Multi-process spinning energy consumption optimization decision system

Abstract

The present invention relates to the technical field of spinning production, and discloses a multi-process spinning energy consumption optimization decision system, which includes a data acquisition module, an energy consumption analysis module, a strategy generation module, a scheduling optimization module, a benchmark comparison module, and a decision output module. The data acquisition module acquires equipment parameters and energy consumption data and sets monitoring intervals; the energy consumption analysis module divides process nodes and generates energy consumption feature vectors; the strategy generation module extracts key indicators and establishes optimization rules; the scheduling optimization module identifies operating modes and calculates the amplitude of energy consumption fluctuations; the benchmark comparison module derives the optimal energy consumption threshold and generates a deviation sequence; and the decision output module integrates the deviation sequence to form an optimized decision plan. This system realizes the refined management and dynamic optimization of energy consumption in all spinning processes, improves energy utilization efficiency, reduces production costs, and is suitable for energy consumption optimization control in the spinning production process.

Description

Technical Field

[0001] The invention relates to the technical field of spinning production, in particular to a multi-process spinning energy consumption optimization decision-making system. Background Art

[0002] In the spinning production process, energy consumption management of multiple collaborative processes has always been a key focus of the industry. Traditional spinning energy consumption management methods have many shortcomings and cannot meet the modern textile industry's demand for efficient, energy-saving, and low-carbon production.

[0003] From the perspective of data collection and monitoring, traditional systems are often unable to achieve comprehensive and real-time collection of multi-process equipment parameters and energy consumption data. They lack refined energy consumption monitoring intervals set according to the characteristics of different processes, resulting in the inability to accurately grasp the actual energy consumption status of each process and difficulty in discovering energy consumption anomalies and high-energy-consuming links.

[0004] When it comes to energy consumption analysis, traditional methods lack a scientific division of process nodes and in-depth analysis of energy consumption data. This makes it difficult to build an effective process feature library and identify the inherent connections between process parameter vectors through similar process matching and cluster analysis. Consequently, analysis of energy consumption data remains superficial, and it is impossible to generate vectors that accurately reflect process energy consumption characteristics, which in turn affects the formulation of subsequent optimization strategies.

[0005] In the strategy generation stage, traditional methods make it difficult to accurately extract key energy consumption indicators from complex energy consumption data. There is a lack of systematic energy consumption optimization rules associated with process nodes, and the optimization rules cannot be dynamically adjusted according to the number of process nodes. As a result, the optimization strategy is not targeted and effective enough, and it is impossible to achieve accurate control of the energy consumption of each process node.

[0006] During the scheduling optimization process, traditional systems are unable to effectively identify the timing factors and fluctuation factors in the equipment operation mode, making it difficult to build a state transition network and dynamically schedule key energy consumption indicators based on it. They are also unable to accurately calculate the energy consumption fluctuation range of each process node under different scheduling strategies, making energy consumption scheduling lack of scientific basis and difficult to achieve dynamic optimization of energy consumption.

[0007] In terms of benchmark comparison and decision output, traditional methods lack a scientific method for deriving the optimal energy consumption threshold, and are unable to generate an accurate process energy consumption deviation sequence by comparing with historical data. Moreover, when integrating the deviation sequence to form an optimized decision-making plan, there is a lack of comprehensive analysis of the energy consumption fluctuation direction, convergence rate and diffusion rate, resulting in poor scientificity and operability of the decision-making plan. Summary of the Invention

[0008] The purpose of the present invention is to provide a multi-process spinning energy consumption optimization decision-making system to solve the problems raised in the above background technology.

[0009] To achieve the above-mentioned object, the present invention provides the following technical solution: a multi-process spinning energy consumption optimization decision system, the system comprising:

[0010] The data acquisition module is used to obtain equipment parameters and energy consumption data in the spinning production process and set the energy consumption monitoring interval corresponding to the process;

[0011] The energy consumption analysis module is used to divide the energy consumption monitoring interval into multiple process nodes, perform feature analysis on the energy consumption data of each process node, and generate the energy consumption feature vector corresponding to the process node;

[0012] The strategy generation module is used to extract key energy consumption indicators from the energy consumption feature vector, establish energy consumption optimization rules associated with process nodes, and obtain process control parameters corresponding to the rules;

[0013] The scheduling optimization module is used to identify the equipment operation mode in the process control parameters, dynamically schedule key energy consumption indicators according to the operation mode, and calculate the energy consumption fluctuation range of each process node under different scheduling strategies;

[0014] The benchmark comparison module is used to derive the optimal energy consumption threshold according to the energy consumption fluctuation amplitude, and generate the process energy consumption deviation sequence by comparing the current process energy consumption value with the optimal energy consumption threshold;

[0015] The decision output module is used to analyze the process energy consumption deviation sequence and integrate the process energy consumption deviation sequence into an energy consumption optimization decision plan based on the energy consumption fluctuation trend of the process nodes.

[0016] Preferably, the energy consumption analysis module is implemented by: constructing a process feature library corresponding to the process node, the process feature library including process parameter vectors mapped with equipment parameters and energy consumption data;

[0017] The process parameter vectors are matched with similar processes, and the process parameter vectors are divided into process cluster groups according to the matching results; the distribution center points of the energy consumption data are extracted from the process cluster groups, and the distribution center points are set as process nodes.

[0018] Preferably, the process clustering groups for dividing the process parameter vectors further include:

[0019] According to the equipment type and production stage in the process parameter vector, the equipment speed, yarn count, and ambient temperature and humidity parameters are extracted, and the process feature labels are generated based on these parameters;

[0020] The process feature labels are associated with the process parameter vectors. By calculating the process similarity between the feature labels, the process parameter vectors with similarity higher than the preset process threshold are screened to form a process cluster group.

[0021] Preferably, the implementation method of generating the energy consumption characteristic vector corresponding to the process node includes:

[0022] For each process node, according to the node's time sequence position in the energy consumption monitoring interval, the node's energy consumption fluctuation data within the preset period is obtained, and the node's energy consumption fluctuation coefficient is calculated;

[0023] When the energy consumption fluctuation coefficient exceeds the first fluctuation threshold, the node is marked as a high-energy-consuming node, and its energy consumption data is extracted to form an energy consumption feature vector; when the energy consumption fluctuation coefficient is lower than the first fluctuation threshold, the node is marked as a steady-state node, and the energy consumption data of the node's adjacent nodes are weighted and superimposed, and the superimposed data is reconstructed into an energy consumption feature vector.

[0024] Preferably, the implementation of the strategy generation module includes:

[0025] Separate the equipment energy consumption proportion, idling energy consumption proportion and load fluctuation parameters from the energy consumption feature vector, and generate energy consumption optimization rules for process nodes based on the above parameters;

[0026] If the number of process nodes covered by the current energy consumption optimization rule is less than the preset node threshold, the energy consumption feature vectors of adjacent process nodes are traversed, and the energy consumption indicators not included in the optimization rules of the adjacent nodes are added to the current rule.

[0027] Preferably, the implementation of the scheduling optimization module includes: obtaining a timing factor of the start-stop frequency and a fluctuation factor of the load variation amplitude in the equipment operation mode;

[0028] A state transition network associated with timing factors and fluctuation factors is constructed, and the energy consumption fluctuation amplitude under different scheduling strategies is determined based on the transition probability of each path in the network.

[0029] Preferably, building a state transfer network further includes:

[0030] Identify the periodic characteristics of the timing factor. If the current periodic characteristics completely match the preset production cycle, set the timing factor as the starting node of the state transition network.

[0031] Calculate the transfer correlation between the timing factor and the volatility factor, and generate the intermediate nodes and terminal nodes of the state transition network in descending order of correlation;

[0032] The state of the terminal node is backtracked. When the correlation degree of the terminal node is lower than the preset correlation threshold, it is output as the final path of the state transition network.

[0033] Preferably, the implementation method of calculating the energy consumption fluctuation amplitude includes:

[0034] Statistically calculate the mean value and range of the timing factor of each terminal node in the state transition network, and calculate the global variance of the factors of all nodes;

[0035] The difference between the mean of the timing factor of a single termination node and the mean of the timing factor of the adjacent nodes is divided by the global variance to obtain the timing fluctuation coefficient; at the same time, the ratio of the fluctuation factor range to the global variance is calculated, and the weighted sum of the two is taken as the energy consumption fluctuation amplitude of the node.

[0036] Preferably, the implementation method of deriving the optimal energy consumption threshold includes:

[0037] Extract the fluctuation pattern in the historical data that is closest to the current energy consumption fluctuation amplitude, and calculate the Manhattan distance between the two in the time series distribution as the first benchmark reference value;

[0038] Count the difference in the number of extreme points between the current energy consumption fluctuation amplitude and the historical fluctuation pattern, and use the difference as the second benchmark reference value;

[0039] Based on a linear combination of the first benchmark reference value and the second benchmark reference value, an optimal energy consumption threshold in a preset energy consumption threshold table is matched.

[0040] Preferably, the implementation of the decision output module includes: dividing the energy consumption fluctuation direction of each node in the process energy consumption deviation sequence into a positive fluctuation interval and a negative fluctuation interval;

[0041] The convergence rate of energy consumption deviation in the positive fluctuation range and the diffusion rate of energy consumption deviation in the negative fluctuation range are extracted, and the two are weighted and fused according to the production weights of the process nodes to generate the core adjustment parameters of the energy consumption optimization decision-making plan.

[0042] Compared with the prior art, the present invention has the following beneficial effects:

[0043] The data acquisition module can obtain equipment parameters and energy consumption data in the spinning production process, and set energy consumption monitoring intervals corresponding to the process, realizing real-time, comprehensive collection and refined monitoring of energy consumption data of each process, providing an accurate and reliable data basis for subsequent energy consumption analysis and optimization.

[0044] The energy consumption analysis module builds a process feature library, performs similar process matching and cluster analysis on process parameter vectors, scientifically divides process nodes, and uses different methods to generate energy consumption feature vectors for different types of nodes (high-energy-consuming nodes and steady-state nodes). It achieves in-depth feature analysis of energy consumption data, accurately reveals the energy consumption characteristics and patterns of each process node, and provides strong support for formulating targeted energy consumption optimization strategies.

[0045] The strategy generation module extracts key energy consumption indicators from the energy consumption feature vector, establishes energy consumption optimization rules associated with process nodes, and can dynamically expand the optimization rules according to the number of process nodes, ensuring the comprehensiveness and pertinence of the optimization rules, effectively covering the energy consumption optimization needs of each process node, and improving the efficiency and effectiveness of energy consumption optimization.

[0046] The scheduling optimization module obtains the timing factors and fluctuation factors in the equipment operation mode, constructs a state transition network, and calculates the energy consumption fluctuation amplitude of each process node under different scheduling strategies. It realizes dynamic scheduling and scientific analysis of key energy consumption indicators, provides a scientific basis for selecting the optimal energy consumption scheduling strategy, and can effectively reduce energy consumption fluctuations and improve the stability and efficiency of energy utilization.

[0047] The benchmark comparison module extracts fluctuation patterns from historical data, calculates the Manhattan distance and the difference in the number of extreme points, derives the optimal energy consumption threshold, and generates a process energy consumption deviation sequence. This provides a scientific benchmark for accurately evaluating the current process energy consumption status, can promptly detect energy consumption deviations, and provide a basis for subsequent decision-making adjustments.

[0048] The decision output module divides the fluctuation range according to the process energy consumption deviation sequence, extracts the convergence rate and diffusion rate, generates core adjustment parameters, and integrates the deviation sequence into an energy consumption optimization decision plan, realizing the intelligent transformation from data to decision. It can provide production management personnel with scientific and operational energy consumption optimization decisions, guide production practice, significantly reduce the energy consumption cost of spinning production, and improve the economic and environmental benefits of production. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a working principle diagram of the multi-process spinning energy consumption optimization decision system of the present invention;

[0050] Figure 2 Flowchart for clustering process parameter vectors;

[0051] Figure 3 Flowchart generated for energy optimization rules. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0053] See also Figure 1-Figure 3 The present invention relates to a multi-process spinning energy consumption optimization decision system, which includes: a data acquisition module, an energy consumption analysis module, a strategy generation module, a scheduling optimization module, a benchmark comparison module and a decision output module. The specific implementation steps are as follows:

[0054] The data acquisition module acquires equipment parameters and energy consumption data from the spinning process and sets energy consumption monitoring intervals corresponding to the process. For example, in the carding process, equipment parameters such as carding machine speed and cotton feed rate are collected, along with real-time energy consumption data for the process. A specific energy consumption monitoring interval is set for the carding process to limit the scope and time period of data collection.

[0055] The energy consumption analysis module divides the energy consumption monitoring interval into multiple process nodes, analyzes the energy consumption data for each process node, and generates an energy consumption feature vector corresponding to the process node. For example, within the energy consumption monitoring interval of the drawing process, several process nodes are divided according to the chronological order or process stage of the production process. Each node corresponds to a specific production time period or process operation step. The energy consumption data for each node is then analyzed to extract key information that reflects the node's energy consumption characteristics, forming an energy consumption feature vector.

[0056] The strategy generation module extracts key energy consumption indicators from the energy consumption feature vector, establishes energy consumption optimization rules associated with the process nodes, and obtains the process control parameters corresponding to the rules. Taking the roving process as an example, key indicators such as the equipment energy consumption ratio and the idle energy consumption ratio are extracted from the energy consumption feature vector of this process node. Based on these indicators, corresponding energy consumption optimization rules are established. For example, when the equipment energy consumption ratio exceeds a certain threshold, the equipment operating parameters are adjusted. At the same time, the process control parameters corresponding to these rules are determined, such as the equipment speed adjustment value and the feed rate adjustment value.

[0057] The scheduling optimization module identifies the equipment operating mode within the process control parameters, dynamically schedules key energy consumption indicators based on the operating mode, and calculates the energy consumption fluctuation range of each process node under different scheduling strategies. In the spinning process, the module identifies equipment operating modes, including normal operation mode and frequent start-stop mode. Based on these different operating modes, key energy consumption indicators, such as load fluctuation parameters, are dynamically adjusted and scheduled. The module then calculates the energy consumption fluctuation range of each process node under different scheduling strategies to evaluate the effectiveness of different scheduling strategies.

[0058] The benchmark comparison module derives the optimal energy consumption threshold based on the energy consumption fluctuation range. By comparing the current process energy consumption value with the optimal energy consumption threshold, the module generates a process energy consumption deviation sequence. For example, in the winding process, the module derives the optimal energy consumption threshold based on the energy consumption fluctuation range of the process node, combined with historical data and production process requirements. The module then compares the current process's actual energy consumption with the optimal energy consumption threshold, calculates the difference between the two, and forms a process energy consumption deviation sequence, reflecting the difference between the current energy consumption and the optimal energy consumption.

[0059] The decision output module analyzes the process energy consumption deviation sequence and, based on the energy consumption fluctuation trends of the process nodes, integrates this process energy consumption deviation sequence into an energy consumption optimization decision plan. Throughout the spinning production process, the decision output module analyzes the process energy consumption deviation sequence of each process node to determine whether the energy consumption fluctuation trend of each node is increasing, decreasing, or remaining stable. Based on these trends and deviation sequences, the module then integrates the optimization suggestions for each process into a complete energy consumption optimization decision plan to guide energy consumption optimization adjustments during the production process.

[0060] The technical solution of the present invention is further described in detail below with reference to specific embodiments. Example 1:

[0061] The energy consumption analysis module is specifically implemented by constructing a process feature library corresponding to each process node. The process feature library contains process parameter vectors mapped to equipment parameters and energy consumption data. In a spinning production system, different process nodes have unique equipment operating states and energy consumption characteristics. For example, the carding process involves the operation of key components of the carding machine, such as the licker-in, cylinder, and doffer. Equipment parameters such as the rotational speed and vibration frequency of these components directly affect the energy consumption of the carding process. Furthermore, the energy consumption data for the carding process includes not only the total energy consumption value but also the time distribution characteristics of energy consumption, such as the instantaneous power variation curve. These equipment parameters and energy consumption data are integrated and mapped into a high-dimensional process parameter vector. Each dimension represents a specific equipment parameter or energy consumption characteristic. For example, the first dimension of the vector could represent the rotational speed of the carding machine licker-in, the second dimension represents the vibration frequency of the cylinder, and the third dimension represents the energy consumption per unit time. This mapping method converts complex equipment operating state and energy consumption data into a mathematically manageable vector form, facilitating subsequent analysis and processing.

[0062] Similar process matching is performed on process parameter vectors, and the matching results are used to divide the process parameter vectors into process cluster groups. In actual spinning production, process nodes of the same type may vary due to factors such as production batch and equipment model, but generally have similar energy consumption characteristics and equipment operation modes. The process of similar process matching first requires defining an appropriate similarity metric. Euclidean distance can be used as a similarity metric. For two process parameter vectors, the Euclidean distance in high-dimensional space is calculated, with smaller distances indicating higher similarity. Cosine similarity can also be considered, as it measures the directional similarity between two vectors and is suitable for processing high-dimensional sparse data. After calculating the similarity, a hierarchical clustering algorithm is used to perform cluster analysis on the process parameter vectors. The hierarchical clustering algorithm forms a tree-like cluster structure by continuously merging or splitting cluster groups. In this system, process parameter vectors with similarity exceeding a preset similarity threshold are merged into a single process cluster group. For example, for multiple nodes in the combing process, by calculating the similarity of their process parameter vectors, the nodes with higher similarity are divided into the same cluster group. In this way, the nodes of the combing process can be classified according to energy consumption characteristics and equipment operation modes, which facilitates subsequent targeted analysis and optimization of different types of nodes.

[0063] The distribution center point of the energy consumption data is extracted from the process cluster group, and the distribution center point is set as the process node. In each process cluster group, the energy consumption data shows certain distribution characteristics. By calculating the mean, median and other statistics of the energy consumption data dimension in all process parameter vectors in the cluster group, the distribution center point of the energy consumption data can be determined. For example, for a cluster group containing multiple carding process nodes, the mean of the energy consumption data of all nodes in the group is calculated, and the mean is used as the energy consumption distribution center point of the cluster group. Setting the distribution center point as the process node can represent the typical energy consumption characteristics and equipment operation mode of the cluster group. The advantage of doing this is that it can simplify the subsequent analysis and processing process, and transform the analysis of a large number of process nodes into the analysis of a few representative nodes, while retaining the main characteristics and information of the original data.

[0064] Furthermore, the process clustering grouping for process parameter vectors also involves extracting equipment speed, yarn count, and ambient temperature and humidity parameters based on the equipment type and production stage in the process parameter vectors, and generating process feature labels based on these parameters. Equipment type is an important criterion for distinguishing different processes, as different types of equipment have different energy consumption characteristics and operating modes. In spinning production, equipment such as carding machines, draw frames, roving frames, and spinning frames exhibit significant structural and functional differences, resulting in distinct energy consumption characteristics. The production stage reflects the spinning process flow, from opening and cleaning, carding, draw frames, to roving and spinning. Each stage has different production tasks and process requirements, resulting in varying energy consumption characteristics. Equipment speed directly affects its energy consumption. Generally speaking, higher speeds increase energy consumption, but also potentially higher production efficiency. Yarn count measures yarn thickness, and yarns of different counts have different equipment requirements and energy consumption during production. Ambient temperature and humidity parameters have a significant impact on the spinning production process. Excessively high or low temperatures and humidity can cause unstable equipment operation, increase energy consumption, and also affect yarn quality. Based on these parameters, process feature labels are generated, with each label representing a specific equipment type, production stage, and parameter combination. For example, a process feature label could be represented as "Combing Process - High Speed ​​- High Count Yarn - Suitable Temperature and Humidity." Such a label can accurately describe the characteristics and status of the process nodes, providing richer information for subsequent process clustering and analysis.

[0065] Process feature labels are associated with process parameter vectors. The process similarity between feature labels is calculated, and process parameter vectors with similarities exceeding a preset process threshold are selected to form process clusters. Each process parameter vector is associated with one or more process feature labels, which describe the characteristics and status of the process node represented by the vector. Semantic similarity can be used to calculate process similarity between feature labels. For equipment type labels, a score is assigned based on the functional and structural similarity of the equipment. For example, carding machines and combing machines have certain functional similarities, and their equipment type label similarity can be scored as 0.7. For production stage labels, a score is assigned based on the production process sequence and process relevance. For example, the carding and drawing stages are consecutive stages in spinning production, and their production stage label similarity can be scored as 0.8. For parameter labels, such as equipment speed, yarn count, and ambient temperature and humidity, a score is assigned based on the proximity of the parameter values. The similarities of these labels are combined to determine the process similarity between feature labels. When the similarity between two process feature labels exceeds a preset process threshold, their corresponding process parameter vectors are grouped into the same process cluster. In this way, process nodes with similar characteristics and status can be more accurately divided into the same group, improving the accuracy and effectiveness of process clustering.

[0066] In practical applications, the process of building a process feature library requires consideration of data integrity and accuracy. The collection of equipment parameters and energy consumption data requires the use of high-precision sensors and data acquisition systems to ensure data authenticity and reliability. When mapping data into process parameter vectors, data preprocessing, including data cleaning and normalization, is required to eliminate noise and outliers and improve the quality of the vectors. When matching similar processes within process parameter vectors, the selection of a similarity metric is crucial. Different similarity metrics are suitable for different types of data and application scenarios and need to be selected and adjusted based on the specific situation. During the clustering process, the preset similarity threshold can also affect the quality of the clustering results. A threshold that is too high may result in overly dispersed cluster groups, while a threshold that is too low may result in overly concentrated cluster groups, rendering the clustering meaningless. Therefore, it is necessary to determine an appropriate similarity threshold through experimentation and analysis.

[0067] When extracting the distribution center of energy consumption data from process clusters, the data's distribution characteristics must be considered. For normally distributed data, the mean is a good proxy for the distribution center; however, for non-normally distributed data, the median may better represent the data's central tendency. Therefore, when calculating the distribution center, it is necessary to analyze the data's distribution characteristics and select an appropriate statistic as the distribution center. When generating process feature labels, it is necessary to ensure that the labels accurately reflect the characteristics and status of the process nodes. Labels should be concise and clear while providing sufficient information. When associating process feature labels with process parameter vectors, an effective mapping mechanism must be established to ensure that each vector is accurately associated with the corresponding label. When calculating process similarity between feature labels, the accuracy of the semantic similarity calculation method directly impacts the process clustering results. It is necessary to comprehensively consider all aspects of the labels and adopt an appropriate scoring method to improve the accuracy of similarity calculations.

[0068] Through the above implementation, the energy consumption analysis module can effectively analyze and process energy consumption data in the spinning production process, classify complex process nodes according to energy consumption characteristics and equipment operation modes, extract representative process nodes, and provide strong support for subsequent energy consumption optimization decisions. This method not only improves the efficiency and accuracy of energy consumption analysis, but also provides more targeted energy consumption optimization suggestions for spinning production companies, helping them reduce production costs, improve production efficiency, and achieve sustainable development. Example 2:

[0069] The energy consumption feature vector corresponding to a process node is generated as follows: For each process node, based on its temporal position within the energy consumption monitoring interval, the system obtains energy consumption fluctuation data for the node within a preset period and calculates the node's energy consumption fluctuation coefficient. In the spinning process, each process node is arranged chronologically within the energy consumption monitoring interval, forming a continuous production flow. The preset period should be determined based on the specific process's production rhythm and data collection frequency. For example, for the high-speed spinning process, the preset period can be set to 15 minutes, while for the slower opening and cleaning process, the preset period can be extended to 1 hour. For example, a node in the roving process is located in the third hour of the energy consumption monitoring interval, and the preset period is 1 hour. Energy sensors installed on the roving frame collect real-time energy consumption data for this node during this hour, including instantaneous power and cumulative energy consumption, to form an energy consumption fluctuation data series. The energy consumption fluctuation coefficient is calculated based on the statistical characteristics of this data series. Common calculation methods include standard deviation and coefficient of variation. The standard deviation reflects the dispersion of energy consumption data relative to the mean, while the coefficient of variation is the ratio of the standard deviation to the mean. This factor is used to eliminate incomparable fluctuations in energy consumption across different process nodes due to differences in mean energy consumption. For example, if the mean energy consumption data for the roving process node is 100 kW·h and the standard deviation is 15 kW·h, the coefficient of variation is 0.15, which serves as the energy consumption fluctuation coefficient for that node.

[0070] When the energy consumption fluctuation coefficient exceeds the first fluctuation threshold, the node is marked as a high-energy-consuming node, and its energy consumption data is extracted to form an energy consumption feature vector. The setting of the first fluctuation threshold requires comprehensive consideration of the process requirements of spinning production and historical energy consumption data. It is usually determined through statistical analysis of the energy consumption fluctuation coefficients of similar process nodes, for example, taking the upper quartile of the energy consumption fluctuation coefficients of all similar nodes as the threshold. If the energy consumption fluctuation coefficient of 0.15 for the roving process node exceeds the preset first fluctuation threshold of 0.12, the node is determined to be a high-energy-consuming node. At this point, the system extracts key characteristic parameters from the energy consumption fluctuation data of the node, including energy consumption peak value, valley value, mean, variance, and the slope of the rising and falling edges of energy consumption. These parameters describe the energy consumption characteristics of the node from different dimensions: the energy consumption peak value reflects the energy consumption demand of the equipment under extreme operating conditions, the valley value indicates the energy consumption level of the equipment when it is lightly loaded or in standby mode, the mean reflects the average energy consumption intensity, the variance indicates the severity of energy consumption fluctuations, and the slope reflects the rate of change of energy consumption. These parameters are arranged in a preset order to form a multi-dimensional energy consumption feature vector, such as [peak value, valley value, mean value, variance, rising slope, falling slope]. This vector comprehensively records the energy consumption characteristics of high-energy-consuming nodes and provides a detailed basis for subsequent energy consumption analysis and optimization strategy formulation.

[0071] When the energy consumption fluctuation coefficient falls below the first fluctuation threshold, the node is marked as a steady-state node. The energy consumption data of its neighboring nodes are weighted and superimposed, reconstructing the superimposed data into an energy consumption feature vector. While the energy consumption fluctuations of a steady-state node are relatively stable, analyzing its data alone may not fully reflect the node's actual energy consumption characteristics within the production process. Therefore, a comprehensive analysis requires combining data from neighboring nodes. The range of neighboring nodes can be determined based on the continuity of the process. For example, for a continuous production process, one or two nodes before and after the node can be selected as neighboring nodes. Weights are assigned based on the principle of "closer distance, greater weight": the closer the neighboring node's temporal position to the steady-state node, the higher the weight assigned to its energy consumption data when superimposed. For example, for a steady-state node in the drawing process, its neighboring nodes are the preceding and following nodes. The weight of the preceding node is set to 0.4, the weight of the current node is set to 0.2, and the weight of the following node is set to 0.4. The system obtains energy consumption data for these three nodes within their respective preset periods: energy consumption sequence A for the preceding node, energy consumption sequence B for the current node, and energy consumption sequence C for the following node. First, each sequence is normalized to eliminate dimensionality effects. Then, they are superimposed according to their weights, resulting in a superimposed energy consumption data sequence D = 0.4 × A + 0.2 × B + 0.4 × C. Next, feature extraction is performed on this superimposed sequence D. The extracted characteristic parameters are similar to those for high-energy-consuming nodes, including mean, variance, peak, and valley values. However, because the data from adjacent nodes are superimposed, these characteristic parameters better reflect the overall energy consumption trends of the steady-state node and its surrounding production links. These characteristic parameters are combined into an energy consumption feature vector, such as [superimposed mean, superimposed variance, superimposed peak, superimposed valley]. This vector not only contains the energy consumption information of the steady-state node itself but also incorporates the influence of adjacent nodes, making the description of energy consumption characteristics more comprehensive and accurate.

[0072] In actual applications, the adjustment of the preset cycle needs to be made according to the actual situation of the production site. For example, when the equipment is maintained or the process parameters are changed, the preset cycle can be shortened appropriately to more intensively monitor the changes in energy consumption data. The calculation method of the energy consumption fluctuation coefficient can also be selected according to the characteristics of the data. For non-normally distributed data, robust statistics such as the absolute mean deviation can be used instead of the standard deviation to reduce the impact of outliers on the fluctuation coefficient. The setting of the first fluctuation threshold is a dynamic optimization process. With the improvement of production processes and the updating of equipment, the threshold needs to be re-evaluated and adjusted regularly to ensure that the division of high-energy-consuming nodes and steady-state nodes always conforms to the actual production situation.

[0073] When superimposing the data of adjacent nodes on a steady-state node, the distribution of weights can also take into account the correlation of the equipment's operating status. For example, if the adjacent node and the steady-state node belong to the continuous operation phase of the same equipment, the weight can be appropriately increased; if they belong to the processes of different equipment, the weight can be reduced accordingly. In addition, normalization processing before data superposition is crucial. The energy consumption data of different process nodes may have different dimensions and numerical ranges. Normalization processing can unify the data to the same scale to avoid distortion of the superposition results due to dimensional differences. The dimension of the reconstructed energy consumption feature vector can be adjusted according to actual needs. If a more detailed analysis of the energy consumption trend is required, time series-related feature parameters such as autocorrelation coefficient and partial autocorrelation coefficient can be added.

[0074] Through the above method, the system can flexibly generate different types of energy consumption feature vectors based on the energy consumption fluctuation characteristics of the process nodes. For high-energy-consuming nodes, their own key energy consumption characteristics are directly extracted, which facilitates targeted analysis of the causes of abnormal energy consumption and the formulation of optimization strategies; for steady-state nodes, by fusing the data of adjacent nodes to generate feature vectors, it is possible to capture the potential energy consumption changes caused by process connection or equipment linkage in the production process, providing a more comprehensive perspective for overall energy consumption optimization. This differentiated processing method not only ensures a rapid response to abnormal energy consumption states, but also enables in-depth exploration of energy consumption characteristics under normal production conditions, improving the comprehensiveness and accuracy of the system's energy consumption analysis of the spinning production process. Example 3:

[0075] The strategy generation module is implemented by separating the equipment energy consumption percentage, idling energy consumption percentage, and load fluctuation parameters from the energy consumption feature vector. Based on these parameters, energy consumption optimization rules for process nodes are generated. In a spinning production system, the energy consumption feature vector contains multidimensional data reflecting the energy consumption characteristics of process nodes. The equipment energy consumption percentage reflects the proportion of energy actually consumed by the production equipment during operation and is an important indicator for measuring equipment energy efficiency. For example, during the carding process of cotton fibers, the proportion of energy directly consumed by the motor and licker-in rollers in the total energy consumption is the equipment energy consumption percentage. The idling energy consumption percentage reflects the energy consumption of the equipment when it is not actively loaded, such as when the equipment is started but not yet in production or during production intervals. For roving frames, the idling state while waiting for sliver feeding or bobbin replacement generates a certain amount of energy consumption. The proportion of this energy consumption in the total energy consumption is the idling energy consumption percentage. The load fluctuation parameter describes the severity of load fluctuations during equipment operation. Excessive load fluctuations not only increase energy consumption but may also affect equipment life and product quality. During the spinning process of the spinning frame, changes in yarn tension or uneven feed amount will cause fluctuations in the motor load. The amplitude and frequency of this fluctuation are the specific manifestations of the load fluctuation parameters.

[0076] The system uses a pre-set algorithm to accurately separate these three key parameters from the energy consumption feature vector. For the equipment energy consumption contribution, the system analyzes the correlation between energy consumption data and equipment operating status, identifies energy consumption directly related to production, and calculates its contribution to total energy consumption. During this identification process, the system considers the equipment's process parameters and operating modes, such as the energy consumption distribution of a carding machine at different speeds, to accurately separate equipment energy consumption from other energy consumption. For the idling energy consumption contribution, the system monitors the equipment's start / stop status and load conditions. When the equipment is operating but without a valid load, it records energy consumption data at this time and calculates the contribution of this energy consumption to the entire production cycle. The system then builds a characteristic model for equipment idling conditions and accurately identifies idling conditions by comparing real-time energy consumption data with the model's characteristics. For the load fluctuation parameter, the system analyzes time series data of equipment loads, calculating statistics such as the standard deviation and coefficient of variation to quantify the extent of load fluctuations. The system also analyzes the frequency characteristics of load fluctuations to identify the primary frequency components, providing a deeper understanding of the causes of load fluctuations.

[0077] Based on the separated equipment energy consumption ratio, idling energy consumption ratio and load fluctuation parameters, the system generates energy consumption optimization rules for process nodes. If the equipment energy consumption ratio is low, it means that the energy efficiency of the equipment has not been fully utilized, and there may be problems such as equipment aging and unreasonable process parameters. In response to this situation, the generated optimization rules may include recommendations for equipment maintenance and adjustment of process parameters to improve equipment operating efficiency. If the idling energy consumption ratio is high, it indicates that the equipment has a lot of ineffective energy consumption. The optimization rules may recommend optimizing the production process and reducing the equipment idling time, such as by reasonably arranging the production plan so that the equipment can be shut down in time during production breaks. If the load fluctuation parameters are large, it means that the equipment is unstable, which may lead to increased energy consumption and reduced product quality. The optimization rules may propose adjusting the equipment's control strategy, such as adopting closed-loop control, adding buffer devices, etc., to reduce load fluctuations.

[0078] If the number of process nodes covered by the current energy consumption optimization rule is less than the preset node threshold, the energy consumption feature vectors of adjacent process nodes are traversed, and the energy consumption indicators not included in the optimization rules of the adjacent nodes are added to the current rule. The preset node threshold is a value set according to the production scale and management requirements. It represents the minimum number of process nodes that the energy consumption optimization rule should cover. If the system finds that the currently generated energy consumption optimization rule only covers a few process nodes and does not reach the preset node threshold, it means that the current rule's coverage is not comprehensive enough and may not meet the overall energy consumption optimization needs. At this time, the system will traverse and analyze the energy consumption feature vectors of adjacent process nodes.

[0079] Adjacent process nodes refer to the previous and next process nodes that are closely connected to the current process node in the production process. In spinning production, the adjacent process nodes of the carding process may be the opening and cleaning process and the drawing process. The system extracts the energy consumption feature vectors of these adjacent nodes and analyzes the energy consumption indicators contained in the corresponding optimization rules. Energy consumption indicators not included in the current rules are added to the current rules to expand the coverage of the rules. If the optimization rules for the adjacent drawing process include the energy consumption indicator "the impact of roller speed fluctuations on energy consumption", and the current optimization rules for the carding process do not involve this indicator, the system will incorporate this indicator into the optimization rules for the carding process. In this way, the energy consumption optimization rules are continuously improved and expanded so that they can more comprehensively cover the energy consumption characteristics of each process node and improve the effect of energy consumption optimization.

[0080] In practical applications, the algorithm for separating key parameters from energy consumption feature vectors needs to be continuously optimized and adjusted to adapt to the characteristics of different types of equipment and production processes. For new spinning equipment, its energy consumption characteristics may be different from those of traditional equipment, and the separation algorithm needs to be improved accordingly. When generating energy consumption optimization rules, the feasibility and operability of the rules need to be considered. The proposed optimization suggestions should be specific and clear, and be feasible under existing production conditions. When traversing the energy consumption feature vectors of adjacent process nodes, an effective data association mechanism needs to be established to ensure that relevant information of adjacent nodes can be accurately obtained. The system also needs to verify the rationality of the newly added energy consumption indicators to avoid introducing irrelevant or erroneous indicators.

[0081] To improve the performance and efficiency of the strategy generation module, the system employs parallel computing technology to simultaneously analyze and process energy consumption feature vectors for multiple process nodes. A knowledge base of energy optimization rules can also be established to store and manage historically generated effective rules for reference and reuse in subsequent optimization processes. This approach enables the rapid generation of more comprehensive and effective energy optimization rules, providing strong support for energy management in spinning production. The system can also be integrated with production management systems to directly convert generated optimization rules into production instructions, achieving automated and intelligent energy optimization.

[0082] When handling complex production scenarios, the strategy generation module can also consider the interrelationships between multiple energy consumption indicators. Equipment energy consumption percentage, idle energy consumption percentage, and load fluctuation parameters may interact with each other. For example, excessive load fluctuations may lead to increased equipment idle energy consumption. The system can use multivariate statistical analysis methods to further explore the inherent connections between these indicators, thereby generating more comprehensive and effective energy optimization rules. The system can also prioritize optimization rules based on different production needs and goals, ensuring that energy consumption is minimized while meeting production requirements.

[0083] Through this implementation, the strategy generation module extracts key parameters from energy consumption feature vectors to generate targeted energy optimization rules. By expanding the scope of these rules, the module ensures their comprehensiveness and effectiveness. This approach not only helps spinning companies accurately identify energy consumption issues but also provides specific optimization recommendations, guiding them in implementing energy management measures, improving energy efficiency, and reducing production costs. Example 4:

[0084] The scheduling optimization module is implemented as follows: the timing factor of the start-stop frequency and the fluctuation factor of the load variation amplitude in the equipment operation mode are obtained. In the spinning production system, the operation mode of the equipment has a significant impact on energy consumption. The timing factor of the start-stop frequency reflects the number of times the equipment starts and stops within a certain time period and the time distribution pattern. For example, during the production process of a single shift, the spinning frame may start and stop multiple times due to shift changes, equipment maintenance, or raw material supply. The time points and frequency of these start-stop operations constitute the main content of the timing factor. By analyzing the historical operation data of the equipment, the periodic pattern of the start-stop frequency can be identified, such as the high frequency of start-stops at the start-up of the machine every morning and the low frequency of start-stops during the lunch break.

[0085] The load fluctuation factor describes the degree of load variation during equipment operation. For roving frames, load fluctuations can occur when the fed sliver is unevenly thick or when spinning process parameters are adjusted. This fluctuation not only affects energy consumption but can also negatively impact product quality. The calculation of the fluctuation factor considers multiple dimensions, including the magnitude, frequency, and duration of load variations. The degree of fluctuation can be quantified by calculating statistical quantities such as the standard deviation and range of the load data. Furthermore, the frequency distribution of the fluctuations can be analyzed to identify the primary frequency components.

[0086] A state transition network is constructed, associated with the timing factor and the fluctuation factor. Based on the transition probabilities of each path in the network, the energy consumption fluctuation range under different scheduling strategies is determined. A state transition network is a mathematical model used to describe the transition relationships between system states. In this embodiment, the timing factor and the fluctuation factor are used as state nodes in the network, and the connections between the nodes represent the transition relationships between states. Each transition relationship is assigned a transition probability, which reflects the likelihood of transitioning from one state to another under specific conditions.

[0087] When building a state transition network, the first step is to determine the network's node set. The timing factor can be divided into different states based on the frequency of start and stop, such as high-frequency start and stop states, medium-frequency start and stop states, and low-frequency start and stop states. The volatility factor can similarly be divided into different states, such as high-volatility states, medium-volatility states, and low-volatility states. These states are combined to form the node set of the state transition network. For example, one node can be represented as a "high-frequency start and stop - high volatility" state, while another node can be represented as a "low-frequency start and stop - low volatility" state.

[0088] Determine the transition relationships and transition probabilities between nodes. The determination of transition relationships must be based on the physical laws of equipment operation and historical data. In spinning production, high-frequency start-stop states often lead to increased load fluctuations, creating the possibility of transitioning from a "high-frequency start-stop - low fluctuation" state to a "high-frequency start-stop - high fluctuation" state. Calculating transition probabilities requires statistical analysis of historical data. By mining a large amount of historical operating data and counting the number of transitions between states under different conditions, the corresponding transition probabilities can be calculated.

[0089] Based on the transition probabilities of each path in the state transition network, the energy consumption fluctuations under different scheduling strategies can be evaluated. If a new scheduling strategy is adopted to reduce the frequency of device starts and stops, the probability of the system transitioning from a high-frequency start-stop state to a low-frequency start-stop state in the state transition network will increase. By analyzing the impact of these state transitions on the energy consumption fluctuations, the effectiveness of the new scheduling strategy can be predicted. Specifically, for each possible state transition path, the product of its transition probability and the energy consumption fluctuation corresponding to that path is calculated. The results for all paths are then added together to obtain the expected energy consumption fluctuation under the scheduling strategy.

[0090] Building a state transition network also involves identifying the periodic characteristics of the timing factor. If the current periodic characteristics fully match the preset production cycle, the timing factor is set as the starting node of the state transition network. In spinning production, the operation of many equipment exhibits distinct periodic characteristics. Spinning frames typically operate in shifts, with the production process and equipment operation modes essentially identical for each shift, resulting in periodic variations in the start and stop frequencies. Spectral analysis or autocorrelation analysis of historical data on the timing factor can identify its periodic characteristics.

[0091] The preset production cycle is a time period set according to the company's production plan and process requirements, such as the length of a shift, the production time of a day, etc. When the identified periodic characteristics of the timing factor completely match the preset production cycle, it means that the change pattern of the equipment's start and stop frequency is highly consistent with the production plan. In this case, setting the timing factor as the starting node of the state transition network can more accurately reflect the changing process of the equipment's operating status. Taking an 8-hour shift as an example, if the periodic characteristics of the timing factor are also 8 hours, and the start and stop frequency distribution within each cycle is similar, then using this timing factor as the starting node can better describe the transition of the equipment state from the start to the end of the shift.

[0092] The transition correlation between the timing factor and the fluctuation factor is calculated, and the intermediate and terminal nodes of the state transition network are generated in descending order of correlation. The transition correlation is a measure of the degree of mutual influence between the timing factor and the fluctuation factor. In spinning production, changes in the start-stop frequency often cause changes in the load fluctuation amplitude, and there is a certain correlation between the two. Statistical methods such as correlation analysis and Granger causality test can be used to calculate the transition correlation.

[0093] By calculating the correlation between the timing factor and the fluctuation factor in different states, the strength of the correlation between them is determined. If the correlation between the high-frequency start-stop state and the high-load fluctuation state is high, it indicates that frequent start-stop operations are likely to increase the amplitude of load fluctuations. The intermediate nodes and terminal nodes of the state transition network are generated in descending order of correlation. State combinations with high correlation are prioritized as intermediate nodes, reflecting the main change path of the device's operating state; state combinations with low correlation are designated as terminal nodes, indicating the stable or abnormal state that the device may enter.

[0094] The state of the terminal node is backtracked. When the connection degree of the terminal node falls below a preset connection threshold, it is output as the final path of the state transition network. State backtracking involves tracing back along the transition path from the terminal node to the starting node to verify the rationality and validity of the entire path. In a state transition network, some terminal nodes may have a low connection degree with the starting node due to data noise or accidental factors. These nodes are not very useful for predicting the amplitude of energy consumption fluctuations.

[0095] The preset correlation threshold is a pre-set value used to determine whether the correlation between the ending node and the starting node is sufficiently strong. When the correlation of the ending node is below this threshold, it indicates that the state transition path represented by the node is unlikely to occur in actual production or has a minimal impact on energy consumption fluctuations. Using such an ending node as the final path of the state transition network can eliminate some unreasonable paths and improve the accuracy of energy consumption fluctuation prediction.

[0096] Calculating energy consumption fluctuations involves calculating the mean and range of the timing factor for each terminal node in the state transition network and calculating the global variance of the factors across all nodes. The mean timing factor reflects the average level of device start / stop frequency in the state represented by the terminal node. For a terminal node representing "low frequency start / stop - low fluctuation," calculating the mean timing factor in historical data yields the average device start / stop frequency in that state. The range of the fluctuation factor represents the difference between the maximum and minimum load variations in that state, describing the range of load fluctuations.

[0097] The global variance is the combined variance of the timing factor and fluctuation factor across all nodes. It reflects the degree of dispersion of device operating states across the entire state transition network. By calculating the global variance, we can understand the operational stability of devices under different states. A large global variance indicates that device operating states are dispersed, and energy consumption fluctuations are more likely. Conversely, a small global variance indicates that device operating states are relatively concentrated, and energy consumption fluctuations are relatively small.

[0098] The timing fluctuation coefficient is calculated by subtracting the mean timing factor of a single terminating node from the mean timing factor of its adjacent nodes and dividing the difference by the global variance. The ratio of the fluctuation factor range to the global variance is also calculated, and the weighted sum of the two is used as the energy consumption fluctuation amplitude for that node. The timing fluctuation coefficient measures the degree of variation in the timing factor of a single terminating node relative to that of its adjacent nodes. It reflects the fluctuation in the device's start-stop frequency between different states. The ratio of the fluctuation factor range to the global variance indicates the relative magnitude of load fluctuations within the global scope.

[0099] These two metrics are combined into a comprehensive energy consumption fluctuation index through a weighted summation method. The choice of weights needs to be adjusted based on the specific production situation and equipment characteristics. If the equipment's start-stop frequency has a significant impact on energy consumption, the weight of the time series fluctuation coefficient can be appropriately increased. If the load fluctuation amplitude has a more significant impact on energy consumption, the weight of the ratio of the fluctuation factor range to the global variance can be increased. The energy consumption fluctuation amplitude calculated in this way can fully account for the impact of equipment start-stop frequency and load fluctuation amplitude on energy consumption, providing an accurate basis for optimizing scheduling strategies.

[0100] In practical applications, data quality and quantity must be considered when building state transition networks. The accuracy and completeness of historical data directly impact the reliability of the state transition network. Noisy or missing data can lead to inaccurate state partitioning and biased transition probability calculations. Therefore, data preprocessing, including data cleaning and interpolation, is necessary before network construction.

[0101] The complexity of the state transition network also needs to be properly controlled. Excessive nodes or overly complex transition relationships increase the computational effort and analysis difficulty, and may also lead to overfitting. In practical applications, the state transition network can be appropriately simplified based on the characteristics of the equipment and the complexity of the production process, retaining the key states and transition relationships.

[0102] When calculating energy consumption fluctuations, weight selection is a key issue. Different equipment and production processes may have varying sensitivities to start / stop frequency and load fluctuations, so adjustments need to be made based on specific circumstances. Appropriate weights can be determined through expert experience or analysis of historical data. Alternatively, adaptive weighting can be employed to dynamically adjust weights based on real-time production data to improve the accuracy of energy consumption fluctuation calculations.

[0103] Through this implementation, the scheduling optimization module can accurately identify key factors in equipment operating modes, construct an effective state transition network, and use this network to calculate energy consumption fluctuations under different scheduling strategies. This approach not only helps companies predict energy consumption trends but also provides a scientific basis for optimizing scheduling strategies, thereby achieving the goals of reducing energy consumption and improving production efficiency. Example 5:

[0104] The optimal energy consumption threshold is derived as follows: the fluctuation pattern in historical data that most closely matches the current energy consumption fluctuation amplitude is extracted and the Manhattan distance between the two in a time-series distribution is calculated as the first benchmark reference value. In spinning production systems, historical data stores a large number of energy consumption fluctuation patterns over different time periods. These patterns reflect the energy consumption characteristics of equipment under various operating conditions. The current energy consumption fluctuation amplitude refers to the degree to which the energy consumption of a piece of equipment or process changes over time during the current production process. By analyzing the fluctuation patterns in the historical data and identifying the pattern that most closely matches the current fluctuation amplitude, a reference can be provided for determining the optimal energy consumption threshold.

[0105] Manhattan distance is an indicator used to measure the distance between two vectors in multidimensional space. In this embodiment, it is used to measure the difference in time series distribution between the current fluctuation pattern and the historical fluctuation pattern. Specifically, the current fluctuation pattern and the historical fluctuation pattern are represented as time series vectors, and the elements of each vector correspond to the energy consumption values ​​at different time points. The sum of the absolute values ​​of the energy consumption value differences of the two vectors at each time point is calculated, and the result is the Manhattan distance. For example, if the energy consumption values ​​of the current fluctuation pattern at three consecutive time points are 、 、 , the energy consumption values ​​of the historical fluctuation patterns at the same time point are 、 、 , then their Manhattan distance is:

[0106]

[0107] The smaller the distance value is, the more similar the two fluctuation patterns are in their time series distribution.

[0108] The difference in the number of extreme points between the current energy consumption fluctuation amplitude and the historical fluctuation pattern is calculated and used as a second benchmark reference value. Extreme points refer to the maximum and minimum points in the energy consumption time series, reflecting the severity and trend of energy consumption fluctuations. The number of extreme points in the current energy consumption fluctuation amplitude is compared with the number of extreme points in the historical fluctuation pattern. The smaller the difference, the more similar the changing trends of the two fluctuation patterns. In spinning production, certain process adjustments or equipment failures may cause changes in the number of extreme points in energy consumption fluctuations. Therefore, by comparing the number of extreme points, we can more accurately determine the similarity between the current fluctuation pattern and the historical pattern.

[0109] Based on the linear combination of the first and second reference values, the optimal energy consumption threshold in the preset energy consumption threshold table is matched. The preset energy consumption threshold table is pre-established based on a large amount of historical data and production experience. It records the optimal energy consumption threshold corresponding to different reference value combinations. The first and second reference values ​​are linearly combined according to certain weights to obtain a comprehensive reference value. For example, the comprehensive reference value The calculation formula is:

[0110]

[0111] in: is the first benchmark reference value, is the second benchmark reference value, and is the weight coefficient and satisfies .

[0112] By adjusting the weight coefficient, the importance of the two benchmark reference values ​​can be weighed according to actual production conditions. Based on the calculated comprehensive benchmark value, the closest corresponding value is found in the preset energy consumption threshold table. This value is the optimal energy consumption threshold for the current production situation.

[0113] The implementation of the decision output module involves dividing the energy consumption fluctuation direction of each node in the process energy consumption deviation sequence into positive and negative fluctuation intervals. The process energy consumption deviation sequence is the difference between the actual energy consumption value of each process node and the optimal energy consumption threshold. The energy consumption fluctuation direction refers to the positive or negative sign of the deviation value. A positive deviation indicates that the actual energy consumption of the node is higher than the optimal energy consumption threshold, indicating excessive energy consumption. A negative deviation indicates that the actual energy consumption is lower than the optimal energy consumption threshold, indicating possible problems such as inefficient equipment operation or insufficient production.

[0114] By traversing the process energy consumption deviation sequence, nodes with consecutive positive deviation values ​​are divided into positive fluctuation intervals, and nodes with consecutive negative deviation values ​​are divided into negative fluctuation intervals. In a spinning production process, if the energy consumption deviation values ​​of the first three process nodes are all positive, the deviation values ​​of the next two nodes are negative, and the deviation values ​​of the next four nodes are positive again, the first three nodes can be divided into a positive fluctuation interval, the middle two nodes into a negative fluctuation interval, and the last four nodes into another positive fluctuation interval. This division method helps to intuitively identify abnormal energy consumption areas and provides a basis for subsequent analysis and processing.

[0115] The convergence rate of energy consumption deviation within the positive fluctuation range and the diffusion rate of energy consumption deviation within the negative fluctuation range are extracted and weighted together according to the production weight of the process node to generate the core adjustment parameters of the energy consumption optimization decision-making plan. The convergence rate refers to the rate at which the energy consumption deviation value decreases over time or as the process progresses within the positive fluctuation range, reflecting the system's ability to regulate excessive energy consumption. The diffusion rate refers to the rate at which the absolute value of the energy consumption deviation value increases over time or as the process progresses within the negative fluctuation range, indicating the degree of deterioration of the energy consumption shortage.

[0116] When calculating the convergence rate and diffusion rate, linear regression or other methods can be used to fit the energy consumption deviation values ​​within the fluctuation range. This can produce a trend line showing the deviation values ​​over time or process steps. The slope of the trend line represents the corresponding rate. For positive fluctuation ranges, if the trend line slope is negative and the absolute value is large, it indicates a rapid convergence rate and the system can quickly reduce the degree of excessive energy consumption. For negative fluctuation ranges, if the trend line slope is negative and the absolute value is large, it indicates a rapid diffusion rate and a rapidly deteriorating energy consumption situation.

[0117] The production weight of a process node reflects its importance within the overall production process and its impact on final product quality. Key process nodes have higher production weights, while non-key process nodes have lower production weights. The convergence rate and diffusion rate are weighted and fused according to the production weight of each process node to produce a comprehensive adjustment parameter. For example, for a positive fluctuation interval, the convergence rate of each node is multiplied by the production weight, and the results for all nodes are summed to obtain the comprehensive convergence index for that interval. Similarly, for negative fluctuation intervals, a comprehensive diffusion index is calculated. These two indices are further fused to generate the final core adjustment parameter.

[0118] This core adjustment parameter comprehensively considers the severity and changing trends of both energy consumption exceeding and energy consumption insufficient, as well as the importance of each process node in production. Based on this parameter, targeted energy optimization decisions can be formulated, such as adjusting equipment operating parameters, optimizing production processes, and reallocating resources. In this way, the decision output module transforms complex energy consumption deviation information into specific, actionable optimization recommendations, providing strong support for enterprise energy management.

[0119] In practical applications, extracting similar fluctuation patterns from historical data requires effective organization and indexing of the data. Clustering algorithms can be used to categorize historical fluctuation patterns and establish a classification index to quickly identify the pattern closest to the current fluctuation amplitude. To improve computational efficiency when calculating the Manhattan distance, dimensionality reduction techniques can be used to preprocess the time series vectors and reduce the computational effort.

[0120] When counting extreme points, it's necessary to determine an appropriate extreme point detection method. A simple approach is to set a threshold. When the energy consumption value changes beyond this threshold, an extreme point is identified. More complex algorithms, such as wavelet transforms and sliding windows, can also be used to improve the accuracy of extreme point detection. The establishment of a preset energy consumption threshold table requires careful consideration of various possible production scenarios and equipment states. By analyzing historical data and using machine learning algorithms, the inherent relationship between baseline reference values ​​and optimal energy consumption thresholds can be identified, thereby establishing an accurate and comprehensive threshold table.

[0121] When dividing positive and negative fluctuation intervals, it is important to handle boundary conditions. When adjacent positive and negative deviation value nodes appear alternately, it is necessary to determine whether to divide them into independent intervals or merge them into a single interval based on the specific situation. You can set a minimum interval length threshold; when the interval length is less than this threshold, it is merged with the adjacent interval.

[0122] While linear regression is simple and easy to use when calculating convergence and diffusion rates, it may not be accurate enough for complex nonlinear fluctuation patterns. In such cases, methods such as nonlinear regression or time series analysis can be considered to improve the accuracy of rate calculations. Determining the production weight of a process node requires comprehensive consideration of multiple factors, such as its position in the production process, its impact on product quality, and its contribution to energy consumption. Multi-criteria decision-making methods such as the Analytic Hierarchy Process (AHP) and the Delphi method can be used to scientifically and rationally determine the production weight of each process node.

[0123] Through the above implementation, the system can accurately derive the optimal energy consumption threshold and generate effective energy consumption optimization decision-making plans based on the process energy consumption deviation sequence. This approach not only considers the temporal characteristics and extreme value characteristics of energy consumption fluctuations, but also incorporates the production importance of process nodes, making optimization decisions more scientific and reasonable. In actual production, this method can help companies promptly identify energy consumption anomalies, formulate targeted improvement measures, improve energy utilization efficiency, reduce production costs, and achieve sustainable development.

[0124] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0125] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A multi-process spinning energy consumption optimization decision system, characterized in that: include: The data acquisition module is used to obtain equipment parameters and energy consumption data in the spinning production process and set the energy consumption monitoring interval corresponding to the process; The energy consumption analysis module is used to divide the energy consumption monitoring interval into multiple process nodes, perform feature analysis on the energy consumption data of each process node, and generate the energy consumption feature vector corresponding to the process node; The strategy generation module is used to extract key energy consumption indicators from the energy consumption feature vector, establish energy consumption optimization rules associated with process nodes, and obtain process control parameters corresponding to the rules; The scheduling optimization module is used to identify the equipment operation mode in the process control parameters, dynamically schedule key energy consumption indicators according to the operation mode, and calculate the energy consumption fluctuation range of each process node under different scheduling strategies; The benchmark comparison module is used to derive the optimal energy consumption threshold according to the energy consumption fluctuation amplitude, and generate the process energy consumption deviation sequence by comparing the current process energy consumption value with the optimal energy consumption threshold; The decision output module is used to analyze the process energy consumption deviation sequence and integrate the process energy consumption deviation sequence into an energy consumption optimization decision plan based on the energy consumption fluctuation trend of the process nodes; The implementation of the energy consumption analysis module includes: building a process feature library corresponding to the process node, the process feature library contains process parameter vectors mapped with equipment parameters and energy consumption data; Match similar processes to process parameter vectors, and divide process parameter vectors into process cluster groups based on the matching results; extract the distribution center point of energy consumption data from the process cluster group, and set the distribution center point as the process node; The process clustering groups that divide the process parameter vectors also include: According to the equipment type and production stage in the process parameter vector, the equipment speed, yarn count, and ambient temperature and humidity parameters are extracted, and the process feature labels are generated based on these parameters; The process feature labels are associated with the process parameter vectors. By calculating the process similarity between the feature labels, the process parameter vectors with similarity higher than the preset process threshold are screened to form a process cluster group.

2. A multi-process spinning energy consumption optimization decision system according to claim 1, characterized in that: The implementation methods of generating the energy consumption characteristic vector corresponding to the process node include: For each process node, according to the node's time sequence position in the energy consumption monitoring interval, the node's energy consumption fluctuation data within the preset period is obtained, and the node's energy consumption fluctuation coefficient is calculated; When the energy consumption fluctuation coefficient exceeds the first fluctuation threshold, the node is marked as a high-energy-consuming node, and its energy consumption data is extracted to form an energy consumption feature vector; when the energy consumption fluctuation coefficient is lower than the first fluctuation threshold, the node is marked as a steady-state node, and the energy consumption data of the node's adjacent nodes are weighted and superimposed, and the superimposed data is reconstructed into an energy consumption feature vector.

3. The multi-process spinning energy consumption optimization decision system according to claim 1, characterized in that: The implementation of the strategy generation module includes: Separate the equipment energy consumption proportion, idling energy consumption proportion and load fluctuation parameters from the energy consumption feature vector, and generate energy consumption optimization rules for process nodes based on the above parameters; If the number of process nodes covered by the current energy consumption optimization rule is less than the preset node threshold, the energy consumption feature vectors of adjacent process nodes are traversed, and the energy consumption indicators not included in the optimization rules of the adjacent nodes are added to the current rule.

4. The multi-process spinning energy consumption optimization decision system according to claim 1, characterized in that: The implementation of the scheduling optimization module includes: obtaining the timing factor of the start and stop frequency in the equipment operation mode and the fluctuation factor of the load change amplitude; A state transition network associated with timing factors and fluctuation factors is constructed, and the energy consumption fluctuation amplitude under different scheduling strategies is determined based on the transition probability of each path in the network.

5. A multi-process spinning energy consumption optimization decision system according to claim 4, characterized in that: Building a state transfer network also includes: Identify the periodic characteristics of the timing factor. If the current periodic characteristics completely match the preset production cycle, set the timing factor as the starting node of the state transition network. Calculate the transfer correlation between the timing factor and the volatility factor, and generate the intermediate nodes and terminal nodes of the state transition network in descending order of correlation; The state of the terminal node is backtracked. When the correlation degree of the terminal node is lower than the preset correlation threshold, it is output as the final path of the state transition network.

6. A multi-process spinning energy consumption optimization decision system according to claim 5, characterized in that: The implementation methods for calculating the energy consumption fluctuation range include: Statistically calculate the mean value and range of the timing factor of each terminal node in the state transition network, and calculate the global variance of the factors of all nodes; The difference between the mean of the timing factor of a single termination node and the mean of the timing factor of the adjacent nodes is divided by the global variance to obtain the timing fluctuation coefficient; at the same time, the ratio of the fluctuation factor range to the global variance is calculated, and the weighted sum of the two is taken as the energy consumption fluctuation amplitude of the node.

7. The multi-process spinning energy consumption optimization decision system according to claim 1, characterized in that: The implementation methods for deriving the optimal energy consumption threshold include: Extract the fluctuation pattern in the historical data that is closest to the current energy consumption fluctuation amplitude, and calculate the Manhattan distance between the two in the time series distribution as the first benchmark reference value; Count the difference in the number of extreme points between the current energy consumption fluctuation amplitude and the historical fluctuation pattern, and use the difference as the second benchmark reference value; Based on a linear combination of the first benchmark reference value and the second benchmark reference value, an optimal energy consumption threshold in a preset energy consumption threshold table is matched.

8. The multi-process spinning energy consumption optimization decision system according to claim 1, characterized in that: The implementation method of the decision output module includes: dividing the energy consumption fluctuation direction of each node in the process energy consumption deviation sequence into positive fluctuation intervals and negative fluctuation intervals; The convergence rate of energy consumption deviation in the positive fluctuation range and the diffusion rate of energy consumption deviation in the negative fluctuation range are extracted, and the two are weighted and fused according to the production weights of the process nodes to generate the core adjustment parameters of the energy consumption optimization decision-making plan.

Images