CN114281945A - Carbon reduction strategy knowledge base construction method based on green product case base - Google Patents

Abstract

The invention discloses a carbon reduction strategy knowledge base construction method based on a green product case base, which comprises the following steps: 1, the life cycle of the green product comprises five stages; 2, dividing each paragraph in the case text a into five stages of a life cycle; 3 forming a target vocabulary set; 4, preprocessing each statement in the D paragraphs; 5 screening out K simple sentences; 6, recording a piece of strategy knowledge as { operation, object and effect }; 7, establishing a mapping relation between the operation and the effect; 8, calculating the semantic similarity of the two strategies, and combining the two strategies into one strategy when the semantic similarity is greater than a threshold value; and 9, correspondingly storing all the carbon reduction strategies into each strategy set, and marking labels and numbers to form a strategy knowledge base. The invention carries out similar retrieval by establishing the carbon reduction strategy knowledge base, thereby improving the reuse efficiency of the strategy knowledge and the feasibility of the carbon reduction technology.

Description

Carbon reduction strategy knowledge base construction method based on green product case base

Technical Field

The invention belongs to the field of strategy library construction, and particularly relates to a construction method of a carbon reduction strategy knowledge base based on a green product case library.

Background

With the global economic development, the carbon emission of each industry is greatly increased, the harm to the environment is increased day by day, and meanwhile, the greenhouse effect is caused, and the global temperature is increased. In order to reduce carbon emissions, a number of measures must be taken. The carbon reduction strategy is a relative concept, and reduction of carbon emission data relative to that before implementation of the strategy can be achieved by taking a series of strategic actions in terms of materials, structure, technology, manufacturing processes, etc.

Most of the existing methods are used for constructing a strategy knowledge base from the technical perspective, and the requirements in an actual scene are difficult to meet, so that the carbon reduction strategy is obtained by combining the existing case analysis, and the construction of the strategy knowledge base becomes a current hot research problem. Most of the existing strategy libraries are constructed by simply accumulating strategy knowledge, a storage standard is not formed, and the efficiency of strategy knowledge reuse in the future is not high.

Disclosure of Invention

The invention aims to overcome the defects in the prior art, and provides a construction method of a carbon reduction strategy knowledge base based on a green product case base, so that similar retrieval can be carried out by establishing the carbon reduction strategy knowledge base, and the reuse efficiency of strategy knowledge and the feasibility of a carbon reduction technology can be improved.

In order to achieve the purpose, the invention adopts the following technical scheme:

the invention relates to a construction method of a carbon reduction strategy knowledge base based on a green product case base, which is characterized by comprising the following steps of:

step 1: let the different phases of the life cycle of the green product include: the method comprises a material selection stage, a manufacturing process stage, a service performance stage, a recycling stage and a logistics maintenance stage;

step 2: separating a case text a in a green product case library according to paragraphs to obtain D paragraphs, and preliminarily dividing each paragraph into five stages of a life cycle according to a catalogue, a chapter title and a picture name table name in the case text a; finally dividing each paragraph into five stages of a life cycle according to the word frequency TF of the set feature words in the five stages respectively appearing in each paragraph; thus, the case text a is divided according to different stages of the life cycle;

and step 3: giving the query vocabulary as carbon emission and energy consumption vocabulary, and vectorizing to obtain a word vector A of carbon emission and energy consumption₁Calculating a word vector A₁Similarity with each word vector in an external text library is respectively obtained, and vocabularies corresponding to the word vectors with the similarity larger than a set threshold value are screened out, so that a target vocabulary set b ═ b is formed₁,b₂,...b_i...b_n}，b_iRepresenting the ith vocabulary in the target vocabulary set; n represents the total number of vocabularies in the target vocabulary set;

and 4, step 4: performing word segmentation on each sentence in D paragraphs in the case library text a, replacing pronouns after word segmentation with real words, and performing part-of-speech tagging to obtain each sentence in the D paragraphs after preprocessing;

and 5: dividing each sentence in the preprocessed D paragraphs into a plurality of simple sentences containing one object, and recording the object set in all the simple sentences in the preprocessed D-th paragraph as w_d＝{w_d1,w_d2,...,w_dj,...,w_dm}，w_djExpressing the jth object in the d paragraph, calculating the sum of the similarity between the object in each simple sentence of the d paragraph and each word in the target word set, then taking an average value, and screening out the simple sentences corresponding to the objects with the average values larger than a set average threshold value, thereby obtaining K screened simple sentences; d is equal to [1, D ∈]；

Step 6: for the predicate verb p in the k-th simple sentence after screening_kMarking and analyzing to obtain a subject s in the kth simple sentence_kObject o_kAnd the remaining components, thereby forming a kth predecessor-predicate triple SA_k＝{s_k,p_k,o_k}；

Let a policy knowledge be { operation, object, effect }, and use the predicate verb p_kAs an operation, object o_kAs an object, carbon emission reduction or energy saving is taken as an effect, thereby forming a kth model of operation-object-effect, K ∈ [1, K ∈]；

And 7: establishing a mapping relation f between the operation and the effect according to K models of the operation, the object and the effect_R：P₁，P₂,···,P_n→ R and as rule of "IF-THEN": IF P₁，P₂,···,P_nOf THENR, wherein P₁，P₂,···,P_nRepresents n operations, → represents causes, R represents carbon emission reduction or energy saving;

and 8: calculating any two strategies SA of the same stage of the life cycle by using the formula (1)₁And SA₂Sentence similarity S (SA)₁,SA₂)：

s(SA₁,SA₂)＝q₁×s(s₁,s₂)+q₂×s(p₁,p₂)+q₃×s(o₁,o₂) (1)

In the formula (1), q₁,q₂,q₃Represents the weight of the subject, predicate, object, respectively, and q₁+q₂+q₃＝1，s(s₁,s₂) Representation policy SA₁Subject of (1) s₁And strategy SA₂Subject of (1) s₂Similarity between them, s (p)₁,p₂) Representation policy SA₁Predicate p in (1)₁And strategy SA₂Predicate p in (1)₂Similarity between, s (o)₁,o₂) Representation policy SA₁Object of (1)₁And strategy SA₂Object of (1)₂The similarity between them; when S (SA)₁,SA₂) When the set threshold value is larger than the set threshold value, two strategies SA are used₁And SA₂Merging into a carbon reduction strategy; otherwise, two strategies SA₁And SA₂Respectively as a carbon reduction strategy;

and step 9: and finally dividing each paragraph into five stages of a life cycle, correspondingly storing all carbon reduction strategies into a material selection strategy set I, a manufacturing process strategy set II, a service performance strategy set III, a recycling strategy set IV and a logistics maintenance strategy set V, acquiring all feature vocabularies of green features based on the green features of the given five stages, and sequentially searching in each carbon reduction strategy according to each feature vocabulary, so that the carbon reduction strategies with the searched green features are labeled with corresponding green features, thereby forming the carbon reduction strategies with labels and sequentially giving index numbers to the strategy sets I-V to form a strategy knowledge base.

Compared with the prior art, the invention has the beneficial effects that:

1. the method is combined with a green product case base and starts from an actual scene, and a carbon reduction strategy knowledge base is constructed; existing cases in a green product case library are matched in a keyword search mode, so that the requirements in an actual scene can be better met, and carbon reduction strategies can be better collected; and labeling the obtained carbon reduction strategy knowledge according to the green characteristics, giving index numbers, correspondingly storing the carbon reduction strategy knowledge into five carbon reduction strategy sets, and directly carrying out similar retrieval from a strategy knowledge base when similar problems are encountered subsequently, so that the reuse efficiency of the strategy knowledge and the feasibility of the carbon reduction technology are improved.

2. According to the invention, the target vocabulary set is formed by calculating the word vector similarity, so that the keyword can be better retrieved in the text case a, and a more accurate result can be obtained. And in step 8, strategy knowledge with high similarity is merged by calculating semantic similarity of any two strategies, so that the carbon reduction strategies are more orderly indexed.

Drawings

FIG. 1 is a schematic flow diagram of the present invention;

FIG. 2 is a diagram of the strategy knowledge base form of the present invention;

FIG. 3 is a diagram of a policy knowledge unit of the present invention;

FIG. 4 is a graph of carbon footprint over the life cycle of the present invention.

Detailed Description

In this embodiment, a method for constructing a carbon reduction strategy knowledge base based on a green product case base, as shown in fig. 1, includes the following steps:

step 1: let the different phases of the life cycle of the green product include: the method comprises a material selection stage, a manufacturing process stage, a service performance stage, a recycling stage and a logistics maintenance stage;

as shown in fig. 2, the carbon footprint is the footprint of carbon emissions over a certain green product lifecycle.

Step 2: separating a case text a in a green product case library according to paragraphs to obtain D paragraphs, and preliminarily dividing each paragraph into five stages of a life cycle according to a catalogue, a chapter title and a picture name table name in the case text a; and then according to the set word frequency TF of the feature words of the five stages appearing in each paragraph respectively, wherein,

n_ijthe number of times a feature word representing a certain stage appears in this paragraph, n_kjRepresents the sum of the times of all the words appearing in this paragraph. Each paragraph is finally divided into five stages of the life cycle; thus completing the division of the case text a according to different stages of the life cycle;

in this embodiment, the green product case library is a database composed of a plurality of groups of green data and corresponding green policies. After the green data and the corresponding green policies are numbered and stored, in each case text, a set of green data includes encoded data for the green characteristics, green requirements, and green characteristics, and a set of green data corresponds to at least one green policy, such that the associated set of green data and green policy form a case text.

And step 3: giving the query vocabulary as carbon emission and energy consumption vocabulary, and vectorizing to obtain a word vector A of carbon emission and energy consumption₁Calculating a word vector A₁Of each word vector in the external text corpus respectivelySimilarity, wherein the similarity

s(A₁,B_i) Representing the Euclidean distance, i.e. the degree of similarity, A, of two word vectors₁Word vectors representing the vocabulary of carbon emissions, energy consumption, B_iA word vector representing any word in the external text corpus. And screening out the vocabulary corresponding to the word vector with similarity greater than the set threshold value, thereby forming a target vocabulary set b ═ b₁,b₂,...b_i...b_n}，b_iRepresenting the ith vocabulary in the target vocabulary set; n represents the total number of vocabularies in the target vocabulary set;

and 4, step 4: performing word segmentation on each sentence in D paragraphs in the case library text a, replacing pronouns after word segmentation with real words, and performing part-of-speech tagging to obtain each sentence in the D paragraphs after preprocessing;

and 5: dividing each sentence in the preprocessed D paragraphs into a plurality of simple sentences containing one object, and recording the object set in all the simple sentences in the preprocessed D-th paragraph as w_d＝{w_d1,w_d2,...,w_dj,...,w_dm}，w_djRepresenting the jth object in the jth paragraph, calculating the sum of the similarity between the object in each simple sentence of the jth paragraph and each vocabulary in the target vocabulary set, and averaging, wherein the similarity is

s(B_i,W_dj) Representing the Euclidean distance, i.e. the degree of similarity, B, of two word vectors_iWord vectors, W, representing any of the words in the target set of words_djA word vector representing any vocabulary in the set of objects; mean value of similarity

Wherein the content of the first and second substances,

the sum of the similarity of the word vector representing any vocabulary in the object set and the word vector of each vocabulary in the target vocabulary set. Screening out the simple sentences corresponding to the objects with the average values larger than the set average threshold value, thereby obtaining K screened simple sentences; d is equal to [1, D ∈]；

Step 6: for the predicate verb p in the k-th simple sentence after screening_kMarking and analyzing to obtain a subject s in the kth simple sentence_kObject o_kAnd the remaining components, thereby forming a kth predecessor-predicate triple SA_k＝{s_k,p_k,o_k}；

Let a policy knowledge be { operation, object, effect }, and use the predicate verb p_kAs an operation, object o_kAs an object, carbon emission reduction or energy saving is taken as an effect, thereby forming a kth model of operation-object-effect, K ∈ [1, K ∈]；

And 7: establishing a mapping relation f between the operation and the effect according to K models of the operation, the object and the effect_R：P₁，P₂,···,P_n→ R and as rule of "IF-THEN": IF P₁，P₂,···,P_nOf THENR, wherein P₁，P₂,···,P_nRepresents n operations, → represents causes, R represents carbon emission reduction or energy saving;

as shown in fig. 3, a piece of policy knowledge is { operation, object, effect }, and a policy knowledge unit is formed by numbering a piece of policy knowledge.

And 8: calculating any two strategies SA of the same stage of the life cycle by using the formula (1)₁And SA₂Sentence similarity S (SA)₁,SA₂)：

s(SA₁,SA₂)＝q₁×s(s₁,s₂)+q₂×s(p₁,p₂)+q₃×s(o₁,o₂) (1)

In the formula (1), q₁,q₂,q₃Represents the weight of the subject, predicate, object, respectively, and q₁+q₂+q₃＝1，s(s₁,s₂) Representation policy SA₁Subject of (1) s₁And strategy SA₂Subject of (1) s₂Similarity between them, s (p)₁,p₂) Representation policy SA₁Predicate p in (1)₁And strategy SA₂Predicate p in (1)₂Similarity between, s (o)₁,o₂) Representation policy SA₁Object of (1)₁And strategy SA₂Object of (1)₂The similarity between them; when S (SA)₁,SA₂) When the set threshold value is larger than the set threshold value, two strategies SA are used₁And SA₂Merging into a carbon reduction strategy; otherwise, two strategies SA₁And SA₂Respectively as a carbon reduction strategy;

and step 9: and finally dividing each paragraph into five stages of a life cycle, correspondingly storing all carbon reduction strategies into a material selection strategy set I, a manufacturing process strategy set II, a service performance strategy set III, a recycling strategy set IV and a logistics maintenance strategy set V, acquiring all feature vocabularies of green features based on the green features of the given five stages, and sequentially searching in each carbon reduction strategy according to each feature vocabulary, so that the carbon reduction strategies with the searched green features are labeled with corresponding green features, thereby forming the carbon reduction strategies with labels and sequentially giving index numbers to the strategy sets I-V to form a strategy knowledge base.

As shown in fig. 4, the finally formed policy knowledge base includes five policy sets i to v, each policy set includes a plurality of tags, and the policy numbers corresponding to the tags are stored together.

For the green characteristics in the step 9, the green characteristics selected by given materials comprise the strength of the materials, the cost of the materials and the toxic harmfulness of the materials, the green characteristics of the manufacturing process comprise material processing and component assembly, the green characteristics of service performance comprise structural strength and performance indexes, the green characteristics of recycling include material recycling, part recycling and recycling economy, and the green characteristics of logistics maintenance comprise product packaging and logistics transportation.

And encoding each piece of strategy knowledge according to lca-label-ops-obj-cer, wherein lca indicates the stage of five stages of the life cycle, label represents label data, ops represents behavior operation, obj represents the object of operation implementation, and cer represents reduction data of carbon emission. And collecting and storing each piece of strategy knowledge obtained by sorting to form a strategy knowledge base.

Claims (1)

1. A construction method of a carbon reduction strategy knowledge base based on a green product case base is characterized by comprising the following steps:

step 1: let the different phases of the life cycle of the green product include: the method comprises a material selection stage, a manufacturing process stage, a service performance stage, a recycling stage and a logistics maintenance stage;

step 2: separating a case text a in a green product case library according to paragraphs to obtain D paragraphs, and preliminarily dividing each paragraph into five stages of a life cycle according to a catalogue, a chapter title and a picture name table name in the case text a; finally dividing each paragraph into five stages of a life cycle according to the word frequency TF of the set feature words in the five stages respectively appearing in each paragraph; thus, the case text a is divided according to different stages of the life cycle;

and step 3: giving the query vocabulary as carbon emission and energy consumption vocabulary, and vectorizing to obtain a word vector A of carbon emission and energy consumption₁Calculating a word vector A₁Similarity with each word vector in an external text library is respectively obtained, and vocabularies corresponding to the word vectors with the similarity larger than a set threshold value are screened out, so that a target vocabulary set b ═ b is formed₁,b₂,...b_i...b_n}，b_iRepresenting the ith vocabulary in the target vocabulary set; n represents the total number of vocabularies in the target vocabulary set;

and 4, step 4: performing word segmentation on each sentence in D paragraphs in the case library text a, replacing pronouns after word segmentation with real words, and performing part-of-speech tagging to obtain each sentence in the D paragraphs after preprocessing;

and 5: dividing each sentence in the preprocessed D paragraphs into a plurality of simple sentences containing one object, and recording the object set in all the simple sentences in the preprocessed D-th paragraph as w_d＝{w_d1,w_d2,...,w_dj,...,w_dm}，w_djExpressing the jth object in the d paragraph, calculating the sum of the similarity between the object in each simple sentence of the d paragraph and each word in the target word set, then taking an average value, and screening out the simple sentences corresponding to the objects with the average values larger than a set average threshold value, thereby obtaining K screened simple sentences; d is equal to [1, D ∈]；

Step 6: for the predicate verb p in the k-th simple sentence after screening_kMarking and analyzing to obtain a subject s in the kth simple sentence_kObject o_kAnd the remaining components, thereby forming a kth predecessor-predicate triple SA_k＝{s_k,p_k,o_k}；

Let a policy knowledge be { operation, object, effect }, and use the predicate verb p_kAs an operation, object o_kAs an object, carbon emission reduction or energy saving is taken as an effect, thereby forming a kth model of operation-object-effect, K ∈ [1, K ∈]；

And 7: establishing a mapping relation f between the operation and the effect according to K models of the operation, the object and the effect_R：P₁，P₂,···,P_n→ R and as rule of "IF-THEN": IF P₁，P₂,···,P_nTHEN R, wherein P₁，P₂,···,P_nRepresents n operations, → represents causes, R represents carbon emission reduction or energy saving;

and 8: calculating any two strategies SA of the same stage of the life cycle by using the formula (1)₁And SA₂Sentence similarity S (SA)₁,SA₂)：

s(SA₁,SA₂)＝q₁×s(s₁,s₂)+q₂×s(p₁,p₂)+q₃×s(o₁,o₂) (1)

In the formula (1), q₁,q₂,q₃Represents the weight of the subject, predicate, object, respectively, and q₁+q₂+q₃＝1，s(s₁,s₂) Representation policy SA₁Subject of (1) s₁And strategy SA₂Subject of (1) s₂Similarity between them, s (p)₁,p₂) Representation policy SA₁Predicate p in (1)₁And strategy SA₂Predicate p in (1)₂Similarity between, s (o)₁,o₂) Representation policy SA₁Object of (1)₁And strategy SA₂Object of (1)₂The similarity between them; when S (SA)₁,SA₂) When the set threshold value is larger than the set threshold value, two strategies SA are used₁And SA₂Merging into a carbon reduction strategy; otherwise, two strategies SA₁And SA₂Respectively as a carbon reduction strategy;

and step 9: and finally dividing each paragraph into five stages of a life cycle, correspondingly storing all carbon reduction strategies into a material selection strategy set I, a manufacturing process strategy set II, a service performance strategy set III, a recycling strategy set IV and a logistics maintenance strategy set V, acquiring all feature vocabularies of green features based on the green features of the given five stages, and sequentially searching in each carbon reduction strategy according to each feature vocabulary, so that the carbon reduction strategies with the searched green features are labeled with corresponding green features, thereby forming the carbon reduction strategies with labels and sequentially giving index numbers to the strategy sets I-V to form a strategy knowledge base.

Images