Optimization Method and System for Process Parameters of Manure and Straw Turning Composting

Abstract

The present invention belongs to the technical field of manure treatment, and provides an optimization method and system for the process parameters of manure and straw turning composting. The method includes: collecting and constructing a piece of training data based on each piece of real turning composting data; dividing the training data corresponding to each piece of the turning composting data into multiple groups according to the compost evaluation data, and sorting the training data in each group according to the first compost volume; selecting a number of training data from each group in the sorting order to form a training data group, obtaining multiple training data groups, and training the process parameter optimization model with a number of training data in each training data group until the training reaches the standard; determining the second manure and straw data and the second compost volume involved in the current composting, and processing the second manure and straw data and the second compost volume with the trained process parameter optimization model to predict the second process parameters. The present invention can realize the optimization of the composting process parameters.

Description

Technical Field

[0001] The present invention relates to the technical field of manure treatment, and more specifically, to an optimization method and system for the process parameters of manure and straw turning composting. Background Art

[0002] With the large generation of agricultural waste (such as livestock and poultry manure, crop straw), how to efficiently and environmentally treat these organic wastes has become one of the key issues for the sustainable development of agriculture. Traditional incineration or landfill methods are prone to cause environmental pollution and resource waste, while manure and straw turning composting, as an economically viable organic waste treatment technology, can convert organic matter into stable humus through microbial fermentation, realizing the resource utilization of waste and improving soil fertility at the same time.

[0003] However, in the actual composting process, the composting efficiency and quality are affected by various process parameters, such as temperature, carbon-nitrogen ratio, pH, auxiliary materials, moisture content, ventilation method, ventilation volume, etc. If the parameter regulation is improper, problems such as extended composting cycle, incomplete decomposition, nutrient loss or odor emission may occur. At present, domestic and foreign research mainly focuses on optimizing the composting process through the synergistic effect of temperature-oxygen-microorganisms, but there are still the following technical bottlenecks:

[0004] Imprecise turning strategy: Frequent turning is prone to heat dissipation, while insufficient turning leads to local anaerobic conditions, affecting the uniformity of decomposition. Low ventilation and oxygen supply efficiency: The composting cycle of natural ventilation is long, and the energy consumption of forced ventilation is high. It is necessary to balance the relationship between oxygen supply and temperature control. Imperfect maturity evaluation system: A single index (such as temperature or C / N ratio) is difficult to comprehensively reflect the compost quality, and multi-dimensional dynamic monitoring is required. Therefore, optimizing the process parameters of turning composting is crucial for improving composting efficiency, reducing energy consumption, and ensuring the decomposition quality.

[0005] In summary, how to obtain optimized composting process parameters to achieve a high-grade composting effect is a technical problem that needs to be solved urgently at present. Summary of the Invention

[0006] To solve at least one of the above technical problems, the present invention specifically provides an optimization method, system, electronic device, computer storage medium, and computer program product for the process parameters of manure and straw turning composting.

[0007] The present invention provides an optimization method for the process parameters of manure and straw turning composting, including the following steps:

[0008] Collect a number of real turning composting data, and each piece of the turning composting data includes first manure and straw data, first process parameters, compost evaluation data, and first compost volume;

[0009] Derive a piece of training data based on the first manure-straw data, the first process parameters, the compost evaluation data, and the first compost volume in each piece of the turning and composting data; wherein, the compost evaluation data is used to construct the label of the training data.

[0010] Divide the training data corresponding to each piece of the turning and composting data into multiple groups according to the compost evaluation data, and sort the training data in each group according to the first compost volume.

[0011] Select a number of the training data from each of the groups in the sorted order to form a training data set, obtain multiple training data sets, and use a number of the training data in each training data set to train the process parameter optimization model until the training reaches the standard.

[0012] Determine the second manure-straw data and the second compost volume involved in the current composting, and use the trained process parameter optimization model to process the second manure-straw data and the second compost volume to predict the second process parameters.

[0013] The present invention also provides an optimization system for the process parameters of manure-straw turning and composting. The system includes a collection unit, a training unit, and an optimization unit.

[0014] The collection unit is used to collect a number of real turning and composting data. Each piece of the turning and composting data includes the first manure-straw data, the first process parameters, the compost evaluation data, and the first compost volume.

[0015] The training unit is used to: Derive a piece of training data based on the first manure-straw data, the first process parameters, the compost evaluation data, and the first compost volume in each piece of the turning and composting data; wherein, the compost evaluation data is used to construct the label of the training data.

[0016] Divide the training data corresponding to each piece of the turning and composting data into multiple groups according to the compost evaluation data, and sort the training data in each group according to the first compost volume.

[0017] Select a number of the training data from each of the groups in the sorted order to form a training data set, obtain multiple training data sets, and use a number of the training data in each training data set to train the process parameter optimization model until the training reaches the standard.

[0018] The optimization unit is used to determine the second manure and straw data and the second compost volume involved in the current composting, and process the second manure and straw data and the second compost volume using the trained process parameter optimization model to predict the second process parameters.

[0019] The present invention also provides an electronic device, including: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the method as described in any one of the preceding items.

[0020] The present invention also provides a computer storage medium, on which a computer program is stored, and the computer program, when run by a processor, executes the method as described in any one of the above items.

[0021] The present invention also provides a computer program product, which is used to be called and run by a processor of an electronic device to implement the method as described in any one of the above items.

[0022] The beneficial effects of the present invention are as follows:

[0023] The solution of the present invention uses the training data constructed based on the real turning and composting data to train the process parameter optimization model, and improves the diversity of the training data through the decision-making strategy of the training data group, so that the training effect and the training effect of the process parameter optimization model can be significantly improved, which is beneficial to obtaining better process parameters, thereby improving the effect of turning and composting of manure and straw. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can be obtained based on these drawings.

[0025] Figure 1 is a flowchart showing an optimization method for process parameters of manure and straw turning and composting disclosed in an embodiment of the present invention.

[0026] Figure 2 is a schematic diagram showing the grouping of training data and the composition of training data groups disclosed in an embodiment of the present invention.

[0027] Figure 3 is a schematic diagram showing the structure of an optimization system for process parameters of manure and straw turning and composting disclosed in an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0028] The following describes exemplary embodiments of the present disclosure with reference to the accompanying drawings. Various details of the embodiments of the present disclosure are included to facilitate understanding, and they should be considered merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, descriptions of well-known functions and structures are omitted below for clarity and conciseness.

[0029] Referring to Figure 1 the flow schematic diagram shown, an embodiment of the present invention provides an optimization method for the process parameters of manure and straw turning composting, including the following steps:

[0030] S10. Collect a number of real turning composting data. Each piece of the turning composting data includes first manure and straw data, first process parameters, compost evaluation data, and first compost volume.

[0031] The present invention pre-constructs a process parameter optimization model to predict the optimal turning composting process parameters. However, the process parameter optimization model needs to be pre-trained to have the ability to predict the above process parameters. For this, the present invention collects an appropriate number (for example, 200 pieces) of real turning composting data. The turning composting data at least includes the following data:

[0032] First manure and straw data: Since different manure and straw (such as the composition differences of different livestock manures and the characteristics of different crop straws) will affect the composting process and results, relevant data thereof needs to be collected. This data includes information such as the type and component content of manure and straw.

[0033] First process parameters: Such as temperature, carbon-nitrogen ratio, pH value, auxiliary materials, moisture content, ventilation method, ventilation volume, etc. These parameters will affect the efficiency and quality of composting in actual composting.

[0034] Compost evaluation data: The compost evaluation data can be scores or grades used to characterize the quality. It can be manually scored or automatically evaluated by an automatic evaluation model. The present invention does not make specific limitations on this. The compost evaluation data is at least comprehensively evaluated based on physical indicators, chemical indicators, maturity indicators, biological and sanitary indicators, and other indicators. For example, the compost evaluation data is obtained by weighted calculation of multiple sub-scores. The explanations of each indicator are as follows:

[0035] Physical indicators: The mature compost should have no odor, be dark brown, loose, and have no visible original materials (such as straw and fecal residues).

[0036] Chemical indicators: After composting, the degradation rate of organic matter should be reduced by 30% - 50%, which is evaluated by the reduction of total organic carbon (TOC) or volatile solids (VS). After composting, the carbon-nitrogen ratio should be less than 20:1, that is, close to the soil level of 10 - 15:1. The pH value range should be 6.5 - 8.5. The electrical conductivity (EC) of the mature compost should be < 4 mS / cm, and too high value may cause salt damage.

[0037] Degree of maturity indicator: The ratio of humic acid (HA) / fulvic acid (FA) > 1.5 indicates good humification.

[0038] Biological and sanitation indicators

[0039] Microbial community: The microbial activity is evaluated by PCR or enzyme activities (such as dehydrogenase, urease).

[0040] Pathogens and parasite eggs: It should meet the organic fertilizer standards, such as the mortality rate of Ascaris eggs ≥ 95%, and the coliform value ≤ 0.01 g⁻¹.

[0041] Other indicators: After composting, the concentration of volatile organic compounds (VOCs, such as ammonia, hydrogen sulfide) should be significantly reduced. The heavy metal content should meet the organic fertilizer standards, for example, Cd ≤ 3 mg / kg, Pb ≤ 100 mg / kg.

[0042] The volume of the first compost: Different volumes of compost may have different internal environmental conditions (such as oxygen distribution, temperature diffusion, etc.), which may affect the composting process, so it is also one of the data that needs to be collected.

[0043] S20, according to the first manure-straw data, the first process parameters, the compost evaluation data, and the volume of the first compost in each piece of the turning and composting data, one training data is constructed; among them, the compost evaluation data is used to construct the label of the training data.

[0044] After collecting a sufficient number of turning and composting data, the compost evaluation data among them is used as the data label, and the first manure-straw data, the first process parameters, and the volume of the first compost are used as data entities. The integration of the data entity and the data label gives one training data.

[0045] It can be understood that the first manure-straw data, the first process parameters, and the volume of the first compost, as input features, reflect the initial conditions and process control factors of composting. The compost evaluation data is a comprehensive evaluation of the composting result. The training goal of the model is to learn the relationship between the input features and the label, so as to be able to predict appropriate process parameters according to new inputs to achieve better composting effects.

[0046] S30. Divide the training data corresponding to each of the turning and composting data into multiple groups according to the compost evaluation data, and sort the training data in each group according to the first compost volume.

[0047] Different compost evaluation results (such as different degrees of maturity, quality grades, etc.) represent different composting effects. In the present invention, the training data is grouped according to the compost evaluation data. For example, it is divided into three groups with low, medium, and high compost evaluation data respectively. Each group contains a corresponding number of training data, as Figure 2 shown. Subsequently, different training strategies can be formulated when training the process parameter optimization model to improve the adaptability and accuracy of the model. After completing the grouping, the training data in each group is sorted within the group according to the first compost volume.

[0048] Thus, the preparation work of the training data is completed.

[0049] S40. Select a number of the training data from each of the groups in the sorted order to form a training data group, obtain multiple training data groups, and use a number of the training data in each of the training data groups to train the process parameter optimization model until the training reaches the standard.

[0050] Next, as Figure 2 shown, select a number of training data at the front end of the sorting from each group, and form these training data selected each time into a training data group. After multiple selections, multiple training data groups are obtained.

[0051] Such a setting can make the training data included in each training data group have more significant diversification characteristics, which is beneficial to improving the training effect and training efficiency of the process parameter optimization model, enabling the model to learn more comprehensive composting rules.

[0052] Taking the training data group as a unit, use the training data it contains to train the process parameter optimization model to continuously adjust the parameters of the model, so that the model can accurately predict the appropriate process parameters according to information such as the input manure and straw data and the compost volume until the training of the model reaches the preset standard (such as indicators such as accuracy rate and error meet the requirements). At this time, the model has a certain prediction ability.

[0053] S50. Determine the second manure and straw data and the second compost volume involved in this composting, and use the trained process parameter optimization model to process the second manure and straw data and the second compost volume to predict the second process parameters.

[0054] In practical applications, after determining the second manure-straw data (specific types and component contents of manure and straw, etc.) and the second composting volume for this composting process, these data are input into the pre-trained process parameter optimization model. The process parameter optimization model can then predict the second process parameters suitable for this composting process based on the relationship between the input features learned previously and the composting results, thereby optimizing the composting process, improving the composting efficiency and quality, and solving some technical problems existing in the current composting process.

[0055] It can be understood that the first process parameters and the second process parameters should include the same parameter content.

[0056] The solution of the present invention uses the training data constructed based on the real turning and composting data to train the process parameter optimization model, and improves the diversity of the training data through the decision-making strategy of the training data group, so that the training effect and the training effect of the process parameter optimization model can be significantly improved, which is beneficial to obtaining better process parameters, thereby improving the effect of turning and composting manure and straw.

[0057] Further, dividing the training data corresponding to each of the turning and composting data into multiple groups according to the composting evaluation data, and sorting the training data in each group according to the first composting volume, includes:

[0058] Sorting the training data corresponding to each of the turning and composting data from high to low according to the composting evaluation data, and dividing the sorted training data into corresponding groups based on a preset division ratio;

[0059] Calculating the absolute value of the volume difference between the first composting volume and the second composting volume, and determining the sorting serial number of each of the training data within the corresponding group according to the absolute value; wherein, the smaller the absolute value, the smaller the corresponding sorting serial number, that is, the more forward.

[0060] In this example, sorting the training data from high to low according to the composting evaluation data is to arrange the training data corresponding to the better-quality compost in the front and the worse-quality compost in the back. In this way, the better-quality training data can be extracted first when forming the training data group later. It can be understood that during the training process of the process parameter optimization model, it may not be necessary to use all the training data. That is, after using some training data for training, the model has already converged. Therefore, sorting the better-quality training data forward is of practical significance.

[0061] The division ratio is, for example, preset and used to divide these sorted training data into different groups. For example, it is preset that the first 30% of the training data is divided into a high-grade group, 40% of the medium grade is divided into a medium-quality group, and the last 30% is divided into a low-grade group. The purpose of doing this is to be able to formulate different training strategies for different quality levels of composting when optimizing the process parameter model later, so that the model can learn the characteristics of the process parameters corresponding to different quality composts, improving the adaptability and accuracy of the model.

[0062] The process parameter optimization model in the present invention can be dedicated to one batch of turning and composting, and the composting volume of the same batch of turning and composting is the same. Therefore, the present invention sets the training data closer to the second composting volume to be ranked higher within the group (such as Figure 2 the training data m1, n1, t1 in), which can increase the probability of selecting this type of training data, and the probability of the process parameter optimization model trained with the training data under more "similar" conditions predicting better second process parameters corresponding to the turning and composting of this batch will increase.

[0063] It should be noted that when performing the turning and composting operation for each batch, the sorting method of each training data within the group can be re-determined in the foregoing manner, and then the process parameter optimization model can be retrained or re-trained. However, the training data relied on at this time may not change significantly in proportion. This enables the process parameter optimization model to always adapt to the turning and composting operations with different composting volumes in the current batch.

[0064] Furthermore, before constructing a piece of training data according to the first manure and straw data, the first process parameters, the compost evaluation data, and the first composting volume in each piece of the turning and composting data, the method further includes:

[0065] Obtain the data source information of each piece of the turning and composting data, evaluate the process compliance of the data source information, and obtain a process compliance value;

[0066] If the process compliance value is lower than the preset value and the compost evaluation data of the turning and composting data is rated at medium grade or above, then determine a tuning factor according to the difference between the process compliance value and the preset value, and use the tuning factor to optimize the compost evaluation data.

[0067] In this example, to ensure the training effect of the process parameter optimization model, it is necessary to obtain as much turning and composting data as possible to construct training data. However, in reality, it is very difficult to obtain turning and composting data. Therefore, it is necessary to make the best use of the available turning and composting data.

[0068] The sources of the data for turning and composting are different, that is, they come from turning and composting factories in different regions and of different scales. There are differences in the strict implementation of the pre-established technological parameters for turning and composting in these factories. For example, in some small-scale turning and composting factories, there may be insufficient ventilation time for composting, resulting in the ventilation volume being lower than the standard value.

[0069] In view of the above inevitable actual situations, the present invention first determines the data source information of each turning and composting data (i.e., the turning and composting factory), and further obtains the relevant information of these turning and composting factories. For example, the actual operation data of composting is obtained through on-site visits, personnel interviews, checking work records, etc. Based on these relevant information, the compliance of the turning and composting factory with the process is evaluated, and it can be judged how strictly the turning and composting factory adheres to the composting process standards in actual operation.

[0070] Next, select those with a process compliance value lower than the preset value and the compost evaluation data in the turning and composting data being rated as medium level or above (for example, the score is from 1 to 10, and a score of 6 or above is considered medium level or above), and optimize and adjust the compost evaluation data in these turning and composting data. Specifically, calculate the difference between the process compliance value and the preset value, and then match the corresponding increase factor according to this difference. There is a negative correlation between the increase factor and the difference. The increase factor is, for example, 1.2, 1.5. In this way, the increase factor can appropriately increase the compost evaluation data.

[0071] The reason for the present invention to be set like this is that for those situations where the process compliance value is not high enough, but the actual effect of turning and composting is at an upper-middle level, it indicates that the probability of its process parameters being more reasonable is relatively high, that is, excellent process parameters can overcome the adverse effects brought by the lower process compliance value. In such a case, the present invention sets to use the increase factor to appropriately increase the compost evaluation data of this turning and composting data (in order to achieve a more objective evaluation of the corresponding process parameters), so as to use the data label with a higher evaluation to guide the training direction of the process parameter optimization model, thereby improving the training effect of the process parameter optimization model. At the same time, this can also make the training data corresponding to this turning and composting data be assigned to a more forward group, thereby increasing the probability of its being selected for training. However, the increase factor should not be too high, otherwise it is easy to cause the adjusted compost evaluation data to deviate too much from the real situation, resulting in a poor training effect of the model.

[0072] It should be noted that for those data with a compost evaluation level below the medium level, it is not suitable to use the above method for optimization and adjustment, because the reasons for the poor composting effect may be complex, and the correlation probability with the suitability of the process parameters is not clear. Blindly increasing it may cause the training direction of the process parameter optimization model to deviate.

[0073] Further, according to the sorting order, a number of the training data are respectively selected from each of the groups to form a training data group, including:

[0074] In the first stage of training the process parameter optimization model, a number of the training data are respectively selected from each of the groups according to the first selection rule to form a training data group; wherein, the first selection rule refers to respectively selecting the training data accounting for the first ratio, the second ratio, and the third ratio of the total from the groups with high-grade, medium-grade, and low-grade compost evaluation data, and the first ratio is higher than the second ratio, and the second ratio is higher than the third ratio;

[0075] In the second stage of training the process parameter optimization model, a number of the training data are respectively selected from each of the groups according to the second selection rule to form a training data group; wherein, the second selection rule refers to respectively selecting the training data accounting for the fourth ratio, the fifth ratio, and the sixth ratio of the total from the groups with high-grade, medium-grade, and low-grade compost evaluation data, and the fourth ratio is higher than the fifth ratio, the fifth ratio is higher than the sixth ratio, and the fourth ratio is lower than the first ratio.

[0076] In this example, when training the process parameter optimization model, the training process is divided into different stages, and different selection rules are used to select the training data. In the initial stage of training, it is set to select more training data from the groups with higher compost evaluation data; in the later stage of training, it is set to increase the proportion of the training data selected from the groups with lower compost evaluation data. Specifically:

[0077] Assume that the first ratio is 50%, the second ratio is 30%, and the third ratio is 20%. In this way, during the first-stage training, a relatively large number of training data will be selected from the high-grade group, and a relatively small number of training data will be selected from the low-grade group. In the initial stage of training, the process parameter optimization model has less knowledge about the relationship between the compost process parameters and the composting effect. The training data corresponding to the high-grade compost reflects a relatively better combination of compost process parameters and a better composting result. Therefore, the present invention sets the process parameter optimization model to learn more of these high-grade data in the initial stage of training, which can help the model quickly capture the process parameter characteristics that can produce a better composting effect, thereby quickly establishing a preliminary and relatively reasonable model framework and laying a foundation for further optimizing the model subsequently.

[0078] Suppose the fourth ratio is 30%, the fifth ratio is 35%, and the sixth ratio is 35%. Compared with the first stage, in the second stage, the proportion of data selected from the high-level group decreases, while the proportion of data selected from the low-level group increases. After the training in the first stage, the process parameter optimization model has already had a certain understanding of the better composting process parameters. Therefore, in the present invention, it is set to increase the proportion of training data selected from the low-level group in the second stage, so that the model can be exposed to more data of non-ideal composting situations, learn the characteristics of process parameters that lead to poor composting effects, and further optimize the model, enabling it to more comprehensively understand the impact of different process parameters on the composting effect, improve the generalization ability and accuracy of the model, and thus more accurately predict and optimize the process parameters for various composting situations.

[0079] It should be noted that the total number of training data in the training data set can be fixed, such as 20 or 30. Of course, the total number can also be dynamic. For example, the total number in the first stage is 20, and the total number in the second stage is 15. The present invention does not make specific limitations on this.

[0080] Furthermore, the division ratio is determined by the following method:

[0081] Obtain a number of training record data of the process parameter optimization model. The training data includes model test indicators and the historical division ratios used. Screen a number of training record data with the best model test indicators among them, and fuse the historical division ratios corresponding to these training record data to obtain the division ratio for this training.

[0082] In this example, as previously explained, the process parameter optimization model can be one-time, that is, a process parameter optimization model is trained for each batch of composting operations, and the relevant data for each training is recorded. These training record data contain two key pieces of information. One is the model test indicator, which can reflect the performance of the model after a certain training, such as accuracy, error rate, recall rate, etc. Through these indicators, the quality of the process parameter optimization model in handling the composting process parameter prediction task can be measured. The other is the historical division ratio used, that is, the ratio used to divide the training data into different groups according to the compost evaluation data during the previous training process. For example, there may have been a situation where the training data was divided into high, medium, and low levels at ratios of 30%, 40%, and 30% respectively.

[0083] From the numerous training record data obtained, several training record data with the best model test metrics are screened out. Here, "the best" means that among all the training records, the model test metrics corresponding to these data perform the best overall under a certain standard, such as the highest accuracy rate, the smallest error, etc. According to the historical division ratios corresponding to the screened training record data, through a fusion method (for example, it can be a weighted average of these historical division ratios, and the weights can be determined according to the quality of the corresponding model test metrics, the better the metrics, the higher the weights; or it can also be other reasonable fusion methods) to obtain the division ratio for this training. The division ratio obtained in this way is comprehensively derived based on the historical division situation that previously made the model performance reach the best, so there is a high possibility that it can also help the model obtain better training effects in this training, enabling the model to more effectively learn the process parameter characteristics corresponding to different compost evaluation levels, thereby improving the accuracy and reliability of the model in predicting compost process parameters and optimizing the compost process.

[0084] In addition, the present invention also compares the training effect of the above-mentioned grouping input method of the training data of the present invention with the conventional scheme. The comparison metrics involve the training cycle, accuracy rate, and generalization ability. Among them, the conventional scheme refers to not grouping all the training data, but randomly selecting training data for training, and when the input ratio or input quantity is reached, using the test set to test the model to decide whether to continue training.

[0085] After actual measurement, compared with the conventional method, the training cycle of the scheme of the present invention is shortened by 20% - 30%; the overall prediction accuracy rate of the model after training reaches 85% - 90%, while the overall prediction accuracy rate of the conventional method is only 60% - 70%, and the effect is obvious.

[0086] As Figure 3 shown in the structural schematic diagram, the embodiment of the present invention also discloses an optimization system for the process parameters of manure and straw turning composting. The system includes a collection unit, a training unit, and an optimization unit;

[0087] The collection unit is used to collect a number of real turning composting data. Each piece of the turning composting data includes first manure and straw data, first process parameters, compost evaluation data, and first compost volume;

[0088] The training unit is used to: construct a piece of training data according to the first manure and straw data, the first process parameters, the compost evaluation data, and the first compost volume in each piece of the turning composting data; wherein, the compost evaluation data is used to construct the label of the training data;

[0089] Dividing the training data corresponding to each of the compost turning and fermentation data into a plurality of groups according to the compost evaluation data, and sorting the training data in each group according to the first compost volume;

[0090] According to the sorting order, a number of the training data are respectively selected from each of the groups to form a training data group, and a plurality of the training data groups are obtained, and a process parameter optimization model is trained using a number of the training data in each of the training data groups until the training meets the standard;

[0091] The optimization unit is used to determine the second manure straw data and the second compost volume involved in this composting, use the trained process parameter optimization model to process the second manure straw data and the second compost volume, and predict the second process parameters.

[0092] Furthermore, the training unit is used to:

[0093] The training data corresponding to each of the compost turning and composting data are sorted from high to low according to the compost evaluation data, and each of the sorted training data is divided into corresponding groups based on a preset division ratio;

[0094] The absolute value of the volume difference between the first compost volume and the second compost volume is calculated, and the sorting sequence number of each training data in the corresponding group is determined according to the absolute value; wherein, the smaller the absolute value is, the smaller the corresponding sorting sequence number is, that is, the closer it is to the front.

[0095] Furthermore, the training unit is also used for:

[0096] Obtaining data source information of each of the compost turning and fertilizer fermentation data, evaluating the process compliance of the data source information, and obtaining a process compliance value;

[0097] If the process compliance value is lower than a preset value and the compost evaluation data of the compost turning and fermentation data is evaluated as medium level or above, an increase factor is determined according to the difference between the process compliance value and the preset value, and the compost evaluation data is optimized using the increase factor.

[0098] Furthermore, the training unit is used to:

[0099] In the first stage of training the process parameter optimization model, a number of the training data are selected from each of the groups according to a first selection rule to form a training data group; wherein the first selection rule refers to selecting training data accounting for a first proportion, a second proportion, and a third proportion of the total number from the groups with high-level, medium-level, and low-level compost evaluation data, respectively, and the first proportion is higher than the second proportion, and the second proportion is higher than the third proportion;

[0100] In the second stage of training the process parameter optimization model, a number of the training data are respectively selected from each of the groups according to a second selection rule to form a training data set; wherein, the second selection rule means that the training data accounting for the fourth ratio, the fifth ratio, and the sixth ratio are respectively selected from the groups with high-grade, medium-grade, and low-grade compost evaluation data, and the fourth ratio is higher than the fifth ratio, the fifth ratio is higher than the sixth ratio, and the fourth ratio is lower than the first ratio.

[0101] Further, the division ratio is determined by the following method:

[0102] Obtain a number of training record data for the process parameter optimization model. The training data includes model test metrics and the historical division ratio adopted. Screen a number of training record data with the optimal model test metrics among them, and fuse the historical division ratios corresponding to these training record data to obtain the division ratio for this training.

[0103] An embodiment of the present invention also discloses an electronic device, including: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory and executes the method as described in the foregoing embodiment.

[0104] An embodiment of the present invention also discloses a computer storage medium, on which a computer program is stored. When the computer program is run by a processor, it executes the method as described in the foregoing embodiment.

[0105] An embodiment of the present invention also discloses a computer program product, which is used to be called and run by a processor of an electronic device to implement the method as described in any one of the foregoing.

[0106] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable load balancing devices, so that when the program codes are executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, executed partially on the machine and partially on a remote machine as an independent software package, or executed entirely on a remote machine or server.

[0107] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and no limitations are imposed herein.

[0108] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub - combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. An optimization method for the process parameters of manure and straw turning composting, characterized in that, It includes the following steps: Collect a number of real turning composting data, where each piece of the turning composting data includes first manure and straw data, first process parameters, compost evaluation data, and first compost volume; Construct a piece of training data based on the first manure and straw data, the first process parameters, the compost evaluation data, and the first compost volume in each piece of the turning composting data; among them, the compost evaluation data is used to construct the label of the training data; Divide the training data corresponding to each piece of the turning composting data into multiple groups according to the compost evaluation data, and sort the training data in each group according to the first compost volume; Select a number of the training data from each of the groups in the order of sorting to form a training data group, obtain multiple training data groups, and use a number of the training data in each training data group to train the process parameter optimization model until the training reaches the standard; the total number of the training data in the training data group is fixed or dynamic; Determine the second manure and straw data and the second compost volume involved in this composting, and use the trained process parameter optimization model to process the second manure and straw data and the second compost volume to predict the second process parameters; The dividing the training data corresponding to each piece of the turning composting data into multiple groups according to the compost evaluation data, and sorting the training data in each group according to the first compost volume includes: Sort the training data corresponding to each piece of the turning composting data from high to low according to the compost evaluation data, and divide the sorted training data into corresponding groups based on a preset division ratio; Calculate the absolute value of the volume difference between the first compost volume and the second compost volume, and determine the sorting serial number of each piece of the training data in the corresponding group according to the absolute value; where the smaller the absolute value, the smaller the corresponding sorting serial number, that is, the more forward; Selecting a number of the training data from each of the groups in the order of sorting to form a training data group includes: In the first stage of training the process parameter optimization model, select a number of the training data from each of the groups according to the first selection rule to form a training data group; where the first selection rule means selecting the training data accounting for the first ratio, the second ratio, and the third ratio of the total number from the groups with high-grade, medium-grade, and low-grade compost evaluation data respectively, and the first ratio is higher than the second ratio, and the second ratio is higher than the third ratio; In the second stage of training the process parameter optimization model, select a number of the training data from each of the groups according to the second selection rule to form a training data group; where the second selection rule means selecting the training data accounting for the fourth ratio, the fifth ratio, and the sixth ratio of the total number from the groups with high-grade, medium-grade, and low-grade compost evaluation data respectively, and the fourth ratio is lower than the sixth ratio, and the fourth ratio is lower than the first ratio.

2. The optimization method for the process parameters of manure and straw turning composting according to claim 1, wherein: Before constructing a piece of training data based on the first manure and straw data, the first process parameters, the compost evaluation data, and the first compost volume structure in each piece of the turning and composting data, the method further includes: Obtain the data source information of each piece of the turning and composting data, evaluate the process compliance of the data source information, and obtain a process compliance value; If the process compliance value is lower than a preset value and the compost evaluation data of the turning and composting data is rated at medium level or above, determine a boosting factor according to the difference between the process compliance value and the preset value, and use the boosting factor to optimize the compost evaluation data.

3. An optimization method for the process parameters of manure and straw turning composting according to claim 1, characterized in that: The division ratio is determined by the following method: Obtain a number of training record data for the process parameter optimization model. The training data includes model test indicators and historical division ratios adopted. Screen a number of training record data with the best model test indicators among them, and fuse the historical division ratios corresponding to these training record data to obtain the division ratio for this training.

4. An optimization system for the process parameters of manure and straw turning composting, characterized in that: The system includes a collection unit, a training unit, and an optimization unit; The collection unit is used to collect a number of real turning and composting data. Each piece of the turning and composting data includes first manure and straw data, first process parameters, compost evaluation data, and a first compost volume; The training unit is used to: construct a piece of training data based on the first manure and straw data, the first process parameters, the compost evaluation data, and the first compost volume in each piece of the turning and composting data; wherein, the compost evaluation data is used to construct the label of the training data; Divide the training data corresponding to each piece of the turning and composting data into multiple groups according to the compost evaluation data, and sort the training data in each group according to the first compost volume; Select a number of the training data from each of the groups in the order of sorting to form a training data group, obtain multiple training data groups, and use a number of the training data in each training data group to train the process parameter optimization model until the training reaches the standard; The optimization unit is used to determine the second manure and straw data and the second compost volume involved in this composting, and use the trained process parameter optimization model to process the second manure and straw data and the second compost volume to predict the second process parameters; The training unit is used to: Sort the training data corresponding to each piece of the turning and composting data from high to low according to the compost evaluation data, and divide the sorted training data into corresponding groups based on a preset division ratio; Calculate the absolute value of the volume difference between the first compost volume and the second compost volume, and determine the sorting serial number of each piece of the training data within the corresponding group according to the absolute value; wherein, the smaller the absolute value, the smaller the corresponding sorting serial number, that is, the more forward; The training unit is used to: In the first stage of training the process parameter optimization model, a number of the training data are respectively selected from each of the groups according to the first selection rule to form a training data set; wherein, the first selection rule means that the training data accounting for the first ratio, the second ratio, and the third ratio of the total are respectively selected from the groups with high-level, medium-level, and low-level compost evaluation data, and the first ratio is higher than the second ratio, and the second ratio is higher than the third ratio; In the second stage of training the process parameter optimization model, a number of the training data are respectively selected from each of the groups according to the second selection rule to form a training data set; wherein, the second selection rule means that the training data accounting for the fourth ratio, the fifth ratio, and the sixth ratio of the total are respectively selected from the groups with high-level, medium-level, and low-level compost evaluation data, and the fourth ratio is lower than the sixth ratio, and the fourth ratio is lower than the first ratio.

5. An electronic device, comprising: A memory storing executable program code; A processor coupled to the memory; characterized in that: the processor calls the executable program code stored in the memory and executes the method according to any one of claims 1-3.

6. A computer storage medium, on which a computer program is stored, characterized in that: When the computer program is run by the processor, it executes the method according to any one of claims 1-3.

7. A computer program product, characterized in that: The computer program product is used to be called and run by the processor of an electronic device to implement the method according to any one of claims 1-3.

Images