsequences, the resulting control disorder coefficient can intuitively reflect the degree of control disorder in the entire state sequence, providing a reference for the formulation of control strategies.

[0016] By constructing a feature matrix of the state sequence and performing eigenvalue decomposition, the proportion of key components and the difference index are obtained based on various cases of eigenvalues, enabling analysis of the system's intrinsic characteristics from the perspective of data features. The system stability index is obtained by combining the correlation of data in each row of the feature matrix, the control disorder coefficient, the proportion of key components, and the difference index. This index is then used to control the system, enabling precise regulation of an intelligent automatic coal feeding system with multi-PLC collaborative control. This ensures the system maintains good collaborative performance in complex operating environments, reduces operational failures caused by poor coordination, and improves the overall operational efficiency of the system. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the working principle of the intelligent automatic coal feeding control method based on multi-PLC collaborative control described in this invention. Figure 2 A flowchart for controlling the subsequence acquisition method; Figure 3A flowchart for the method of obtaining instruction interval differences; Figure 4 A flowchart illustrating the method for obtaining the proportion of key components and the difference index; Figure 5 The flowchart shows the method for obtaining the system stability index. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figures 1-5 This invention provides an intelligent automatic coal feeding system and control method based on multi-PLC collaborative control. The specific implementation steps are as follows: This embodiment discloses an intelligent automatic coal feeding control method based on multi-PLC collaborative control. The method will be described in detail below with reference to the specific process.

[0020] The status sequence is composed of status parameter signals collected during the coordinated operation of multiple PLCs in the coal feeding system. The status parameter signals include, but are not limited to, the operating frequency of each PLC, communication delay, response time of executed instructions, operating speed of the coal feeding equipment, and coal quantity detection value. These parameter signals are arranged in chronological order to form the status sequence.

[0021] The distribution of nodes in the state sequence is used to divide the state sequence into cooperative stages. Based on the distribution of cooperative stages, the cooperative stages in the state sequence are clustered to obtain each control subsequence. Nodes are the time points in the state sequence where the parameter signals change significantly. By identifying these nodes, the state sequence is divided into different cooperative stages. Then, these stages are clustered to obtain control subsequences with similar characteristics.

[0022] Based on the fluctuations in local state change trends and the differences between local states within the control subsequences, the cooperative fluctuation index of each control subsequence is obtained. By analyzing the parameter fluctuation amplitudes in local regions within the control subsequences and the parameter differences between different local regions, an index reflecting the cooperative stability of the subsequences is calculated.

[0023] Based on the differences in the changing trends of instructions within the control subsequence and the differences in the timing trends of instructions within the control subsequence, we obtain the differences in instruction descent and instruction intervals. The changing trends of instructions include the magnitude of increase or decrease in instruction values, and the changing trends of timing include the changes in the time interval between instruction issuances. By quantifying these differences, we obtain the corresponding indicators.

[0024] The control disorder coefficient of the state sequence is obtained based on the differences in command descent, command interval, the average of the cooperative fluctuation index of all control subsequences, and the average of the similarity between control subsequences. This coefficient comprehensively reflects the impact of command changes, cooperative fluctuations, and subsequence similarity on the system's control order.

[0025] A feature matrix of the state sequence is constructed, and eigenvalues ​​are obtained by eigenvalue decomposition. Based on the distribution, average, and differences of the eigenvalues, the proportion of key components and the difference index of the eigenvalue sequence are obtained. The feature matrix consists of the eigenvectors of each parameter in the state sequence. After eigenvalue decomposition, the proportion of key components and the difference index are obtained by analyzing the characteristics of each eigenvalue.

[0026] The system stability index is obtained based on the correlation between data in each row of the feature matrix, the control disorder coefficient, the proportion of key components, and the difference index. The system stability index is then used to control an intelligent automatic coal feeding system with multi-PLC collaborative control. By comprehensively analyzing the correlation of the feature matrix and the above-mentioned indices, a system stability index reflecting the overall stability of the system is obtained, and the system is dynamically adjusted based on this index.

[0027] Example 1: A state node detection algorithm is used to process the state sequence to obtain all instructions and nodes. The state sequence consists of various state parameter signals from multiple PLCs operating collaboratively in the coal feeding system, arranged chronologically. These parameter signals cover the operating status parameters of each PLC and the working parameters of the coal feeding equipment. During the operation of the state node detection algorithm, the state sequence is analyzed segment by segment, and nodes are identified by monitoring changes in parameter signals. Specifically, the algorithm sets a continuous analysis window, the length of which is adjusted according to the time span of the state sequence and the frequency of parameter changes. As the window moves, the algorithm calculates the amplitude of parameter signal changes within the current window. When the amplitude of parameter signal changes exceeds the normal fluctuation range at a certain moment, that moment is marked as a node. Simultaneously, the algorithm extracts all instructions contained in the state sequence. These instructions include control instructions issued by each PLC and equipment operation instructions. Instruction extraction is based on preset instruction recognition rules, which are completed by matching the unique signal characteristics of the instructions.

[0028] The state sequence is divided into stage sequences at each node. Each node corresponds to a point in time in the state sequence, and the portion between two adjacent nodes forms a stage sequence. Each stage sequence contains all state parameter signals and related instructions within that time period. During the segmentation process, it is necessary to ensure the integrity of each stage sequence, that is, all data between the start time of one node and the start time of the next node are included in the corresponding stage sequence, without any data omissions or duplications.

[0029] The eigenvalues ​​of elements in each stage sequence are calculated. Each stage sequence contains elements representing parameter signal values ​​across multiple dimensions. Through multi-dimensional analysis of these elements, numerical values ​​reflecting the overall characteristics of the stage sequence are extracted as eigenvalues. The analysis comprehensively considers factors such as the changing trends of the parameter signals, the range of numerical distribution, and the degree of correlation between parameters, transforming these factors into quantifiable eigenvalues. For each stage sequence, a corresponding eigenvalue is obtained, which can represent the characteristics of that stage sequence to a certain extent.

[0030] The feature values ​​of all stage sequences are used as input to a clustering algorithm, which outputs clusters. The algorithm groups the input feature values, grouping stage sequences with similar feature values ​​into the same cluster. During clustering, the algorithm calculates the similarity between different feature values, based on the numerical closeness of the feature values ​​and the similarity in parameter variation patterns of the stage sequences they represent. After multiple rounds of iterative calculations, the clustering results reach a stable state, meaning there are no longer significant changes in intra-cluster elements. At this point, each output cluster contains a group of stage sequences with similar feature values.

[0031] Calculate the mean of the internal elements of each cluster. For each cluster, summarize the feature values ​​of all its stage sequences, and then calculate the arithmetic mean of these feature values ​​to obtain the mean of the internal elements of the cluster. During the calculation, it is necessary to ensure that all feature values ​​are included in the calculation and that no element is ignored, so that the mean can accurately reflect the overall level of feature values ​​within the cluster.

[0032] The cluster with the largest internal element mean is selected as the control cluster. By comparing the internal element means of each cluster, the cluster with the largest mean is chosen as the control cluster. This process relies solely on the magnitude of the internal element mean and does not introduce any other additional evaluation metrics.

[0033] The state subsequences corresponding to elements within a control cluster are used as control subsequences. Each element in a control cluster corresponds to a stage sequence in the state sequence. The position of these stage sequences within the state sequence and the parameter information they contain together constitute the corresponding state subsequence. These state subsequences are extracted and used as control subsequences for subsequent analysis and calculation. During the extraction process, the original time order and data integrity of the state subsequences must be maintained to ensure that they accurately reflect the state characteristics of the corresponding stage. These control subsequences will be used to further analyze various indicators in the multi-PLC collaborative control process, providing basic data for subsequent system stability index calculation and control strategy formulation.

[0034] By collecting state parameter signals from multiple PLCs operating collaboratively in a coal handling system to form a state sequence, and then performing multi-dimensional analysis and processing, we can deeply uncover the potential patterns and characteristics of the system's operation. Dividing the system into collaborative stages and clustering them to obtain control sub-sequences helps to refine the study of different operating stages, making subsequent analysis more targeted.

[0035] Example 2: Based on the fluctuations in the local state change trends and the differences between local states in the control subsequence, the bidirectional response coefficient and bidirectional adjustment coefficient of each control subsequence are obtained, and then the coordinated fluctuation index of each control subsequence is obtained by combining these two coefficients.

[0036] For each control subsequence, the first step is to obtain its maximum value. The control subsequence consists of status parameter signals collected during the operation of the coal feeding system, arranged in chronological order. These signals may involve the PLC's operating frequency, the rotational speed of the coal feeding equipment, and coal quantity detection values, among others. The maximum value is the largest value among all parameter signals in the control subsequence. By traversing each element in the subsequence and comparing them, the maximum value and its location can be determined.

[0037] After determining the maximum value, the subsequence consisting of the maximum value and all its predecessor elements is used as the predecessor subsequence for each control subsequence. The predecessor subsequence contains all elements from the starting element of the control subsequence to the maximum value element (inclusive), and its length is determined by the position of the maximum value in the control subsequence. Simultaneously, the subsequence consisting of all elements following the maximum value is used as the successor subsequence for each control subsequence. The successor subsequence contains all elements from the first element after the maximum value element to the last element of the control subsequence.

[0038] Obtain the first-order difference sequence of the predecessor subsequence for each control subsequence. The first-order difference sequence is obtained by calculating the difference between two adjacent elements in the predecessor subsequence, i.e., the value of the latter element minus the value of the former element. The resulting new sequence reflects the amplitude and trend of the parameter signal changes in the predecessor subsequence. Then, a response function is used to process all elements in the first-order difference sequence to obtain the difference response sequence. The response function can be set according to the actual application scenario; its function is to transform the elements in the first-order difference sequence, such as smoothing abnormal fluctuations in the difference, so that the processed sequence better reflects the overall change characteristics.

[0039] The absolute value of the difference between the absolute value of the sum of elements in the difference response sequence and the absolute value of the length of the first-order difference sequence of the precursor subsequence is calculated and used as the volatility index of the precursor subsequence for each control subsequence. The absolute value of the sum of elements is obtained by adding the absolute values ​​of all elements in the difference response sequence, and the length of the first-order difference sequence is the number of elements in the sequence. Taking the absolute value of the difference between the two yields an index that reflects the volatility of the precursor subsequence.

[0040] The volatility index of the successor subsequence of each control subsequence is obtained using the same method as for the predecessor subsequence. Specifically, the first-order difference sequence of the successor subsequence is first calculated, then the corresponding difference response sequence is obtained through the response function, and finally the absolute value of the difference between the absolute value of the sum of elements and the length of the first-order difference sequence is calculated as the volatility index of the successor subsequence.

[0041] The product of the absolute difference between the volatility indices of the predecessor and successor subsequences and their mean is calculated and used as the bidirectional response coefficient for each control subsequence. The mean is the sum of the volatility indices of the predecessor and successor subsequences, divided by 2. The product of the absolute difference and the mean comprehensively reflects the degree of difference in volatility response between the predecessor and successor subsequences.

[0042] Calculate the variances of the elements in the predecessor and successor subsequences of each control subsequence. The variance is obtained by averaging the squared differences between each element in the sequence and the mean of the sequence, reflecting the dispersion of the elements in the sequence. The mean of the variances of the predecessor and successor subsequences is used as the two-way adjustment coefficient for each control subsequence. This mean is also obtained by adding the two variances and dividing by 2, reflecting the combined situation of the dispersion of element distribution in the predecessor and successor subsequences.

[0043] The coordinated fluctuation index of each control subsequence is obtained by combining the two-way response coefficient and the two-way adjustment coefficient. The calculation of the coordinated fluctuation index needs to comprehensively consider the two-way response coefficient and the two-way adjustment coefficient. The specific combination method can be determined according to actual needs. For example, by setting different weights to perform a weighted sum of the two coefficients, the final coordinated fluctuation index can comprehensively reflect the overall characteristics of the control subsequence in terms of local state change trend fluctuation and local state differences.

[0044] The coordinated fluctuation index is obtained by controlling the state change trend and the differences between states of the subsequence. It can accurately reflect the coordinated operation status within each subsequence, allowing operators to clearly understand the coordinated performance of each local link.

[0045] Example 3: All instructions in all control sub-sequences are sorted according to their temporal positions in the coal feeding system to construct an instruction sequence. The control sub-sequences originate from the state sequences of multiple PLCs operating collaboratively within the coal feeding system. Each control sub-sequence contains multiple instructions, which are control signals transmitted between PLCs or operation commands to the equipment. The temporal positions are determined based on the system's timeline. Each instruction is recorded with its corresponding time information upon issuance. Instructions in all control sub-sequences are arranged in ascending order to form a continuous instruction sequence. During this arrangement, the time stamp of each instruction is checked one by one to ensure accurate sequence and prevent time reversal. Even if instructions in different control sub-sequences are similar in content, they are strictly ordered according to their respective temporal positions.

[0046] The timing of all instructions in the control subsequence is sorted by size to construct an instruction timing sequence. Here, timing refers to the relative position of an instruction within its control subsequence. For example, in a control subsequence with 10 instructions, the first instruction is marked with timing 1, the second with timing 2, and so on up to 10. For each control subsequence, the timing tags of all instructions are extracted and then arranged in ascending order to form the corresponding instruction timing sequence. For multiple control subsequences, their respective instruction timings are processed separately, and then all processed timing sequences are integrated to ensure that the order within each timing sequence conforms to the order of instructions in the original control subsequence.

[0047] Obtain the first-order difference sequences of the instruction sequence and instruction timing sequence. For the instruction sequence, the first-order difference sequence is calculated by subtracting the parameter value of the previous instruction from the parameter value of the subsequent instruction, and then forming a new sequence from the resulting differences. The parameter values ​​can be specific operational values ​​contained in the instructions, such as equipment operating speed, coal quantity setpoints, etc. This calculation reflects the magnitude of parameter changes between adjacent instructions. For the instruction timing sequence, the first-order difference sequence is calculated by subtracting the previous timing value from the subsequent timing value, and then forming a new sequence from the resulting differences. This sequence reflects the interval between the relative positions of adjacent instructions in the control subsequence.

[0048] The variance of all elements in the first-order difference sequence of the instruction sequence is used as the instruction descent variance. The variance is calculated based on all differences in the first-order difference sequence. By calculating the degree of deviation of these differences from the mean of the sequence, a value reflecting the dispersion of the differences is obtained. If the differences between the elements in the first-order difference sequence of the instruction sequence are large, the corresponding variance value will be large, and vice versa. This variance value is defined as the instruction descent variance, used to reflect the degree of instability of the instruction parameter changes.

[0049] The variance of all elements in the first-order difference sequence of the instruction timing sequence is used as the instruction interval difference. Similarly, by calculating the variance of each element in this first-order difference sequence, the dispersion of the relative positional intervals of instructions within the control sub-sequence is reflected. When the variance of the first-order difference sequence of the instruction timing sequence is large, it indicates that the distribution intervals of instructions in the control sub-sequence are uneven, sometimes dense and sometimes sparse; when the variance is small, it indicates that the distribution intervals of instructions are relatively stable. This variance value is called the instruction interval difference, used to describe the fluctuation characteristics of the instruction's temporal distribution.

[0050] Throughout the process, the construction of the sequence and the calculation of differences require multiple verifications to ensure data accuracy. For example, when sorting the instruction sequence, the time stamps are repeatedly checked to avoid sorting errors caused by time recording mistakes; when calculating the first-order difference sequence, any omissions or miscalculations are checked to ensure that each difference is accurate. For variance calculation, all elements in the sequence must be included, without omitting any difference, to ensure that the final instruction descent difference and instruction interval difference truly reflect the characteristics of the sequence. This data will be used for subsequent calculation of control disorder coefficients, providing fundamental information for analyzing the coordinated control state of the system.

[0051] By considering the differences in instruction change trends and timing trends in the control subsequences, and obtaining differences in instruction descent and instruction intervals, we can comprehensively capture the changing characteristics of instructions during transmission and execution, providing an effective way to judge the consistency of instruction execution.

[0052] Example 4: The mean of the cooperative volatility indices of all control subsequences is calculated as the average volatility index. A control subsequence is a subsequence with specific characteristics divided from the state sequence. Each control subsequence has a corresponding cooperative volatility index, which comprehensively reflects the volatility of state changes within the subsequence and the differences between states. To calculate the average volatility index, the cooperative volatility indices of all control subsequences are collected, their values ​​are summed, and then divided by the total number of control subsequences. The result is the average volatility index. During the calculation, it is essential to ensure that all control subsequences are covered, without omitting any, to reflect the overall level of cooperative volatility.

[0053] The average of the similarity coefficients between each control subsequence and all other control subsequences is used as the sequence similarity of each control subsequence. The similarity coefficient measures the degree of similarity between two control subsequences and is obtained by comparing and analyzing the state parameter signals and command information of the two subsequences. For each control subsequence, the similarity coefficients need to be calculated separately with all other control subsequences. These similarity coefficients are then summed and divided by the number of similarity coefficients used in the calculation to obtain the sequence similarity of that control subsequence. This process requires processing each control subsequence individually to ensure that each subsequence is compared with all other subsequences, avoiding duplicate calculations or omissions in comparison.

[0054] The control disorder coefficient is calculated by combining the mean sequence similarity of all control subsequences with the instruction descent difference, instruction interval difference, and average volatility index. First, the mean sequence similarity of all control subsequences is calculated by summing the sequence similarity of each control subsequence and then dividing by the total number of control subsequences. Then, the four indicators—mean sequence similarity, instruction descent difference, instruction interval difference, and average volatility index—are processed. The processing method involves converting the value of each indicator to a range of 0 to 1. The conversion process is determined based on the value range of each indicator. For example, if the maximum value of an indicator is 100 and the minimum value is 0, then a value of 50 for that indicator will be converted to 0.5.

[0055] After normalization, the four indicators are combined according to preset weights. The weights are set based on the degree of influence of each indicator in reflecting the degree of control disorder in the system; different weight allocations will result in different emphases in the final control disorder coefficient. The combination method is to multiply the value of each normalized indicator by its corresponding weight, and then add all the products together; the sum is the control disorder coefficient.

[0056] Throughout the calculation process, attention must be paid to the accuracy and consistency of the data. For example, when calculating the similarity coefficient, ensure that the time span and parameter types of the two control subsequences used for comparison are consistent to avoid deviations in the similarity coefficient calculation due to data mismatch. During normalization, accurately determine the value range of each indicator to ensure that the converted values ​​accurately reflect the relative magnitude of the original indicators. When setting weights, comprehensively consider the system's operating characteristics and control requirements to ensure that the weight allocation conforms to the actual situation.

[0057] The magnitude of the control disorder coefficient reflects the overall degree of control disorder in the state sequence. A larger value indicates lower similarity between control sub-sequences during multi-PLC collaborative control, larger variations and intervals in commands, higher overall level of collaborative fluctuation, and a relatively disordered control state. Conversely, a smaller value indicates a more orderly control state. This coefficient will serve as an important basis for subsequently calculating the system stability index, incorporating it into the calculation process to provide a quantitative indicator for assessing the system's stability.

[0058] By combining the differences in instruction correlation, the average of the cooperative fluctuation index, and the average of the similarity among control subsequences, the resulting control disorder coefficient can intuitively reflect the degree of control disorder of the entire state sequence, providing a reference for the formulation of control strategies.

[0059] Example 5: All eigenvalues ​​are sorted in descending order to construct an eigenvalue sequence. The eigenvalues ​​are derived from the eigenvalue decomposition of the feature matrix of the state sequence, and each eigenvalue corresponds to the importance of a certain feature in the feature matrix. During the sorting process, starting with the largest eigenvalue, the other eigenvalues ​​are arranged in descending order of numerical value, forming a continuous eigenvalue sequence. The numerical values ​​of each eigenvalue must be precisely compared during sorting to ensure that the relationship between adjacent eigenvalues ​​in the sequence is accurate. Even if the numerical differences between eigenvalues ​​are extremely small, they must be arranged strictly according to their actual numerical order.

[0060] The eigenvalue sequence is used as input to the threshold segmentation method, and the output is the threshold. The threshold segmentation method analyzes the distribution of each eigenvalue in the eigenvalue sequence to determine a critical value that can distinguish the importance of different features. During processing, the algorithm performs multiple rounds of calculations on the eigenvalue sequence, considering the distribution characteristics of eigenvalues ​​at different segmentation points, and finally determines a suitable threshold. This threshold divides the eigenvalue sequence into two parts: one part contains eigenvalues ​​that have a more significant impact on the system's characteristics, and the other part contains eigenvalues ​​with relatively weaker impact.

[0061] All feature values ​​greater than or equal to a threshold are designated as key feature values, and all feature values ​​less than the threshold are designated as auxiliary feature values. The segmentation is strictly based on the comparison between the feature values ​​and the threshold, without introducing other judgment criteria, ensuring the objectivity of the segmentation results. The number of key and auxiliary feature values ​​depends on the threshold value; a higher threshold results in fewer key feature values, while a lower threshold results in more key feature values.

[0062] The ratio of the number of key eigenvalues ​​to the total number of eigenvalues ​​is taken as the proportion of key components in the eigenvalue sequence. Accurate counting of both key eigenvalues ​​and the total number of eigenvalues ​​is necessary to ensure the precision of the ratio calculation. This ratio reflects the proportion of key eigenvalues ​​in the entire eigenvalue sequence, demonstrating the proportion of components that play a major role in the system's characteristics.

[0063] The mean and variance of all key eigenvalues ​​in the eigenvalue sequence are calculated, and the product of the mean and variance is used as the volatility index of the key eigenvalues. The mean is obtained by summing all key eigenvalues ​​and dividing by the number of key eigenvalues, while the variance is obtained by calculating the deviation of each key eigenvalue from the mean. The product of the two can comprehensively reflect the overall level and dispersion of the key eigenvalues, reflecting their volatility.

[0064] The volatility index of auxiliary features is calculated using the same method as that used for the volatility index of key features. Specifically, the mean and variance of the auxiliary features are first calculated, and then multiplied together to obtain the volatility index. This calculation process is consistent with that for the volatility index of key features, ensuring the comparability of the two volatility indices.

[0065] The absolute value of the difference between the volatility indices of key eigenvalues ​​and auxiliary eigenvalues ​​is used as the difference index of the eigenvalue sequence. This absolute value of the difference directly reflects the degree of difference in volatility between key eigenvalues ​​and auxiliary eigenvalues; the larger the difference, the more obvious the difference in volatility characteristics between the two types of eigenvalues.

[0066] Calculate the absolute value of the similarity coefficient between each row of data in the feature matrix and all other rows, and then take the mean of the absolute values ​​of all similarity coefficients for each row as the co-correlation coefficient for that row. The similarity coefficient is obtained by analyzing the degree of correlation between two rows of data; the larger the absolute value, the stronger the correlation between the two rows. To calculate the co-correlation coefficient for each row, add the absolute values ​​of the similarity coefficients of that row to all other rows, and then divide by the number of other rows. The result is the co-correlation coefficient for that row.

[0067] The mean of the correlation coefficients of all rows in the feature matrix is ​​used as the feature matrix's own correlation coefficient. The calculation involves summing the correlation coefficients of all rows and then dividing by the total number of rows; the resulting mean reflects the overall correlation between the rows in the feature matrix.

[0068] The system stability index is calculated by combining the self-coordination coefficient with the proportion of key components, the difference index, and the control disorder coefficient. In the calculation process, the four indicators—self-coordination coefficient, proportion of key components, difference index, and control disorder coefficient—are first processed to adjust the values ​​of each indicator to a similar range. The processing method is determined based on the original value range of each indicator to ensure that the processed values ​​accurately reflect the relative magnitudes of each indicator. Then, these four processed indicators are combined according to a preset ratio. The combination method involves multiplying the value of each indicator by its corresponding proportional coefficient and then summing all the products; the resulting sum is the system stability index.

[0069] When controlling an intelligent automatic coal feeding system with multi-PLC collaborative control based on the system stability index, the default thresholds of the control parameters are first obtained based on the system stability index of the state sequence. The thresholds are determined according to the numerical range of the system stability index; different system stability indices correspond to different thresholds, and the threshold settings must be combined with the system's operating parameters and control requirements. The multi-PLC collaborative operating parameters are then optimized based on the default thresholds of the control parameters. These operating parameters include the PLC communication cycle, data transmission rate, and instruction execution priority. During optimization, each operating parameter must be adjusted to within the range allowed by the thresholds. The collaborative control method is used to calculate the coal feeding control parameters in the coal feeding system. This method coordinates the instruction transmission and execution between the PLCs, ensuring that the operating parameters of the coal feeding equipment are compatible with the system stability index, thus achieving intelligent control of the coal feeding process.

[0070] By constructing a feature matrix of the state sequence and performing eigenvalue decomposition, the proportion of key components and the difference index are obtained based on various cases of eigenvalues, enabling analysis of the system's intrinsic characteristics from the perspective of data features. The system stability index is obtained by combining the correlation of data in each row of the feature matrix, the control disorder coefficient, the proportion of key components, and the difference index. This index is then used to control the system, enabling precise regulation of an intelligent automatic coal feeding system with multi-PLC collaborative control. This ensures the system maintains good collaborative performance in complex operating environments, reduces operational failures caused by poor coordination, and improves the overall operational efficiency of the system.

[0071] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0072] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An intelligent automatic coal feeding control method based on multi-PLC collaborative control, characterized in that, The method includes the following steps: The state sequence is composed of state parameter signals collected during the coordinated operation of multiple PLCs in the coal feeding system. The collaborative phases are divided according to the distribution of nodes in the state sequence to obtain the sequence of each phase. The collaborative phases in the state sequence are clustered according to the distribution of the collaborative phases to obtain each control subsequence. The feature values ​​of the elements in each phase sequence are calculated and input into the clustering algorithm to obtain each cluster. The mean of the internal elements of each cluster is calculated. The cluster with the largest mean of internal elements is taken as the control cluster. The state subsequence corresponding to the internal elements of the control cluster is taken as the control subsequence. Based on the fluctuation of local state change trends and the differences between local states in the control subsequence, the bidirectional response coefficient and bidirectional adjustment coefficient of each control subsequence are obtained, and the coordinated fluctuation index of each control subsequence is obtained. The bidirectional response coefficient is the product of the absolute value and the mean of the difference between the fluctuation indices of the predecessor subsequence and the successor subsequence, and the bidirectional adjustment coefficient is the mean of the variances of the predecessor subsequence and the successor subsequence. Based on the differences in the changing trends of instructions in the control subsequence and the differences in the timing of instructions in the control subsequence, the differences in instruction descent and instruction interval are obtained. Based on the difference in command descent, the difference in command interval, the average of the cooperative volatility index of all control subsequences, and the average of the similarity between control subsequences, the control disorder coefficient of the state sequence is obtained. The mean of the sequence similarity of all control subsequences is calculated. The mean of the sequence similarity, the difference in command descent, the difference in command interval, and the average volatility index are normalized. Each normalized index value is multiplied by its corresponding weight, and all products are added together. The sum is the control disorder coefficient. A feature matrix of the state sequence is constructed, and key feature values ​​and auxiliary feature values ​​are divided by threshold segmentation. The fluctuation index of the key feature values ​​and the fluctuation index of the auxiliary feature values ​​are obtained. The proportion of key components and the difference index of the feature value sequence of the state sequence are obtained. The proportion of key components is the ratio of the number of key feature values ​​to the total number of all feature values. The difference index is the absolute value of the difference between the fluctuation index of the key feature values ​​and the fluctuation index of the auxiliary feature values. The system stability index of the state sequence is obtained based on the correlation between data in each row of the feature matrix, the control disorder coefficient, the proportion of key components, and the difference index; the intelligent automatic coal feeding system with multi-PLC collaborative control is controlled based on the system stability index.

2. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 1, characterized in that, The method for obtaining the control subsequence is as follows: The state node detection algorithm is used to obtain all instructions and nodes in the state sequence. The state sequence is divided into stage sequences from each node. The feature values ​​of the elements in each stage sequence are calculated. The feature values ​​of all stage sequences are used as input to the clustering algorithm, and the output is each cluster. Calculate the mean of internal elements in each cluster, select the cluster with the largest mean of internal elements as the control cluster, and select the state subsequence corresponding to the internal elements of the control cluster as the control subsequence.

3. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 1, characterized in that, The method for obtaining the coordinated volatility index is as follows: Based on the fluctuations in the local state change trends and the differences between local states in the control subsequence, the bidirectional response coefficient and bidirectional adjustment coefficient of each control subsequence are obtained. The coordinated fluctuation index of each control subsequence is obtained by combining the bidirectional response coefficient and bidirectional adjustment coefficient of each control subsequence.

4. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 3, characterized in that, The method for obtaining the bidirectional adjustment coefficient is as follows: For each control subsequence, obtain the maximum value in each control subsequence, take the subsequence consisting of the maximum value and all elements preceding the maximum value as the predecessor subsequence of each control subsequence, and take the subsequence consisting of all elements following the maximum value as the successor subsequence of each control subsequence. Obtain the first-order difference sequence of the predecessor subsequence of each control subsequence, and use the response function to process all elements in the first-order difference sequence to obtain the difference response sequence; The absolute value of the difference between the sum of elements in the differential response sequence and the length of the first-order difference sequence of the predecessor subsequence is calculated and used as the fluctuation index of the predecessor subsequence of each control subsequence. Using the same method as for the predecessor subsequence, the volatility index of the successor subsequence of each control subsequence is obtained; the product of the absolute value of the difference between the volatility indices of the predecessor subsequence and the successor subsequence and the mean value is calculated as the bidirectional response coefficient of each control subsequence. Calculate the variance of the elements in the predecessor subsequence and the variance of the elements in the successor subsequence for each control subsequence, and use the mean of the variances of the predecessor and successor subsequences as the bidirectional adjustment coefficient for each control subsequence.

5. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 2, characterized in that, The method for obtaining the instruction interval difference is as follows: Sort all instructions in all control subsequences according to their timing position in the coal supply system to construct an instruction sequence; sort the timing of all instructions in the control subsequences according to their size to construct an instruction timing sequence. Obtain the first-order difference sequences of the instruction sequence and the instruction timing sequence. Use the variance of all elements in the first-order difference sequence of the instruction sequence as the instruction descent difference; use the variance of all elements in the first-order difference sequence of the instruction timing sequence as the instruction interval difference.

6. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 1, characterized in that, The method for obtaining the control disorder coefficient is as follows: The mean of the co-variance indices of all control subsequences is calculated as the average volatility index. The mean of the similarity coefficients between each control subsequence and all other control subsequences is taken as the sequence similarity of each control subsequence; The control disorder coefficient is calculated by combining the mean of sequence similarity of all control subsequences with the instruction descent difference, instruction interval difference, and average volatility index.

7. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 1, characterized in that, The method for obtaining the proportion of key components and the difference index is as follows: Sort all feature values ​​in descending order to construct a feature value sequence; The feature value sequence is used as the input of the threshold segmentation method, and the output of the threshold segmentation method is the threshold. All feature values ​​greater than or equal to the threshold are used as key feature values, and all feature values ​​less than the threshold are used as auxiliary feature values. The ratio of the number of key feature values ​​to the total number of all feature values ​​is taken as the proportion of key components in the feature value sequence; Calculate the mean and variance of all key eigenvalues ​​in the eigenvalue sequence, and use the product of the mean and variance as the volatility index of the key eigenvalues; use the same method as the volatility index of the key eigenvalues ​​to calculate the volatility index of the auxiliary eigenvalues, and use the absolute value of the difference between the volatility index of the key eigenvalues ​​and the volatility index of the auxiliary eigenvalues ​​as the difference index of the eigenvalue sequence.

8. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 1, characterized in that, The method for obtaining the system stability index is as follows: Calculate the absolute value of the similarity coefficient between each row of data in the feature matrix and all other rows of data, and take the mean of the absolute values ​​of all the similarity coefficients of each row of data as the co-correlation coefficient of each row of data; The mean of the synergy coefficients of all rows in the feature matrix is ​​taken as the synergy coefficient of the feature matrix itself. The system stability index is calculated by combining its own synergy coefficient with the proportion of key components, the difference index, and the control disorder coefficient.

9. The intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in claim 1, characterized in that, The intelligent automatic coal feeding system based on system stability index and multi-PLC collaborative control includes: The default threshold for control parameters is obtained from the system stability index based on the state sequence; The parameters for multi-PLC collaborative operation are optimized based on the default thresholds of the control parameters, and the coal feeding control parameters in the coal feeding system are calculated using the collaborative control method.

10. An intelligent automatic coal feeding system based on multi-PLC collaborative control, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent automatic coal feeding control method based on multi-PLC collaborative control as described in any one of claims 1-9.

Images