Efficiency evaluation system and method for biomass-to-natural gas based on multi-source data fusion

Abstract

The application discloses a biomass-to-natural gas efficiency evaluation system and method based on multi-source data fusion, and relates to the technical field of biomass energy sources.Core indexes are extracted from a biomass-to-natural gas process, samples are collected to construct a set, data preprocessing is completed through abnormal value elimination, missing value processing and standardization; the Euclidean distance of the samples is calculated, the initial clustering center is determined and updated iteratively, and the optimal clustering number is screened in combination with a K-SSE curve and an average profile coefficient; the clustering class features are quantified, the index weight is distributed, a three-level scoring model is constructed to divide low, medium and high efficiency grades; subsequently, the index fluctuation range and the safety threshold interval of each grade are set, data is collected, and the condition is met to trigger an early warning; the low efficiency reason is analyzed, a process parameter optimization scheme is formulated, and the core indexes of the low and medium efficiency working conditions are improved to the high efficiency grade G1%-G2%, so that the working condition optimization is realized, and the gas production efficiency and stability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of biomass energy, and specifically to a biomass-to-natural gas efficiency evaluation system and method based on multi-source data fusion. Background Technique

[0002] At present, the biomass-to-natural gas technology, as an important direction of renewable energy utilization, has significant value in alleviating energy shortages and reducing environmental pollution. However, its production process involves multi-dimensional variables such as raw material characteristics, reaction parameters, and equipment status. The core indicators are prone to fluctuations due to the influence of complex working conditions, resulting in a lack of systematicness and accuracy in the evaluation of gas production efficiency; Existing evaluation methods mostly rely on single indicators or empirical judgments, and do not achieve the effective fusion of multi-source data, making it difficult to comprehensively reflect the actual operating status of the process. At the same time, there is a lack of a scientific clustering classification and efficiency level划分 mechanism, and it is impossible to accurately define the boundaries of low-efficiency, medium-efficiency, and high-efficiency working conditions, resulting in模糊定位 of the reasons for inefficiency; In addition, traditional methods do not establish a dynamic threshold warning system, making it difficult to monitor异常波动 of core indicators in real time, which is likely to cause production stability problems. Moreover, the optimization方案 for low-efficiency working conditions lacks data support and is mostly general adjustments, unable to achieve targeted improvement of process parameters, ultimately restricting the efficiency improvement and规模化稳定运行 of the biomass-to-natural gas process. Summary of the Invention

[0003] The purpose of the present invention is to provide a biomass-to-natural gas efficiency evaluation system and method based on multi-source data fusion to solve the problems proposed in the existing technology.

[0004] To achieve the above purpose, the present invention provides the following technical solution: A biomass-to-natural gas efficiency evaluation method based on multi-source data fusion, the method comprising the following steps: Step S1: Extract core indicators from the biomass-to-natural gas process, collect n core indicator samples, construct a core indicator set, and perform data cleaning and standardization processing on the samples in the core indicator set; Step S1-1: Extract core indicators from the biomass-to-natural gas process, collect n core indicator samples, and construct a core indicator set; Step S1-2: For each sample data of the core indicators in the core indicator set, after sorting the sample data of the core indicators in ascending order, use the box plot method to剔除异常值; Step S1-3: For missing values, calculate the sample data missing rate. When the missing rate < x1, fill it with the industry average value of this indicator. When the missing rate is in the interval [x1, x2], fill it with the average value of the sample data of similar core indicators. When the missing rate > x2,剔除 the sample data, where x1 and x2 are missing rate thresholds; Steps S1-4: Standardize the samples using the Z-score standardization method after data cleaning.

[0005] Core indicators are extracted and aggregated. After outlier removal, missing value classification and standardization, data noise is eliminated, gaps are filled and dimensional differences are eliminated, providing high-quality data for subsequent analysis and avoiding evaluation bias caused by poor-quality data.

[0006] Step S2: Calculate the Euclidean distance between the samples, select cluster centers, calculate the shortest distance from the remaining samples to the cluster centers, iteratively calculate the initial cluster centers, perform sample classification and iterative center updates, and determine the optimal number of clusters. Step S2-1: Calculate the Euclidean distance between the samples, specifically: ; In the formula, Let m be the Euclidean distance between the i-th sample and the k-th cluster center, and m be the number of indicators. Let j be the standardized value of the i-th sample and the j-th indicator. For the k-th cluster center, the standardized value of the j-th index; Step S2-2: Randomly select one sample as the first cluster center, calculate the shortest distance from all samples to the randomly selected first cluster center, and select the next cluster center based on the probability that the sample with the greater distance is selected as the cluster center. Repeat this process to select K initial cluster centers. Step S2-3: For each sample, calculate the Euclidean distance from the sample to the K initial cluster centers, and assign the sample to the cluster with the smallest distance. Specifically:

[0007] In the formula, This indicates the final cluster category to which the i-th sample belongs. In mathematics, k is the operator that takes the parameter corresponding to the minimum value. k represents the range of values ​​from 1 to k. Let be the Euclidean distance between the i-th sample and the k-th cluster center. Step S2-4: For each cluster k, calculate the standardized mean of all samples in cluster k on the core metric, and use it as the new cluster center. Specifically: ; In the formula, This represents the standardized cluster center value of the k-th cluster on the j-th metric. n is the reciprocal of the number of samples in this class. k This represents the total number of samples contained in the k-th cluster. This represents the summation of the standardized values ​​of all samples within the k-th cluster on the j-th index; Set a threshold for the change in cluster centers and a maximum number of iterations, and set an iteration termination condition, specifically as follows: When the change in cluster centers is less than the threshold for the change in cluster centers; The number of iterations has reached its maximum value. After d consecutive iterations, the class assignment of the sample remains unchanged; The iteration stops when any one of the iteration stopping conditions is met. Step S2-5: Set the range of values ​​for K. Specifically, calculate the sum of squared Euclidean distances (SSE) from all samples to their respective cluster centers for each K, plot the K-SSE curve, and set the inflection point of the curve as the optimal value of K. Calculate the sum of squared Euclidean distances (SSE) from all samples to their respective cluster centers for each K, plot the K-SSE curve, and set the inflection point of the curve as the optimal value of the corresponding K; Step S2-6: For each candidate K, calculate the silhouette coefficient of all samples, then calculate the average silhouette coefficient, and select the K with the largest average silhouette coefficient as the optimal number of clusters.

[0008] The Euclidean distance of the samples is calculated iteratively to determine the cluster centers. The optimal number of clusters is selected by combining the K-SSE curve and the average profile coefficient, so as to achieve scientific classification of samples, solve the problem of traditional classification relying on experience, and provide a precise basis for efficiency level classification.

[0009] Step S3: Based on the clustering results corresponding to the optimal number of clusters, quantify the category features and classify the efficiency levels; Step S3-1: Set the core statistics as category mean, category standard deviation, and category percentage. For the clustering results corresponding to the optimal K, calculate the core statistics of each category of samples on each indicator to form the category statistics. Step S3-2: Classify the core indicators into three dimensions: efficiency correlation dimension, stability dimension, and resource consumption dimension. Step S3-3: Professionals assign weights to the core indicators of the three dimensions based on the actual business scenario, construct a three-level scoring model, give a comprehensive score to each clustering result, and then use the quantile method to classify the efficiency level into low efficiency, medium efficiency and high efficiency.

[0010] By quantifying clustering features and allocating indicator weights according to three dimensions to construct a scoring model, efficiency levels are divided using the quantile method, clearly defining the boundaries of operating conditions, and shifting efficiency assessment from fuzzy judgment to quantitative grading, thus clarifying the positioning of operating conditions.

[0011] Step S4: For the divided efficiency levels, calculate the fluctuation range of the core indicators, set a safety threshold range based on the fluctuation range of the core indicators, and trigger an early warning when the core indicators exceed the safety threshold range. Step S4-1: Select the core indicators corresponding to each efficiency level from the core indicator set and construct the associated dataset of "efficiency level-core indicator-sample data". Step S4-2: For each core indicator in each efficiency level, sort all the raw data of a core indicator under a certain efficiency level in ascending order, and calculate the 5th percentile Q5 and 95th percentile Q of that indicator data. 95 , with [Q5,Q 95 The normal fluctuation range of this core indicator under this efficiency level is recorded in the "Level-Indicator-Fluctuation Range" correspondence table. Step S4-3: Set the fluctuation range of the high efficiency level as the target threshold range of the optimal operating state. The target threshold range is the high efficiency threshold range. Based on the high efficiency level threshold, the fluctuation of x% is used as the medium efficiency threshold range for optimizing from medium efficiency to high efficiency. The fluctuation range of the low efficiency index is taken as the low efficiency threshold range of the low efficiency condition. The efficiency level threshold range is the high efficiency threshold range, the medium efficiency threshold range, and the low efficiency threshold range. Step S4-4: Extract the extreme boundaries of each core indicator from all valid samples, calculate the 99% confidence interval as the safety threshold interval for all operating conditions, set the collection period to 1 hour, collect the actual operating values ​​of the core indicators during production, and set the early warning trigger conditions: The actual values ​​of the core indicators do not fall within the threshold range corresponding to the current efficiency level; The actual values ​​of key indicators have exceeded the safe threshold range; The change in the indicator within 1 hour exceeded b% of the high-efficiency level threshold range. An alert is triggered when the actual operating value meets any of the above conditions.

[0012] Set fluctuation ranges and safety thresholds for each level of indicators, monitor and issue early warnings on an hourly cycle, detect abnormal indicators in real time, avoid production instability, and ensure the continuous and reliable operation of the gas production process.

[0013] Step S5: Based on the clustering results, identify the causes of inefficiency, specify targeted process parameter optimization schemes, and adjust inefficient and medium-efficiency conditions to high-efficiency conditions.

[0014] Step S5-1: Extract the cluster categories of low efficiency levels, summarize the original sample data of all core indicators under the category, the standardized data and category statistics, and construct the low efficiency sample feature dataset; Step S5-2: Conduct statistical analysis on the low-efficiency sample feature dataset to identify key features, where the key features are abnormal mean features and high-fluctuation features. The abnormal mean features are obtained by calculating the deviation rate of the mean value of each core index of the low-efficiency category from the mean value of the full sample, and the high-fluctuation features are obtained by screening the indicators in the low-efficiency category with a standard deviation > 1.5 times the standard deviation of the full sample as high-fluctuation features; Step S5-3: Based on the clustering categories of the high-efficiency level, calculate the absolute difference value of the low-efficiency and high-efficiency categories on the same index, and screen the indicators with a relative difference rate > 15% as potential offset cause indicators; Step S5-4: Calculate the correlation coefficient r between the candidate cause indicators and the core efficiency indicators, and set relevant thresholds r1 and r2 according to special scenarios. When |r| > r2, determine that the candidate cause indicator is the core cause of low efficiency; when r1 ≤ |r| ≤ r2, determine that the candidate cause indicator is the secondary cause of low efficiency; when |r| < r1, exclude the candidate cause; Step S5-5: Adjust the working conditions by raising the core indicators of low efficiency and medium efficiency to G1% - G2% of the core indicators of high efficiency level.

[0015] Construct a low-efficiency sample set, analyze the differences and correlation coefficients to identify the reasons for low efficiency, formulate a plan to raise the low and medium efficiency indicators to G1% - G2% of the high efficiency level, accurately locate the root cause and optimize it, and improve the gas production efficiency.

[0016] The system includes a core index processing module, a clustering analysis and optimal clustering number determination module, an efficiency level division module, a threshold setting and warning module, and a low-efficiency cause analysis and optimization module; The core index processing module is responsible for extracting core indicators from the biomass-to-natural gas process, completing sample collection, data cleaning and standardization, and laying a data foundation for subsequent analysis; [[ID=!5]] The clustering analysis and optimal clustering number determination module is responsible for calculating the Euclidean distance and iteratively updating the clustering center to screen out the optimal clustering number and achieve reasonable classification of samples; The efficiency level division module is responsible for quantifying the characteristics of clustering categories, constructing a scoring model, and dividing samples into different efficiency levels; The threshold setting and warning module is responsible for determining the fluctuation range and safety threshold interval of the core indicators of each efficiency level, setting warning conditions and triggering warnings; The low-efficiency cause analysis and optimization module is responsible for identifying the core cause of low-efficiency working conditions, formulating a process parameter optimization plan, and improving the operating efficiency.

[0017] The core index processing module includes an index collection and set construction unit and a data cleaning and standardization unit; The indicator collection and set construction unit is responsible for extracting core indicators, collecting n core indicator samples, and constructing a complete set of core indicators. The data cleaning and standardization unit is responsible for removing outliers, filling in or removing missing values, and using the Z-score standardization method to standardize the cleaned data. The cluster analysis and optimal cluster number determination module includes a cluster center iteration unit and an optimal cluster number screening unit; The unit is responsible for calculating the Euclidean distance of the samples, selecting the initial cluster centers, updating the cluster centers iteratively, and terminating the iteration when the stopping condition is met. The unit is responsible for setting the range of K values ​​and determining the optimal number of clusters by combining the inflection point of the K-SSE curve and the average profile coefficient.

[0018] The efficiency level classification module includes a category feature quantification unit and an efficiency level evaluation unit; The category feature quantification unit is responsible for calculating the core statistics of each class of samples, dividing the core indicators into three dimensions, and clarifying the category features; The efficiency level assessment unit is responsible for assigning weights to the three-dimensional indicators, constructing a three-level scoring model, and using the quantile method to divide the efficiency levels into low, medium, and high categories. The threshold setting and early warning module includes a threshold range determination unit and an early warning triggering unit; The threshold range determination unit is responsible for constructing the associated dataset, calculating the quantiles of each indicator to determine the fluctuation range, and setting the threshold ranges for different efficiency levels and the safety threshold range for all operating conditions. The early warning triggering unit is responsible for acquiring actual operating values ​​at a 1-hour collection cycle. An early warning is triggered when any of the following conditions are met: the indicator exceeds the threshold, the safe range is breached, or there is a large fluctuation in a short period of time.

[0019] The inefficiency cause analysis and optimization module includes an inefficiency cause identification unit and an operating condition optimization and adjustment unit; The inefficiency cause identification unit is responsible for constructing an inefficiency sample feature dataset and determining the core and secondary causes of inefficiency through difference analysis and correlation coefficient calculation. The operating condition optimization and adjustment unit is responsible for improving the core indicators of low and medium efficiency operating conditions to G1%-G2% of high efficiency level indicators, thereby achieving operating condition optimization.

[0020] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention achieves accurate evaluation of gasification efficiency by combining multi-source data preprocessing and scientific cluster analysis with a scoring model to classify efficiency levels, thus solving the problem of one-sidedness in traditional evaluation and providing a reliable basis for optimization.

[0021] 2. By setting dynamic thresholds and monitoring on a one-hour cycle, the present invention triggers an early warning when an index is abnormal, establishes a real-time early warning mechanism, effectively ensures stable production, and reduces operation risks.

[0022] 3. By analyzing the reasons for inefficiency and formulating targeted solutions, the present invention improves low and medium efficiency indexes to high efficiency levels G1%-G2%, realizes working condition optimization, and improves the overall gas production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a schematic flow chart of the biomass-to-natural gas efficiency evaluation method based on multi-source data fusion according to the present invention; Figure 2 is a schematic structural diagram of the biomass-to-natural gas efficiency evaluation system based on multi-source data fusion according to the present invention. SPECIFIC EMBODIMENTS

[0024] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] Embodiment 1: As Figures 1-2 shown, the present invention provides a technical solution, a biomass-to-natural gas efficiency evaluation method based on multi-source data fusion, and the method includes the following steps: Step S1: Extract core indexes from the biomass-to-natural gas process, collect n core index samples, construct a core index set, and perform data cleaning and standardization processing on the samples in the core index set; Step S1-1: Extract core indexes from the biomass-to-natural gas process, collect n core index samples, and construct a core index set; Step S1-2: For each sample data of the core indexes in the core index set, after ascending sorting of the sample data of the core indexes, use the box plot method to eliminate outliers; Step S1-3: For missing values, calculate the missing rate of the sample data. When the missing rate < x1, use the industry average value of this index for filling. When the missing rate is in the interval [x1, x2], use the average value of the sample data of the same type of core index for filling. When the missing rate > x2, eliminate the sample data, where x1 and x2 are missing rate thresholds; Step S1-4: For the data after data cleaning, use the Z-score standardization method to standardize the samples.

[0026] Core indicators are extracted and aggregated. After outlier removal, missing value classification and standardization, data noise is eliminated, gaps are filled and dimensional differences are eliminated, providing high-quality data for subsequent analysis and avoiding evaluation bias caused by poor-quality data.

[0027] Step S2: Calculate the Euclidean distance between the samples, select cluster centers, calculate the shortest distance from the remaining samples to the cluster centers, iteratively calculate the initial cluster centers, perform sample classification and iterative center updates, and determine the optimal number of clusters. Step S2-1: Calculate the Euclidean distance between the samples, specifically: ; In the formula, Let m be the Euclidean distance between the i-th sample and the k-th cluster center, and m be the number of indicators. Let j be the standardized value of the i-th sample and the j-th indicator. For the k-th cluster center, the standardized value of the j-th index; Step S2-2: Randomly select one sample as the first cluster center, calculate the shortest distance from all samples to the randomly selected first cluster center, and select the next cluster center based on the probability that the sample with the greater distance is selected as the cluster center. Repeat this process to select K initial cluster centers. Step S2-3: For each sample, calculate the Euclidean distance from the sample to the K initial cluster centers, and assign the sample to the cluster with the smallest distance. Specifically: ; In the formula, This indicates the final cluster category to which the i-th sample belongs. In mathematics, k is the operator that takes the parameter corresponding to the minimum value. k represents the range of values ​​from 1 to k. Let be the Euclidean distance between the i-th sample and the k-th cluster center. Step S2-4: For each cluster k, calculate the standardized mean of all samples in cluster k on the core metric, and use it as the new cluster center. Specifically: ; In the formula, This represents the standardized cluster center value of the k-th cluster on the j-th metric. n is the reciprocal of the number of samples in this class. k This represents the total number of samples contained in the k-th cluster. This represents the summation of the standardized values ​​of all samples within the k-th cluster on the j-th index; Set a threshold for the change in cluster centers and a maximum number of iterations, and set an iteration termination condition, specifically as follows: When the change in cluster centers is less than the threshold for the change in cluster centers; The number of iterations has reached its maximum value. After d consecutive iterations, the class assignment of the sample remains unchanged; The iteration stops when any one of the iteration stopping conditions is met. Step S2-5: Set the range of values ​​for K. Specifically, calculate the sum of squared Euclidean distances (SSE) from all samples to their respective cluster centers for each K, plot the K-SSE curve, and set the inflection point of the curve as the optimal value of K. Calculate the sum of squared Euclidean distances (SSE) from all samples to their respective cluster centers for each K, plot the K-SSE curve, and set the inflection point of the curve as the optimal value of the corresponding K; Step S2-6: For each candidate K, calculate the silhouette coefficient of all samples, then calculate the average silhouette coefficient, and select the K with the largest average silhouette coefficient as the optimal number of clusters.

[0028] The Euclidean distance of the samples is calculated iteratively to determine the cluster centers. The optimal number of clusters is selected by combining the K-SSE curve and the average profile coefficient, so as to achieve scientific classification of samples, solve the problem of traditional classification relying on experience, and provide a precise basis for efficiency level classification.

[0029] Step S3: Based on the clustering results corresponding to the optimal number of clusters, quantify the category features and classify the efficiency levels; Step S3-1: Set the core statistics as category mean, category standard deviation, and category percentage. For the clustering results corresponding to the optimal K, calculate the core statistics of each category of samples on each indicator to form the category statistics. Step S3-2: Classify the core indicators into three dimensions: efficiency correlation dimension, stability dimension, and resource consumption dimension. Step S3-3: Professionals assign weights to the core indicators of the three dimensions based on the actual business scenario, construct a three-level scoring model, give a comprehensive score to each clustering result, and then use the quantile method to classify the efficiency level into low efficiency, medium efficiency and high efficiency.

[0030] By quantifying clustering features and allocating indicator weights according to three dimensions to construct a scoring model, efficiency levels are divided using the quantile method, clearly defining the boundaries of operating conditions, and shifting efficiency assessment from fuzzy judgment to quantitative grading, thus clarifying the positioning of operating conditions.

[0031] Step S4: For the divided efficiency levels, calculate the fluctuation range of the core indicators, set a safety threshold range based on the fluctuation range of the core indicators, and trigger an early warning when the core indicators exceed the safety threshold range. Step S4-1: Select the core indicators corresponding to each efficiency level from the core indicator set and construct the associated dataset of "efficiency level-core indicator-sample data". Step S4-2: For each core indicator in each efficiency level, sort all the raw data of a core indicator under a certain efficiency level in ascending order, and calculate the 5th percentile Q5 and 95th percentile Q of that indicator data. 95 , with [Q5,Q 95 The normal fluctuation range of this core indicator under this efficiency level is recorded in the "Level-Indicator-Fluctuation Range" correspondence table. Step S4-3: Set the fluctuation range of the high efficiency level as the target threshold range of the optimal operating state. The target threshold range is the high efficiency threshold range. Based on the high efficiency level threshold, the fluctuation of x% is used as the medium efficiency threshold range for optimizing from medium efficiency to high efficiency. The fluctuation range of the low efficiency index is taken as the low efficiency threshold range of the low efficiency condition. The efficiency level threshold range is the high efficiency threshold range, the medium efficiency threshold range, and the low efficiency threshold range. Step S4-4: Extract the extreme boundaries of each core indicator from all valid samples, calculate the 99% confidence interval as the safety threshold interval for all operating conditions, set the collection period to 1 hour, collect the actual operating values ​​of the core indicators during production, and set the early warning trigger conditions: The actual values ​​of the core indicators do not fall within the threshold range corresponding to the current efficiency level; The actual values ​​of key indicators have exceeded the safe threshold range; The change in the indicator within 1 hour exceeded b% of the high-efficiency level threshold range. An alert is triggered when the actual operating value meets any of the above conditions.

[0032] Set fluctuation ranges and safety thresholds for each level of indicators, monitor and issue early warnings on an hourly cycle, detect abnormal indicators in real time, avoid production instability, and ensure the continuous and reliable operation of the gas production process.

[0033] Step S5: Based on the clustering results, identify the causes of inefficiency, specify targeted process parameter optimization schemes, and adjust inefficient and medium-efficiency conditions to high-efficiency conditions.

[0034] Step S5-1: Extract the cluster categories of low efficiency levels, summarize the original sample data of all core indicators under the category, the standardized data and category statistics, and construct the low efficiency sample feature dataset; Step S5-2: Perform statistical analysis on the low-efficiency sample feature dataset to identify key features. The key features are abnormal mean features and high volatility features. Abnormal mean features are obtained by calculating the deviation rate between the mean of each core indicator of the low-efficiency category and the mean of the whole sample. High volatility features are identified by screening indicators in the low-efficiency category whose standard deviation is greater than 1.5 times the standard deviation of the whole sample. Step S5-3: Based on the clustering categories of the high-efficiency level, calculate the absolute difference value between the low-efficiency category and the high-efficiency category for the same indicator, and screen out the indicators with a relative difference rate > 15% as potential offset cause indicators; Step S5-4: Calculate the correlation coefficient r between the candidate cause indicator and the core efficiency indicator. According to the special scenarios, set the relevant thresholds r1 and r2. When |r| > r2, determine that the candidate cause indicator is the core cause of low efficiency. When r1 ≤ |r| ≤ r2, determine that the candidate cause indicator is the secondary cause of low efficiency. When |r| < r1, exclude the candidate cause; Step S5-5: Adjust the working conditions by raising the core indicators of low efficiency and medium efficiency to G1% - G2% of the core indicators of high efficiency level.

[0035] Construct a low-efficiency sample set, analyze the differences and correlation coefficients to identify the reasons for low efficiency, formulate a plan to raise the low and medium efficiency indicators to G1% - G2% of the high-efficiency level, accurately locate the root cause and optimize it to improve the gas production efficiency.

[0036] The system includes a core indicator processing module, a clustering analysis and optimal clustering number determination module, an efficiency level division module, a threshold setting and warning module, and a low-efficiency cause analysis and optimization module; The core indicator processing module is responsible for extracting core indicators from the biomass-to-natural gas process, completing sample collection, data cleaning and standardization, and laying a data foundation for subsequent analysis; The clustering analysis and optimal clustering number determination module is responsible for calculating the Euclidean distance, iteratively updating the clustering center, screening out the optimal clustering number, and realizing the reasonable classification of samples; The efficiency level division module is responsible for quantifying the characteristics of clustering categories, constructing a scoring model, and dividing samples into different efficiency levels; The threshold setting and warning module is responsible for determining the fluctuation range and safety threshold interval of the core indicators of each efficiency level, setting warning conditions and triggering warnings; The low-efficiency cause analysis and optimization module is responsible for identifying the core cause of low-efficiency working conditions, formulating a process parameter optimization plan, and improving the operating efficiency.

[0037] The core indicator processing module includes an indicator collection and set construction unit and a data cleaning and standardization unit; The indicator collection and set construction unit is responsible for extracting core indicators, collecting n core indicator samples, and constructing a complete core indicator set; The data cleaning and standardization unit is responsible for removing outliers, filling or removing missing values, and standardizing the cleaned data using the Z-score standardization method; The clustering analysis and optimal clustering number determination module includes a clustering center iteration unit and an optimal clustering number screening unit; The unit is responsible for calculating the Euclidean distance of the samples, selecting the initial cluster centers, updating the cluster centers iteratively, and terminating the iteration when the stopping condition is met. The unit is responsible for setting the range of K values ​​and determining the optimal number of clusters by combining the inflection point of the K-SSE curve and the average profile coefficient.

[0038] The efficiency level classification module includes a category feature quantification unit and an efficiency level evaluation unit; The category feature quantification unit is responsible for calculating the core statistics of each class of samples, dividing the core indicators into three dimensions, and clarifying the category features; The efficiency level assessment unit is responsible for assigning weights to the three-dimensional indicators, constructing a three-level scoring model, and using the quantile method to divide the efficiency levels into low, medium, and high categories. The threshold setting and early warning module includes a threshold range determination unit and an early warning triggering unit; The threshold range determination unit is responsible for constructing the associated dataset, calculating the quantiles of each indicator to determine the fluctuation range, and setting the threshold ranges for different efficiency levels and the safety threshold range for all operating conditions. The early warning triggering unit is responsible for acquiring actual operating values ​​at a 1-hour collection cycle. An early warning is triggered when any of the following conditions are met: the indicator exceeds the threshold, the safe range is breached, or there is a large fluctuation in a short period of time.

[0039] The inefficiency cause analysis and optimization module includes an inefficiency cause identification unit and an operating condition optimization and adjustment unit; The inefficiency cause identification unit is responsible for constructing an inefficiency sample feature dataset and determining the core and secondary causes of inefficiency through difference analysis and correlation coefficient calculation. The operating condition optimization and adjustment unit is responsible for improving the core indicators of low and medium efficiency operating conditions to G1%-G2% of high efficiency level indicators, thereby achieving operating condition optimization.

[0040] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for evaluating the efficiency of biomass-to-natural gas production based on multi-source data fusion, characterized by: The method includes the following steps: Step S1: Extract the core indicators from the biomass-to-natural gas process, collect n core indicator samples, construct a core indicator set, and perform data cleaning and standardization on the samples in the core indicator set; Step S2: Calculate the sample Euclidean distance, select the clustering centers, calculate the shortest distance from the remaining samples to the clustering centers, iteratively calculate the initial clustering centers, perform sample classification and center iterative update, and determine the optimal number of clusters; Step S3: Quantify the category features according to the clustering results corresponding to the optimal number of clusters, and divide the efficiency levels; Step S4: For the divided efficiency levels, calculate the fluctuation range of the core indicators, set the safety threshold interval according to the fluctuation range of the core indicators, and trigger an alarm when the core indicators exceed the safety threshold interval; Step S5: Identify the reasons for inefficiency according to the clustering results, specify a targeted process parameter optimization plan, and adjust the inefficient and medium-efficiency working conditions to high efficiency.

2. The method for evaluating the efficiency of biomass-to-natural gas production based on multi-source data fusion according to claim 1, characterized in that: The specific steps of step S1 are as follows: Step S1-1: Extract the core indicators from the biomass-to-natural gas process, collect n core indicator samples, and construct a core indicator set; Step S1-2: For each sample data of the core indicators in the core indicator set, after sorting the sample data of the core indicators in ascending order, use the box plot method to eliminate outliers; Step S1-3: For missing values, calculate the sample data missing rate. When the missing rate < x1, fill it with the industry average of this indicator. When the missing rate is in the interval [x1, x2], fill it with the average of the sample data of the same type of core indicators. When the missing rate > x2, eliminate the sample data. The x1 and x2 are missing rate thresholds; Step S1-4: For the data after data cleaning, use the Z-score standardization method to standardize the samples.

3. The method for evaluating the efficiency of biomass-to-natural gas production based on multi-source data fusion according to claim 2, characterized in that: The specific steps of step S2 are as follows: Step S2-1: Calculate the Euclidean distance of the samples, specifically: ； In the formula, Let m be the Euclidean distance between the i-th sample and the k-th cluster center, and m be the number of indicators. Let j be the standardized value of the i-th sample and the j-th indicator. For the k-th cluster center, the standardized value of the j-th index; Step S2-2: Randomly select 1 sample as the first clustering center, calculate the shortest distance from all samples to the randomly selected first clustering center, and select the next clustering center according to the higher probability that the farther samples are selected as the clustering center. Repeat this iteration to select K initial clustering centers; Step S2-3: For each sample, calculate the Euclidean distance from the sample to the K initial clustering centers, and assign the sample to the cluster with the smallest distance, specifically: ； In the formula, This indicates the final cluster category to which the i-th sample belongs. In mathematics, k is the operator that takes the parameter corresponding to the minimum value. k represents the range of values ​​from 1 to k. Let be the Euclidean distance between the i-th sample and the k-th cluster center. Step S2-4: For each cluster k, calculate the standardized mean of all samples in cluster k on the core indicators as the new clustering center, specifically: ； In the formula, This represents the standardized cluster center value of the k-th cluster on the j-th metric. n is the reciprocal of the number of samples in this class. k This represents the total number of samples contained in the k-th cluster. This represents the summation of the standardized values ​​of all samples within the k-th cluster on the j-th index; Set the threshold of the change amount of the clustering center and the maximum number of iterations, and set the iteration stop condition, specifically: When the change amount of the clustering center is less than the threshold of the change amount of the clustering center; The number of iterations reaches the maximum number of iterations; For d consecutive iterations, the category attribution of the samples does not change; When any one of the iteration stop conditions is met, the iteration stops; Step S2-5: Set the value range of K, specifically: Calculate the sum of the squares of the Euclidean distances SSE from all samples to the所属 clustering centers corresponding to each K, draw the K-SSE curve, and set the inflection point of the curve as the optimal value of the corresponding K; Step S2-6: For each candidate K, calculate the silhouette coefficient of all samples, then calculate the average silhouette coefficient, and select the K with the largest average silhouette coefficient as the optimal number of clusters.

4. The method for evaluating the efficiency of biomass-to-natural gas production based on multi-source data fusion according to claim 3, characterized in that: The specific steps of step S3 are as follows: Step S3-1: Set the core statistics as category mean, category standard deviation, and category percentage. For the clustering results corresponding to the optimal K, calculate the core statistics of each category of samples on each indicator to form the category statistics. Step S3-2: Classify the core indicators into three dimensions: efficiency correlation dimension, stability dimension, and resource consumption dimension. Step S3-3: Professionals assign weights to the core indicators of the three dimensions based on the actual business scenario, construct a three-level scoring model, give a comprehensive score to each clustering result, and then use the quantile method to classify the efficiency level into low efficiency, medium efficiency and high efficiency.

5. The method for evaluating the efficiency of biomass-to-natural gas production based on multi-source data fusion according to claim 4, characterized in that: The specific steps of step S4 are as follows: Step S4-1: Select the core indicators corresponding to each efficiency level from the core indicator set and construct the associated dataset of "efficiency level-core indicator-sample data". Step S4-2: For each core indicator in each efficiency level, sort all the raw data of a core indicator under a certain efficiency level in ascending order, and calculate the 5th percentile Q5 and 95th percentile Q of that indicator data. 95 , with [Q5,Q 95 The normal fluctuation range of this core indicator under this efficiency level is recorded in the "Level-Indicator-Fluctuation Range" correspondence table. Step S4-3: Set the fluctuation range of the high efficiency level as the target threshold range of the optimal operating state. The target threshold range is the high efficiency threshold range. Based on the high efficiency level threshold, the fluctuation of x% is used as the medium efficiency threshold range for optimizing from medium efficiency to high efficiency. The fluctuation range of the low efficiency index is taken as the low efficiency threshold range of the low efficiency condition. The efficiency level threshold range is the high efficiency threshold range, the medium efficiency threshold range, and the low efficiency threshold range. Step S4-4: Extract the extreme boundaries of each core indicator from all valid samples, calculate the 99% confidence interval as the safety threshold interval for all operating conditions, set the collection period to 1 hour, collect the actual operating values ​​of the core indicators during production, and set the early warning trigger conditions: The actual values ​​of the core indicators do not fall within the threshold range corresponding to the current efficiency level; The actual values ​​of key indicators have exceeded the safe threshold range; The change in the indicator within 1 hour exceeded b% of the high-efficiency level threshold range. An alert is triggered when the actual operating value meets any of the above conditions.

6. The method for evaluating the efficiency of biomass-to-natural gas production based on multi-source data fusion according to claim 5, characterized in that: The specific steps of step S5 are as follows: Step S5-1: Extract the cluster categories of low efficiency levels, summarize the original sample data of all core indicators under the category, the standardized data and category statistics, and construct the low efficiency sample feature dataset; Step S5-2: Perform statistical analysis on the low-efficiency sample feature dataset to identify key features. The key features are abnormal mean features and high volatility features. Abnormal mean features are obtained by calculating the deviation rate between the mean of each core indicator of the low-efficiency category and the mean of the whole sample. High volatility features are identified by screening indicators in the low-efficiency category whose standard deviation is greater than 1.5 times the standard deviation of the whole sample. Step S5-3: Using the high-efficiency cluster category as the benchmark, calculate the absolute difference between the inefficient and high-efficiency categories on the same indicator, and screen indicators with a relative difference rate >15% as potential offsetting factors. Step S5-4: Calculate the correlation coefficient r between the candidate cause index and the core efficiency index. Set relevant thresholds r1 and r2 according to the special agent scenario. When |r| > r2, determine that the candidate cause index is a core cause of low efficiency. When r1 ≤ |r| ≤ r2, determine that the candidate cause index is a secondary cause of low efficiency. When |r| < r1, exclude the candidate cause; Step S5-5: Adjust the operating conditions by raising the core indicators of low and medium efficiency to G1% - G2% of the core indicators of high efficiency level.

7. A biomass-to-natural gas efficiency evaluation system based on multi-source data fusion, characterized in that: The system includes a core indicator processing module, a clustering analysis and optimal clustering number determination module, an efficiency level classification module, a threshold setting and warning module, and a low-efficiency cause analysis and optimization module; The core indicator processing module is responsible for extracting core indicators from the biomass-to-natural gas process, completing sample collection, data cleaning and standardization, and laying a data foundation for subsequent analysis; The clustering analysis and optimal clustering number determination module is responsible for calculating the Euclidean distance, iteratively updating the clustering center, screening out the optimal clustering number, and achieving reasonable classification of samples; The efficiency level classification module is responsible for quantifying the characteristics of clustering categories, constructing a scoring model, and classifying samples into different efficiency levels; The threshold setting and warning module is responsible for determining the fluctuation range and safety threshold interval of the core indicators of each efficiency level, setting warning conditions and triggering warnings; The low-efficiency cause analysis and optimization module is responsible for identifying the core causes of low-efficiency operating conditions, formulating process parameter optimization plans, and improving operating efficiency.

8. The biomass-to-natural gas efficiency evaluation system based on multi-source data fusion according to claim 7, characterized in that: The core indicator processing module includes an indicator collection and set construction unit and a data cleaning and standardization unit; The indicator collection and set construction unit is responsible for extracting core indicators, collecting n core indicator samples, and constructing a complete core indicator set; The data cleaning and standardization unit is responsible for removing outliers, filling or removing missing values, and standardizing the cleaned data using the Z-score standardization method; The clustering analysis and optimal clustering number determination module includes a clustering center iterative unit and an optimal clustering number screening unit; The unit is responsible for calculating the Euclidean distance of samples, selecting the initial clustering center, and iteratively updating the clustering center until the stop condition is met to terminate the iteration; The unit is responsible for setting the range of K values, combining the inflection point of the K-SSE curve and the average silhouette coefficient to determine the optimal clustering number.

9. The biomass-to-natural gas efficiency evaluation system based on multi-source data fusion according to claim 7, characterized in that: The efficiency level classification module includes a category feature quantification unit and an efficiency level assessment unit; The category feature quantification unit is responsible for calculating the core statistics of each class of samples, dividing the core indicators into three dimensions, and clarifying the category features; The efficiency level assessment unit is responsible for assigning weights to the three-dimensional indicators, constructing a three-level scoring model, and using the quantile method to divide the low, medium, and high efficiency levels; The threshold setting and warning module includes a threshold interval determination unit and a warning trigger unit; The threshold interval determination unit is responsible for constructing an associated data set, calculating the quantiles of each indicator to determine the fluctuation range, and setting the threshold intervals of different efficiency levels and the full-condition safety threshold interval; The early warning triggering unit is responsible for acquiring actual operating values ​​at a 1-hour collection cycle. An early warning is triggered when any of the following conditions are met: the indicator exceeds the threshold, the safe range is breached, or there is a large fluctuation in a short period of time.

10. The biomass-to-natural gas efficiency evaluation system based on multi-source data fusion according to claim 7, characterized in that: The inefficiency cause analysis and optimization module includes an inefficiency cause identification unit and an operating condition optimization and adjustment unit; The inefficiency cause identification unit is responsible for constructing an inefficiency sample feature dataset and determining the core and secondary causes of inefficiency through difference analysis and correlation coefficient calculation. The operating condition optimization and adjustment unit is responsible for improving the core indicators of low and medium efficiency operating conditions to G1%-G2% of high efficiency level indicators, thereby achieving operating condition optimization.

Images