Intelligent optimization design method and system for compound formula

Abstract

The invention relates to the technical field of formula optimization, in particular to an intelligent optimization design method and system for a compound formula, and the method comprises the following steps: obtaining a plurality of formula component combinations and performance data, extracting performance values and variation amplitudes, merging performance response offset segments, analyzing component coupling interference, establishing a feasible concentration boundary, and eliminating a negative correlation path. According to the method, the performance change amplitude of the adjacent component combination is extracted, the performance response offset section is identified, the deviation analysis of the concentration disturbance ratio between the components and the performance response is combined, and the actual influence of the process disturbance parameters on the performance is superposed; a feasible concentration boundary with engineering constraint and performance compatibility is established, a combination path with inconsistent directionality is effectively avoided in combination with a performance negative correlation path elimination mechanism, a combination with the minimum performance deviation degree is preferably selected on the premise of ensuring concentration rationality, the pertinence of formula performance regulation and control is improved, and the method is suitable for popularization and application. And the component concentration optimization efficiency and the result reliability are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of formula optimization, in particular to a compound formula intelligent optimization design method and system. BACKGROUND

[0002] The technical field of formula optimization mainly involves modeling, analyzing and optimizing the composition of raw materials and their ratio scheme, with the goal of obtaining the optimal or near-optimal formula that meets the preset indicators through algorithm modeling, simulation testing, historical data mining, etc. under the premise of meeting product performance, cost, safety and environmental protection requirements. This field is widely used in industries such as chemical, material, pharmaceutical, food, daily chemical, etc. It usually combines mathematical modeling, experimental design, machine learning, etc. to solve the problem of multiple variables, strong coupling, high experimental cost in the formula design process. The development trend of technology focuses on introducing intelligent algorithms and automation tools to improve design efficiency, reduce trial and error costs, and improve the consistency and controllability of product performance.

[0003] Among them, the compound formula intelligent optimization design method refers to a method of modeling and optimizing the components and proportions of the compound formula using data modeling, aiming to maximize the performance or minimize the cost of the compound formula under the condition of meeting the performance constraints, improve the efficiency of formula design, reduce experimental costs, shorten the research and development cycle, and realize the directional regulation of performance indicators.

[0004] The traditional design method does not consider the sudden change of performance in the local disturbance path, which leads to the fact that the abnormal fluctuation of performance in some concentration intervals is not identified in time, resulting in the situation that the judgment of the trend of component action is wrong in actual research and development. The coupling effect between the unidentified components makes the interference effect caused by the disturbance of non-independent variables in the performance response not be reasonably corrected to the concentration tolerance range of the components under the condition of process fluctuation, which makes the optimal formula deviate from the target performance under the actual process environment. In the context where multiple performance objectives exist in a negative correlation relationship, the path selection lacks an exclusion mechanism, which may cause a combination with poor compatibility to be selected into the optimization result, reducing the usability and subsequent verification efficiency of the optimization result. SUMMARY

[0005] In order to solve the technical problems existing in the prior art, the compound formula intelligent optimization design method provided by the embodiments of the present application comprises the following steps: In order to achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows: the compound formula intelligent optimization design method comprises the following steps: S1: Obtain a plurality of target compound formula component combinations, extract the performance variation amplitude between adjacent combinations according to the component concentration change trajectory, if any performance value presents window shrinkage, overflow or mutation in continuous combinations, record the concentration interval as a performance response offset section, and generate an offset interval list; S2: According to the offset interval list, the ratio difference between each pair of component concentration disturbance amplitude and target performance response rate of change is calculated, the ratio difference is compared with the ratio under single component independent disturbance, it is judged whether there is deviation, the corresponding concentration range and the influence performance type are recorded, and the response characteristic comparison item set is generated; S3: Obtain the process disturbance parameter value under the current test condition, calculate the numerical difference between the current process state and the target value according to the process fluctuation response sensitivity value of each component in the response characteristic comparison item set corresponding to the performance item, linearly compress the upper and lower limit values of the original concentration, and generate the component feasible concentration boundary set; S4: The concentration combination in the component feasible concentration boundary set is called, the Pearson correlation coefficient between impact strength and ductility is calculated, if any two performance parameters in the path show negative correlation, the component path is recorded as an unsuitable path, and the screened component path set is generated; S5: Based on the combination data in the screened component path set, the deviation degree of the corresponding performance target threshold value is calculated, and the sum of each deviation degree multiplied by the set performance weight vector is calculated, the target number of combinations is screened, and the corresponding component concentration is extracted, the recommended formula information with feasible concentration interval and performance compatibility is integrated, and the compound optimization formula combination is generated.

[0006] As a further scheme of the present application, the offset interval list includes performance trend jump section, component concentration interference section and response interval abnormal overlap section, the response characteristic comparison item set includes high coupling sensitive component pair, influence performance index classification and concentration influence threshold range, the component feasible concentration boundary set includes concentration upper limit adjustment boundary, concentration lower limit compression boundary and component concentration compatibility range, the screened component path set includes positive performance synergy path, negative performance exclusion path and feasibility path screening result, and the compound optimization formula combination includes preferred component concentration scheme, performance score weight list and performance index matching path set.

[0007] As a further scheme of the present application, the specific steps of S1 are: S101: Obtain a plurality of target compound formula component combinations, extract impact strength, ductility and conductivity three performance values according to corresponding performance detection data, call concentration value and performance parameter, screen the combination pairs with disturbance amplitude in the set concentration change interval according to the difference value of adjacent combinations in the concentration on the component disturbance path, and construct the concentration change path sequence to generate the component disturbance trajectory data set; S102: call the concentration path in the component disturbance trajectory dataset, calculate the change amplitude value of three performance parameters in adjacent combination pairs in turn, and jointly judge according to the change direction difference and the change amplitude size difference, mark the path point position where the performance parameter appears the turning direction or the change rate exceeds the performance floating threshold, obtain the performance trend jump node list; S103: based on the node path segment positioned in the performance trend jump node list, backtrack the corresponding component disturbance path interval, mark the concentration section where multiple jump nodes exist continuously as an abnormal response area, extract the concentration boundary range and the performance type label according to the density of the performance change amplitude in the abnormal response area, and generate a performance response offset interval list.

[0008] As a further scheme of the present application, the performance floating threshold is set in the following manner: the change amplitude values of the three performance parameters of impact resistance, ductility and conductivity under the same concentration disturbance path in the component disturbance trajectory dataset are counted, and the sum of the average value and the standard deviation of the change amplitude value of the performance parameter under the path is taken as the performance floating threshold of the corresponding performance parameter. The change direction difference judgment process is specifically comparing the increase and decrease change directions of the same performance parameter in the previous group and the next group combination, and if the change direction is changed from positive to negative or from negative to positive, it is determined as a direction turning point. The change amplitude size difference judgment process is specifically calculating the difference value between the change amplitude values of the same performance parameter in adjacent combination pairs, and if the difference value exceeds the performance floating threshold, the path point position is marked as a jump node.

[0009] As a further scheme of the present application, the specific steps of S2 are as follows: S201: based on the concentration interval marked in the performance response offset interval list, extract the effective component combination in each interval, and pair the components in the combination two by two, call the concentration value sequence of each component pair in the interval and the performance value sequence of the corresponding impact resistance, ductility and conductivity, construct the component concentration pair and performance difference contrast array, and generate the component disturbance and performance difference mapping information; S202: according to each row record in the component disturbance and performance difference mapping information, calculate the ratio between the concentration disturbance amplitude and the target performance response change rate of the corresponding component pair, calculate the difference value between the ratio and the corresponding performance ratio generated when the same component is independently disturbed in a non-interaction path, and compare the difference value with the coupling judgment threshold value, establish the component difference judgment result set, and obtain the coupling interference deviation result list. S203: Based on the coupling interference deviation result list marked as the deviation value exceeding the coupling determination threshold, record the corresponding concentration disturbance interval and the target performance parameter label causing the deviation, and number and classify the component pairs, establish a data structure set with coupling behavior characteristics and performance influence identification, and generate a response characteristic reference item set.

[0010] As a further scheme of the present application, the specific steps of S3 are: S301: Obtain the process disturbance parameter values under the current test conditions, including the mixing temperature, reaction time and shear rate, and calculate the difference value of each parameter and the target set value, call the process sensitivity factor of the component pair performance item in the response characteristic reference item set, multiply the numerical difference between each process parameter and the target set value by the corresponding sensitivity factor, establish the performance influence numerical table of the component corresponding to the current working condition, and generate a working condition adaptability difference value set; S302: Calculate the concentration adjustment proportion factor of the component according to the performance influence value of the component corresponding performance item in the working condition adaptability difference value set, and multiply the proportion factor by the upper and lower limits of the original set concentration of the component respectively, take the result and the original boundary value for numerical comparison and execute linear scaling to obtain the adjustment range of the component concentration under the current working condition, and obtain the concentration adjustment boundary set under the working condition control; S303: Based on the upper and lower limit values of each component in the concentration adjustment boundary set under the working condition control and the combination mode, establish the feasible concentration intersection space structure of the component combination under the current process disturbance, integrate and label the concentration boundary in the combination, and construct a structured output form with upper and lower limit information and ratio constraint relationship, and generate a feasible concentration boundary set of the component.

[0011] As a further scheme of the present application, the specific steps of S4 are: S401: Call the component concentration combination in the feasible concentration boundary set of the component, extract the corresponding impact strength, ductility and electrical conductivity performance values, construct the sequence of performance values with concentration change, and construct the variation trend table of three performance parameters for each group of combination paths according to the variation direction of the performance item between the concentration combinations, and generate performance trend path feature information; S402: According to the impact strength and ductility data columns in the performance trend path feature information, calculate the Pearson correlation coefficient of each group of paths under the two performance items, judge whether the correlation coefficient is less than the negative correlation determination threshold, if yes, record the path as a negative correlation path, and include the index number into the path exclusion list, and obtain a negative correlation path exclusion index set; S403: Based on the negative correlation path elimination index set, the path set corresponding to the index in the component concentration combination is screened out, the remaining path set is renumbered and merged according to the concentration item order, a path set meeting the performance direction consistency is constructed as a candidate scheme, and a screened component path set is generated.

[0012] As a further scheme of the present application, the negative correlation determination threshold is set in the following manner: the mean and standard deviation of the Pearson correlation coefficient between the impact resistance and ductility of the verified secondary correlation path are calculated, the sum of the mean and standard deviation is taken as the negative correlation determination threshold, and if the Pearson correlation coefficient corresponding to the path is less than the negative correlation determination threshold, the path is marked as a negative correlation path.

[0013] As a further scheme of the present application, the specific steps of S5 are as follows: S501: Call each concentration combination in the screened component path set, extract the corresponding impact resistance, ductility and electrical conductivity three performance target values, and calculate the difference between each performance value and the corresponding performance target threshold, take the absolute value of the difference to construct a performance deviation vector, establish the performance deviation distribution of each group combination in three performance dimensions, and generate component combination performance deviation information; S502: According to each performance deviation vector in the component combination performance deviation information, multiply the same performance dimension item in the performance weight vector, and sum the product value to obtain the weighted performance score result corresponding to each concentration combination, and pair the score result with the combination number to construct an index sequence, and obtain a weighted score ranking index set; S503: Based on the top N component combination numbers in the weighted score ranking index set, where N is a positive integer, extract the corresponding concentration combination structure, and aggregate and reconstruct the concentration range, target performance label and combination path number information to construct a formula recommendation list with performance weight fitness and boundary concentration matching, and generate a compound optimization formula combination.

[0014] The compound formula intelligent optimization design system comprises: The concentration deviation analysis module obtains a plurality of target compound formula component combinations, extracts the performance variation amplitude between adjacent combinations according to the component concentration change trajectory, and if any performance value presents window shrinkage, overflow or mutation in continuous combinations, records the concentration interval as a performance response deviation section, and generates a deviation interval list. The characteristic comparison analysis module calculates the ratio difference between the disturbance amplitude of each pair of component concentrations and the target performance response change rate according to the deviation interval list, compares the ratio difference with the ratio under single component independent disturbance, judges whether there is deviation, records the corresponding concentration range and the affected performance type, and generates a response characteristic comparison item set. The concentration boundary adjustment module obtains a process disturbance parameter value under a current test working condition, calculates a numerical difference between a current process state and a target value according to a process fluctuation response sensitivity value of each component in the response characteristic reference set according to a corresponding performance item, linearly compresses original upper and lower limit values of the concentration, and generates a feasible concentration boundary set of the component; The component path screening module calls a concentration combination in the feasible concentration boundary set of the component, calculates a Pearson correlation coefficient between the impact strength and the ductility, and records the component path as an unsuitable path if any two performance parameters in the path are negatively correlated, thereby generating a screened component path set. The optimization formula construction module calculates a deviation degree of the corresponding performance target threshold value based on the combination data in the screened component path set, multiplies each deviation degree by a set performance weight vector, sums the results, screens a target number of combinations, extracts corresponding component concentrations, integrates recommended formula information with feasible concentration intervals and performance compatibility, and generates a compound optimization formula combination.

[0015] Compared with the prior art, the advantages and positive effects of the present application are as follows: In the present application, the performance change amplitude of adjacent component combinations is extracted, and the performance response deviation section is identified, so that the sensitive interval of the component concentration on the performance index can be identified. The deviation analysis of the concentration disturbance ratio between the components and the performance response is combined to determine the concentration coupling interference effect and the performance type, and the actual influence of the process disturbance parameter on the performance is stacked. The feasible concentration boundary with engineering constraints and performance compatibility is established. The performance negative correlation path exclusion mechanism is combined to effectively avoid the directional inconsistency of the combination path. The performance weight score is integrated with multiple performance targets. The combination with the smallest performance deviation degree is optimized under the premise of ensuring the rationality of the concentration. The performance compatible formula is accurately screened and recommended. The performance of the formula is improved. The efficiency and reliability of the component concentration optimization are improved. The matching ability of the compound formula under complex performance indicators is enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application. Those skilled in the art can also obtain other drawings according to these drawings without creating any creative labor.

[0017] Figure 1 The step flowchart of the present application is shown in the figure. Figure 2 The S1 refinement schematic diagram of the present application is shown in the figure. Figure 3 The S2 refinement schematic diagram of the present application is shown in the figure. Figure 4 S3 refinement schematic diagram of the present application; Figure 5 S4 refinement schematic diagram of the present application; Figure 6 S5 refinement schematic diagram of the present application; Figure 7 System module diagram of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the present application will be described below with reference to the drawings.

[0019] In the embodiments of the present application, the words such as "example", "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "example" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0020] In the embodiments of the present application, "image" and "picture" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding" and "corresponding" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0021] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1. When the distinction is not emphasized, the meanings expressed are consistent.

[0022] In order to make the technical problems, technical solutions and advantages of the present application more clear, the following will be described in detail with reference to the drawings and specific embodiments.

[0023] Please refer to Figure 1 The compound formula intelligent optimization design method provided in the embodiments of the present application comprises the following steps: S1: Obtain a plurality of target compound formula component combinations, extract three performance values of impact resistance, ductility and electrical conductivity according to corresponding performance detection data, extract the performance variation amplitude between adjacent combinations according to the component concentration change trajectory, and merge and analyze the performance response window under the disturbance path of the same component. If any performance value presents window shrinkage, overflow or mutation in continuous combinations, record the concentration interval as a performance response offset section, and generate an offset interval list; S2: According to the offset interval list, extract the two-component combination in the same interval and the corresponding performance difference value, calculate the ratio difference between the concentration disturbance amplitude of each pair of components and the target performance response rate of change, compare the ratio difference with the ratio under single-component independent disturbance, judge whether there is deviation, if there is, mark the component pair as coupled interference term, record the corresponding concentration range and the affected performance type, and generate the response characteristic control item set; S3: Obtain the process disturbance parameter value under the current test condition, including mixing temperature, reaction time and shear rate, calculate the numerical difference between the current process state and the target value according to the process fluctuation response sensitivity value of each component in the response characteristic control item set corresponding to the performance item, and generate the concentration adjustment coefficient of each component. Linear compression is performed on the upper and lower limit values of the original concentration to establish the feasible concentration space of the component combination, and the feasible concentration boundary set of the component is generated; S4: Call the concentration combination in the feasible concentration boundary set of the component, calculate the Pearson correlation coefficient between the impact strength and ductility according to the change direction between the three performance corresponding to the combination, judge whether the change direction constitutes a negative correlation, if any two performance parameters in the path show a negative correlation, record the component path as an unsuitable path, exclude the combination path, and keep the remaining combination as a candidate path, and generate the screened component path set; S5: Based on the combination data in the screened component path set, call the three performance target values corresponding to each combination, calculate the deviation degree of the corresponding performance target threshold in turn, and multiply each deviation degree by the set performance weight vector to generate the weighted performance score of each combination, screen the target number of combinations, extract the corresponding component concentration, integrate the recommended formula information with feasible concentration interval and performance compatibility, and generate the compound optimization formula combination; The offset interval list includes performance trend jump section, component concentration interference section and response interval abnormal overlap section, the response characteristic control item set includes high coupling sensitive component pair, performance index classification and concentration influence threshold range, the feasible concentration boundary set of the component includes concentration upper limit adjustment boundary, concentration lower limit compression boundary and concentration compatibility range between components, the screened component path set includes positive performance synergy path, negative performance exclusion path and feasibility path screening result, and the compound optimization formula combination includes preferred component concentration scheme, performance score weight list and performance index matching path set.

[0024] Please refer to Figure 2 , the specific steps of S1 are: S101: Obtain a plurality of target compound formula component combination, according to the corresponding performance detection data, extract the impact strength, ductility and conductivity three performance values, call concentration value and performance parameter, according to the difference of adjacent combination in the component disturbance path, screen the combination pair with the disturbance amplitude in the set concentration change interval, and construct the concentration change path sequence, generate the component disturbance trajectory data set; A plurality of target compound formula component combination is traversed, the concentration value of each component in each component combination is extracted, and is associated with the corresponding three performance values. According to the set component disturbance path, the adjacent combination pairs with small difference in concentration are screened. The specific process of screening is that a concentration change interval is set in advance, for example, the allowed range of concentration change of all components is set to one percentage point to three percentage points. All adjacent two component combinations in the disturbance path are traversed, and the absolute value of the difference value of the same component in the two combinations is calculated. If the absolute value of the difference value of all corresponding component concentration values falls within the preset concentration change interval, the combination pair composed of the two component combinations is determined as an effective combination pair and is retained. For example, the adjacent combination A and combination B on the path, the concentration of component 1 in combination A is ten percentage points, and the concentration of component 2 is twenty percentage points; The concentration of component 1 in combination B is eleven point two percentage points, and the concentration of component 2 is twenty-one percentage points. The concentration difference of component 1 is calculated to be one point two percentage points, and the concentration difference of component 2 is one percentage point. Since the two difference values are within the set interval of one percentage point to three percentage points, the combination pair composed of combination A and combination B is screened. On the contrary, if the concentration difference of any one component exceeds the interval, the combination pair is rejected. All screened effective combination pairs are connected according to the order in the original disturbance path to form one or more continuous concentration change path sequences. Each node in the path sequence represents a specific component combination, and the connection between the nodes represents the small disturbance of the component concentration. All constructed concentration change path sequences are summarized and stored together with the impact strength, ductility and conductivity performance values corresponding to each combination, to generate the component disturbance trajectory data set.

[0025] S102: Call the concentration path in the component disturbance trajectory data set, calculate the change amplitude value of the three performance parameters in the adjacent combination pair in turn, and judge according to the difference of the change direction and the difference of the change amplitude of the performance parameter, mark the path point position where the performance parameter appears the turning point of increase and decrease or the change rate exceeds the performance floating threshold, obtain the performance trend jump node list; The performance floating threshold is set in the following manner: the variation amplitude values of the three performance parameters of impact strength, ductility and conductivity under the same concentration perturbation path in the component perturbation trajectory data set are respectively counted, and the sum of the average value and the standard deviation of the variation amplitude values of the performance parameters under the path is taken as the performance floating threshold of the corresponding performance parameter; The judgment process of the change direction difference is specifically comparing the increase and decrease change directions of the same performance parameters in the previous group and the next group combination. If the change direction is changed from positive to negative or from negative to positive, it is determined that the direction is turned. The judgment process of the change amplitude size difference is specifically calculating the difference between the change amplitude values of the same performance parameters in the adjacent combination pair. If the difference exceeds the performance floating threshold, the path point is marked as a jump node. For each adjacent combination pair in the sequence, the variation range value of the three performance parameters is calculated in turn. The specific calculation process is as follows: for a combination pair (for example, combination A and combination B), the value of a certain performance parameter of combination B is extracted, and the value of the same performance parameter in combination A is subtracted to obtain the difference value, which is the variation range value of the performance parameter in the path segment from A to B. For example, the impact strength of combination A is fifty megapascals, and the impact strength of combination B is fifty-two megapascals, so the variation range value of the impact strength is positive two megapascals. Repeat this calculation for all adjacent combination pairs in the path to obtain the variation range value sequence of each performance parameter in each path segment. Next, the marking of the performance trend jump node is performed. The marking process includes two steps: variation direction difference judgment and variation range size difference judgment. Before judgment, the performance floating threshold needs to be set. The setting method of the threshold is as follows: for a certain performance parameter in a certain path, for example, the impact strength, the variation range values of the impact strength in all path segments of the path are counted, and the average value and the standard deviation of the variation range values are calculated. The sum of the average value and the standard deviation is used as the performance floating threshold of the impact strength in the path. For example, the average value of the impact strength variation range value in a certain path is one point five megapascals, and the standard deviation is zero point five megapascals, so the performance floating threshold is two megapascals. The performance floating thresholds of the ductility and the conductivity are also calculated in the same way. The judgment process of the variation direction difference is to compare the variation direction of the same performance parameter in two consecutive path segments. For example, the impact strength variation range values in the two consecutive path segments from A to B and B to C are compared. If the first value is positive (representing performance improvement) and the second value is negative (representing performance decline), or the first value is negative and the second value is positive, it is determined that a direction turning point occurs at point B, and point B is marked. The judgment process of the variation range size difference is to calculate the absolute value of the difference between the variation range values of the same performance parameter in two consecutive path segments. The difference value is compared with the preset corresponding performance floating threshold. If the difference value exceeds the performance floating threshold, the intermediate path point is also marked as a jump node. For example, the impact strength variation range from A to B is two megapascals, and the variation range from B to C is five megapascals, the difference between the two is three megapascals, and if the previously calculated impact strength performance floating threshold is two point five megapascals, then since three megapascals is greater than two point five megapascals, point B is also marked as a jump node. All path points marked due to direction turning or variation rate exceeding the threshold are collected to obtain the performance trend jump node list.

[0026] S103: Based on the node path segment located in the performance trend jump node list, backtrack the corresponding component disturbance path interval, mark the concentration segment where multiple jump nodes exist continuously as an abnormal response area, extract the concentration boundary range and performance type label according to the density of performance change amplitude in the abnormal response area, and generate a performance response offset interval list; All paths are checked, and the path segment where multiple jump nodes exist continuously in concentration is marked as an abnormal response area. The specific criterion for marking is that if the number of jump nodes exceeds a preset minimum number threshold (for example, set to three nodes) in a continuous concentration change interval, the entire concentration interval is defined as an abnormal response area. After determining the abnormal response area, further analyze the density of performance change in the area. The evaluation method of the density is to calculate the number of performance jump nodes contained in the unit concentration change range in the abnormal response area. For example, the total concentration change range of an abnormal response area is five percentage points, and the area contains five jump nodes, so the density is one node per percentage point. According to the level of density, the abnormal response area is classified. At the same time, the starting component concentration and the ending component concentration of each abnormal response area are extracted, and these two concentration values constitute the concentration boundary range of the area. In addition, according to the main performance parameter (i.e. the performance parameter with the most jump times) that causes the jump in the area, add a performance type label to the area, such as "impact strength dominant" or "multiple performance parameters jointly affect". Structurally organize the concentration boundary range and performance type label of all identified abnormal response areas to generate a performance response offset interval list.

[0027] Please refer to Figure 3 The specific steps of S2 are as follows: S201: Based on the concentration interval marked in the performance response offset interval list, extract the effective component combination in each interval, and pair the components in the combination two by two, call the concentration value sequence of each component pair in the interval and the corresponding performance value sequence of impact strength, ductility and electrical conductivity, construct a component concentration pair and performance difference pair array, and generate component disturbance and performance difference mapping information; For each combination of components within a concentration interval, all components within the combination are paired two by two. For example, a combination containing component 1, component 2, and component 3 will generate three pairs of components: (component 1, component 2), (component 1, component 3), and (component 2, component 3). For each pair of components, within the concentration interval in which the pair is located, the respective concentration value sequence of the two components is extracted, as well as the performance value sequence of impact resistance, ductility, and electrical conductivity corresponding to each component combination within the interval. These sequences are aligned and integrated to construct a component concentration pair and performance difference control array. Each row record in the array contains the specific concentration values of a pair of components, and the specific values of the three performances exhibited by the entire formulation at this point. For example, one row record can be: component 1 concentration is ten percentage points, component 2 concentration is twenty percentage points, corresponding impact resistance is fifty-five megapascals, ductility is three hundred percent, and electrical conductivity is one hundred siemens per meter. By processing all component combinations within a concentration interval in this way, a detailed control array reflecting the relationship between the concentration of component pairs and the overall performance of the formulation is formed. Repeat the above process for all concentration intervals and all component pairs to ultimately generate complete component perturbation and performance difference mapping information.

[0028] S202: According to each row record in the component perturbation and performance difference mapping information, calculate the ratio between the concentration perturbation amplitude of the corresponding component pair and the target performance response change rate, calculate the difference between the ratio and the corresponding performance ratio generated when the same component is independently perturbed in a non-interaction path, and compare the difference with a coupling determination threshold value, establish a set of component difference determination results, and obtain a list of coupling interference deviation results; In the two consecutive records of the same component pair, the difference of the target performance parameter (e.g. impact strength) is taken, and the difference of the two component concentration changes (e.g. the sum of the absolute values of the two component concentration differences) is taken, and then the performance difference is divided by the sum of the concentration difference, to obtain a response change rate. In order to establish the difference value determination result set, a reference value is needed for comparison. This reference value is the corresponding performance ratio generated by the independent disturbance of the same component in the non-interaction path. The way to obtain the reference value is to change only the concentration of a single component in an independent experiment, measure the change in performance, calculate the ratio of performance change to concentration change, and take the average of multiple experimental results as the performance ratio of the independent disturbance of the component. Then, the difference value is calculated, that is, the response change rate calculated in the interaction path is subtracted from the reference performance ratio. After obtaining the difference value, the difference value is compared with the preset coupling determination threshold value. The basis for setting the coupling determination threshold value is a large amount of historical experimental data of non-coupling action. The difference between the response change rate and the reference performance ratio caused by non-coupling factors such as experimental errors in these data is statistically distributed, and the 95th percentile of the distribution is taken as the coupling determination threshold value. For example, if the historical data shows that the difference value of non-coupling disturbance is mostly less than 0.5, the coupling determination threshold value can be set to 0.5. If the calculated difference value exceeds the coupling determination threshold value, it is determined that the two components have coupling interference at this concentration point. Record the determination results (“coupling interference exists” or “coupling interference does not exist”) of all component pairs at all points, establish the difference value determination result set between components, and select all records determined to have coupling interference from it to form the coupling interference deviation result list.

[0029] S203: Based on the component pairs marked as the deviation value exceeding the coupling determination threshold in the coupling interference deviation result list, record the corresponding concentration disturbance interval and the target performance parameter label that causes the deviation, and number and classify the component pairs to establish a data structure set with coupling behavior characteristics and performance impact identification, and generate a response characteristic reference item set; Each component pair exhibiting coupling behavior is uniquely numbered, for example, the combination of “component 1 and component 2” is numbered as “CP001”. At the same time, it is classified and managed according to the performance parameters that are jointly affected, for example, all coupling component pairs that mainly affect impact strength are divided into the “impact performance coupling class”. Through such arrangement, a data structure set is established, each item of which clearly describes a component pair, its concentration range where coupling occurs, the type of performance affected, and a unique identifier. This set is the response characteristic reference item set, which stores the interaction characteristics between components and their specific impact on the performance of the final product.

[0030] Please refer to Figure 4 , the specific steps of S3 are: S301: Obtain the process disturbance parameter values under the current test working condition, including the mixing temperature, reaction time and shear rate, and calculate the difference value of each parameter and the target set value, call the process sensitivity factor of the corresponding performance item of the component in the response characteristic reference item set, multiply the numerical difference between each process parameter and the target set value by the corresponding sensitivity factor, establish the performance influence numerical table of the component corresponding to the current working condition, and generate the working condition adaptability difference value set; The target value of the parameter is set in the standard process procedure. For each process parameter, calculate the difference value between its current measured value and the target set value. For example, if the current mixing temperature is measured as one hundred and eighty-five degrees Celsius, and the target set value is one hundred and eighty degrees Celsius, then the temperature difference is positive five degrees Celsius. Next, call the response characteristic reference item set generated in the previous step, and extract the process sensitivity factor associated with each component and the target performance item from it. The process sensitivity factor is obtained by statistical analysis of previous experimental data, which quantifies the degree of performance change under the influence of a specific component when the unit process parameter changes. For example, for the performance item "Component A affects impact strength", the "temperature sensitivity factor" may be set to "negative zero point two megapascals per degree Celsius change in impact strength", which is based on historical production data analysis and statistics of the average change rate of the impact strength of a formula containing component A when only the temperature is changed. Then, multiply the process parameter difference value calculated in the first step by the corresponding sensitivity factor. In the previous example, the temperature difference of positive five degrees Celsius multiplied by the sensitivity factor of negative zero point two megapascals per degree Celsius results in an impact strength influence value of component A under the current working condition of negative one megapascal. Repeat this calculation for all process parameters (mixing temperature, reaction time, shear rate) and all affected component performance items, and compile all the calculation results to establish a detailed performance influence numerical table of the component corresponding to the current working condition, which is the working condition adaptability difference value set.

[0031] S302: Calculate the concentration adjustment scaling factor of the component according to the performance influence value of the corresponding performance item of the component in the working condition adaptability difference value set, and multiply the scaling factor by the upper and lower limits of the original set concentration of the component respectively, take the result and perform numerical comparison with the original boundary value and execute linear scaling to obtain the adjustment range of the concentration of the component under the current working condition, and obtain the concentration adjustment boundary set under the working condition control; The calculation logic of the scaling factor is based on the principle of performance compensation, that is, according to the negative or positive impact on a performance item, the concentration of the component that can affect the performance is adjusted inversely. The specific calculation of the scaling factor is carried out according to a preset "concentration-performance" response model, which describes how much performance change a unit concentration change can bring. For example, if the model shows that the concentration of component A increases by one percentage point, the impact strength increases by 0.5 MPa, and the current working condition causes the impact strength to decrease by 1 MPa, in order to compensate for the 1 MPa decrease, the concentration of component A needs to be increased by 2 percentage points. This "2 percentage points" is the concentration adjustment amount calculated based on the performance impact value, which is compared with the original set concentration of the component to obtain a scaling factor. The calculated concentration adjustment scaling factor is multiplied by the upper and lower limits of the concentration set of the component in the original formula design. For example, the original concentration range of component A is 8% to 12%, and if the calculated adjustment scaling factor is 110%, the preliminary adjusted range is 8.8% to 13.2%. Subsequently, linear scaling operation is performed to check the preliminary adjusted boundary values with the constraint condition that the total concentration of the formula system is 100% and the adjustment of other components, to ensure that the adjusted total concentration is still conserved, and the concentration of a single component does not exceed its physical or chemical limit. Through such calculation and scaling, the dynamic adjustment range of each component under the current specific working condition is finally obtained, and these ranges are collected to obtain the concentration adjustment boundary set under the working condition control.

[0032] S303: Based on the upper and lower limit values of each component in the concentration adjustment boundary set under the working condition control and the combination mode, a feasible concentration cross-space structure of the component combination under the current process disturbance is established, and the concentration boundaries in the combination are integrated and labeled to construct a structured output table with upper and lower limit information and ratio constraint relationship, and a feasible concentration boundary set of the component is generated; The concentration adjustment boundary set under the condition regulation sets the upper and lower limit values of the concentration for each component, combined with the combination mode of all components in the formula, to establish a multi-dimensional space structure describing all feasible component concentration combinations under the current process disturbance. This cross-space structure defines all effective formula points that meet the concentration boundary after the current condition adjustment. The specific construction process is to regard the adjusted concentration range of each component as a dimension, and the space formed by all component dimensions is the feasible concentration cross-space. The combinations in this space are integrated and labeled, for example, when the concentration of component A takes its upper limit value, the feasible concentration ranges of components B and C are clearly marked, while ensuring that the sum of all component concentrations always equals a fixed value (such as one hundred percent). This process generates a series of matching constraints. Finally, all adjusted concentration upper and lower limit information of the components and matching constraints are integrated into a structured output table. This table clearly lists the recommended concentration range of each component under the current process disturbance and the concentration linkage between different components. This table is the final generated component feasible concentration boundary set.

[0033] Please refer to Figure 5 The specific steps of S4 are: S401: Call the component concentration combination in the component feasible concentration boundary set, extract the corresponding impact strength, ductility, and electrical conductivity performance values, construct the performance value change sequence with concentration, and construct the performance trend path characteristic information according to the change direction of the performance item between concentration combinations, generate performance trend path characteristic information; For each concentration combination point on each path, extract the predicted or measured values of the three performance values of impact strength, ductility, and electrical conductivity from the original performance database or through the performance prediction model. In this way, each concentration combination path corresponds to three performance value sequences with concentration. Next, according to the change direction of the three performance sequences between adjacent combinations on the path, a three-performance parameter change trend table is constructed. The specific construction method is that for two adjacent combination points (such as combination P1 and combination P2) on the path, compare the three performance values of P2 and P1. If the impact strength value increases from P1 to P2, record "increase" in the table; if it decreases, record "decrease"; if it does not change, record "flat". The same operation is performed for ductility and electrical conductivity. The change trend (such as "increase, decrease, increase") of all adjacent points on each path is concatenated to form the performance trend path characteristic information of the path.

[0034] S402: According to the impact resistance and ductility data column in the performance trend path characteristic information, calculate the Pearson correlation coefficient of each group path under two performance items, judge whether the correlation coefficient is less than the negative correlation determination threshold, if it is, record the path as a negative correlation path, and include the index number into the path exclusion list, and obtain the negative correlation path elimination index set; The setting method of the negative correlation determination threshold is to calculate the mean and standard deviation of the Pearson correlation coefficient between the impact resistance and ductility of the verified auxiliary correlation path, and to take the sum of the mean and standard deviation as the negative correlation determination threshold. If the Pearson correlation coefficient of the path is less than the negative correlation determination threshold, the path is marked as a negative correlation path. The calculation process of Pearson correlation coefficient is a standard statistical method for evaluating the linear correlation degree of two groups of data. The result ranges from negative one to positive one, negative one indicates complete negative correlation, and positive one indicates complete positive correlation. After calculation, the obtained coefficient value is compared with a preset negative correlation determination threshold. The setting method of the negative correlation determination threshold is as follows: first, collect a batch of material formula paths which have been proved to have significant negative correlation in priori knowledge or past experiments, calculate the Pearson correlation coefficient between the impact resistance and ductility of these paths, and obtain a set of coefficient values. Then, calculate the mean and standard deviation of the coefficient values. Add the calculated mean and standard deviation to obtain the sum as the negative correlation determination threshold. For example, if the mean of the coefficients of the verified negative correlation paths is negative zero point seven and the standard deviation is zero point one, the negative correlation determination threshold is set to negative zero point eight. When judging, if the Pearson correlation coefficient calculated for a candidate path is less than negative zero point eight (for example, negative zero point eight five), it is determined that the path is a negative correlation path. Record the index number of all paths determined to be negative correlation paths, and include them in a path exclusion list to finally obtain the negative correlation path elimination index set.

[0035] S403: Based on the negative correlation path elimination index set, remove the path group corresponding to the index from the component concentration combination, and renumber and merge the remaining path set according to the concentration item order to construct a path set meeting the performance direction consistency as a candidate scheme, and generate the screened component path set; All candidate component concentration combination paths are screened. The entire path group corresponding to the path number included in the index set is removed from the candidate set. After screening, the remaining path set meeting the performance direction consistency (i.e. there is no strong negative correlation between impact resistance and ductility) is renumbered and merged according to the original order or new logical order of the concentration items. This screened and recombined path set is constructed as a candidate scheme set meeting the specific performance coordination requirements. The set is finally formatted and output to generate the screened component path set for subsequent optimization selection.

[0036] Please refer to Figure 6 The specific steps of S5 are as follows: S501: For each concentration combination in the filtered component path set, extract the corresponding performance target values of impact resistance, ductility, and electrical conductivity, and calculate the difference between each performance value and the corresponding performance target threshold value. Take the absolute value of the difference to construct a performance deviation vector, establish the performance deviation distribution of each combination in the three performance dimensions, and generate component combination performance deviation information. For each concentration combination, extract the target values of the three performances of impact resistance, ductility, and electrical conductivity. At the same time, obtain the target threshold values set for the three performances in advance, which represent the minimum performance standard or ideal performance value that the product needs to ultimately achieve. Calculate the difference between each performance actual value and its corresponding performance target threshold value. For example, if the impact resistance of a combination is measured as fifty-eight megapascals, and the target threshold value is sixty megapascals, the difference is negative two megapascals. Take the absolute value of this difference, which is two megapascals, as the deviation of this performance item. Perform the same operation on ductility and electrical conductivity to obtain two other deviation values. Combine the three deviation values (e.g., two megapascals, five percent, ten siemens per meter) into a three-dimensional performance deviation vector. Repeat this process for each concentration combination in the path set to establish a vector for each combination that quantifies its deviation from the final performance target. Aggregate all these vectors to generate component combination performance deviation information, which visually displays the deviation distribution of each candidate combination from the expected target in the three key performance dimensions.

[0037] S502: According to each performance deviation vector in the component combination performance deviation information, multiply the same performance dimension item in the performance weight vector, and sum the product values to obtain the weighted performance score result corresponding to each concentration combination. Pair the score result with the combination number to construct an index sequence and obtain a weighted score ranking index set. The performance weight vector is set according to the application requirements and market positioning of the specific product, and different importance is assigned to the three performances. For example, for an application that emphasizes safety, the performance weight vector can be set as: impact strength weight 0.5, ductility weight 0.3, and electrical conductivity weight 0.2, and the sum of the weights of the three is 1. Then, each element in the performance deviation vector is multiplied by the weight value in the performance weight vector in the same performance dimension. Taking the foregoing as an example, the deviation of the impact strength "2 MPa" will be multiplied by the weight "0.5". After performing this multiplication operation on the three performances, the three product values are added. The final sum value is the weighted performance score result corresponding to the concentration combination, and the lower the score represents the closer the comprehensive performance to the weighted target. After calculating the weighted performance score of each concentration combination, the score result is paired with the unique number of the combination, and an index sequence is constructed in the order of score value from low to high. This sequence is the weighted score ranking index set.

[0038] S503: Group the component combinations whose score values are ranked in the top N positions in the weighted score ranking index set, where N is a positive integer, extract the corresponding concentration combination structure, and aggregate and reconstruct the concentration range, target performance label, and combination path number information to construct a formula recommendation list with performance weight matching degree and boundary concentration matching, and generate a compound optimization formula combination; Extract the component combination numbers whose score values are ranked in the top N positions. N is a positive integer set according to actual needs, for example, if three alternatives are needed, N is three. According to these top-ranking numbers, the complete concentration combination structure corresponding to the original data set is accurately extracted, including the specific concentration value or concentration range of each component. Then, the information of these preferred solutions is aggregated and reconstructed, and the concentration information, target performance label (impact strength, ductility, electrical conductivity) relied on in the screening process, and the combination path number are integrated together. Through such aggregation and formatting, a formula recommendation list with high performance weight matching degree and good boundary concentration matching is constructed. This list not only gives the specific formula composition, but also attaches the high matching proof of its performance to the preset target, and finally generates a compound optimization formula combination that can be directly referred to by R&D or production personnel.

[0039] Please refer to Figure 7 , the compound formula intelligent optimization design system, the system comprises: The concentration deviation analysis module obtains a plurality of target compound formula component combinations, extracts the performance variation amplitude between adjacent combinations according to the component concentration change trajectory, and records the concentration interval as a performance response deviation section if any performance value presents window shrinkage, overflow or mutation in continuous combinations, and generates a deviation interval list; The characteristic comparison analysis module calculates the ratio difference between the concentration disturbance amplitude of each component pair and the target performance response change rate according to the offset interval list, compares the ratio difference with the ratio under single component independent disturbance, judges whether there is deviation, records the corresponding concentration range and the affected performance type, and generates a response characteristic comparison item set; The concentration boundary adjustment module obtains the process disturbance parameter value under the current test working condition, calculates the numerical difference between the current process state and the target value according to the process fluctuation response sensitivity value of each component in the response characteristic comparison item set corresponding to the performance item, linearly compresses the upper and lower limit values of the original concentration, and generates a component feasible concentration boundary set; The component path screening module calls the concentration combination in the component feasible concentration boundary set, calculates the Pearson correlation coefficient between the impact resistance and the ductility, and records the component path as an unsuitable path if any two performance parameters in the path are negatively correlated, and generates a screened component path set. The optimization formula construction module calculates the deviation degree of the corresponding performance target threshold based on the combination data in the screened component path set, multiplies each deviation degree by the set performance weight vector, sums the results, screens a target number of combinations, extracts the corresponding component concentrations, integrates the recommended formula information with feasible concentration intervals and performance compatibility, and generates a compound optimization formula combination.

[0040] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for intelligent optimization design of compound formulations, characterized in that, Includes the following steps: S1: Obtain multiple target compound formulation component combinations, extract the performance variation range between adjacent combinations according to the component concentration change trajectory, and if any performance value shows window contraction, overflow or abrupt change in continuous combinations, record the concentration range as the performance response offset segment and generate an offset range list. S2: Based on the offset interval list, calculate the ratio difference between the concentration perturbation amplitude of each pair of components and the target performance response change rate, compare the ratio difference with the ratio under single-component independent perturbation, determine whether there is a deviation, record the corresponding concentration range and the type of performance impact, and generate a response characteristic comparison itemset; S3: Obtain the process disturbance parameter values ​​under the current test conditions, calculate the numerical difference between the current process state and the target value for each component in the response characteristic comparison item set according to the process fluctuation response sensitivity value of the corresponding performance item, linearly compress the original upper and lower limits of concentration, and generate a feasible concentration boundary set of the component. S4: Call the concentration combination in the feasible concentration boundary set of the component, calculate the Pearson correlation coefficient between impact strength and ductility. If any two performance parameters in the path are negatively correlated, record the component path as a path not recommended for adoption, and generate a set of filtered component paths. S5: Based on the combination data in the selected component path set, calculate the deviation of the corresponding performance target threshold, and sum the results by multiplying each deviation by the set performance weight vector to select the combination of target quantity, extract the corresponding component concentration, integrate the recommended formulation information with feasible concentration range and performance compatibility, and generate the optimized compound formulation combination.

2. The intelligent optimization design method for compound formulations according to claim 1, characterized in that, The offset interval list includes performance trend jump segments, component concentration interference segments, and abnormally overlapping response interval segments. The response characteristic comparison itemset includes highly coupled sensitive component pairs, performance index classifications, and concentration influence threshold ranges. The component feasible concentration boundary set includes upper concentration adjustment boundaries, lower concentration compression boundaries, and concentration compatibility ranges between components. The screened component path set includes positive performance synergy paths, negative performance exclusion paths, and feasible path screening results. The compound optimized formulation combination includes preferred component concentration schemes, performance score weight lists, and performance index matching path sets.

3. The intelligent optimization design method for compound formulations according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain multiple target compound formulation component combinations, extract three performance values ​​of impact strength, ductility and electrical conductivity based on the corresponding performance test data, call the concentration value and performance parameters, filter the combination pairs whose perturbation amplitude is within the set concentration change interval based on the difference in concentration between adjacent combinations in the component perturbation path, construct the concentration change path sequence, and generate the component perturbation trajectory dataset. S102: Call the concentration path in the component perturbation trajectory dataset, calculate the change amplitude of the three performance parameters in adjacent combination pairs in turn, and make a joint judgment based on the difference in the direction of change and the difference in the magnitude of change of the performance parameters, mark the path points where the performance parameters show a change in direction of increase or decrease or the rate of change exceeds the performance fluctuation threshold, and obtain a list of performance trend jump nodes. S103: Based on the node path segment located in the performance trend jump node list, backtrack the corresponding component disturbance path interval, and mark the concentration segment with multiple consecutive jump nodes as an abnormal response region. According to the density of performance change amplitude in the abnormal response region, extract the concentration boundary range and performance type label to generate a performance response offset interval list.

4. The intelligent optimization design method for compound formulations according to claim 3, characterized in that, The performance fluctuation threshold is set by statistically analyzing the variation amplitude values ​​of the three performance parameters, impact strength, ductility and conductivity, under the same concentration perturbation path in the component perturbation trajectory dataset, and using the sum of the average value and standard deviation of the variation amplitude values ​​of the performance parameters under the path as the performance fluctuation threshold of the corresponding performance parameter. The specific process for judging the difference in the direction of change is as follows: compare the direction of increase or decrease of the same performance parameter in the previous group and the next group. If the direction of change changes from positive to negative or from negative to positive, it is determined to be a change in direction. The process of determining the magnitude difference of the change is as follows: calculate the difference between the change magnitude values ​​of the same performance parameter in adjacent combination pairs. If the difference exceeds the performance fluctuation threshold, mark the path point as a jump node.

5. The intelligent optimization design method for compound formulations according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the concentration ranges marked in the performance response offset range list, extract the effective component combinations in each range, pair the components in the combination, call the concentration value sequence of each component pair in the range and the corresponding performance value sequence of impact strength, ductility and conductivity, construct the component concentration pair and performance difference comparison array, and generate component perturbation and performance difference mapping information. S202: Based on each record in the component perturbation and performance difference mapping information, calculate the ratio between the concentration perturbation amplitude of the corresponding component pair and the target performance response change rate, calculate the difference between the ratio and the corresponding performance ratio generated when the same component is independently perturbed in a non-interactive path, compare the difference with the coupling judgment threshold, establish a component difference judgment result set, and obtain a coupling interference deviation result list. S203: Based on the component pairs marked as having deviation values ​​exceeding the coupling judgment threshold in the coupling interference deviation result list, record the corresponding concentration disturbance range and the target performance parameter label that caused the deviation, and number and classify the component pairs to establish a data structure set with coupling behavior characteristics and performance impact identifiers, and generate a response characteristic comparison itemset.

6. The intelligent optimization design method for compound formulations according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Obtain the process disturbance parameter values ​​under the current test conditions, including mixing temperature, reaction time and shear rate, and calculate the difference between each parameter and the target set value. Call the process sensitivity factor of the corresponding performance item of the component in the response characteristic comparison item set, and multiply the numerical difference between each process parameter and the target set value by the corresponding sensitivity factor to establish a performance influence table of the component and the current conditions, and generate a set of condition adaptability difference values. S302: Based on the performance impact value of the corresponding performance item of the component in the working condition adaptability difference set, calculate the concentration adjustment ratio factor of the component, and multiply the ratio factor with the original set upper and lower limits of the component concentration respectively. Compare the result with the original boundary value and perform linear scaling to obtain the adjustment range of the component concentration under the current working condition, and obtain the concentration adjustment boundary set under the working condition control. S303: Based on the upper and lower limits and combination methods of each component in the concentration adjustment boundary set under the aforementioned operating condition control, establish a feasible concentration cross space structure for component combinations under the current process disturbance, integrate and label the concentration boundaries in the combination, construct a structured output form with upper and lower limit information and ratio constraint relationship, and generate a feasible concentration boundary set for the components.

7. The intelligent optimization design method for compound formulations according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Call the component concentration combinations in the feasible concentration boundary set of the components, extract the corresponding three performance values ​​of impact strength, ductility and conductivity, construct the sequence of performance values ​​changing with concentration, and construct the change trend comparison table of the three performance parameters for each combination path according to the direction of change of performance items between concentration combinations, and generate performance trend path feature information. S402: Based on the impact strength and ductility data columns in the performance trend path feature information, calculate the Pearson correlation coefficient of each path under the two performance items, determine whether the correlation coefficient is less than the negative correlation judgment threshold, if it is true, record the path as a negative correlation path, and add the index number to the path exclusion list to obtain the negative correlation path removal index set. S403: Based on the negative correlation path elimination index set, filter out the path groups corresponding to the index in the component concentration combination, and renumber and merge the remaining path set according to the order of concentration items to construct a path set that satisfies the consistency of performance direction as a candidate scheme, and generate the filtered component path set.

8. The intelligent optimization design method for compound formulation according to claim 7, wherein the negative correlation judgment threshold is set by calculating the mean and standard deviation of the Pearson correlation coefficient between impact strength and ductility in the verified negative correlation path, and using the sum of the mean and standard deviation as the negative correlation judgment threshold. If the Pearson correlation coefficient corresponding to the path is less than the negative correlation judgment threshold, the path is marked as a negative correlation path.

9. The intelligent optimization design method for compound formulations according to claim 1, characterized in that, The specific steps of S5 are as follows: S501: Call each concentration combination in the filtered component path set, extract the corresponding three performance target values ​​of impact strength, ductility and conductivity, calculate the difference between each performance value and the corresponding performance target threshold, take the absolute value of the difference to construct the performance deviation vector, establish the performance deviation distribution of each combination under the three performance dimensions, and generate component combination performance deviation information. S502: Based on the performance deviation vector of each group in the component combination performance deviation information, multiply it with the same performance dimension item in the performance weight vector, and sum the product values ​​to obtain the weighted performance score result corresponding to each concentration combination. Then, pair the score result with the combination number to construct an index sequence and obtain the weighted score sorting index set. S503: Based on the component combination numbers ranked Nth by the weighted scoring sorting index set, where N is a positive integer, extract the corresponding concentration combination structure, and aggregate and reconstruct the concentration range, target performance label and combination path number information to construct a formulation recommendation list with performance weight fit and boundary concentration matching, and generate compound optimized formulation combinations.

10. A compound formulation intelligent optimization design system, characterized in that, The system is used to implement the intelligent optimization design method for compound formulations according to any one of claims 1-9, and the system includes: The concentration offset analysis module acquires multiple target compound formulation component combinations, extracts the performance variation range between adjacent combinations according to the component concentration change trajectory, and if any performance value shows window contraction, overflow or abrupt change in consecutive combinations, the concentration range is recorded as the performance response offset segment, and a list of offset intervals is generated. The characteristic comparison analysis module calculates the ratio difference between the concentration perturbation amplitude of each pair of components and the target performance response change rate according to the offset interval list, compares the ratio difference with the ratio under single component independent perturbation, determines whether there is a deviation, records the corresponding concentration range and the type of performance impact, and generates a response characteristic comparison itemset. The concentration boundary adjustment module obtains the process disturbance parameter values ​​under the current test conditions, calculates the numerical difference between the current process state and the target value for each component in the response characteristic comparison item set according to the process fluctuation response sensitivity value of the corresponding performance item, linearly compresses the original upper and lower limits of concentration, and generates a feasible concentration boundary set for the component. The component path screening module calls the concentration combinations in the feasible concentration boundary set of the component, calculates the Pearson correlation coefficient between impact strength and ductility, and records the component path as a path not recommended for adoption if any two performance parameters in the path are negatively correlated, and generates a set of screened component paths. The optimized formulation construction module calculates the deviation of the corresponding performance target threshold based on the combination data in the set of screened component paths. It then multiplies each deviation by a set performance weight vector and sums the results to screen the combination of target quantities, extract the corresponding component concentrations, integrate recommended formulation information with feasible concentration ranges and performance compatibility, and generate optimized compound formulation combinations.

Images