A cement proportioning method based on bionic algorithm and solid waste preferential disposal driving

Abstract

The application is a kind of cement batching method based on solid waste preferential assimilation driving of bionic algorithm. The method comprises the following steps: integrating solid waste raw material component data into attribute matrix; obtaining optimal solid waste raw material ratio through bionic algorithm optimization calculation represented by non-dominated sorting multi-objective evolutionary algorithm; simultaneously achieving maximum assimilation target waste residue amount and balancedly approaching expected three rate value target; bionic algorithm optimization calculation obtains optimal batching ratio, guiding actual cement production. In the optimization process, normalized actual cement production feed data is randomly used as initial guess value, accelerating calculation convergence speed. The method meets the maximum assimilation of solid waste raw material in the process of producing cement by calcining calcium carbide method produced chloroethylene waste residue. Effectively solves the problem of unstable cement production performance caused by large fluctuation of solid waste raw material in the process of cement production using industrial solid waste as solid waste raw material, simultaneously realizes preferential assimilation of large solid waste inventory, and is also applicable to effective utilization of large solid waste raw material.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of solid waste resource utilization, and particularly relates to a cement batching method based on solid waste preferential disposal driving of a bionic algorithm. BACKGROUND

[0002] In recent years, with the rapid growth of urban solid waste and large industrial solid waste in China, the use of these solid wastes for cement rotary kiln collaborative treatment and disposal has attracted more and more attention, and the collaborative treatment and disposal amount of household garbage, municipal sludge and general industrial solid waste has also increased day by day. The amount of cement kiln collaborative treatment and disposal of hazardous solid waste accounts for more than 45% of the amount of hazardous waste incineration.

[0003] Taking the cement rotary kiln collaborative treatment of industrial solid waste as an example, in recent years, through in-depth analysis of the raw material composition and physical and chemical properties of various industrial waste residues, it is found that industrial solid waste can be used as raw material for cement production. The production of waste residue cement, taking polyvinyl chloride waste residue as an example, is mainly to use calcium slag as the main calcium raw material, and to add appropriate amounts of various other calcium, clay and iron waste residues for full waste residue cement clinker calcination. Then, desulfurization ash or citric acid residue is used to replace natural gypsum, and fly ash, coal gangue and limestone chips are used as mixed materials to produce full waste residue portland cement. After continuous adjustment of the design scheme, reasonable matching of raw materials, pre-production test, analogy screening, process selection, preliminary homogenization, classification storage, effective metering, production preparation, routine monitoring and two-milling-one-burning process, qualified cement products are obtained, and full waste residue portland cement clinker and full waste residue portland cement that meet national cement standards are successfully obtained. These full waste residue raw materials are mainly derived from industrial solid waste generated by some upstream industries such as chemical industry, calcium industry, thermal power industry and biochemical companies, such as calcium raw materials like calcium slag, calcium furnace dust, lime slag, limestone chips, and purification ash, clay raw materials like fly ash, slag and coal gangue, correction raw materials like sulfuric acid slag, copper slag and iron tailings, and retarder raw materials like desulfurization gypsum and citric acid slag. Compared with traditional cement processes, full waste residue cement has many types of raw materials and complex composition, and the amount of waste residue fluctuates greatly with the operation of upstream enterprises, which also leads to uneven performance and quality of the produced cement, which is a great challenge for cement raw material batching.

[0004] In the field of cement feed ratio, the cement raw material batching often adopts the Excel artificial trial table method, mainly using the constraint conditions of three rate values to calculate the cement compatibility; in Excel, linear / nonlinear programming module can also be used for optimization solution, but in actual situation, the method often cannot obtain the optimal solution. At the same time, after the incorporation of solid waste, the raw material components and types of cement process are more complex than ordinary Portland cement, although technical personnel have explored for many years and have accumulated some practical experience of waste slag cement raw material batching, but the method of using Excel table for cement raw material compatibility calculation is low in efficiency, and cannot well solve the problems of waste slag feed ratio and uneven quality of cement products under the condition of large fluctuation of raw materials. Although there are related reports that genetic algorithm is used for optimization control of traditional Portland cement raw material ratio, there are still many problems. The genetic algorithm reported previously is mainly applied to the optimization of traditional cement batching method, which is mainly for the simple traditional cement production process; and sometimes the consumption of some large inventory solid waste is small in a batch of solutions obtained by Excel linear / nonlinear programming, which is difficult to meet the demand of maximum consumption of large industrial solid waste such as carbide slag.

[0005] The case cited in the application adopts an evolutionary algorithm based on non-dominated sorting multi-objective optimization, which is often used to seek the optimal solution of multiple objective functions. The main difference between the selected algorithm and the ordinary genetic algorithm is:

[0006] (1) The method of fast non-dominated sorting is adopted, which greatly reduces the computational complexity;

[0007] (2) The crowding degree and crowding degree comparison operator are defined, which replaces the need to specify the sharing radius, so that the individuals in the quasi-Pareto domain can be expanded to the entire Pareto domain, and are uniformly distributed, maintaining the diversity of the population;

[0008] (3) The elite strategy is introduced, which expands the sampling space, prevents the loss of the best individual, and improves the operation speed of the algorithm.

[0009] The algorithm is often used to optimize multi-objective problems, and the optimization effect will be greatly improved. There is no case of applying this algorithm to the calculation of industrial solid waste cement raw material compatibility with complex components before, and the application of this bionic algorithm program to the raw material ratio optimization problem of waste slag cement process is the first time. However, there are still many technical problems if this algorithm is simply applied to the calculation of industrial solid waste cement raw material compatibility with more complex components. Therefore, the bionic algorithm program is further improved according to the characteristics of solid waste raw materials, and the bionic optimization of the raw material ratio of waste slag cement process is successfully realized. SUMMARY

[0010] The application provides a cement batching method based on a bionic algorithm and driven by solid waste preferential treatment, and solves the problems of large feed fluctuation, unstable cement production quality and imperfect three rate value calculation method when industrial waste slag is used as raw material to produce cement. Specifically, the following steps are included:

[0011] Step 1, select several groups of data of solid waste raw materials for producing cement, which includes the types of solid waste raw materials, the key component parameters of solid waste raw materials, and the content proportions of the main five oxides CaO, Al2O3, Fe2O3, SiO2 and SO3 in them, including the content percentage of each waste slag in the total feed, integrating these data, integrating the solid waste raw material component data into an attribute matrix, and taking the feed ratio of each raw material as an optimized vector. Specifically, all types of feed are matrix blocks A, and A is a 5*n matrix. For example, the first column of matrix A is carbide slag, the second column is mixed material, and the subsequent columns are other solid wastes, etc. Vector represents the feed percentage of each raw material in the cement raw material, and vector represents the total amount of the five oxides in the cement feed, and the formula is as follows:

[0012]

[0013] Among them, C1 represents the mass percentage of CaO in carbide slag, C2 represents the mass percentage of CaO in mixed material, and so on; S1 represents the mass percentage of SiO2 in carbide slag, S2 represents the mass percentage of SiO2 in mixed material, and so on; F1 represents the mass percentage of Fe2O3 in carbide slag, F2 represents the mass percentage of Fe2O3 in mixed material, and so on; A1 represents the mass percentage of Al2O3 in carbide slag, A2 represents the mass percentage of Al2O3 in mixed material, and so on; s1 represents the mass percentage of SO3 in carbide slag, s2 represents the mass percentage of SO3 in mixed material, and so on.

[0014] Preferably, taking a batch of cement feed as an example:

[0015] Kind (wt%) CaO SiO2 Fe2O3 Al2O3 SO3 Feed ratio Carbide slag 68.32 3.38 0.51 1.87 0.48 40.27 Mixed material 86.27 3.70 0.70 0.70 0.50 25.10 Fly ash 10.00 49.84 7.70 17.91 1.29 7.28 Slag 7.90 48.20 7.43 16.00 0.45 4.44 Steel slag 44.61 16.70 18.49 4.43 0.41 6.78 Silicon powder 0.31 94.52 0.24 0.46 0.49 4.15 Silica 7.56 44.89 7.56 15.93 0.21 7.00 Incineration ash 26.03 26.95 8.37 8.96 5.54 4.99

[0016] Integrating and converting into an attribute matrix, that is:

[0017] · =

[0018] By compiling the historical feed data in recent years into an attribute matrix, these historical data are randomly used as initial guess values in the subsequent algorithm iteration calculation process to obtain the optimal solution.

[0019] Step 2, preferably, a bionic algorithm represented by a non-dominated sorting genetic algorithm for multi-objective optimization (NSGA-II) is selected to optimize the ratio of the solid waste raw materials according to the characteristics of the solid waste raw materials. In the algorithm setting process, the expected tri- rate values (lime saturation factor KH or LSF, silica rate SM and alumina rate IM) of the cement are taken as the fitness function, according to which the adaptability of the generated solid waste raw material ratio solution is evaluated (i.e. the closer the actual tri-rate values calculated to the expected tri-rate values, the stronger the adaptability of the corresponding solid waste raw material ratio solution in this algorithm). After inputting the expected tri-rate values, a new round of population growth and screening is carried out with the expected tri-rate values as the fitness function, and a batch of optimal solutions satisfying the four objective functions , , , , KH* (or LSF*), SM* and IM* are the expected tri-rate values, and KH (or LSF), SM and IM are the actual tri-rate values.

[0020] The calculation formula of the tri-rate value in the waste residue cement process is:

[0021]

[0022]

[0023]

[0024]

[0025] The deviation between the expected tri-rate value and the actual tri-rate value calculated by the tri-rate value formula of the final optimal solution and the ratio of the solid waste raw material is taken as the objective function , , , , the objective function is written as:

[0026]

[0027]

[0028]

[0029]

[0030] or

[0031] wherein , , , The objective functions are respectively , , , The weighting factor, and Much larger , , To balance the constraints on the three rates, preferably, the following settings are adopted: , , , This ensures that y1, y2, y3, and y4 all tend to 0. Under the condition that... , , , Under the premise, Weighting factors Further increases can make , , It can more evenly approach 0, preferably. yes , , 5 to 15 times, , , The ratio of the three is 0.5 to 3, increasing... The values ​​are adjusted so that the overall three ratios can approach the desired values ​​in a balanced manner. In actual cement production, when the cement performance is highly sensitive to a particular three ratio, the weighting factor before the constraint function for that ratio can be modified in the algorithm settings. This method dynamically measures the impact of a specific sensitivity rate on the cement mix design. Compared to the constraints in traditional cement batching methods that calculate the three ratios (by taking the sum of the squares of the differences between the three ratios), this method allows the actual three ratios KH (or LSF), SM, and IM to approach the desired three ratios KH* (or LSF*), SM*, and IM* infinitely, and can scientifically measure the impact of a certain sensitivity rate on the cement mix design.

[0032] Historical data and expected three-rate values ​​are imported into the algorithm, encoded into bit strings to generate the initial population, and then processed through non-dominated sorting, biomimetic operations, and genetic mutation to generate a subpopulation Q. n The offspring and parent populations are merged into R. n After non-dominated sorting and crowding calculation, a new parent population is generated, and then genetic evolution continues to generate a new offspring population. The optimal individual retention mechanism is adopted until the number of generations reaches the maximum set value or the fitness function is less than the threshold. By setting the optimal front-end individual coefficient, preferably set to 0.2~0.5, a batch of optimal solutions are obtained.

[0033] Step 3, sort the batch of optimal solutions obtained in Step 2 according to the target function of maximum consumption of a certain waste residue, and take the top solutions. , , , Step 3, sort the batch of optimal solutions obtained in Step 2 according to the target function of maximum consumption of a certain waste residue, and take the top solutions. Step 3, sort the batch of optimal solutions obtained in Step 2 according to the target function of maximum consumption of a certain waste residue, and take the top solutions.

[0034] Preferably, the feeding composition of a batch of cement plant solid waste mentioned in Step 1 (Table 2) is used as an example to explain the process steps of the cement batching method of the present application.

[0035] Kind (wt%) CaO SiO2 Fe2O3 Al2O3 SO3 Feed ratio Carbide slag 68.32 3.38 0.51 1.87 0.48 40.27 Mixed material 86.27 3.70 0.70 0.70 0.50 25.10 Fly ash 10.00 49.84 7.70 17.91 1.29 7.28 Slag 7.90 48.20 7.43 16.00 0.45 4.44 Steel slag 44.61 16.70 18.49 4.43 0.41 6.78 Silicon powder 0.31 94.52 0.24 0.46 0.49 4.15 Silica 7.56 44.89 7.56 15.93 0.21 7.00 Incineration ash 26.03 26.95 8.37 8.96 5.54 4.99

[0036] The expected three rate values to be satisfied are KH*=0.88, SM*=2.15, and IM*=1.45, and the cement batching is performed with the goal of maximum waste residue consumption of calcium carbide slag.

[0037] The optimal solution of the solid waste feeding ratio obtained by the non-dominated sorting multi-objective evolutionary algorithm based on the bionic algorithm is:

[0038] Calcium carbide slag: 50.0576%, mixed material: 16.0989%, fly ash: 2.4237%, slag: 3.5862%, steel slag: 5.8345%, silicon powder: 4.6222%, silica stone: 11.4113%, and incineration ash: 6.0386%.

[0039] The corresponding three rate values are KH=0.88187, SM=2.1506, and IM=1.4451, which are very close to the expected three rate values KH*=0.88, SM*=2.15, and IM*=1.45.

[0040] Step 4, the optimal solution obtained is used to guide the actual waste residue cement process raw material batching. The cement feeding method of the present application is not only suitable for raw material batching of solid waste residue cement, but also suitable for raw material batching optimization of traditional Portland cement and other cement production processes.

[0041] In this context, the application combines the industrial solid waste cement raw material feeding ratio data accumulated in the actual operation of cement enterprises in the past few years as the initial guess value, uses a bionic algorithm represented by a non-dominated sorting multi-objective evolutionary algorithm to establish a waste residue cement process raw material feeding ratio model, and uses the bionic algorithm optimization model to solve the balanced approach to the expected three rate values, and at the same time, the solid waste raw material represented by carbide slag with high inventory pressure is preferentially consumed to achieve the optimal waste residue-containing cement raw material ratio.

[0042] The patent application optimizes the full waste residue cement batching based on a bionic algorithm of a non-dominated sorting multi-objective evolutionary algorithm driven by solid waste consumption, and the obtained cement raw material ratio can not only relatively evenly meet the expected three rate values, but also can meet the goal of consuming as much solid waste as possible. Compared with the artificial trial and error method, the Excel linear / nonlinear solution method and the traditional genetic algorithm calculation, the method has the advantages of high efficiency and high precision, and is not only suitable for the raw material ratio of solid waste residue cement, but also suitable for the raw material ratio optimization of traditional Portland cement and other cement production processes, and has a wide application prospect in the cement industry. BRIEF DESCRIPTION OF DRAWINGS

[0043] Fig. 1 Flow chart of cement batching algorithm driven by solid waste preferential consumption based on bionic algorithm; Fig. 2 User interaction interface and optimization result display schematic diagram of cement batching optimization system driven by solid waste preferential consumption based on bionic algorithm; Fig. 3 Non-dominated sorting hierarchical diagram, corresponding to the Rank histogram in the middle diagram; Fig. 4 Schematic diagram of Pareto front (the first diagram at the top); wherein Fig. 1 ①、②、③ algorithm steps in Fig. 2 The software interface ①、②、③ function areas and control buttons correspond one by one. DETAILED DESCRIPTION

[0044] Referring to the drawings in the specification Fig. 1 A cement batching method driven by solid waste preferential consumption based on bionic algorithm, preferably, taking the composition of the waste residue cement raw material in Table 2 as an example.

[0045] Select the composition of the waste residue cement raw material in Table 2, that is, the attribute matrix mentioned in step 1 of the invention.

[0046] Preferably, the algorithm parameter setting is performed, the target lime saturation coefficient KH*=0.89, the target silicon rate SM*=2.1, and the target aluminum rate IM*=1.4 are set, and the specific implementation case is as follows.

[0047] (1) The detailed calculation process of the bionic algorithm of the application for solving the cement ratio

[0048] ① Start the cement batching program, initialize the initial parameters:

[0049] Import the oxide content data of each component of the cement raw material, for example:

[0050] Kind (wt%) CaO SiO2 Fe2O3 Al2O3 SO3 Carbide slag 68.32 3.38 0.51 1.87 0.48 Mixed material 86.27 3.70 0.70 0.70 0.50 Fly ash 10.00 49.84 7.70 17.91 1.29 Slag 7.90 48.20 7.43 16.00 0.45 Steel slag 44.61 16.70 18.49 4.43 0.41 Silicon powder 0.31 94.52 0.24 0.46 0.49 Silica 7.56 44.89 7.56 15.93 0.21 Incineration ash 26.03 26.95 8.37 8.96 5.54

[0051] ② After algorithm processing, integrate into attribute matrix:

[0052] By compiling the historical raw material data in recent years into attribute matrix, Table 4 is a part of the historical raw material (waste slag percentage), these historical data will be randomly used as initial guess value in the subsequent algorithm iteration calculation process to find the optimal solution .

[0053] · =

[0054] Kind Feed ratio history data 1 (wt%) Feed ratio history data 2 (wt%) Feed ratio history data 3 (wt%) Feed ratio history data 4 (wt%) Feed ratio history data 5 (wt%) Carbide slag 40.27 39.71 40.87 39.17 37.34 Mixed material 25.10 25.42 26.93 25.61 28.73 Fly ash 7.28 7.34 8.13 7.34 6.09 Slag 4.44 4.51 4.10 4.82 5.29 Steel slag 6.78 6.61 7.07 6.62 6.69 Silicon powder 4.15 4.23 3.91 4.52 4.27 Silica 7.00 6.70 6.59 7.05 7.87 Incineration ash 4.99 4.93 4.25 4.62 5.24

[0055] Import the expected tri-rate values of cement (lime saturation coefficient KH or LSF, silicon rate SM and aluminum rate IM): KH*=0.88, SM*=2.15, IM*=1.45, as fitness function.

[0056] Import various properties of the coal used: burning heat consumption 3014 kJ / kg, coal calorific value 27010 kJ / kg, coal ash content 12.28%.

[0057] Set appropriate parameters of genetic algorithm: including optimal individual proportion 0.3 (i.e. 30%), population size 200, evolution generation number 300, repeated number of genetic algorithm simulation 3 times, first raw material (carbide slag) as priority consumption object, maximum allowed variance between expected value 0.03 (i.e. 3%).

[0058] <L1=1001 0101> [L2=0100 1011] [L3=0101 1001] L4 = 1010 0001 …… …… …… …… L 197 =0110 1101 L 198 =1011 0000]]> L 199 =0011 1000 L 200 =1111 0000]]>

[0059] ④ Non-dominated sorting and crowding calculation are performed on the population P with a size of 200, i.e. after stratification of P by non-dominated sorting algorithm, crowding calculation is performed, as shown in the accompanying drawings Fig. 3 .

[0060] The population P is divided into 9 levels, the first non-dominated layer number is 60, the second non-dominated layer number is 30, the third non-dominated layer is 25, and the ninth non-dominated layer number is 7. Thus, the entire population is stratified, and the algorithm calculates the distance between each individual and the objective function, that is, the crowding degree calculation, so that each individual has two attributes of non-dominated layer and crowding degree. Thus, the complexity of the algorithm calculation is reduced.

[0061] 5. The population P is subjected to bionic operation, selection, crossover, mutation, etc., such as:

[0062] Through the elite strategy of the bionic algorithm, individuals with larger crowding degrees and in the better non-inferior level are selected, and the optimal front coefficient is set to 0.3, then 60 individuals are selected, and the following examples list the first four individuals to illustrate the crossover and mutation.

[0063] L1=1001 0101 L2=0100 1011 L3=0101 1001 L4=1010 0001

[0064] 1001 1011 1001 1001 1001 0001 0100 0101 0100 1001 0100 0001 0101 0101 0101 1011 0101 0001 1010 0101 1010 1011 1010 1001

[0065] 1001 1011 1001 1001 1001 0001 1101 1011 0001 1001 1010 0001 …… …… …… 0100 0101 0100 1001 0100 0001 1100 0101 0000 1001 0100 1001 …… …… …… 0101 0101 0101 1011 0101 0001 0101 1101 1101 1011 0101 0011 …… …… …… 1010 0101 1010 1011 1010 1001 1010 1101 1110 1011 1000 1001 …… …… ……

[0066] 6. P and Qn are combined into Rn, and the bionic operation of 4 and 5 is repeated in turn until the algorithm iteration number is reached or the termination threshold is met.

[0067] 7. The generated optimal solution (Pareto front drawing, see the accompanying drawings Fig. 4 ) that meets the expected three rate value is sorted according to the high and low of the solid waste requirement for consumption, and the algorithm selects the best solution, and the output is the waste residue ratio, and part of the Pareto optimal solution is listed in Table 5:

[0068] Table 5 Cement ratio solution of bionic algorithm

[0069]

[0070] From the part of the Pareto optimal solution listed in Table 5, from 1 to 5, the consumption of calcium carbide slag is sorted from high to low. In order to meet the goal of preferentially consuming calcium carbide slag, solution 1 is selected.

[0071] 7. The optimal solution of each solid waste feed ratio obtained by the non-dominated sorting multi-objective evolutionary algorithm based on the bionic algorithm is:

[0072] Table 6 Optimal solution of bionic algorithm cement ratio

[0073]

[0074] The corresponding actual tri-rate values are KH=0.88187, SM=2.1506, and IM=1.4451, which are very close to the expected tri-rate values KH*=0.88, SM*=2.15, and IM*=1.45.

[0075] (2) Comparison between the artificial trial-and-error method, the Excel linear / nonlinear programming calculation algorithm, and the bionic algorithm of the application in solving the cement proportioning case.

[0076] ① The artificial trial-and-error method is used to calculate the cement proportioning:

[0077] The constraint condition is listed, i.e.,

[0078]

[0079]

[0080]

[0081] Solving the five-monomial equation set has infinite solutions, and a group of solutions is taken, as shown in Table 7:

[0082] Table 7 Artificial trial-and-error method cement proportioning solution

[0083]

[0084] The artificial trial-and-error method is the process of solving the five-monomial equation set, but the process is very tedious and not simple enough, and often no solution exists.

[0085] ② The Excel table linear / nonlinear programming module is used to solve the problem by constraining the expected tri-rate value to be equal to the actual tri-rate value, and using the tri-rate value calculation formula, it is found that there are multiple solutions that satisfy the constraint condition, and a random example is shown in Table 8:

[0086] Table 8 Excel linear / nonlinear programming method cement proportioning solution 1

[0087]

[0088] In the above implementation case, the constraint condition that the expected tri-rate value is equal to the actual tri-rate value is set by the Excel table, and the cement proportioning ratio obtained is shown in the above table, wherein the solutions of ①, ②, and ③ can satisfy this condition, except that solution ① is more suitable for meeting the demand for consumption of waste residue piles based on the cement plant; solutions ② and ③ both contain a raw material proportioning of 0, and the consumption amount of carbide slag is too small to meet the actual demand.

[0089] ③ When the composition of the waste residue raw material changes, the solution obtained by the Excel linear / nonlinear programming may also have no solution.

[0090] Preferably, taking the feed components in Table 9 as an example:

[0091] Kind (wt%) CaO SiO2 Fe2O3 Al2O3 SO3 Carbide slag 58.32 3.38 0.51 1.87 0.48 Mixed material 62.21 21.02 0.70 0.70 0.50 Fly ash 11.00 52.84 7.70 14.21 2.10 Slag 7.90 56.25 9.41 15.87 1.48 Steel slag 15.21 13.63 55.24 5.68 0.61 Silicon powder 0.11 95.21 0.15 0.46 0.21 Silica 3.26 62.12 4.21 8.25 0.36 Incineration ash 26.03 26.95 8.37 8.96 5.54

[0092] Preferably, the expected triage values KH*=0.89, SM*=2.1, and IM*=1.4 are set, and the cement proportioning is solved in this way, as shown in Table 10.

[0093] Table 10 Excel linear / nonlinear programming method cement proportioning solution 2

[0094]

[0095] From the above implementation case, under the feed components in Table 9, the expected triage values can only meet two of them, wherein KH*=KH=0.89, SM*=SM=2.1, and IM=1.35≠1.4=IM*, it can be seen that in the process of solving the cement proportioning ratio by Excel, some feed component data will make the solution obtained invalid, and cannot meet all the constraint conditions (i.e. no solution is found), in addition, the content of carbide slag in the feed proportioning is as high as 69.58%, which does not conform to the actual situation.

[0096] (4) The bionic algorithm of the present application is used to proportion the cement feed under the feed components in Table 9, taking carbide slag as the maximum consumption amount of waste residue, as shown in Table 11.

[0097] Table 11 Bionic algorithm cement proportioning solution

[0098]

[0099] From the above implementation case, under the feed components in Table 9, the solution of the cement proportioning by the Excel linear / nonlinear programming method cannot be found. The bionic algorithm of the present application can solve the problem that the triage values KH, SM and IM cannot all approach KH*, SM* and IM* when the cement proportioning is solved by the Excel linear / nonlinear programming method, and a relatively reasonable waste residue feed proportioning can be obtained.

[0100] (3) Comparison between the traditional genetic algorithm and the bionic algorithm of the present application in solving the cement raw material proportioning case.

[0101] 1. Cement proportioning obtained by the ordinary genetic algorithm (without setting the target of preferential consumption of certain solid waste)

[0102] The solution obtained by the ordinary genetic algorithm is a possible solution after the optimization of the algorithm system from the random initial value, and there are often multiple approximate solutions. For example, after the expected triage values are input into the calculation model, the proportioning ratios of 8 kinds of industrial waste residues are obtained, and some typical solutions are shown in Table 12.

[0103] Table 12 Genetic algorithm cement proportioning solution

[0104]

[0105] From the above implementation cases, the KH, SM, and IM values of each solution can generally meet the target of KH*=KH=0.89, SM*=SM=2.1, and IM*=IM=1.4. In solutions ① and ②, it can be seen that the feeding ratios of carbide slag and mixed materials differ greatly, and the comparison between solutions ② and ③ can observe that the differences between the two items of incineration ash and silica are also large, and it is difficult to obtain the optimal matching data required for stable production.

[0106] Two cement matching solutions with large fluctuations in the feeding of carbide slag obtained by the traditional genetic algorithm are listed in Table 13.

[0107] Table 13: Cement matching solutions obtained by the genetic algorithm

[0108]

[0109] From the above implementation cases, it can be seen in solutions ① and ② that the matching of all feeding items differs greatly. The calculation of cement matching by this method has large fluctuations in the matching of each solid waste raw material. For our waste slag cement production process, it is also impossible to achieve the goal of preferential consumption of waste slag with large inventory pressure in the cement raw material yard and stable production.

[0110] ② Preferably, the maximum extent of carbide slag is used as the consumption amount for the cement feeding matching optimization of this full industrial waste slag. The unique optimal solution obtained by the non-dominated sorting multi-objective evolutionary algorithm based on the bionic algorithm of the present application takes into account the expected three rate values and the maximum inventory amount of waste slag consumption, as shown in Table 14.

[0111] Table 14: Cement matching solutions obtained by the bionic algorithm

[0112]

[0113] From the above implementation cases, the actual calculated three rate values KH=0.89076, SM=2.0998, and IM=1.3985 have very small errors with the expected three rate values KH*=0.89, SM*=2.1, and IM*=1.4, and there is no 0% situation for each waste slag feeding, which is consistent with the actual situation. Compared with the feeding of each waste raw material in Table 4 calculated by the traditional simple genetic algorithm, historical feeding data can be used as the initial guess value, the feeding range is stable, and it is consistent with the actual situation of waste slag inventory. In this case, the goal of using carbide slag as the main consumption of waste slag can be achieved.

[0114] (4) Comparison of the calculation process of cement raw material matching by the traditional genetic algorithm and the bionic algorithm of the present application.

[0115] ① Traditional genetic algorithm calculation case.

[0116] Preferably, taking the feed components in Table 2 as the attribute matrix, setting the expected tri-rate values KH*=0.89, SM*=2.1, IM*=1.4, and setting the maximum iteration number of the genetic algorithm to 300 times.

[0117] The cement proportioning ratio calculated by the conventional genetic algorithm, the calculation time, and the iteration number at the time of convergence are shown in Table 15.

[0118] Table 15 Solution of cement proportioning by genetic algorithm and solution time

[0119]

[0120] In the above embodiment, the cement proportioning ratio KH=0.88994, SM=2.1001, IM=1.4002 obtained by the conventional genetic algorithm and the expected tri-rate values KH*=0.89, SM*=2.1, IM*=1.4 are basically equal, the entire calculation process consumes 118 s, and the maximum genetic iteration number set to 300 times is reached, but the convergence condition is still not met, the calculation is ended, and the calculation converges.

[0121] 2. The bionic algorithm calculation case of the present application.

[0122] Preferably, taking the feed components in Table 2 as the attribute matrix.

[0123] Preferably, setting the expected tri-rate values KH*=0.89, SM*=2.1, IM*=1.4, and setting the maximum iteration number of the bionic algorithm to 300 times.

[0124] The cement proportioning ratio calculated by the bionic algorithm of the present application, the calculation time, and the iteration number at the time of convergence are shown in Table 16.

[0125] Table 16 Solution of cement proportioning by bionic algorithm and solution time

[0126]

[0127] In the above embodiment, the cement proportioning ratio KH=0.89021, SM=2.99007, IM=1.39913 obtained by the conventional genetic algorithm and the expected tri-rate values KH*=0.89, SM*=2.1, IM*=1.4 are basically equal, the optimal solution of the cement proportioning ratio is the proportioning ratio of the historical feed data. The entire calculation process consumes 30.8 s, and the iteration is 80 times, and the calculation reaches the convergence condition.

[0128] Compared with the traditional genetic algorithm and the bionic algorithm, the traditional genetic algorithm uses random data as the initial guess value, according to Table 14, the calculation time and the iteration number are relatively long to reach convergence. The algorithm of the application uses historical feed data as the initial population, when the expected tri value is close to the tri value calculated by the feed data, the historical feed data can be used as the initial guess value, which can greatly reduce the calculation time and complete the convergence of the fitness function iteration in a smaller number of iterations.

[0129] (5) The bionic algorithm of the application is used to calculate the feed ratio of the traditional Portland cement, the cement co-processing sludge and the incineration fly ash.

[0130] ① The calculation case of the traditional Portland cement raw material ratio method by the algorithm program of the application is as follows:

[0131] The expected tri value KH*=0.89, SM*=2.5, IM*=1.5 is set, and the optimal solution obtained is shown in Table 17.

[0132] Serial number / Kind (%) Limestone Clay Sandstone Iron slag Incineration ash KH SM IM ① 80.05 12.02 6.55 1.38 2.12 0.88961 2.50123 1.50014

[0133] As can be seen from the above implementation cases, KH, SM and IM are very close to the values of KH*, SM* and IM*, which shows that the cement batching method based on the bionic algorithm is also applicable to the configuration of the traditional Portland cement raw material.

[0134] ② The batching calculation case of the leather-making sludge co-processing by the algorithm is as follows:

[0135] The proportion of each oxide in the leather-making sludge is CaO: 12.81%, Al2O3: 5.69%, Fe2O3: 4.51%, SiO2: 3.85%, SO3: 0.35%.

[0136] The batching scheme of the traditional Portland cement raw material mixed with the leather-making sludge is KH*=0.89, SM*=2.1, IM*=1.4, and the preparation of the cement is shown in Table 18.

[0137] Serial number / Kind (%) Limestone Clay Sandstone Copper slag Tannery sludge KH SM IM ① 86.82 2.74 9.12 0.82 0.5 0.89024 2.10232 1.49987

[0138] As can be seen from the above implementation cases, KH, SM and IM are very close to the values of KH*, SM* and IM*, and the cement batching method based on the bionic algorithm in the application is also applicable to the cement co-processing sludge.

[0139] ③ The batching calculation case of the incineration fly ash co-processing by the algorithm is as follows:

[0140] The incineration fly ash belongs to a kind of hazardous waste, one of its disposal methods is to be mixed into cement raw materials for co-disposal. Preferably, in the certain domestic waste incineration fly ash, the proportion of each oxide is CaO: 10.95%, Al2O3: 23.19%, Fe2O3: 6.62%, SiO2: 42.22%, SO3: 4.22%.

[0141] The expected tri-rate value of the tannery sludge mixed into the traditional Portland cement raw material is KH*=0.89, SM*=2.1, IM*=1.4, and the ratio scheme is shown in Table 19.

[0142] Serial number / Kind (%) Limestone Quartz sand Shale Copper slag Household waste incineration fly ash KH SM IM ① 82.90 9.99 0.00 2.10 5.01 0.89064 2.09957 1.50002

[0143] As can be seen from the above implementation cases, KH, SM and IM are very close to KH*, SM* and IM* values, and the cement batching method based on the bionic algorithm in the application is also applicable to the co-disposal of domestic waste incineration fly ash (hazardous waste) in cement.

[0144] Those skilled in the art will readily understand that the above description is only a preferred embodiment of the present application and is not intended to limit the present application, and any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A cement batching method driven by the priority disposal of solid waste based on a biomimetic algorithm, characterized in that, This method is implemented through a hierarchical, two-stage optimization process, including the following steps: The chemical composition data of several solid waste raw materials are integrated into an attribute matrix; the proportions of these solid waste raw materials are used as the vectors to be optimized. The first stage employs a biomimetic optimization algorithm, represented by the non-dominated sorting multi-objective evolutionary algorithm, to generate a Pareto optimal solution set that meets cement quality requirements by solving a set of objective functions y1 or y′1, y2, y3, y4 based on the deviations between actual and expected three-ratio values. The expected three-ratio values ​​are calculated based on the solid waste raw material input data used in existing cement production and the corresponding performance data of the produced cement clinker. The specific objective function is as follows: or Where KH* or LSF*, SM*, and IM* are the expected three-rate values, and KH or LSF, SM, and IM are the actual three-rate values ​​calculated based on the vector to be optimized. φ1 or φ′1, φ2, φ3, and φ4 are the target weight factors. The objective functions y1, y2, y3, and y4, which characterize the deviation of the three-rate values, are respectively weighted by φ1 or φ′1, φ2, φ3, and φ4. Among them, φ4 is much larger than φ1 or φ′1, φ2, and φ3, and φ4 is 5 to 15 times larger than φ1, φ2, and φ3. The ratio between φ1, φ2, and φ3 is 0.5 to 3. Under the premise of ensuring lim y1→0, lim y2→0, lim y3→0, and lim y4→0, the increase of the weight factor of y4 makes y1, y2, and y3 approach 0 in a balanced manner, thereby ensuring that the actual three-rate values ​​are infinitely close to the expected three-rate values. The second stage: In the Pareto optimal solution set generated in the first stage, all solutions in the solution set are sorted according to the fifth objective function y4. The fifth objective function y4 is the amount of specific target solid waste raw material represented by carbide slag. The solution with the highest y4 value is selected as the final optimal batching ratio, which is used to guide the solid waste raw material ratio of cement production raw materials. The fifth objective function is the maximum amount of solid waste raw material represented by carbide slag, i.e., the objective function y4. After inputting historical feed data, an initial population P is generated. After non-dominated sorting, biomimetic operations, and genetic mutation, a subpopulation Q is generated. n The offspring population and the parent population are merged into R. n After non-dominated sorting and crowding calculation, a new parent population is generated, and then genetic evolution continues to generate a new offspring population until the number of generations reaches the maximum set value or the fitness function is less than the threshold. By setting the optimal front-end individual coefficient, which is set to 0.1 to 0.5, a batch of optimal solutions are obtained. These optimal solutions are sorted from high to low according to the amount of carbide slag consumed, i.e., the objective function y4. The group with the highest ranking is taken as the optimal solution, which is the best solution that ultimately satisfies all objective functions in a balanced way.

2. The cement batching method driven by the priority disposal of solid waste based on a biomimetic algorithm as described in claim 1, characterized in that... The attribute matrix is ​​matrix block A, which is a 5×n matrix; the first column of matrix A is carbide slag, the second column is the mixing material, and the subsequent columns are other solid waste raw materials, vector... This represents the percentage of each solid waste raw material fed into the cement raw meal, expressed as a vector. This represents the total amount of five oxides in cement raw materials, specifically in the following form: Where A represents several sets of chemical composition data for cement raw materials, C1 represents the mass percentage of CaO in carbide slag, C2 represents the mass percentage of CaO in the mixture, and so on; S1 represents the mass percentage of SiO2 in carbide slag, S2 represents the mass percentage of SiO2 in the mixture, and so on; F1 represents the mass percentage of Fe2O3 in carbide slag, F2 represents the mass percentage of Fe2O3 in the mixture, and so on; A1 represents the mass percentage of Al2O3 in carbide slag, A2 represents the mass percentage of Al2O3 in the mixture, and so on; s1 represents the mass percentage of SO3 in carbide slag, s2 represents the mass percentage of SO3 in the mixture, and so on.

3. The cement batching method driven by the priority disposal of solid waste based on a biomimetic algorithm as described in claim 1, characterized in that... The proportions of several solid waste raw materials are compiled into an attribute matrix. When finding the optimal solution, these historical data can be randomly used as initial guesses, thereby accelerating the calculation convergence speed.

4. The cement batching method driven by the priority disposal of solid waste based on a biomimetic algorithm as described in claim 1, characterized in that... It is applicable not only to the optimization of solid waste raw material ratios in cement co-firing of industrial solid waste, domestic waste or hazardous waste, but also to the optimization of raw material ratios in the production of traditional silicate cement using non-solid waste raw materials.

5. A cement batching method driven by the priority disposal of solid waste based on a biomimetic algorithm as described in claim 1, characterized in that... The solid waste raw materials include several of the following: carbide slag, mixing material, fly ash, furnace slag, steel slag, silicon powder, silica, and incineration ash.

Images