Fertilizer production process optimization method and system

Abstract

The invention relates to the technical field of process optimization, in particular to a fertilizer production process optimization method and system. The method comprises the following steps: obtaining original data composed of process parameters and evaluation indexes of a plurality of production batches, carrying out standardization processing, processing interval type evaluation indexes by using a Gaussian membership function, calculating an initial weight through an entropy weight method in combination with a standardization value and a time sequence fluctuation coefficient, carrying out correlation correction on the initial weight, and carrying out correlation correction on the initial weight; and when the absolute value of the correlation coefficient of any two evaluation indexes is greater than the corresponding correlation threshold value, reducing the initial weight according to the variance contribution degree, redistributing other indexes by using the reduced weight to obtain a final weight, constructing a comprehensive evaluation model based on the final weight, and determining an optimization scheme. According to the scheme, the interval type evaluation indexes can be standardized, the problems of information redundancy and weight distribution distortion can be eliminated, then the accuracy of the evaluation result is improved, and a more reliable optimization scheme is provided for the fertilizer production process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of process optimization. More particularly, the present application relates to a method and system for optimizing a fertilizer production process. BACKGROUND

[0002] As a key production material in modern agricultural production, the product quality and production efficiency of fertilizers are of great importance to the protection of agricultural yield increase and food security. Therefore, how to scientifically and objectively comprehensively evaluate the fertilizer production process and find the optimal combination of process parameters based on the evaluation results to achieve quality improvement, efficiency increase, cost reduction and energy saving has become a core challenge faced by the current fertilizer production field. However, the traditional process optimization method not only has limited optimization effect and poor stability when facing complex and variable working conditions, but also is difficult to accurately measure the comprehensive influence of each process parameter on the overall production efficiency, which cannot meet the needs of modern production fine management.

[0003] To overcome the limitations of the traditional process optimization method, the existing technology introduces objective weighting methods such as entropy weight method into the multi-index comprehensive evaluation system. The entropy weight method determines the weight according to the information entropy of the index data itself. Specifically, the greater the dispersion degree of the index data, the smaller the information entropy, and the higher the weight. Using this method can effectively eliminate the subjective interference of human factors and more objectively determine the weight.

[0004] However, the standardization processing method in the above entropy weight method is usually only applicable to benefit-type or cost-type indicators. For evaluation indicators with optimal target intervals, i.e. not the greater / smaller the better, the simple linear mapping relationship cannot truly reflect the degree of excellence, which is likely to cause deviation in the evaluation results, thereby affecting the accuracy of the evaluation results and making it difficult to effectively optimize and adjust the production process. In addition, the entropy weight method implies the theoretical assumption that the evaluation indicators are independent of each other, but in the actual fertilizer production process, there is often a high correlation between multiple evaluation indicators. This correlation can cause some characteristic information to be repeatedly calculated, leading to characteristic information redundancy and distortion of weight distribution, thereby affecting the accuracy of the evaluation results and causing the evaluation results to not truly reflect the running characteristics of the process, which makes it difficult to effectively guide the optimization and adjustment of the production process. SUMMARY

[0005] The present application aims to provide a method and system for optimizing a fertilizer production process to solve the problem that the existing technology cannot handle evaluation indicators with optimal target intervals and ignores the high correlation between evaluation indicators, thereby leading to inaccurate evaluation results and making it difficult to effectively guide the optimization and adjustment of the production process. To this end, the present application provides solutions in the following two aspects.

[0006] In the first aspect, the present application provides a method for optimizing a fertilizer production process, comprising: The process involves acquiring raw data from multiple production batches in the fertilizer production process. This raw data consists of process parameters and evaluation indicators. The raw data is then standardized, including: when the evaluation indicator is an interval-type indicator with an optimal target interval, calculating the standardized value of the evaluation indicator using a Gaussian membership function based on the deviation between the measured value of the interval indicator and the center value of the optimal target interval; when the evaluation indicator is a benefit-type or cost-type evaluation indicator, calculating the standardized value of the evaluation indicator using an interval mapping method; calculating the time-series fluctuation coefficient of each evaluation indicator within a preset time period; and calculating the result using the entropy weight method based on the standardized value of the evaluation indicator and the time-series fluctuation coefficient. The process involves: determining the initial weights for each evaluation indicator; performing correlation correction on the initial weights, including: constructing a correlation coefficient matrix between evaluation indicators; calculating a correlation threshold for each pair of evaluation indicators; when the absolute value of the correlation coefficient between any two evaluation indicators is greater than the corresponding correlation threshold, reducing the initial weights based on their variance contribution, and redistributing the reduced weights proportionally to the inverse of the average correlation with the remaining evaluation indicators to obtain the final weights; constructing a comprehensive evaluation model based on the final weights and the standardized values; and determining the optimal combination of process parameters corresponding to the optimal comprehensive evaluation value through the comprehensive evaluation model, which serves as the optimal fertilizer production process optimization scheme.

[0007] Preferably, the process parameters include synthesis tower temperature, steam flow rate, and shift converter pressure; the evaluation indicators include product conversion rate, energy consumption per ton of product, nitrogen, phosphorus, and potassium content of the product, and product purity.

[0008] Preferably, when the evaluation index is an interval-type evaluation index with an optimal target interval, the standardized value of the evaluation index is calculated using a Gaussian membership function based on the deviation between the measured value of the interval index and the center value of the optimal target interval, including: In the formula, For the first The first batch The standardized values ​​of each evaluation indicator For the first The first batch The measured values ​​of each evaluation indicator For the first The optimal target interval center value of each evaluation index For the first The standard deviation of each evaluation indicator It is an exponential function with the natural constant e as its base.

[0009] Preferably, the optimal target interval center value and standard deviation are calculated in the following manner: for the interval-type evaluation index... The optimal target range for each evaluation indicator is set as follows: The formula for calculating the center value of the optimal target interval is: ; wherein, is the optimal target interval center value of the jth evaluation index, , are the lower limit and the upper limit of the optimal target interval of the jth evaluation index, respectively; the calculation formula of the standard deviation is: ; wherein, is the standard deviation of the jth evaluation index.

[0010] Preferably, when the evaluation index is a benefit type or a cost type evaluation index, the interval mapping method is used to calculate the standardized value of the evaluation index, including: for the jth evaluation index which is a benefit type evaluation index, the calculation formula of the standardized value is: ; for the jth evaluation index which is a cost type evaluation index, the calculation formula of the standardized value is: ; wherein, ; wherein, is the standardized value of the jth evaluation index under the ith batch, is the measured value of the jth evaluation index under the ith batch, , , are the maximum value and the minimum value of the measured value of the jth evaluation index in all batches, respectively.

[0011] Preferably, based on the standardized value of the evaluation index and the time series fluctuation coefficient, the initial weight of each evaluation index is calculated by the entropy weight method, including: normalizing the standardized value of the evaluation index, and calculating the information entropy of the evaluation index according to the normalization result; calculating the information entropy weight of the evaluation index based on the information entropy of the evaluation index, and the formula is: ; wherein, is the information entropy weight of the jth evaluation index, is the information entropy of the jth evaluation index, is the information entropy of the jth evaluation index, is the information entropy of the jth evaluation index, is the information entropy of the jth evaluation index, is the information entropy of the jth evaluation index, is the total number of evaluation indexes; multiplying the information entropy weight and the time series fluctuation coefficient to obtain a composite weight; normalizing the composite weight to obtain the initial weight of the evaluation index.

[0012] ​​​​​​​​Preferably, the step of calculating the correlation threshold for each pair of evaluation indicators includes: for any pair of evaluation indicators, obtaining the time-series fluctuation coefficients of the two indicators respectively; calculating the arithmetic square root of the product of the two time-series fluctuation coefficients as the geometric mean of the two time-series fluctuation coefficients; and taking the reciprocal of the geometric mean of the two time-series fluctuation coefficients as the correlation threshold.

[0013] Preferably, the step of reducing the initial weight based on the variance contribution of any two evaluation indicators when the absolute value of their correlation coefficients is greater than the corresponding correlation threshold includes: when the evaluation indicators and evaluation indicators The absolute value of the correlation coefficient is greater than the evaluation index. and evaluation indicators The correlation threshold between them, and the evaluation metrics The variance contribution is less than the evaluation index. When considering the variance contribution, then the evaluation index... The initial weights are reduced; the reduced weights satisfy: In the formula, For the first The weights of each evaluation indicator after reduction from their initial weights. For the first The initial weights of each evaluation indicator For the first The evaluation index and the first Correlation thresholds among evaluation indicators For the first The evaluation index and the first The correlation coefficient between the evaluation indicators It is the absolute value symbol.

[0014] Preferably, the step of redistributing the reduced weights proportionally to the inverse of the average correlation with the remaining evaluation indicators to obtain the final weights includes: redistributing the evaluation indicators... The difference between the initial weights and the weights after reduction is taken as the reduction amount; the reduction amount is then distributed proportionally to all indicators except the evaluation index. Other evaluation indicators besides those mentioned above; for any given evaluation indicator ( The obtained weight allocation satisfies: In the formula, For the first The weight allocation of each evaluation indicator , The first , The absolute value of the average correlation between one evaluation indicator and all other evaluation indicators. , They are respectively excluding evaluation indicators The outside One evaluation indicator, and the total number of remaining evaluation indicators; after adjustment, the evaluation indicators... The final weight is the evaluation index. The sum of the initial weights and the obtained weight allocations.

[0015] In the second aspect, the fertilizer production process optimization system includes: The processor and memory, wherein the memory stores computer program instructions for optimizing fertilizer production processes, and when the computer program instructions are executed by the processor, the above-described fertilizer production process optimization method is implemented.

[0016] The beneficial effects of this invention are as follows: This invention standardizes interval-type evaluation indicators with optimal target intervals using Gaussian membership functions, ensuring that the standardization results accurately reflect the quality of the data and thus improve the accuracy of the evaluation results. When calculating the initial weights of the evaluation indicators using the entropy weight method, this invention introduces the time-series fluctuation coefficient of the evaluation indicators, ensuring that the weight allocation simultaneously considers the data dispersion and the stability of the process. Furthermore, this invention sets correlation judgment criteria related to the fluctuation characteristics of each pair of evaluation indicators and reduces the initial weights based on the variance contribution of the evaluation indicator pair. The reduced weights are then rationally redistributed according to the average correlation with the remaining indicators, thereby eliminating information redundancy and weight allocation distortion caused by the high correlation between evaluation indicators. This improves the objectivity of the evaluation model, ensures the accuracy of the evaluation results, and provides a more reliable optimization scheme for fertilizer production processes. Attached Figure Description

[0017] Figure 1 This schematically illustrates the flowchart of the fertilizer production process optimization method in this embodiment; Figure 2 The schematic diagram illustrates the structural block diagram of the fertilizer production process optimization system in this embodiment. Detailed Implementation

[0018] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0019] like Figure 1 As shown, the fertilizer production process optimization method in this embodiment includes steps S1 to S4: Step S1: Obtain raw data from multiple production batches in the fertilizer production process. The raw data consists of process parameters and evaluation indicators.

[0020] In one embodiment, the process parameters include synthesis tower temperature, steam flow rate, and shift converter pressure; the evaluation indicators include product conversion rate, energy consumption per ton of product, nitrogen, phosphorus, and potassium content of the product, and product purity.

[0021] Specifically, in this embodiment, continuous data is collected through a manufacturing execution system or a distributed control system. The operational data for each production batch, i.e., the raw data consisting of process parameters and evaluation indicators. Among them, process parameters include, but are not limited to, synthesis tower temperature, steam flow rate, and shift converter pressure, while evaluation indicators include, but are not limited to, product conversion rate, energy consumption per ton of product, nitrogen, phosphorus, and potassium content of the product, and product purity. Energy consumption per ton of product is an indicator that measures the energy consumed in producing each ton of fertilizer.

[0022] A batch and evaluation index corresponding to process parameters constitute a OK Column data matrix ,in Total number of batches This represents the total number of evaluation indicators.

[0023] Step S2 involves standardizing the raw data, including: when the evaluation indicator is an interval-type evaluation indicator with an optimal target interval, calculating the standardized value of the evaluation indicator using a Gaussian membership function based on the deviation between the measured value of the interval indicator and the center value of the optimal target interval; when the evaluation indicator is a benefit-type or cost-type evaluation indicator, calculating the standardized value of the evaluation indicator using an interval mapping method; calculating the time-series fluctuation coefficient of each evaluation indicator within a preset time period; and calculating the initial weight of each evaluation indicator using the entropy weight method based on the standardized value and time-series fluctuation coefficient of the evaluation indicator.

[0024] In one embodiment, when the evaluation index is an interval-type evaluation index with an optimal target interval, the standardized value of the evaluation index is calculated using a Gaussian membership function based on the deviation between the measured value of the interval index and the center value of the optimal target interval, including: ; In the formula, For the first The first batch The standardized values ​​of each evaluation indicator For the first The first batch The measured values ​​of each evaluation indicator For the first The optimal target interval center value of each evaluation index For the first The standard deviation of each evaluation indicator It is an exponential function with the natural constant e as its base.

[0025] The optimal target interval center value and the standard deviation are calculated according to the following manner: For the jth evaluation index which is an interval type evaluation index, the optimal target interval is set as ; ; The calculation formula of the optimal target interval center value is: ; wherein, is the optimal target interval center value of the jth evaluation index, , , are the lower limit and the upper limit of the optimal target interval of the jth evaluation index, respectively; ; The calculation formula of the standard deviation is: ; wherein, is the standard deviation of the jth evaluation index.

[0026] Specifically, the optimal target interval of each interval type evaluation index is pre-set by experts according to the requirements of the specific production process and experience. Taking the product purity as an example, the optimal target interval is set as , at this time, the optimal target interval center value is 99% and the standard deviation is about 0.167. If the measured value of the product purity of one batch is 99%, which is the center value, the standardized value calculated by the Gaussian membership function is 1, representing the optimal state. If the measured value of the product purity of another batch is 98%, the standardized value calculated by the Gaussian membership function is a value close to 0, representing a serious deviation. If the standardized value of the product purity of another batch is a value close to 1 but less than 1, such as 0.487, it represents a slight deviation.

[0027] The degree of deviation of the measured value of the interval type evaluation index from the optimal target interval center value is smoothed by the curve similar to the normal distribution, so that the score is higher when closer to the center value and lower when deviating from the center value.

[0028] In one embodiment, when the evaluation index is a benefit type or a cost type evaluation index, the interval mapping method is used to calculate the standardized value of the evaluation index, comprising: For the jth evaluation index which is a benefit type evaluation index, the calculation formula of the standardized value is: ; For the jth evaluation index which is a cost type evaluation index, the calculation formula of the standardized value is: ; ; ; wherein, is the standardized value of the jth batch of the jth evaluation index. ​​​ The standardized values ​​of each evaluation indicator For the first The first batch The measured values ​​of each evaluation indicator , The first The maximum and minimum values ​​of each evaluation indicator among all measured values ​​across all batches.

[0029] Specifically, benefit-based evaluation indicators reflect the gains, advantages, or effectiveness of the evaluated object; a higher value generally indicates better benefits and superior performance. Cost-based evaluation indicators reflect the inputs, consumption, or losses of the evaluated object; a lower value generally indicates lower costs and more efficient resource utilization. By measuring the advantage of the actual value of a benefit-based evaluation indicator relative to its minimum value in a given batch, a higher actual value of the benefit-based evaluation indicator in that batch results in a higher standardized value for that indicator. Conversely, by measuring the difference between the actual value of a cost-based evaluation indicator and its maximum value in a given batch, a lower actual value of the cost-based evaluation indicator in that batch results in a higher standardized value for that indicator.

[0030] By employing specific standardization methods for different types of evaluation indicators, standardized values ​​for each type of evaluation indicator can be obtained. This can unify evaluation indicators with different dimensions and optimal target ranges onto a 0 to 1 evaluation scale, and can also truly reflect the quality of each data point.

[0031] In one embodiment, calculating the time-series fluctuation coefficient of each evaluation index within a preset time period includes: obtaining all measured values ​​of the evaluation index in all batches; calculating the standard deviation and average value of all measured values; and using the ratio of the standard deviation to the average value as the time-series fluctuation coefficient of the evaluation index within the preset time period.

[0032] Specifically, by calculating the first Each evaluation indicator is in The ratio of the standard deviation to the mean within each batch is used to obtain the first batch. The time series fluctuation coefficient of each evaluation indicator can be obtained.

[0033] In one embodiment, the initial weights of each evaluation indicator are calculated using the entropy weight method based on the standardized values ​​and time-series fluctuation coefficients of the evaluation indicators, including: The standardized values ​​of the evaluation indicators are normalized, and the information entropy of the evaluation indicators is calculated based on the normalization results. The information entropy weight of an evaluation indicator is calculated based on its information entropy, using the following formula: In the formula, For the first The information entropy weight of each evaluation indicator For the first Information entropy of each evaluation indicator For the first Information entropy of each evaluation indicator The total number of evaluation indicators; The information entropy weight is multiplied by the time series fluctuation coefficient to obtain the composite weight; the composite weight is then normalized to obtain the initial weight of the evaluation index.

[0034] In one embodiment, the normalization process for the standardized values ​​of the evaluation index is performed as follows: In the formula, For the first The first batch The result after normalizing the standardized values ​​of each evaluation indicator. For the first The first batch The standardized values ​​of each evaluation indicator Total number of batches For the first Each evaluation indicator is in The sum of all standardized values ​​within a batch.

[0035] The information entropy of the evaluation index is calculated based on the normalization result, satisfying the expression: In the formula, For the first Information entropy of each evaluation indicator It is the logarithmic exponent with the natural constant e as the base.

[0036] Specifically, the standardized values ​​of the evaluation indicators are analyzed using the entropy weight method. If the standardized values ​​of the evaluation indicators are within... Significant differences across batches indicate that the product conversion rate, as an evaluation indicator, carries a large amount of information, resulting in a lower information entropy and thus a higher information entropy weight. Conversely, if the standardized value of the evaluation indicator is within a certain range... If the batches are very similar, their information entropy will be high, and the resulting information entropy weight will be low.

[0037] In one embodiment, the composite weight is expressed by the formula: In the formula, For composite weights, For the first The information entropy weight of each evaluation indicator For the first The time-series fluctuation coefficient of each evaluation indicator.

[0038] The composite weights are normalized to obtain the initial weights of the evaluation indicators, which satisfy the following relationship: In the formula, For the first The initial weights of each evaluation indicator For composite weights, For the first The composite weight of each evaluation indicator.

[0039] The time-series fluctuation coefficient reflects the historical importance or instability of the evaluation indicator. A composite weight is obtained by multiplying the information entropy weight of the evaluation indicator by its own time-series fluctuation coefficient. This composite weight is then normalized to obtain the initial weight of the corresponding evaluation indicator.

[0040] Therefore, based on the standardized values ​​and time-series fluctuation coefficients of the evaluation indicators, the initial weights of each evaluation indicator are calculated using the entropy weight method, which takes into account both the degree of differentiation of the evaluation indicators in the current batch and the historical situation of the evaluation indicators.

[0041] Step S3 involves relevance correction of the initial weights, including: constructing a correlation coefficient matrix between evaluation indicators; calculating a correlation threshold for each pair of evaluation indicators; when the absolute value of the correlation coefficient between any two evaluation indicators is greater than the corresponding correlation threshold, reducing the initial weights according to their variance contribution, and redistributing the reduced weights proportionally to the inverse of the average correlation with the remaining evaluation indicators to obtain the final weights.

[0042] In this embodiment, calculation The Pearson correlation coefficients between each pair of evaluation indicators constitute a correlation coefficient matrix. .

[0043] The calculation of the correlation threshold for each pair of evaluation indicators includes: For any pair of evaluation indicators, obtain their time-series fluctuation coefficients respectively; Calculate the arithmetic square root of the product of the two time series fluctuation coefficients, and use it as the geometric mean of the two time series fluctuation coefficients; The reciprocal of the geometric mean of the two time-series fluctuation coefficients is used as the correlation threshold.

[0044] In one embodiment, the geometric mean of the two time-series fluctuation coefficients is expressed by the formula: In the formula, For the first The time-series fluctuation coefficient of the first evaluation indicator and the second The geometric mean of the time-series fluctuation coefficients of the evaluation indicators For the first The time-series fluctuation coefficient of each evaluation indicator For the first The time-series fluctuation coefficient of each evaluation indicator.

[0045] Specifically, the correlation threshold is inversely proportional to the geometric mean of the time-series fluctuation coefficients of each pair of evaluation indicators. For the evaluation indicators product conversion rate and product purity, if their time-series fluctuation coefficients are 1.5 and 1.2 respectively, it indicates that these two evaluation indicators have been historically active or important. The geometric mean between the two time-series fluctuation coefficients is approximately 1.34, and the corresponding correlation threshold is their reciprocal, approximately 0.75. For the evaluation indicator product nitrogen, phosphorus, and potassium content, the product nitrogen, phosphorus, and potassium content is relatively stable, with a time-series fluctuation coefficient of only 0.6. When paired with product conversion rate, the geometric mean between the two time-series fluctuation coefficients is approximately 0.95, and the correlation threshold is 1.05. Therefore, for two historically active evaluation indicators, if their correlation coefficient exceeds 0.75, it is considered that there may be information overlap between them. For combinations of evaluation indicators with relatively stable indicators in a pair, the correlation threshold is high, and redundant judgments are unlikely to be triggered, thus allowing for the identification and handling of overlap issues between key information.

[0046] In one embodiment, the step of reducing the initial weights based on the variance contribution of any two evaluation indicators when the absolute value of their correlation coefficients is greater than the corresponding correlation threshold includes: When evaluation indicators and evaluation indicators The absolute value of the correlation coefficient is greater than the evaluation index. and evaluation indicators The correlation threshold between them, and the evaluation metrics The variance contribution is less than the evaluation index. When considering the variance contribution, then the evaluation index... The initial weights are reduced; The reduced weights satisfy: In the formula, For the first The weights of each evaluation indicator after reduction from their initial weights. For the first The initial weights of each evaluation indicator For the first The evaluation index and the first Correlation thresholds among evaluation indicators For the first The evaluation index and the first The correlation coefficient between the evaluation indicators It is the absolute value symbol.

[0047] The reduced weight is proportionally redistributed according to the inverse of the average correlation of the rest of the evaluation indexes to obtain the maximum weight, comprising: taking the difference between the initial weight of the evaluation index and the weight after being reduced as the generated reduction amount; proportionally distributing the generated reduction amount to the rest of the evaluation indexes except the evaluation index ; for any evaluation index ( ), the obtained weight distribution amount satisfies: ; in the formula, is the weight distribution amount obtained by the th evaluation index, , are the average correlation absolute values of the , th evaluation index and all the rest of the evaluation indexes, , are the th evaluation index except the evaluation index and the total number of the rest of the evaluation indexes; After adjustment, the maximum weight of the evaluation index is the sum of the initial weight of the evaluation index and the obtained weight distribution amount.

[0048] Specifically, when the correlation coefficient absolute value of any pair of evaluation indexes is greater than the correlation threshold value of the two, and the variance contribution degree of one of the evaluation indexes is less than the variance contribution degree of the other, the evaluation index with the smaller variance contribution degree can largely explain the other evaluation index, and thus the initial weight of the evaluation index with the smaller variance contribution degree needs to be reduced to obtain the reduced weight. Then, the initial weight of the evaluation index with the smaller variance contribution degree is subtracted by the reduced weight to obtain the generated reduction amount, that is, the reduced weight. The average correlation absolute value , of each of the rest of the evaluation indexes except the reduced evaluation index and all the rest of the evaluation indexes is calculated; for any evaluation index ( ), the proportion of the allocated reduction amount of the evaluation index is obtained by dividing the average correlation absolute value of the evaluation index by the sum of the average correlation absolute values of all the rest of the evaluation indexes except the evaluation index ; wherein, the lower the average correlation of the evaluation index with all the rest of the evaluation indexes, the more independent the evaluation index is, the more unique information the evaluation index provides, and the more weight the evaluation index gets. For any evaluation index , the initial weight of the th evaluation index is added by the obtained weight distribution amount to obtain the maximum weight of the evaluation index. The initial weights of the evaluation indicators are obtained using the same method, and the calculation of the first evaluation indicator is performed using the same method. The sum of the initial weights and the obtained weight allocations for each evaluation indicator is used as the adjusted evaluation indicator. The final weights; after adjustment, the evaluation indicators The final weight is Therefore, by redundancy judgment and weight redistribution, a new weight system is obtained, which can reduce the impact of redundant information and increase the importance of unique evaluation indicators, making the evaluation results more objective.

[0049] Step S4: Construct a comprehensive evaluation model based on the final weights and the standardized values. Determine the process parameter combination corresponding to the optimal comprehensive evaluation value through the comprehensive evaluation model, which serves as the optimal fertilizer production process optimization scheme.

[0050] In one embodiment, for Each batch in each production batch , and Standardized values ​​of each evaluation indicator Multiply the values ​​by their corresponding final weights and sum them to obtain the comprehensive evaluation value for each batch. Compare the comprehensive evaluation values ​​of all batches and find the maximum value. The production batch corresponding to this maximum value contains all process parameters, which represents the optimal fertilizer production process optimization scheme.

[0051] In another embodiment, the comprehensive evaluation model is the TOPSIS model. Specifically, for Each batch in each production batch The standardized value of each evaluation indicator is multiplied by its corresponding final weight to obtain a weighted value. From the weighted values ​​of all batches, the optimal value for each evaluation indicator is selected, with the maximum value for benefit-type evaluation indicators and the minimum value for cost-type evaluation indicators, forming a virtual optimal solution. The worst value for each evaluation indicator is selected, with the minimum value for benefit-type evaluation indicators and the maximum value for cost-type evaluation indicators, forming a virtual worst solution. The weighted Euclidean distance between each actual production batch solution and the optimal ideal value (positive ideal solution) and the weighted Euclidean distance between the actual production batch solution and the worst ideal value is calculated. The weighted Euclidean distance between each actual production batch solution and the worst ideal value is divided by the sum of the weighted Euclidean distances between each actual production batch solution and the optimal ideal value (positive ideal solution) and the worst ideal value (negative ideal solution), and the result is taken as the relative closeness. The relative closeness is taken as the final comprehensive evaluation value. The set of process parameters that maximizes the relative closeness value is the optimal fertilizer production process optimization scheme.

[0052] This invention also provides a fertilizer production process optimization system. For example... Figure 2As shown, the system comprises a processor and a memory having stored therein computer program instructions for optimizing a fertilizer production process, which when executed by the processor implement the method for optimizing a fertilizer production process according to the present application.

[0053] The system further comprises a communication bus and a communication interface and other components well known to those skilled in the art, the arrangement and function of which are known in the art and thus will not be described here in detail.

[0054] In the present application, the aforementioned memory can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device. For example, the computer-readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random-Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc., or any other medium that can be used to store the desired information and that can be accessed by an application, module, or both. Any such computer storage media can be part of the device or accessible or connectable thereto. Any application or module described in the present application can be implemented by computer-readable / executable instructions stored or otherwise held by such computer-readable media.

[0055] In the description of the present application, the meaning of "a plurality of" is at least two, such as two, three or more, etc., unless otherwise explicitly specified.

[0056] Although the present application has been shown and described with respect to several embodiments thereof, it will be apparent that those skilled in the art will be able to make many modifications, changes and alterations to the present application without departing from the spirit and scope thereof.

Claims

1. A method for optimizing a fertilizer production process, characterized in that, The method comprises the following steps: Obtaining original data of multiple production batches in a fertilizer production process, the original data being composed of process parameters and evaluation indexes; Standardizing the original data, including: when the evaluation index is an interval type evaluation index with an optimal target interval, calculating the standardization value of the evaluation index by using a Gaussian membership function based on the deviation degree of the measured value of the interval index from the center value of the optimal target interval; when the evaluation index is a benefit type or cost type evaluation index, calculating the standardization value of the evaluation index by using an interval mapping method; calculating the time sequence fluctuation coefficient of each evaluation index within a preset time period; calculating the initial weight of each evaluation index by using an entropy weight method based on the standardization value and the time sequence fluctuation coefficient of the evaluation index; Correcting the correlation of the initial weight, including: constructing a correlation coefficient matrix between the evaluation indexes; calculating a correlation threshold value for each pair of evaluation indexes; when the absolute value of the correlation coefficient of any two evaluation indexes is greater than the corresponding correlation threshold value, reducing the initial weight according to the variance contribution degree of the two, and proportionally redistributing the reduced weight according to the inverse of the average correlation with the remaining evaluation indexes to obtain the final weight; Constructing a comprehensive evaluation model based on the final weight and the standardization value, and determining the process parameter combination corresponding to the optimal comprehensive evaluation value by the comprehensive evaluation model as the optimal fertilizer production process optimization scheme.

2. The method of optimizing a fertilizer production process according to claim 1, wherein, The process parameters include the temperature of a synthesis tower, the steam flow and the pressure of a shift converter; the evaluation indexes include the product conversion rate, the energy consumption per ton of product, the product nitrogen, phosphorus and potassium content, and the product purity.

3. The method of optimizing a fertilizer production process according to claim 1, wherein, When the evaluation index is an interval type evaluation index with an optimal target interval, the standardization value of the evaluation index is calculated by using a Gaussian membership function based on the deviation degree of the measured value of the interval index from the center value of the optimal target interval, including: ； In the formula, For the first The first batch The standardized values ​​of each evaluation indicator For the first The first batch The measured values ​​of each evaluation indicator For the first The optimal target interval center value of each evaluation index For the first The standard deviation of each evaluation indicator It is an exponential function with the natural constant e as its base.

4. The method of optimizing a fertilizer production process according to claim 3, wherein, The center value and the standard deviation of the optimal target interval are calculated according to the following method: For the first interval-type evaluation index The optimal target range for each evaluation indicator is set as follows: ; The calculation formula of the optimal target interval center value is: ; in the formula, is the optimal target interval center value of the first evaluation index, is the optimal target interval center value of the second evaluation index, , is the lower limit of the optimal target interval of the first evaluation index, and is the upper limit of the optimal target interval of the first evaluation index. The formula for calculating the standard deviation is: In the formula, For the first The standard deviation of each evaluation indicator.

5. The method of optimizing a fertilizer production process according to claim 1, wherein, When the evaluation index is a benefit type or cost type evaluation index, the standardization value of the evaluation index is calculated by using an interval mapping method, including: For the first evaluation index which is a benefit type evaluation index, the calculation formula of the standardized value is: ; and ; For the first evaluation index which is a cost type evaluation index, the calculation formula of the standardized value is: ; and ; In the formula, For the first The first batch The standardized values ​​of each evaluation indicator For the first The first batch The measured values ​​of each evaluation indicator , The first The maximum and minimum values ​​of each evaluation indicator among all measured values ​​across all batches.

6. The method of optimizing a fertilizer production process according to claim 1, wherein, The initial weight of each evaluation index is calculated by using an entropy weight method based on the standardization value and the time sequence fluctuation coefficient of the evaluation index, including: Normalizing the standardization value of the evaluation index, and calculating the information entropy of the evaluation index according to the normalization result; The information entropy weight of the evaluation index is calculated based on the information entropy of the evaluation index, and the formula is: ; in the formula, the information entropy weight of the first evaluation index, the information entropy of the first evaluation index, the information entropy of the first evaluation index, the total number of evaluation indexes; Multiplying the information entropy weight and the time sequence fluctuation coefficient to obtain a composite weight; normalizing the composite weight to obtain the initial weight of the evaluation index.

7. The method of optimizing a fertilizer production process according to claim 1, wherein, The correlation threshold value for each pair of evaluation indexes is calculated, including: For any pair of evaluation indexes, the time sequence fluctuation coefficients of the two are obtained respectively; The arithmetic square root of the product of the two time sequence fluctuation coefficients is calculated as the geometric mean of the two time sequence fluctuation coefficients; The inverse of the geometric mean of the two time sequence fluctuation coefficients is taken as the correlation threshold value.

8. The method of optimizing a fertilizer production process according to claim 1, wherein, When the absolute value of the correlation coefficient of any two evaluation indexes is greater than the corresponding correlation threshold value, the initial weight is reduced according to the variance contribution degree of the two, including: When evaluation indicators and evaluation indicators The absolute value of the correlation coefficient is greater than the evaluation index. and evaluation indicators The correlation threshold between them, and the evaluation metrics The variance contribution is less than the evaluation index. When considering the variance contribution, then the evaluation index... The initial weights are reduced; The reduced weights satisfy: In the formula, For the first The weights of each evaluation indicator after reduction from their initial weights. For the first The initial weights of each evaluation indicator For the first The evaluation index and the first Correlation thresholds among evaluation indicators For the first The evaluation index and the first The correlation coefficient between the evaluation indicators It is the absolute value symbol.

9. The method of optimizing a fertilizer production process according to claim 8, wherein, The step of redistributing the reduced weights proportionally to the inverse of the average correlation with the remaining evaluation indicators to obtain the final weights includes: redistributing the evaluation indicators... The difference between the initial weights and the weights after reduction is taken as the reduction amount; the reduction amount is then distributed proportionally to all indicators except the evaluation index. Other evaluation indicators besides those mentioned above; for any given evaluation indicator ( The obtained weight allocation satisfies: In the formula, For the first The weight allocation of each evaluation indicator , The first , The absolute value of the average correlation between one evaluation indicator and all other evaluation indicators. , They are respectively excluding evaluation indicators The outside One evaluation indicator, and the total number of other evaluation indicators; After adjustment, the evaluation index has the maximum weight which is the sum of the initial weight of the evaluation index and the obtained weight distribution amount.

10. A system for optimizing a fertilizer production process, characterized by ​ a processor and a memory, said memory storing computer program instructions for a fertilizer production process optimization method according to any one of claims 1-9, when said computer program instructions are executed by said processor.

Images