A method and system for resource processing of kitchen waste and municipal sludge
By calculating the importance and correlation of multi-dimensional environmental parameters of kitchen waste and municipal sludge, principal component analysis, dimensionality reduction and cluster matching are carried out, the problem of inaccurate matching of temperature parameters in the existing technology is solved and the resource conversion rate is improved.
Patent Information
- Application Number
- CN202410370933.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-29
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-03-29
AI Technical Summary
In the prior art, the dimensionality reduction effect is poor using principal component analysis, resulting in inaccurate matching of temperature parameters during the resource treatment of kitchen waste and municipal sludge, which reduces the resource conversion rate.
By obtaining the multi-dimensional environmental parameters and resourced outputs in the historical time period, calculating the importance and correlation of each dimension, performing random sorting and principal component analysis to reduce the dimension, obtaining the optimal dimensionality reduction data, and performing cluster matching. Obtain the reference temperature for the real-time resource processing process.
More accurate temperature parameter matching is achieved, resource conversion rate is improved, and the effectiveness of resource processing is ensured.
Smart Images

Figure CN118260580B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of garbage resource treatment, and in particular to a method and system for resource treatment of kitchen waste and municipal sludge. Background Art
[0002] In the city's waste resource management, both restaurant waste and municipal sludge are identified as wet waste for treatment. Wet waste refers to waste that contains a lot of water, including food scraps, fruit peels, vegetable residues, etc., which belong to restaurant waste, and sludge waste containing organic matter, inorganic matter, and microorganisms produced by sewage treatment plants after treating wastewater, which belongs to municipal sludge. Therefore, resource management is mainly achieved through resource treatment plants using biodegradation technology to convert wet waste into biogas, organic fertilizers and other substances.
[0003] There are many environmental parameters in the process of resource treatment of wet garbage, including humidity, oxygen concentration, pH value and carbon-nitrogen ratio, among which the parameter with the greatest impact on biological decomposition technology is the temperature parameter. Because the environmental parameters of different batches of wet garbage are different, the corresponding most suitable temperature parameters are also different. In order to accurately control the temperature parameters in each resource treatment process, it is necessary to find the most suitable temperature parameters as a reference in combination with the resource treatment process in the historical time period. Because the environmental parameters contain multiple dimensions, the environmental parameters need to be reduced in dimension during the matching process, so as to quickly and accurately achieve the matching purpose. In the prior art, the order of different dimensions in the process of reducing the dimension of multidimensional environmental parameters using principal component analysis will lead to different principal component analysis results. If the dimensionality reduction effect is not good, the selected principal component information will not accurately reflect the influence of each environmental parameter on the resource treatment, which will cause errors in the matching results, and the reference temperature of real-time resource treatment is not suitable, which reduces the resource conversion rate. Summary of the invention
[0004] In order to solve the technical problem that the prior art uses principal component analysis to reduce the dimension, which leads to the failure of resource processing matching and the inability to obtain accurate temperature reference information, the purpose of the present invention is to provide a method and system for resource processing of kitchen waste and municipal sludge, and the technical scheme adopted is as follows:
[0005] The present invention proposes a method for resource processing of kitchen waste and municipal sludge, the method comprising:
[0006] Obtain the multi-dimensional environmental parameters and resource output of each resource recovery process of restaurant kitchen waste and municipal sludge by resource recovery plants in the historical time period;
[0007] Under each dimension, the dimensional importance of each dimension is obtained according to the distribution disorder of environmental parameters and the correlation between environmental parameters and resource output;
[0008] Randomly sort all dimensions to obtain multiple dimension sorting results; perform principal component analysis dimensionality reduction according to each dimension sorting result, and obtain the dimensionality reduction priority between the two dimensions according to the dimensional importance between the two dimensions and the data distribution of the environmental parameters after the two dimensions are combined; obtain the optimal dimension sorting result according to all dimensionality reduction priorities in the dimensionality sorting results, as well as the number of dimensions involved in dimensionality reduction and their dimensionality reduction priorities;
[0009] According to the optimal dimension sorting result, the multi-dimensional environmental parameters are reduced in dimension to obtain optimal dimension reduction data; according to the optimal dimension reduction data, each resource processing process is clustered to obtain clusters of similar processing processes;
[0010] The real-time resource recovery process is matched with the similar process cluster according to the similarity between the environmental parameters after dimensionality reduction, and the temperature parameter of the resource recovery process in the successfully matched similar process cluster is used as the reference temperature of the real-time resource recovery process.
[0011] Furthermore, the method for obtaining the resource output includes:
[0012] The volume of biogas and the mass of organic fertilizer produced in each resource recovery process are obtained, and the Euclidean norm of the biogas volume and the mass of organic fertilizer is used as the resource recovery output.
[0013] Furthermore, the method for obtaining the distribution disorder includes:
[0014] In each dimension, the information entropy of the environmental parameters is taken as the distribution disorder.
[0015] Furthermore, the method for obtaining the correlation includes:
[0016] Under each dimension, the difference between the environmental parameters and the resource output in each resource processing process is normalized to obtain the first output impact of each resource processing process under the corresponding dimension; the average first output impact of all resource processing processes is used as the second output impact under the corresponding dimension;
[0017] In each dimension, the correlation coefficient between the environmental parameter sequence and the resource output sequence formed by all resource recovery processes is obtained, and the correlation coefficient is multiplied by the second output influence to obtain the correlation between the environmental parameter and the resource recovery output in the corresponding dimension.
[0018] Furthermore, the method for obtaining the dimension importance includes:
[0019] Under each dimension, the correlation is multiplied by the distribution disorder to obtain the dimension importance of each dimension.
[0020] Furthermore, the method for obtaining the dimensionality reduction priority includes:
[0021] The dimensionality reduction priority is obtained according to the dimensionality reduction priority formula, and the dimensionality reduction priority formula includes:
[0022] where λ i,k is the dimensionality reduction priority between dimension i and dimension k in the dimension combination participating in dimensionality reduction, norm() is the normalization function, and E i is the dimension importance of dimension i, E k is the dimension importance of dimension k, N i,k is the number of data points in the dimension combination of dimension i and dimension k, D(a,b) is the distance between data point a and data point b in the dimension combination of dimension i and dimension k, maxσ i,k It is the maximum projection variance of the data points in the dimension combination of dimension i and dimension k.
[0023] Furthermore, the method for obtaining the optimal dimension sorting result includes:
[0024] In the process of principal component analysis dimensionality reduction for each dimension sorting result, the dimensionality reduction is stopped according to a preset contribution rate threshold, and the first dimensionality reduction priority between the dimensions involved in the dimensionality reduction and their dimensional combinations is obtained;
[0025] Assuming that the dimensionality reduction does not stop, the second dimensionality reduction priority between dimensions in the entire dimensionality reduction process is obtained;
[0026] The discriminant value formula is used to obtain the discriminant value of each dimension sorting result, and the dimension sorting result corresponding to the maximum discriminant value is selected as the optimal dimension sorting result; the discriminant value formula includes:
[0027] f is the discriminant value, y is the number of dimension sorting results participating in the dimension reduction process of principal component analysis, μ is the preset contribution rate threshold, λ 1,p is the first dimensionality reduction priority of the pth dimension combination participating in dimensionality reduction, g is the total number of dimensions, λ 2,l It is the second dimensionality reduction priority of the lth dimension combination in the entire dimensionality reduction process.
[0028] Furthermore, the method for obtaining the clusters of similar processing processes includes:
[0029] Select one principal component dimension in the optimal dimensionality reduction data as the target principal component dimension, combine the target principal component dimension with any other dimension of the environmental parameter to obtain a third dimensionality reduction priority, and use the average third dimensionality reduction priority of all dimension combinations corresponding to the target principal component dimension as a similarity weight; obtain the initial similarity between the environmental parameters corresponding to the target principal component dimension of the two resource processing processes, multiply the initial similarity by the similarity weight, and obtain the optimized similarity between the two resource processing processes under the target principal component dimension;
[0030] The optimized similarities corresponding to each principal component dimension in the optimal dimensionality reduction data are squared and accumulated to obtain a first accumulated value, the first accumulated value is squared and normalized to obtain a final similarity between two resource processing processes; the resource processing processes are classified according to the final similarity to obtain clusters of similar processing processes.
[0031] Furthermore, matching the real-time resource processing process with the cluster of similar processing processes according to the similarity between the environmental parameters includes:
[0032] The real-time resource processing process is reduced in dimension according to the principal component direction of the optimal dimensionality reduction data to obtain real-time dimensionality reduction data; the real-time dimensionality reduction data has the same principal component dimension as the optimal dimensionality reduction data;
[0033] One principal component dimension in the optimal dimensionality reduction data is randomly selected as the target principal component dimension, the data average value of the optimal dimensionality reduction data under the target principal component dimension in the similar processing process cluster is used as the reference data of the corresponding similar processing process cluster, and the similarity between the real-time dimensionality reduction data under the target principal component dimension and the reference data is used as the initial matching degree under the target principal component dimension; the initial matching degree of each principal component dimension is squared and then accumulated to obtain a second accumulated value, the second accumulated value is squared to obtain the matching degree between the real-time resource processing process and the similar processing process cluster, and the similar processing process cluster corresponding to the maximum matching degree is selected as the matching cluster of the real-time resource processing process.
[0034] The present invention proposes a system for resource processing of food waste and municipal sludge, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the processor implements any one of the steps of a method for resource processing of food waste and municipal sludge.
[0035] The present invention has the following beneficial effects:
[0036] The embodiment of the present invention first obtains the dimensional importance of each dimension of the environmental parameter. The dimensional importance can characterize the degree of influence of the dimension in the resource processing process, and then serve as a reference for the subsequent evaluation of the dimensionality reduction order, that is, the greater the degree of influence, the more it is necessary to make the final principal component direction deviate from the dimension, that is, the information of the dimension needs to be retained, and the priority of the dimensionality reduction of the dimension should be greater. The principal component analysis and dimensionality reduction are further performed on the sorting results of each dimension, and the optimal dimension sorting result is determined by obtaining information such as the dimensionality reduction priority between the dimensions in the sorting results, and the optimal dimensionality reduction data is obtained. That is, the optimal dimensionality reduction data is the result with a better dimensionality reduction effect. According to the information in the optimal dimensionality reduction data, the environmental information in the resource processing process can be effectively characterized, and then accurate classification can be achieved. The real-time resource processing process can be matched with the cluster of similar processing processes through the similarity between the environmental parameters after dimensionality reduction to obtain the reference temperature of the real-time resource processing process. The resource conversion rate can be improved by adjusting the parameters of the resource processing process through the reference temperature. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0038] Figure 1 This is a flow chart of a method for resource processing of kitchen waste and municipal sludge provided in one embodiment of the present invention. DETAILED DESCRIPTION
[0039] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a method and system for resource processing of kitchen waste and municipal sludge proposed by the present invention, its specific implementation method, structure, characteristics and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
[0041] The specific scheme of the method and system for resource processing of kitchen waste and municipal sludge provided by the present invention is described in detail below with reference to the accompanying drawings.
[0042] See also Figure 1, which shows a flow chart of a method for resource processing of kitchen waste and municipal sludge provided by an embodiment of the present invention, the method comprising:
[0043] Step S1: Obtain multi-dimensional environmental parameters and resource output during each resource recovery process of food waste and municipal sludge by a resource recovery treatment plant in a historical time period.
[0044] During the biological decomposition of wet garbage such as food waste and municipal sludge in resource recovery plants, there will be many types of environmental parameters, such as humidity, oxygen concentration, pH value, carbon-nitrogen ratio, etc. Each type of environmental parameter can be regarded as a dimension. Therefore, multidimensional environmental parameters will be generated in a resource recovery process. Multidimensional environmental parameters can characterize the type of the current resource recovery process.
[0045] The embodiment of the present invention aims to determine the reference information in the real-time information by matching the real-time information with the historical information. Therefore, it is necessary to obtain the multi-dimensional environmental parameters in multiple resource processing processes in the historical time period, and because it is the resource processing process at the historical moment, the resource output after each resource processing process is completed can be counted. In the subsequent process, the importance of each dimension can be analyzed according to the relationship between the resource output and the environmental parameters in each dimension.
[0046] In one embodiment of the present invention, one day's operation of the resource recovery plant is regarded as one resource recovery process, and the historical time period is set to 30 days, that is, data from 30 resource recovery processes are collected.
[0047] Preferably, in one embodiment of the present invention, the method for obtaining resource output includes:
[0048] The volume of biogas and the mass of organic fertilizer produced in each resource recovery process are obtained, and the Euclidean norm of the volume of biogas and the mass of organic fertilizer is taken as the resource recovery output. That is, the resource recovery output is expressed by the formula: Among them, H j is the resource output of the jth resource recovery process, V j is the volume of biogas in the jth resource recovery process, M j is the quality of organic fertilizer in the jth resource recovery process.
[0049] It should be noted that in the embodiment of the present invention, in order to ensure the simplicity and accuracy of the calculation project in the subsequent steps, all collected data need to be normalized. According to the data range of the same latitude data, the value range of each collected data is limited to between 0 and 1 to avoid the influence of dimension in the subsequent calculation process.
[0050] Step S2: Under each dimension, the dimensional importance of each dimension is obtained according to the distribution disorder of the environmental parameters and the correlation between the environmental parameters and the resource output.
[0051] For the principal component analysis dimensionality reduction algorithm, it achieves dimensionality reduction by sequentially constructing dimension combinations and calculating contribution rates, and retaining multiple data dimensions with the largest contribution rates. Therefore, the order of combination between dimensions in the principal component analysis process is more important. In the dimensionality reduction process, important environmental parameters should be reduced first, and then unimportant dimensions should be removed, and important environmental parameters should be retained to form dimensionality reduction results. Therefore, it is necessary to first obtain the importance of environmental parameters in each dimension.
[0052] For an environmental parameter, the greater its change in different resource processing processes, the more features the data shows, and the more important the information is; if the change is large and the resource output has the same change, it means that the environmental parameter is an important parameter affecting the resource output, and the corresponding dimension is more important. Therefore, the importance of each dimension can be obtained based on the distribution disorder of the environmental parameters and the correlation between the environmental parameters and the resource output.
[0053] Preferably, in one embodiment of the present invention, in each dimension, the information entropy of the environmental parameters is used as the distribution chaos. That is, the environmental parameters generated by each resource processing process are constructed into a sequence, and each dimension corresponds to a sequence. The distribution chaos can be obtained by analyzing the information entropy of the elements in the environmental parameter sequence. That is, the greater the information entropy, the greater the change of the elements in the sequence, and the greater the distribution chaos. The expression of distribution chaos is:
[0054] Where V i is the distribution disorder of the i-th dimension, G represents the number of categories of data in the environmental parameters of the i-th dimension, and P i,z is the frequency of the zth category data of the environmental parameter under the ith dimension, and ln is the logarithmic function with a natural constant as the base.
[0055] It should be noted that the information entropy formula is a technical means well known to those skilled in the art, and the formula composition and meaning will not be elaborated here.
[0056] Preferably, in one embodiment of the present invention, the method for obtaining the correlation includes:
[0057] In each dimension, the difference between the environmental parameters and the resource output in each resource processing process is normalized to obtain the first output impact of each resource processing process in the corresponding dimension; the average first output impact of all resource processing processes is used as the second output impact in the corresponding dimension. That is, the size relationship between the environmental parameters and the resource output is characterized by differences. The larger the difference, the larger the data of the environmental parameters relative to the resource output, or the larger the data of the resource output relative to the environmental parameters. Because the data are all normalized, the larger the difference, the greater the possibility that the environmental parameters in this dimension will affect the resource output. Therefore, the average first output impact is used as the second output impact in the corresponding dimension. In one embodiment of the present invention, the formula for the second output impact is expressed as:
[0058] Among them, A i is the second output impact under the i-th dimension, N is the number of resource processing processes, T i,j is the environmental parameter of the i-th dimension in the j-th resource recovery process, H j is the resource output under the j-th resource recovery process, min() is the minimum selection function, max() is the maximum selection function, and norm() is the normalization function.
[0059] In the formula for the second output impact, the difference between the two data is represented in the form of a ratio. That is, the larger the ratio, the greater the relative difference between the two data, and the greater the corresponding first output impact and second output impact.
[0060] In each dimension, the correlation coefficient between the environmental parameter sequence and the resource output sequence formed by all resource recovery processes is obtained. The larger the correlation coefficient, the more likely the environmental parameters and resource recovery output in the corresponding dimension have the same change relationship, and the more likely the environmental parameters are to affect the resource recovery output. Therefore, the correlation coefficient is multiplied by the second output influence to obtain the correlation between the environmental parameters and resource recovery output in the corresponding dimension.
[0061] In one embodiment of the present invention, the correlation coefficient is represented by the absolute value of the Pearson correlation coefficient, that is, the environmental parameters of the corresponding dimensions in each resource processing process constitute an environmental parameter sequence, and the corresponding resource output constitutes a resource output sequence. Because the lengths of the two sequences are the same and the elements are normalized, the absolute value of the Pearson correlation coefficient can be directly calculated to obtain the correlation coefficient, that is, the larger the correlation coefficient, the more similar the change trend of the two data. It should be noted that the calculation method of the Pearson correlation coefficient is a technical means well known to those skilled in the art and will not be elaborated here.
[0062] In one embodiment of the present invention, because the greater the distribution disorder, the more obvious the characteristic information of the environmental parameters under this dimension, the more important the environmental parameters of this dimension are; the greater the correlation between the environmental parameters and the resource output, the more likely the environmental parameters under this dimension are to affect the resource output, so the correlation is multiplied by the distribution disorder to obtain the dimension importance of each dimension.
[0063] Step S3: Randomly sort all dimensions to obtain multiple dimension sorting results; perform principal component analysis dimensionality reduction based on the sorting results of each dimension, and obtain the dimensionality reduction priority between the two dimensions based on the dimensional importance between the two dimensions and the data distribution of the environmental parameters after the two dimensions are combined; obtain the optimal dimension sorting result based on all dimensionality reduction priorities in the dimension sorting results, as well as the number of dimensions involved in dimensionality reduction and their dimensionality reduction priorities.
[0064] In the process of principal component analysis and dimensionality reduction, as the dimensions are combined and reduced in sequence, the proportion of the principal component formed by any dimension in each dimensionality reduction process gradually decreases. Therefore, the earlier the order of dimensionality reduction, the greater the proportion of the principal component it occupies, and the more data features are retained. Therefore, in the process of dimensionality reduction, it is necessary to determine the effective dimension sorting results for dimensionality reduction. In order to find the optimal dimension sorting result, all dimensions can be randomly sorted first to obtain multiple dimension sorting results. In one embodiment of the present invention, the dimensions are randomly sorted to obtain 50 dimensional sorting results for subsequent analysis.
[0065] According to the sorting results of each dimension, the principal component analysis can be used for dimensionality reduction. In the principal component analysis dimensionality reduction process, the data of one dimension is first decentralized and then divided by the standard deviation of the data, so that the mean of the seed data is 0 and the standard deviation is 1; the adjacent two dimensions in the dimension sorting results are further combined, and the dimension combination formed by the two dimensions corresponds to a two-dimensional coordinate system, and the maximum projection variance can be calculated, that is, the contribution rate in the dimensionality reduction process. In the traditional principal component analysis dimensionality reduction process, the contribution rate can be used to control the dimensionality reduction process. When the contribution rate reaches a certain threshold, the dimensionality reduction is stopped, that is, the information of the subsequent dimensions is eliminated, and the current dimensionality reduction result is retained. It should be noted that because the traditional principal component analysis dimensionality reduction does not take into account the dimensional importance of each dimension, the maximum projection variance cannot be directly used as the contribution rate when reducing the dimensionality of environmental parameters. It is necessary to further obtain the dimensionality reduction priority between the two dimensions based on the dimensional importance between the two dimensions in the dimensional combination and the data distribution of the environmental parameters after the two dimensions are combined. If a dimension sorting result is a better sorting result, the dimension reduction priority between the dimensions designed in the dimensionality reduction process will be larger, and the number of dimensions involved in the dimensionality reduction will be smaller. Therefore, the optimal dimension sorting result can be screened out based on all dimensionality reduction priorities in the dimension sorting results, the number of dimensions involved in the dimensionality reduction and their dimensionality reduction priorities.
[0066] It should be noted that in the process of dimension combination by dimensionality reduction of principal component analysis, two dimensions form a dimension combination, which will be reduced to a new dimension after component analysis. This new dimension and the subsequent original dimension in the dimension sorting result form a new dimension combination, and so on. In the dimension sorting result, except for the first pair of adjacent dimensions, the other dimension combinations are a combination of a new dimension after dimensionality reduction and the original dimension. And the new dimension obtained by dimensionality reduction needs to recalculate the dimension importance, which will not be repeated in the subsequent description for the convenience of description.
[0067] Preferably, in one embodiment of the present invention, a method for obtaining dimensionality reduction priority includes:
[0068] The dimensionality reduction priority is obtained according to the dimensionality reduction priority formula, which includes:
[0069] where λ i,k is the dimensionality reduction priority between dimension i and dimension k in the dimension combination participating in dimensionality reduction, norm() is the normalization function, and E i is the dimension importance of dimension i, E k is the dimension importance of dimension k, N i,k is the number of data points in the dimension combination of dimension i and dimension k, D(a,b) is the distance between data point a and data point b in the dimension combination of dimension i and dimension k, maxσ i,k It is the maximum projection variance of the data points in the dimension combination of dimension i and dimension k.
[0070] In the dimensionality reduction priority formula, the greater the importance of the dimension between two dimensions, the more important the two dimensions are, the more they need to be prioritized for dimensionality reduction, and the greater the need for this dimension combination to be at the front of the dimensionality reduction order, that is, the greater the dimensionality reduction priority; the smaller the difference in dimensionality importance between the two dimensions, the closer the influence of the two dimensions on the resource output, the more suitable the two dimensions are as a dimensional combination, and the greater the corresponding dimensionality reduction priority should be. It represents the cumulative sum of the distances between data points in the two-dimensional coordinate system formed by the dimension combination. This item can characterize the distribution of data points under the dimension combination. The larger the item is, the more dispersed the data points are distributed, which means that the data information in the dimension combination is richer, and the corresponding dimensionality reduction priority should be greater; the maximum projection variance is the contribution rate calculated in the traditional principal component analysis process. The larger the value is, the stronger the dimensionality reduction priority of the dimension combination is.
[0071] Preferably, in one embodiment of the present invention, a method for obtaining the optimal dimension sorting result includes:
[0072] In the process of principal component analysis dimensionality reduction for each dimension sorting result, the dimensionality reduction is stopped according to the preset contribution rate threshold, and the first dimensionality reduction priority between the dimensions involved in the dimensionality reduction and their dimensionality combinations is obtained. In one embodiment of the present invention, the contribution rate threshold is set to 0.7. It should be noted that this process is the process of controlling the start and stop of dimensionality reduction in the conventional principal component analysis dimensionality reduction process. The dimensionality reduction process can also continue without stopping to analyze the combination information between subsequent dimensions. Assuming that the dimensionality reduction is not stopped, the second dimensionality reduction priority between the dimensions in the entire dimensionality reduction process is obtained. It should be noted that the methods for obtaining the first dimensionality reduction priority and the second dimensionality reduction priority are the same and will not be repeated.
[0073] The discriminant value formula is used to obtain the discriminant value of each dimension sorting result, and the dimension sorting result corresponding to the maximum discriminant value is selected as the optimal dimension sorting result; the discriminant value formula includes:
[0074] f is the discriminant value, y is the number of dimension sorting results participating in the dimension reduction process of principal component analysis, μ is the preset contribution rate threshold, and λ 1,p is the first dimensionality reduction priority of the pth dimension combination participating in dimensionality reduction, g is the total number of dimensions, λ 2,l It is the second dimensionality reduction priority of the lth dimension combination in the entire dimensionality reduction process.
[0075] In the discriminant value formula, It represents the cumulative value of the second dimensionality reduction priority of the dimension combination in the entire dimensionality reduction process. The larger the cumulative value, the better the sorting effect of the dimension sorting result, and the larger the discriminant value should be. In the left item, the smaller the number of participants in dimensionality reduction should be, the better, that is, y is located in the denominator, and the smaller the value, the greater the discriminant value; while the smaller the number of participants in dimensionality reduction, the closer the cumulative value of the first dimensionality reduction priority between the involved dimensional combinations should be to the contribution rate threshold, the better, that is, The smaller it is, the better the dimension combination involved in dimensionality reduction can meet the contribution rate threshold requirement, and the larger the corresponding discriminant value is. Therefore, the dimension sorting result with the largest discriminant value can be selected as the optimal dimension sorting result.
[0076] Step S4: reducing the dimensions of the multi-dimensional environmental parameters according to the optimal dimension sorting result to obtain optimal dimension reduction data; clustering each resource processing process according to the optimal dimension reduction data to obtain clusters of similar processing processes.
[0077] According to the optimal dimension sorting results, the multidimensional environmental parameters can be reduced in dimension to obtain the optimal reduced dimension data. It should be noted that the dimensions in the optimal reduced dimension data are all new principal component dimensions obtained by dimensionality reduction, and corresponding data information exists for each resource processing process. After effective dimensionality reduction, useless environmental parameter information can be eliminated, and the component information with distinctive characteristics for the resource processing process can be retained. Therefore, based on the optimal reduced dimension data, each resource processing process can be effectively clustered to obtain clusters of similar processing processes, that is, the resource processing processes in the clusters of similar processing processes can be regarded as resource processing processes with the same working mode, while the working modes of resource processing processes between different clusters of similar processing processes are different.
[0078] Preferably, in one embodiment of the present invention, the method for obtaining clusters of similar processing processes includes:
[0079] One of the principal component dimensions in the optimal dimensionality reduction data is selected as the target principal component dimension, and the target principal component dimension is combined with any other dimension of the environmental parameters to obtain the third dimensionality reduction priority, that is, the calculation method of the third dimensionality reduction priority is the same as the above-mentioned dimensionality reduction priority calculation method, and the designed dimension importance and other features are all obtained according to the same acquisition method, which will not be elaborated here. The average third dimensionality reduction priority of all dimensional combinations corresponding to the target principal component dimension is used as the similarity weight. The larger the similarity weight, the higher the credibility of the similar features reflected by the information of this dimension. Therefore, the initial similarity between the environmental parameters corresponding to the target principal component dimension of the two resource processing processes is obtained, and the initial similarity is multiplied by the similarity weight to obtain the optimized similarity between the two resource processing processes under the target principal component dimension.
[0080] The optimized similarities corresponding to each principal component dimension in the optimal dimensionality reduction data are squared and accumulated to obtain a first accumulated value, and the first accumulated value is normalized after taking the square root to obtain the final similarity between the two resource processing processes. The resource processing processes are classified according to the final similarity to obtain clusters of similar processing processes. In one embodiment of the present invention, the process of classifying resource processing processes according to the final similarity includes: first determining the number of clusters of similar processing processes by the elbow method, setting a similarity threshold, identifying resource processing processes whose final similarity is greater than the similarity threshold as the same category, and counting the final similarities between all resource processing processes until all clusters of similar processing processes are obtained.
[0081] In one embodiment of the present invention, the final similarity formula is expressed as:
[0082] Where F(c,d) is the final similarity between resource processing process c and resource processing process d, norm() is the normalization function, z is the number of principal component dimensions, g is the number of environmental parameters, λ q,t is the dimensionality reduction priority between the qth principal component dimension and the tth environmental parameter dimension, T q,c is the data of resource processing process c on the qth principal component dimension, T q,d The resource processing process d is the data on the qth principal component dimension, c is a preset constant, and c=0.1 is set.
[0083] The final similarity formula is based on the deformation of the Euclidean norm, and the initial similarity is weighted by introducing similarity weights, where It represents the initial similarity, that is, the smaller the difference between the two data, the greater the initial similarity, where the role of c is to prevent the denominator from being 0.
[0084] Step S5: Match the real-time resource recovery process with the cluster of similar process according to the similarity between the environmental parameters after dimensionality reduction, and use the temperature parameter of the resource recovery process in the successfully matched cluster of similar process as the reference temperature of the real-time resource recovery process.
[0085] After the resource utilization processes in the historical time period are grouped in step S4, clusters of similar processes can be matched based on the environmental parameters of the real-time resource utilization processes, and then the temperature parameters of the resource utilization processes in the successfully matched clusters of similar processes are used as the reference temperature of the real-time resource utilization processes. In one embodiment of the present invention, the average temperature of all resource utilization processes in the successfully matched clusters of similar processes is used as the reference temperature of the real-time resource utilization process. In another embodiment of the present invention, the temperature of the resource utilization process corresponding to the cluster center point of the cluster of similar processes can also be used as the reference temperature.
[0086] Preferably, in one embodiment of the present invention, matching the real-time resource processing process with the cluster of similar processing processes according to the similarity between the environmental parameters includes:
[0087] The real-time resource processing process is reduced in dimension according to the principal component direction of the optimal dimensionality reduction data to obtain real-time dimensionality reduction data; the real-time dimensionality reduction data has the same principal component dimension as the optimal dimensionality reduction data. That is, the multi-dimensional environmental parameter information obtained in the real-time resource processing process is mapped to the principal component direction of the optimal dimensionality reduction data, and then the real-time dimensionality reduction data of the real-time resource processing process is determined, and the dimensional data with unclear information characteristics is eliminated.
[0088] One of the principal component dimensions in the optimal dimensionality reduction data is selected as the target principal component dimension. The average value of the optimal dimensionality reduction data under the target principal component dimension in the cluster of similar processing processes is used as the reference data of the corresponding cluster of similar processing processes. The similarity between the real-time dimensionality reduction data and the reference data under the target principal component dimension is used as the initial matching degree under the target principal component dimension. The initial matching degree of each principal component dimension is squared and then accumulated to obtain a second accumulated value. The second accumulated value is squared to obtain the matching degree between the real-time resource processing process and the cluster of similar processing processes. The cluster of similar processing processes corresponding to the maximum matching degree is selected as the matching cluster of the real-time resource processing process. That is, the matching degree is also a deformation of the Euclidean norm formula, which is expressed as follows:
[0089]
[0090] Among them, F ′ is the matching degree, T q,0 is the data of the real-time dimension reduction data under the qth principal component dimension, is the reference data under the qth principal component dimension.
[0091] It should be noted that after the real-time resource processing process is completed, it can be used as a historical resource processing process to update the cluster of similar processing processes to avoid the impact of random changes in data on temperature parameters.
[0092] In summary, the embodiment of the present invention determines the dimensional importance of environmental parameters under each dimension according to the relationship between environmental parameters and resource output. Multiple dimensional sorting results are obtained by random sorting, and the dimensionality reduction priority is determined according to the dimensional importance in the dimensional combination and the data distribution. The optimal dimensional sorting result is screened out from the multiple dimensional sorting results according to the dimensionality reduction priority to obtain the optimal dimensionality reduction data. The resource processing process is classified according to the optimal dimensionality reduction data to obtain clusters of similar processing processes. The reference temperature of the real-time resource processing process can be obtained by matching. The present invention obtains an accurate resource processing process category through the optimal dimensional sorting result, so that an effective reference temperature can be provided for the real-time resource processing process.
[0093] The present invention also proposes a system for resource processing of food waste and municipal sludge, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements any one of the steps of a method for resource processing of food waste and municipal sludge.
[0094] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0095] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.
Claims
1. A method for recycling kitchen waste and municipal sludge, characterized in that: The method comprises: Obtain the multi-dimensional environmental parameters and resource output of each resource recovery process of restaurant kitchen waste and municipal sludge by resource recovery plants in the historical time period; Under each dimension, the dimensional importance of each dimension is obtained according to the distribution disorder of environmental parameters and the correlation between environmental parameters and resource output; Randomly sort all dimensions to obtain multiple dimension sorting results; perform principal component analysis dimensionality reduction according to each dimension sorting result, and obtain the dimensionality reduction priority between the two dimensions according to the dimensional importance between the two dimensions and the data distribution of the environmental parameters after the two dimensions are combined; obtain the optimal dimension sorting result according to all dimensionality reduction priorities in the dimensionality sorting results, as well as the number of dimensions involved in dimensionality reduction and their dimensionality reduction priorities; According to the optimal dimension sorting result, the multi-dimensional environmental parameters are reduced in dimension to obtain optimal dimension reduction data; according to the optimal dimension reduction data, each resource processing process is clustered to obtain clusters of similar processing processes; Matching the real-time resource recovery process with the cluster of similar process according to the similarity between the environmental parameters after dimensionality reduction, and using the temperature parameter of the resource recovery process in the cluster of similar process that has been successfully matched as the reference temperature of the real-time resource recovery process; The method for obtaining the optimal dimension sorting result includes: In the process of principal component analysis dimensionality reduction for each dimension sorting result, the dimensionality reduction is stopped according to a preset contribution rate threshold, and the first dimensionality reduction priority between the dimensions involved in the dimensionality reduction and their dimensional combinations is obtained; Assuming that the dimensionality reduction does not stop, the second dimensionality reduction priority between dimensions in the entire dimensionality reduction process is obtained; The discriminant value formula is used to obtain the discriminant value of each dimension sorting result, and the dimension sorting result corresponding to the maximum discriminant value is selected as the optimal dimension sorting result; the discriminant value formula includes: f is the discriminant value, y is the number of dimension sorting results participating in the dimension reduction process of principal component analysis, μ is the preset contribution rate threshold, λ 1,p is the first dimensionality reduction priority of the pth dimension combination participating in dimensionality reduction, g is the total number of dimensions, λ 2,l It is the second dimensionality reduction priority of the lth dimension combination in the entire dimensionality reduction process.
2. A method for recycling kitchen waste and municipal sludge according to claim 1, characterized in that: The method for obtaining the resource output includes: The volume of biogas and the mass of organic fertilizer produced in each resource recovery process are obtained, and the Euclidean norm of the biogas volume and the mass of organic fertilizer is used as the resource recovery output.
3. A method for recycling kitchen waste and municipal sludge according to claim 1, characterized in that: The method for obtaining the distribution chaos includes: In each dimension, the information entropy of the environmental parameters is taken as the distribution disorder.
4. A method for recycling kitchen waste and municipal sludge according to claim 1, characterized in that: The method for obtaining the correlation includes: Under each dimension, the difference between the environmental parameters and the resource output in each resource processing process is normalized to obtain the first output impact of each resource processing process under the corresponding dimension; the average first output impact of all resource processing processes is used as the second output impact under the corresponding dimension; In each dimension, the correlation coefficient between the environmental parameter sequence and the resource output sequence formed by all resource recovery processes is obtained, and the correlation coefficient is multiplied by the second output influence to obtain the correlation between the environmental parameter and the resource recovery output in the corresponding dimension.
5. A method for recycling kitchen waste and municipal sludge according to claim 1, characterized in that: The method for obtaining the dimension importance includes: Under each dimension, the correlation is multiplied by the distribution disorder to obtain the dimension importance of each dimension.
6. A method for recycling kitchen waste and municipal sludge according to claim 1, characterized in that: The method for obtaining the dimensionality reduction priority includes: The dimensionality reduction priority is obtained according to the dimensionality reduction priority formula, and the dimensionality reduction priority formula includes: where λ i,k is the dimensionality reduction priority between dimension i and dimension k in the dimension combination participating in dimensionality reduction, norm() is the normalization function, and E i is the dimension importance of dimension i, E k is the dimension importance of dimension k, N i,k is the number of data points in the dimension combination of dimension i and dimension k, D(a,b) is the distance between data point a and data point b in the dimension combination of dimension i and dimension k, maxσ i,k It is the maximum projection variance of the data points in the dimension combination of dimension i and dimension k.
7. A method for recycling kitchen waste and municipal sludge according to claim 1, characterized in that: The method for obtaining clusters of similar processing processes includes: Select one principal component dimension in the optimal dimensionality reduction data as the target principal component dimension, combine the target principal component dimension with any other dimension of the environmental parameter to obtain a third dimensionality reduction priority, and use the average third dimensionality reduction priority of all dimension combinations corresponding to the target principal component dimension as a similarity weight; obtain the initial similarity between the environmental parameters corresponding to the target principal component dimension of the two resource processing processes, multiply the initial similarity by the similarity weight, and obtain the optimized similarity between the two resource processing processes under the target principal component dimension; The optimized similarities corresponding to each principal component dimension in the optimal dimensionality reduction data are squared and accumulated to obtain a first accumulated value, the first accumulated value is squared and normalized to obtain a final similarity between two resource processing processes; the resource processing processes are classified according to the final similarity to obtain clusters of similar processing processes.
8. The method for recycling kitchen waste and municipal sludge according to claim 1, characterized in that: The matching of the real-time resource processing process with the cluster of similar processing processes according to the similarity between the environmental parameters includes: The real-time resource processing process is reduced in dimension according to the principal component direction of the optimal dimensionality reduction data to obtain real-time dimensionality reduction data; the real-time dimensionality reduction data has the same principal component dimension as the optimal dimensionality reduction data; One principal component dimension in the optimal dimensionality reduction data is randomly selected as the target principal component dimension, the data average value of the optimal dimensionality reduction data under the target principal component dimension in the similar processing process cluster is used as the reference data of the corresponding similar processing process cluster, and the similarity between the real-time dimensionality reduction data under the target principal component dimension and the reference data is used as the initial matching degree under the target principal component dimension; the initial matching degree of each principal component dimension is squared and then accumulated to obtain a second accumulated value, the second accumulated value is squared to obtain the matching degree between the real-time resource processing process and the similar processing process cluster, and the similar processing process cluster corresponding to the maximum matching degree is selected as the matching cluster of the real-time resource processing process.
9. A system for recycling kitchen waste and municipal sludge, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for resource processing of food waste and municipal sludge as claimed in any one of claims 1 to 8 are implemented.
