Product production quality abnormity recommendation method and device, equipment, medium and product

Abstract

The invention discloses a product production quality abnormity recommendation method and device, equipment, a medium and a product. The method comprises the steps of determining a topic distribution probability of each abnormal topic in a to-be-processed production abnormal text; and determining a target recommended production anomaly based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the to-be-processed production anomaly text and the topic distribution probability of each abnormal topic in the to-be-processed production anomaly text. According to the method, word ambiguity in the prior art and semantic problems caused by multi-word phrases and complex syntactic structures are solved, effective recommendation of target recommendation production exceptions is effectively ensured, and the exception handling efficiency is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of quality management, and in particular to a method, device, equipment, medium and product for recommending abnormal product production quality. Background Art

[0002] A Quality Management System (QMS) is used to manage and control quality-related activities to ensure that products or services meet quality standards and customer requirements.

[0003] Quality exception management is the core process function of QMS, which mainly includes the following processes: initiation of exception tickets, problem analysis, temporary processing solutions, cause analysis, improvement measures, prevention strategies and problem library archiving.

[0004] QMS can be used for the management of different categories. During the management process, there are phenomena such as different meanings of the same words in different categories and different expressions by different people, which lead to problems such as word ambiguity, multi-word phrases and complex syntactic structures. Summary of the invention

[0005] The present invention provides a method, device, equipment, medium and product for recommending abnormal product production quality, so as to solve the word ambiguity and semantic problems caused by multi-word phrases and complex syntactic structures in the prior art.

[0006] According to one aspect of the present invention, there is provided a method for recommending abnormal product production quality, comprising:

[0007] Determine the topic distribution probability of each abnormal topic in the production abnormality text to be processed; wherein the abnormal topic is used to characterize the abnormality type to which the production abnormality text to be processed belongs; the abnormal topic includes but is not limited to one of the following: product quality abnormality, operation process abnormality, production tool abnormality;

[0008] The target recommended production anomaly is determined based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the production anomaly text to be processed and the topic distribution probability of each abnormal topic in the production anomaly text to be processed.

[0009] According to another aspect of the present invention, there is provided a device for recommending abnormal product production quality, comprising:

[0010] A distribution probability determination module is used to determine the topic distribution probability of each abnormal topic in the production abnormality text to be processed; wherein the abnormal topic is used to characterize the abnormality type to which the production abnormality text to be processed belongs; the abnormal topic includes but is not limited to one of the following: product quality abnormality, operation process abnormality, production tool abnormality;

[0011] An anomaly determination module is used to determine a target recommended production anomaly based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the production anomaly text to be processed and the topic distribution probability of each abnormal topic in the production anomaly text to be processed.

[0012] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0013] at least one processor; and

[0014] a memory communicatively connected to the at least one processor; wherein,

[0015] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for recommending abnormal product production quality described in any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the method for recommending abnormal product production quality described in any embodiment of the present invention when executed.

[0017] According to another aspect of the present invention, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, the method for recommending abnormal product production quality according to any embodiment of the present invention is implemented.

[0018] The technical solution of the embodiment of the present invention determines the topic distribution probability of each abnormal topic in the production abnormality text to be processed; determines the target and recommends the actual production abnormalities in the production process to the user based on the topic distribution probability of each abnormal topic in each trained production abnormality associated with the production abnormality text to be processed and the topic distribution probability of each abnormal topic in the production abnormality text to be processed, thereby solving the word ambiguity that occurs in the prior art and the semantic problems caused by multi-word phrases and complex syntactic structures, effectively ensuring the effective recommendation of the target recommended production abnormalities, and thus improving the accuracy of exception processing.

[0019] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 creative work.

[0021] Figure 1 It is a flow chart of a recommended method for abnormal product production quality provided by an embodiment of the present invention;

[0022] Figure 2 is a flow chart of another recommended method for abnormal product production quality provided by an embodiment of the present invention;

[0023] Figure 3 It is a structural schematic diagram of a recommended device for abnormal product production quality provided by an embodiment of the present invention;

[0024] Figure 4 It is a structural block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] In the actual management process, word segmentation search algorithms, vector retrieval technology, and bag-of-words model technology can be used to recommend product quality anomalies:

[0028] Among them, the implementation principles of the word segmentation search algorithm include: it can perform matching searches based on keywords, but it cannot solve the impact of word ambiguity and unregistered words, and the text relevance is weak; the existing word segmentation also lacks evolutionary algorithms, and cannot perform semantic self-iteration based on the actual usage of end users; it cannot provide a text association recommendation algorithm that is sustainable and self-optimized based on the actual data of the enterprise and without human intervention.

[0029] Vector retrieval technology is a technology that realizes data retrieval and matching by calculating the similarity between vectors. Vector retrieval can effectively process natural language text without too much structural processing. By calculating the similarity between vectors, the similarity and relevance between documents can be evaluated, thereby obtaining more accurate results. However, the accuracy and scalability of vector retrieval technology largely depends on feature selection and parameter adjustment. This requires people with in-depth understanding and experience to judge and maintain, and it cannot be automated and use data feedback from ordinary users.

[0030] The bag-of-words model is a concise and easy-to-calculate text representation method based on the frequency analysis principle in statistics. This method ignores the order and grammatical relationship between words and can build a model without annotated data. It is suitable for preliminary analysis of unannotated text data. However, the bag-of-words model ignores the order relationship between words and may not be able to capture the semantics that depend on the order, understand the semantic association between words, or effectively process the semantic information brought by multi-word phrases and complex syntactic structures.

[0031] In view of this, the present invention proposes a recommendation method for abnormal product production quality, which can adopt the abnormal topic distribution model of fitting reinforcement learning as a solution to the above problems. Topic modeling is a technology used to extract topics or topics from text data, and is used to mine potential topics and key information from text data. Topics can be regarded as general descriptions of text data, which cover the key abstractions about the business in the text, and is a probabilistic graphical model for topic modeling. Its basic idea is that each document is a mixture of a set of topics, and each topic is composed of a set of words. The LDA model attempts to find the best combination of topics and words to explain the given text data.

[0032] In one embodiment, Figure 1 This is a flowchart of a method for recommending abnormal product production quality provided by an embodiment of the present invention. This embodiment is applicable to the case of recommending abnormal production in an industrial production process. The method can be executed by a device for recommending abnormal product production quality. The device can be implemented in the form of hardware and / or software. The device can be configured in an electronic device. Figure 1 As shown, the method includes:

[0033] S110, determining the topic distribution probability of each abnormal topic in the production abnormality text to be processed.

[0034] Among them, the abnormal topic is used to characterize the abnormal type to which the production abnormal text to be processed belongs; the abnormal topic includes but is not limited to one of the following: product quality abnormality, operation process abnormality and production tool abnormality. Among them, the production abnormal text to be processed refers to the text filled in the production process that causes the abnormality of product quality, production efficiency and / or production process failure. It is a production abnormal text that needs to be processed for quality management / production efficiency management. Generally speaking, the production abnormal text to be processed may include a phrase, a complete paragraph or a sentence, and the phrase, paragraph or sentence may include multiple industrial production words, that is, a text to be processed consisting of multiple industrial production words. The abnormal topic refers to the abnormal type to which the production abnormal text to be processed belongs. The abnormal topic contained in each production abnormal text to be processed is related to the field, operation process, etc. of the production abnormal text to be processed. Exemplarily, the abnormal topic may include but is not limited to one of the following: product quality abnormality, operation process abnormality and production tool abnormality; for example, product quality abnormality may include: hole bad spots and product damage; operation process abnormality may include but is not limited to one of the following: vulcanization abnormality, injection molding abnormality and welding temperature abnormality; production tool abnormality may include but is not limited to one of the following: mold wear and abrasive debris. In actual operation, each production abnormality text to be processed may belong to one abnormal topic or multiple abnormal topics. When a production abnormality text to be processed belongs to one abnormal topic, the topic distribution probability of the abnormal topic is directly determined to be 1; when a text to be processed belongs to multiple abnormal topics, the topic distribution probability of each abnormal topic needs to be determined, and the sum of the topic distribution probabilities of each abnormal topic is 1. For example, a production abnormality text to be processed belongs to three abnormal topics, namely abnormal topic 1, abnormal topic 2 and abnormal topic 3, then the topic distribution probabilities of abnormal topic 1, abnormal topic 2 and abnormal topic 3 are a, b and c respectively, then the sum of a, b and c is 1. Generally speaking, a production abnormality text to be processed can be composed of multiple abnormal topics, and each abnormal topic can be composed of multiple industrial production words according to a certain probability distribution.

[0035] In an embodiment, the topic distribution probability of each abnormal topic in the to-be-processed production abnormal text can be statistically calculated by a pre-built target abnormal topic distribution model. Exemplarily, the target abnormal topic distribution model can include but is not limited to one of the following: Latent Dirichlet Allocation (LDA) model; Probabilistic Latent Semantic Analysis (PLSA).

[0036] S120 , determining a target recommended production anomaly based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the production anomaly text to be processed and the topic distribution probability of each abnormal topic in the production anomaly text to be processed.

[0037] Among them, the trained production anomaly refers to a production anomaly that has the same abnormal topic as the production anomaly text to be processed, and the topic distribution probability of the abnormal topic contained has been determined. In the actual operation process, the number of trained production anomalies associated with each production anomaly text to be processed can be one or more. Of course, when the number of trained production anomalies associated with the production anomaly text to be processed is one, the target recommended production anomaly is directly determined to be the trained production anomaly; when the number of trained production anomalies associated with the production anomaly text to be processed is at least two, the target recommended production anomaly can be determined based on the topic distribution probability of each abnormal topic in each trained production anomaly and the topic distribution probability of each abnormal topic in the production anomaly text to be processed. Specifically, the difference in topic distribution probability between the production anomaly text to be processed and the same abnormal topic in each trained production anomaly can be statistically analyzed to obtain the corresponding matching probability difference. For example, the production anomaly text to be processed and the three trained production anomalies all contain three abnormal topics, then the matching probability difference of the three abnormal topics can be obtained, and then the matching probability differences of the three abnormal topics are added together to obtain the sum of the matching probability differences; then the trained production anomaly with the smallest difference between 1 and the sum of the matching probability differences is selected as the target recommended production anomaly.

[0038] The technical solution of this embodiment determines the topic distribution probability of each abnormal topic in the production abnormality text to be processed; determines the target recommended production abnormality based on the topic distribution probability of each abnormal topic in each trained production abnormality associated with the production abnormality text to be processed and the topic distribution probability of each abnormal topic in the production abnormality text to be processed, and compares the distribution probability of the abnormal topic in the text with the distribution probability of the abnormal training, so that the recommended abnormality is more in line with the actual production abnormality, and can avoid the problem that the judgment result is inconsistent with the actual production abnormality due to the expression habits of the filler in the production abnormality text due to the high probability of certain abnormal words. At the same time, it solves the word ambiguity in the prior art and the semantic problems caused by multi-word phrases and complex syntactic structures, effectively ensuring the effective recommendation of the target recommended abnormality, thereby improving the accuracy of abnormality processing.

[0039] In one embodiment, Figure 2 1 is a flowchart of another method for recommending product production quality anomalies provided by an embodiment of the present invention. This embodiment further illustrates the process of determining the topic distribution probability and the process of determining the target recommended production anomalies based on the above embodiment. Figure 2 As shown, the method includes:

[0040] S210, segmenting the production abnormality text to be processed to obtain a corresponding actual initial industrial production word set.

[0041] Among them, the actual initial industrial production word set refers to the set of all industrial production words contained in the production exception text to be processed. In an embodiment, the number of industrial production words contained in the actual initial industrial production word set is the same as the number of industrial production words contained in the production exception text to be processed. In an embodiment, a special word segmentation tool can be used to segment the production exception text to be processed, obtain all the industrial production words contained therein, and form the corresponding actual initial industrial production word set.

[0042] S220: Filter out invalid words in the actual initial industrial production word set to obtain a corresponding actual target industrial production word set.

[0043] Among them, invalid words refer to stop words, punctuation and other irrelevant related words contained in the actual initial industrial production word set; the actual target industrial production word set refers to the set of other industrial production words except invalid words contained in the production abnormality text to be processed. Among them, the actual target industrial production word set includes but is not limited to words in at least one of the following aspects: production tools, product structure, product quality and process. For example, the words in production tools may include but are not limited to one of the following: molds and molds, etc.; the words in product structure may include but are not limited to one of the following: fixed points and holes, etc.; the words in product quality may include but are not limited to one of the following: bad points, bad, damage and scratches, etc.; the words in process may include but are not limited to one of the following: vulcanization, welding, injection molding, heat treatment, etc. In general, the actual target industrial production word set can be a subset of the actual initial industrial production word set, that is, the number of industrial production words contained in the actual target industrial production word set is less than the number of industrial production words contained in the actual initial industrial production word set. In an embodiment, irrelevant related words such as stop words and punctuation marks in the actual initial industrial production word set are identified and extracted, and these irrelevant related words are removed from the actual initial industrial production word set, and the remaining related industrial production words constitute the corresponding actual target industrial production word set.

[0044] S230, de-duplicating and vectorizing the industrial production words in the actual target industrial production word set to obtain a corresponding actual industrial production word matrix.

[0045] The actual industrial production word matrix refers to a matrix composed of all industrial production words contained in the actual target industrial production word set and the corresponding number of each industrial production word. In an embodiment, the process of deduplicating industrial production words in the actual target industrial production word set refers to the process of filtering out repeated industrial production words contained in the actual target industrial production word set; the process of vectorizing industrial production words in the actual target industrial production word set refers to the process of numbering each industrial production word contained in the actual target industrial production word set.

[0046] Exemplarily, assuming that the actual target industrial production word set includes: word 1, word 2, word 3, word 4, word 1, word 3, word 5, word 6, word 7 and word 8, and the word 1 and word 3 contained in the actual target industrial production word set are repeated, both of which are two, then the process of deduplicating the words in the actual target industrial production word set is to obtain word 1, word 2, word 3, word 4, word 5, word 6, word 7 and word 8 from a word 1 and a word 3 contained in the actual target industrial production word set, and then vectorize word 1, word 2, word 3, word 4, word 5, word 6, word 7 and word 8, that is, number word 1, word 2, word 3, word 4, word 5, word 6, word 7 and word 8 as 0, 1, 2, 3, 4, 5, 6 and 7 respectively, and the actual word matrix obtained is [0: word 1, 1: word 2, 2: word 3, 3: word 4, 4: word 5, 5: word 6, 6: word 7, 7: word 8].

[0047] S240. Input the actual industrial production word matrix, the actual abnormal topic set, the number of industrial production words contained in the actual industrial production word matrix, and the number of abnormal topics contained in the actual abnormal topic set associated with the production abnormal text to be processed into a pre-created target abnormal topic distribution model to obtain the topic distribution probability of each abnormal topic in the production abnormal text to be processed.

[0048] The actual abnormal topic set refers to the set of all abnormal topics to which the production abnormal text to be processed belongs, that is, the number of abnormal topics contained in the actual abnormal topic set is equal to the number of abnormal topics to which the production abnormal text to be processed belongs. The target abnormal topic distribution model refers to a statistical model created in advance for analyzing text data, which is used to characterize the potential abnormal topics in the production abnormal text to be processed and the degree of association with each abnormal topic.

[0049] In an embodiment, the actual industrial production word matrix, the actual abnormal topic set, the number of industrial production words contained in the actual industrial production word matrix, and the number of abnormal topics contained in the actual abnormal topic set associated with the production abnormal text to be processed can be input into a pre-created target abnormal topic distribution model to obtain the topic distribution probability of each abnormal topic in the production abnormal text to be processed.

[0050] In one embodiment, the process of constructing a target abnormal topic distribution model includes: determining a pre-acquired training industrial production word matrix for each production anomaly to be trained; randomly initializing the abnormal topic to which each industrial production word in the training industrial production word matrix associated with each production anomaly to be trained to obtain an abnormal topic matrix corresponding to each production anomaly to be trained; determining the number of times each industrial production word in the training industrial production word matrix corresponds to each abnormal topic, and the number of times each production anomaly to be trained corresponds to each abnormal topic based on the abnormal topic matrix of each production anomaly to be trained; determining the total number of corresponding abnormal topics based on the number of times all production anomalies to be trained correspond to each abnormal topic; inputting the total number of each abnormal topic, the number of times each production anomaly to be trained corresponds to each abnormal topic, the number of times each industrial production word corresponds to each abnormal topic, and the number of abnormal topics into a pre-created initial abnormal topic distribution model for iterative training until the output topic distribution probability of each industrial production word tends to be stable, thereby obtaining the corresponding target abnormal topic distribution model. In one embodiment, a training industrial production word matrix for each production anomaly to be trained that is obtained in advance is determined, including: performing word segmentation on each production anomaly to be trained that is obtained in advance to obtain a corresponding training initial industrial production word set; filtering out invalid words in the training initial industrial production word set to obtain a corresponding training target industrial production word set; deduplicating and vectorizing the industrial production words in the training target industrial production word set to obtain a training industrial production word matrix corresponding to each production anomaly to be trained.

[0051] Among them, the production anomaly to be trained refers to the sample used for training the initial abnormal topic distribution model; the training initial industrial production word set refers to the set of all industrial production words contained in all production anomalies to be trained; the training target industrial production word set refers to the set of other industrial production words except invalid words contained in all production anomalies to be trained; the training industrial production word matrix refers to the matrix composed of all industrial production words except invalid words contained in each production anomaly to be trained and the corresponding number of each industrial production word. In the actual training process, in order to ensure that the determination accuracy of the topic distribution probability of the constructed target abnormal topic distribution model is higher, the number to be trained can be increased as much as possible, that is, more production anomalies to be trained are used to iteratively train the initial abnormal topic distribution model. In an example, the industrial production words in the training target industrial production word set can be deduplicated and vectorized in turn to obtain the corresponding training industrial production word list, and the overlapping words between all industrial production words except invalid words contained in a production anomaly to be trained and the training industrial production word list, and the corresponding numbers are used as the training industrial production word matrix corresponding to the production anomaly to be trained.

[0052] Among them, the abnormal topic matrix refers to the matrix composed of abnormal topics associated with each industrial production word in the training industrial production word matrix associated with each production abnormality to be trained. In the embodiment, each production abnormality to be trained is looped, and the abnormal topic to which each industrial production word in the training industrial production word matrix associated with the production abnormality to be trained is randomly initialized, that is, the abnormal topic matrix corresponding to the production abnormality to be trained can be obtained, and the number of times each industrial production word in each training industrial production word matrix corresponds to the abnormal topic, and the number of times each production abnormality to be trained corresponds to each abnormal topic is counted; then the number of times all production abnormalities to be trained correspond to each abnormal topic is added to obtain the total number of times for each abnormal topic.For example, the production anomaly to be trained is represented in the form of a document, and there are 6 documents (d0, d1, d2, d3, d4 and d5), and there are 3 abnormal topics, namely topic 1, topic 2 and topic 3; the training industrial production word matrix contains 10 words, namely word 1, word 2, word 3, word 4, word 5, word 6, word 7, word 8, word 9 and word 10, and the corresponding training industrial production word matrix is ​​[0: word 1, 1: word 2, 2: word 3, 3: word 4, 4: word 5, 5: word 6, 6: word 10] 7, 7: word 8, 8: word 9, 9: word 10]; then the number of times each industrial production word corresponds to each abnormal topic is counted: word 1: [topic 1: a1; topic 2: b1; topic 3: c1], word 2: [topic 1: a2; topic 2: b2; topic 3: c2], word 3: [topic 1: a3; topic 2: b3; topic 3: c3], word 4: [topic 1: a4; topic 2: b4; topic 3: c4], word 5: [topic 1: a5; topic 2: b5; topic 3: c5], word 6: [topic 1: a6; topic 2: b6; topic 3: c6] Topic 2: b6; Topic 3: c6], word 7: [topic 1: a7; topic 2: b7; topic 3: c7], word 8: [topic 1: a8; topic 2: b8; topic 3: c8], word 9: [topic 1: a9; topic 2: b9; topic 3: c9] word 10: [topic 1: a10; topic 2: b10; topic 3: c10]; and count the number of times each production anomaly to be trained corresponds to each abnormal topic: d0: [topic 1: A1; topic 2: B1; topic 3: C1], d1: [topic 1: A2; topic 2: B2; topic 3: C1] 3: C2], d2: [Topic 1: A3; Topic 2: B3; Topic 3: C3], d3: [Topic 1: A4; Topic 2: B4; Topic 3: C4], d4: [Topic 1: A5; Topic 2: B5; Topic 3: C5], d5: [Topic 1: A6; Topic 2: B6; Topic 3: C6], and then determine the total number of corresponding abnormal topics based on the number of times all production anomalies to be trained correspond to each abnormal topic: [Topic 1: A1+A2+A3+A4+A5+A6; Topic 2: B1+B2+B3+B4+B5+B6; Topic 3:.

[0053] C1+C2+C3+C4+C5+C6].

[0054] S250, determining a matching probability difference between the topic distribution probability of each abnormal topic in the to-be-processed production abnormality text and the topic distribution probability of the corresponding abnormal topic in each trained production abnormality.

[0055] Among them, the matching probability difference refers to the difference between the topic distribution probability of each abnormal topic in the production abnormality text to be processed and the topic distribution probability of the corresponding abnormal topic in the trained production abnormality; the trained production abnormality refers to the topic distribution probability of the abnormal topic that has been determined in the material library. Exemplarily, it is assumed that the production abnormality text to be processed and the three trained production abnormality documents (document a, document b and document c) all contain three abnormal topics, namely topic 1, topic 2 and topic 3; and the topic distribution probabilities corresponding to topic 1, topic 2 and topic 3 in the production abnormality text to be processed are D1, probability D2 and probability D3 respectively, the topic distribution probabilities corresponding to topic 1, topic 2 and topic 3 in document a are E1, probability E2 and probability E3 respectively, the topic distribution probabilities corresponding to topic 1, topic 2 and topic 3 in document b are F1, probability F2 and probability F3 respectively, and the topic distribution probabilities corresponding to topic 1, topic 2 and topic 3 in document c are The rate is G1, the probability is G2, and the probability is G3, then the matching probability difference of topic 1 between the production abnormal text to be processed and document a is D1-E1, the matching probability difference of topic 2 is D2-E2, and the matching probability difference of topic 3 is D3-E3; the matching probability difference of topic 1 between the production abnormal text to be processed and document b is D1-F1, the matching probability difference of topic 2 is D2-F2, and the matching probability difference of topic 3 is D3-F3; the matching probability difference of topic 1 between the production abnormal text to be processed and document c is D1-G1, the matching probability difference of topic 2 is D2-G2, and the matching probability difference of topic 3 is D3-G3.

[0056] S260: Determine an actual matching probability value between the production exception text to be processed and the corresponding trained production exception based on the matching probability difference value associated with each trained production exception.

[0057] Among them, the actual matching probability value is used to characterize the degree of matching of the subject between the production abnormality text to be processed and each trained production abnormality; the higher the actual matching probability value, the higher the degree of matching of the subject between the production abnormality text to be processed and the trained production abnormality. In an embodiment, the actual matching probability value can add the matching probability difference of each abnormal subject to obtain the sum of the matching probability difference; then the difference between 1 and the sum of the matching probability difference is used as the actual matching probability value. Exemplarily, the actual matching probability value between the production abnormality text to be processed and document a can be 1-(D1-E1)-(D2-E2)-(D3-E3); the actual matching probability value between the production abnormality text to be processed and document b can be 1-(D1-F1)-(D2-F2)-(D3-F3); the actual matching probability value between the production abnormality text to be processed and document c can be 1-(D1-G1)-(D2-G2)-(D3-G3).

[0058] S270. Taking the trained production anomaly with the largest actual matching probability value as the target recommended production anomaly.

[0059] In an embodiment, a trained production anomaly with the largest actual matching probability value among multiple trained production anomalies is used as a target recommended production anomaly for the production anomaly text to be processed, so that the relevant fields in the target recommended production anomaly can be directly used to fill in the fields of the form associated with the production anomaly text to be processed, thereby avoiding the tedious process of manually filling in all fields in the production anomaly text to be processed, thereby improving the user form filling efficiency.

[0060] In one embodiment, the method for recommending product production quality abnormalities further includes: filling in a form with the production abnormality text to be processed based on the target recommended production abnormality to obtain a target production abnormality form corresponding to the production abnormality text to be processed. The target production abnormality form refers to a list in which the relevant industrial production words in the production abnormality text to be processed are represented in a table. In an embodiment, the industrial production words in the production abnormality text to be processed fill in the relevant fields in the associated production abnormality form, and some unfilled fields can be directly filled with the fields in the target recommended production abnormality to generate the corresponding target production abnormality form.

[0061] In one embodiment, a form is filled for the production exception text to be processed based on the target recommended production exception to obtain the target production exception form corresponding to the production exception text to be processed, including: identifying and extracting each unfilled field in the production exception text to be processed, and the actual value of the unfilled field in the target recommended production exception; filling the actual value of each unfilled field into the corresponding field of the production exception form associated with the production exception text to be processed, and obtaining the target production exception form corresponding to the production exception text to be processed. Among them, the unfilled field refers to a field that cannot be valued based on the industrial production words contained in the production exception text to be processed; generally speaking, the unfilled field can be a field that can be shared by the production exception text to be processed and the trained production exception. In an embodiment, the unfilled fields in the production exception text to be processed are identified and extracted, and the actual value of each unfilled field in the production exception form associated with the target recommended production exception is obtained, and these actual values ​​are filled into the corresponding unfilled fields in the production exception text to be processed, so that the unfilled fields in the target production exception form corresponding to the production exception text to be processed can be simply and quickly completed, thereby greatly reducing the job requirements and professional capabilities of relevant personnel, and greatly improving the processing efficiency of the production exception form.

[0062] In one embodiment, as a preferred embodiment, the process of modeling and recommendation is described. The specific process of modeling and recommendation is as follows:

[0063] Step 1: Preprocessing of production anomalies to be trained.

[0064] Wherein, step 1 includes step 11, step 12 and step 13.

[0065] Step 11: Prepare the content for training production exceptions

[0066] Step 111, extract the relevant content of the production exception order, including the subject description of the exception order, detailed description of the problem, related document type, temporary processing measures, root cause classification, root cause analysis, permanent improvement measures, standardized measures to prevent recurrence, and process attachment related data, and remove styles and irrelevant characters, merge into documents as a modeling corpus.

[0067] Step 112: remove non-text tags from the document in step 111.

[0068] For example, the training production exception text is: Due to the bad point of the mold hole, the mold fixed point is damaged, resulting in poor pressure damage at the fixed position after vulcanization. The mold damage of the mold hole is confirmed to be unrepairable, so the mold hole is knocked as a dead point. The subsequent production of the mold hole products will be treated as defective products, and the next step of edge tearing will not be carried out.

[0069] Step 12: Perform word segmentation on the production anomalies to be trained to obtain the training industrial production word matrix.

[0070] Before modeling, it is necessary to preprocess the training production abnormal text. The specific steps are as follows:

[0071] Step 121, segment the training production abnormal text in step 11 to obtain a training initial industrial production word set, for example, the training initial industrial production word set is: ["due to", "mold", "on", "hole", "bad point", "caused by", "mold", "fixed point", "damage", "," "lead to", "vulcanization", "after", "fixed", "position", "pressed", "bad"...];

[0072] Step 122, remove stop words and punctuation marks from the word segmentation results to obtain a training target industrial production word set, for example, the training target industrial production word set is: ["mold", "hole", "bad point", "mold", "fixed point", "damage", "vulcanization", "fixed", "position", "pressed", "bad", ...];

[0073] Step 123: De-duplicate and vectorize the industrial production words in the training target industrial production word set, and create a training industrial production word list and a training industrial production word matrix;

[0074] According to the processing result of step 12, a training industrial production vocabulary is established, and then the industrial production words in the training industrial production vocabulary are vectorized to obtain a training industrial production word matrix.

[0075] Among them, the training industrial production vocabulary is the words after all industrial production words in all production anomalies to be trained are deduplicated; then the training industrial production vocabulary is vectorized to obtain a training industrial production word matrix. For example, the training industrial production word matrix is: [0: "mold", 1: "hole", 2: "bad point", 3: "fixed point", 4: "damage", 5: "vulcanization", 6: "fixed", 7: "position", 8: "pressing", 9: "bad"].

[0076] Step 2: Modeling based on the training industrial production word matrix.

[0077] Wherein, step 2 includes step 21, step 22 and step 23.

[0078] Step 21, initializing modeling parameters;

[0079] Set the number of topics = abnormal problem (completed) category * category * problem handling method;

[0080] Set the number of production anomalies to the number of production anomalies to be trained, and count the number of words contained in each production anomaly to be trained as the length of each word;

[0081] Set the vocabulary to the vocabulary of production anomalies to be trained and count the number of vocabulary;

[0082] For example, the production anomaly to be trained is represented in the form of documents, and the documents are 6, specifically [d0, d1, d2, d3, d4, d5];

[0083] For example, the preprocessed industrial production word matrices to be trained for production anomalies are [

[0084] d0:[1:"hole",4:"damage",7:"position",2:"bad pixel",3:"fixed point",1:"hole",4:"damage",3:"fixed point",2:"bad pixel",3:"fixed point",1:"hole",4:"damage",3:"fixed point",2:"bad pixel",3:"fixed point",6:"fixed"],

[0085] d1:[2:"bad pixel",2:"bad pixel",4:"damage",2:"bad pixel",4:"damage",2:"bad pixel",2:"bad pixel",2:"bad pixel",2:"bad pixel",4:"damage",2:"bad pixel",2:"bad pixel"],

[0086] d2:[1:"hole",6:"fixed",5:"vulcanization",6:"fixed",0:"mold",1:"hole",6:"fixed",5:"vulcanization",6:"fixed",0:"mold",1:"hole",6:"fixed",5:"vulcanization",6:"fixed",9:"bad"0:"mold"],

[0087] d3:[5:"vulcanization",6:"fixed",6:"fixed",2:"bad point",8:"pressed",3:"fixed point",6:"fixed",5:"vulcanization",6:"fixed",2:"bad point",2:"bad point",6:"fixed",5:"vulcanization",6:"fixed",6:"fixed",6:"fixed",0:"mold"],

[0088] d4:[2:"bad pixel",2:"bad pixel",4:"damage",4:"damage",4:"damage",4:"damage",1:"hole",5:"vulcanization",7:"position",5:"vulcanization",9:"bad",5:"vulcanization",5:"vulcanization",1:"hole",1:"hole",1:"hole",1:"hole",0:"mold"],

[0089] d5:[5:"vulcanization",4:"damage",2:"bad point",3:"fixed point",4:"damage",5:"vulcanization",8:"compression",6:"fixed",5:"vulcanization",4:"damage",3:"fixed point",2:"bad point"] ]

[0091] There are 6 production anomaly documents to be trained. After merging the industrial production words of the 6 production anomaly documents to be trained and removing the duplicates, the training industrial production vocabulary is [0: "mold", 1: "hole", 2: "bad point", 3: "fixed point", 4: "damage", 5: "vulcanization", 6: "fixed", 7: "position", 8: "pressing", 9: "bad"], a total of 10 industrial production words;

[0092] The above 6 production anomalies to be trained are a category in a classification, and there are 3 problem-solving methods. Therefore, the number of initialized anomaly topics is 3, specifically [t0: "hole bad point", t1: "mold wear", t2: "abrasive chip"].

[0093] Step 22: Initialize abnormal topic data according to the production abnormality to be trained;

[0094] Loop through each production anomaly to be trained, and randomly initialize the abnormal topic of each industrial production word in the training industrial production word matrix of the production anomaly to be trained. The results are as follows:

[0095] d0:[t0:"hole bad point",t2:"mold debris",t2:"mold debris",t1:"mold wear",t0:"hole bad point",t0:"hole bad point",t1:"mold wear",t0:"hole bad point",t1:"mold wear",t0:"hole bad point",t1:"mold wear",t0:"hole bad point",t0:"hole bad point",t0:"hole bad point",t0:"hole bad point",t1:"mold wear",t0:"hole bad point",t0:"hole bad point"],

[0096] d1:[t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear",t1:"mold wear"],

[0097] d2:[t0:"hole bad point",t2:"abrasive tool debris",t2:"abrasive tool debris",t2:"abrasive tool debris",t2:"abrasive tool debris",t0:"hole bad point",t2:"abrasive tool debris",t2:"abrasive tool debris",t2:"abrasive tool debris",t2:"abrasive tool debris",t2:"abrasive tool debris",t2:"abrasive tool debris",t0:"hole bad point",t2:"abrasive tool debris",t2:"abrasive tool debris",t2:"abrasive tool debris"],

[0098] d3:[t2:"Mold chip",t2:"Mold chip",t2:"Mold chip",t1:"Mold wear",t2:"Mold chip",t0:"Hole damage",t2:"Mold chip",t2:"Mold chip",t2:"Mold chip",t2:"Mold chip",t1:"Mold wear",t1:"Mold wear",t2:"Mold chip",t2:"Mold chip",t2:"Mold chip",t2:"Mold chip",t2:"Mold chip",t2:"Mold chip",t2:"Mold chip"],

[0099] d4:[t1:"Mold wear",t1:"Mold wear",t1:"Mold wear",t1:"Mold wear",t1:"Mold wear",t1:"Mold wear",t0:"Bad hole point",t2:"Mold debris",t0:"Bad hole point",t0:"Bad hole point",t1:"Mold wear",t2:"Mold debris",t0:"Bad hole point",t0:"Bad hole point",t0:"Bad hole point",t0:"Bad hole point",t1:"Mold wear",t0:"Bad hole point"],

[0100] d5:[t0:"hole bad point",t1:"mold wear",t1:"mold wear",t0:"hole bad point",t1:"mold wear",t0:"hole bad point",t2:"mold debris",t2:"mold debris",t0:"hole bad point",t1:"mold wear",t0:"hole bad point",t1:"mold wear"];

[0101] Then, count the number of times each industrial production word is classified into the abnormal topic:

[0102] 0:"Mold":[t0:"Broken hole":1,t1:"Mold wear":0,t2:"Mold debris":4],

[0103] 1: "hole": [t0: "hole bad point": 9, t1: "mold wear": 1, t2: "mold debris": 1],

[0104] 2:"bad spots":[t0:"hole bad spots":0,t1:"mold wear":19,t2:"mold debris":0],

[0105] 3: "Fixed point": [t0: "hole bad point": 8, t1: "mold wear": 0, t2: "mold debris": 0],

[0106] 4:"Damage":[t0:"Hole damage":1,t1:"Mold wear":11,t2:"Mold debris":1],

[0107] 5: "Vulcanization": [t0: "hole damage": 6, t1: "mold wear": 0, t2: "mold debris": 7],

[0108] 6: "Fixed": [t0: "Broken hole": 1, t1: "Mold wear": 0, t2: "Mold debris": 15],

[0109] 7: "Position": [t0: "Broken hole": 1, t1: "Mold wear": 0, t2: "Mold debris": 1],

[0110] 8: "Crushing": [t0: "Damage point": 0, t1: "Mold wear": 0, t2: "Mold debris": 2],

[0111] 9: "bad": [t0: "hole bad point": 0, t1: "mold wear": 1, t2: "mold debris": 1];

[0112] Then, count the number of times each production anomaly to be trained corresponds to each anomaly topic:

[0113] d0:"[t0:"hole damage":10,t1,:"mold wear":4,t2:"mold debris":2],

[0114] d1:"[t0:"hole damage":0,t1,:"mold wear":12,t2:"mold debris":0],

[0115] d2:"[t0:"hole damage":3,t1,:"mold wear":0,t2:"mold debris":13],

[0116] d3:"[t0:"hole damage":1,t1,:"mold wear":3,t2:"mold debris":13],

[0117] d4:"[t0:"hole damage":8,t1,:"mold wear":8,t2:"mold debris":2],

[0118] d5:"[t0:"hole damage":5,t1,:"mold wear":5,t2:"mold debris":2];

[0119] Then count the total number of each abnormal topic nwsum = [t0: "hole bad point": 27, t1,: "mold wear": 32, t2: "mold debris": 32];

[0120] Step 23: Modeling Calculation

[0121] The iteration calculation starts from 1, and stops until the number of abnormal topics calculated for multiple times and the topic distribution probability of each industrial production word in the abnormal topic tend to be stable (unchanged for two consecutive times).

[0122] Loop every production anomaly, every industrial production word

[0123] {Calculate the most likely abnormal topic for each industrial production word, as follows:

[0124] {First remove the industrial production term from the technical device,

[0125] {Loop through each exception topic,

[0126] Calculate the distribution probability (multinomial distribution) of industrial production words in each abnormal topic. The specific formula is (the number of words in this topic + document topic parameter) / (the number of words in this topic * document topic parameter)*(this document. The number of words in this topic + topic word parameter) / (the total number of words in the document + topic word parameter);

[0127]

[0128] The document topic parameter refers to a pre-configured fixed parameter, for example, it may include but is not limited to: the prior probability of the document corresponding to the topic and the prior probability of the topic corresponding to the word, etc.

[0129] The calculation results are as follows: The data of each production anomaly to be trained corresponds to the abnormal topic:

[0130] [t0:"hole damage":0.33166458072590743,t1,:"mold wear":0.33166458072590704,t2:"mold debris":0.33667083854818486],

[0131] [t0:"hole damage":0.31108329856765427,t1,:"mold wear":0.31713252676957376,t2:"mold debris":0.3717841746627707],

[0132] [t0:"hole damage":0.33826373876436444,t1,:"mold wear":0.34986915462509766,t2:"mold debris":0.3118671066105362],

[0133] [t0:"hole damage":0.324644936605538,t1,:"mold wear":0.3572944441421309,t2:"mold debris":0.31806061925232504],

[0134] [t0:"hole damage":0.3505944931163959,t1,:"mold wear":0.3151856487275764,t2:"mold debris":0.33421985815602856],

[0135] [t0:"hole damage":0.3245723821443472,t1,:"mold wear":0.3310387984981228,t2:"mold debris":0.3443888193575311];

[0136] The probability that each abnormal topic contains each industrial production word is:

[0137] t0:"hole bad point":[0:"mold":0.06305070437026689,1:"hole":0.13970284750794706,2:"bad point":0.1734090477739382,3:"fixed point":0.0911761589991968,4:"damage":0.134282685182 0512,5:"Vulcanization":0.13919013972552696,6:"Fixed":0.1544743336813598,7:"Position":0.035754861753402,8:"Crush":0.03245892847649513,9:"Bad":0.036500292529816174],

[0138] t1,:"Mold wear":[0:"Mold":0.06357301792941572,1:"Hole":0.11679550243360731,2:"Bad point":0.18479298303698274,3:"Fixed point":0.09432102621863601,4:"Damage":0.121618979728 05673,5:"Vulcanization":0.1341232232409735,6:"Fixed":0.18157804737153427,7:"Position":0.03470533697390825,8:"Crush":0.03526939392007947,9:"Bad":0.03322248914680634],

[0139] t2:"Mold debris":[0:"Mold":0.0579475374902893,1:"Hole":0.11151765731509657,2:"Bad point":0.21616831299950795,3:"Fixed point":0.08878430158976604,4:"Damage":0.150191479184 49408,5:"Vulcanization":0.1287867191843146,6:"Fixed":0.14821450776934267,7:"Position":0.03270289409443318,8:"Crush":0.0332868513086044,9:"Bad":0.03239973906415115];

[0140] Step 3: Determination of target recommendation production anomalies

[0141] Step 31: Segment and vectorize the production abnormality text to be processed to obtain an actual industrial production word matrix;

[0142] Step 32: Use the thesaurus and the subject library of the calculated abnormal subjects to calculate the subject distribution probability of each abnormal subject in the production abnormal text to be processed. The result is as follows:

[0143] [t0:"hole damage":0.32847309135420585,t1,:"mold wear":0.311526908635794,t2:"mold debris":0.36000000001000015];

[0144] Step 33, selecting the most likely abnormal topic according to the topic distribution probability tp of each abnormal topic in each trained production abnormality;

[0145] Traverse the abnormal topic array of the trained production anomaly, find the maximum probability and its corresponding topic index, and return the probability mapping of the industrial production words of the abnormal topic with the highest probability;

[0146] The specific calculation formula is: actual matching probability value = 1-abs(new article t0-theme t0 of training article)-abs(new article t1-theme t1 of training article)-abs(new article t2-theme t2 of training article);

[0147] The best fit of the above theme is d2, and the specific calculation is as follows: 1-abs(0.32847309135420585-0.31108329856765427)-abs(0.311526908635794-0.31713252676957376)-abs(0.36000000001000015-0.3717841746627707)=1-0.01738979278655158-0.00560561813377976-0.01178417465277055=0.96522041442689811;

[0148] From this, we can see that d2 is the target recommended production exception.

[0149] The method for recommending product production quality anomalies applicable to production anomalies of multiple categories proposed in the present invention can effectively resolve word ambiguity and process semantic information brought by multi-word phrases, complex syntactic structures, etc.; and effectively resolve semantic differences in related words used by different personnel backgrounds and different business domains, and identify the expression intentions of different people, so as to avoid semantic problems such as different categories using the same word with different meanings and different personnel using the same word with different meanings; and can model rules through the content of historical anomaly libraries, fully identify differences in different categories, different problems, different processing methods, etc., and make effective recommendations; effectively use historical data to quickly locate anomalies and provide effective solutions, thereby improving the efficiency of end-user exception processing and reducing the experience and ability requirements of relevant personnel.

[0150] In one embodiment, Figure 3 Schematic diagram of a recommended device for detecting abnormal product production quality provided by an embodiment of the present invention. Figure 3 As shown, the device includes: a distribution probability determination module 310 and an abnormality determination module 320.

[0151] The distribution probability determination module 310 is used to determine the topic distribution probability of each abnormal topic in the production abnormality text to be processed; the abnormal topic includes but is not limited to one of the following: product quality abnormality, operation process abnormality, production tool abnormality;

[0152] The anomaly determination module 320 is used to determine the target recommended production anomaly based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the production anomaly text to be processed and the topic distribution probability of each abnormal topic in the production anomaly text to be processed.

[0153] In one embodiment, the distribution probability determination module 310 includes:

[0154] The word segmentation unit is used to segment the production abnormality text to be processed to obtain the corresponding actual initial industrial production word set;

[0155] A filtering unit is used to filter out invalid words in the actual initial industrial production word set to obtain a corresponding actual target industrial production word set; wherein the actual target industrial production word set includes but is not limited to words in at least one of the following aspects: production tools, product structure, product quality and process;

[0156] A deduplication and vectorization unit, used to deduplicate and vectorize the words in the actual target industrial production word set to obtain a corresponding actual industrial production word matrix;

[0157] The first determination unit is used to input the actual industrial production word matrix, the actual abnormal topic set, the number of industrial production words contained in the actual industrial production word matrix, and the number of abnormal topics contained in the actual abnormal topic set associated with the production abnormal text to be processed into a pre-created target abnormal topic distribution model to obtain the topic distribution probability of each abnormal topic in the production abnormal text to be processed.

[0158] In one embodiment, the abnormality determination module 320 includes:

[0159] A second determination unit is used to determine a matching probability difference between a topic distribution probability of each abnormal topic in the production abnormality text to be processed and a topic distribution probability of a corresponding abnormal topic in each trained production abnormality;

[0160] A third determination unit, configured to determine an actual matching probability value between the production exception text to be processed and the corresponding trained production exception based on the matching probability difference value associated with each trained production exception;

[0161] The fourth determining unit is configured to use the trained production anomaly with the largest actual matching probability value as a target recommended production anomaly.

[0162] In one embodiment, the process of constructing the target abnormal topic distribution model includes:

[0163] Determine a pre-acquired training industrial production word matrix for each production anomaly to be trained;

[0164] Randomly initialize the abnormal topic of each industrial production word in the training industrial production word matrix associated with each production abnormality to be trained, and obtain the abnormal topic matrix corresponding to each production abnormality to be trained;

[0165] Determine the number of times each industrial production word in the training industrial production word matrix corresponds to each abnormal topic, and the number of times each production abnormality to be trained corresponds to each abnormal topic based on the abnormal topic matrix of each production abnormality to be trained;

[0166] Determine the total number of corresponding abnormal topics based on the number of times all production abnormalities to be trained correspond to each abnormal topic;

[0167] The total number of times each abnormal topic occurs, the number of times each production abnormality to be trained corresponds to each abnormal topic, the number of times each industrial production word corresponds to each abnormal topic, and the number of abnormal topics are input into the pre-created initial abnormal topic distribution model for iterative training until the output topic distribution probability of each industrial production word tends to be stable, and the corresponding target abnormal topic distribution model is obtained.

[0168] In one embodiment, determining a pre-acquired training industrial production word matrix for each production anomaly to be trained is specifically used for:

[0169] Segment each pre-acquired production anomaly to be trained to obtain a corresponding initial industrial production word set for training;

[0170] Filter out invalid words in the initial training industrial production word set to obtain the corresponding training target industrial production word set;

[0171] The words in the training target industrial production word set are deduplicated and vectorized to obtain a training industrial production word matrix corresponding to each production anomaly to be trained.

[0172] In one embodiment, the device for recommending abnormal product production quality further includes:

[0173] The form generation module is used to fill in the form of the production exception text to be processed based on the target recommended production exception, and obtain the target production exception form corresponding to the production exception text to be processed.

[0174] In one embodiment, the form generation module includes:

[0175] An identification and extraction unit, used to identify and extract each unfilled field in the production exception text to be processed, and the actual value of the unfilled field in the target recommended production exception;

[0176] The form generating unit is used to fill the actual value of each unfilled field into the corresponding field of the production exception form associated with the production exception text to be processed, so as to obtain the target production exception form corresponding to the production exception text to be processed.

[0177] The device for recommending abnormal product production quality provided in the embodiment of the present invention can execute the method for recommending abnormal product production quality provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0178] In one embodiment, Figure 4 is a structural block diagram of an electronic device provided by an embodiment of the present invention, such as Figure 4As shown, a schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0179] like Figure 4 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0180] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0181] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a recommended method for abnormal product production quality.

[0182] In some embodiments, the method for recommending abnormal product production quality may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the method for recommending abnormal product production quality described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the method for recommending abnormal product production quality by any other appropriate means (e.g., by means of firmware).

[0183] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0184] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0185] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0186] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0187] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0188] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0189] An embodiment of the present invention further provides a computer program product, including a computer program, which, when executed by a processor, can implement a method for recommending abnormal product production quality as provided in any embodiment of the present application.

[0190] In the process of implementation, the computer program product can be written in one or more programming languages ​​or a combination thereof to perform the computer program code of the present application, and the programming language includes an object-oriented programming language, such as Java, Smalltalk, C++, and also includes a conventional procedural programming language, such as "C" language or similar programming language. The program code can be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on the remote computer, or completely on the remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (for example, using an Internet service provider to connect through the Internet).

[0191] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0192] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for recommending abnormal product production quality, characterized in that: include: Determine the topic distribution probability of each abnormal topic in the production abnormality text to be processed; wherein the abnormal topic is used to characterize the abnormality type to which the production abnormality text to be processed belongs; the abnormal topic includes but is not limited to one of the following: product quality abnormality, operation process abnormality, production tool abnormality; The target recommended production anomaly is determined based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the production anomaly text to be processed and the topic distribution probability of each abnormal topic in the production anomaly text to be processed.

2. The method according to claim 1, characterized in that Determining the topic distribution probability of each abnormal topic in the production abnormal text to be processed includes: Perform word segmentation on the production abnormality text to be processed to obtain the corresponding actual initial industrial production word set; Filter out invalid words in the actual initial industrial production word set to obtain a corresponding actual target industrial production word set; wherein the actual target industrial production word set includes but is not limited to words in at least one of the following aspects: production tools, product structure, product quality and process; De-duplication and vectorization are performed on the industrial production words in the actual target industrial production word set to obtain a corresponding actual industrial production word matrix; The actual industrial production word matrix, the actual abnormal topic set, the number of industrial production words contained in the actual industrial production word matrix and the number of abnormal topics contained in the actual abnormal topic set associated with the production abnormal text to be processed are input into a pre-created target abnormal topic distribution model to obtain the topic distribution probability of each abnormal topic in the production abnormal text to be processed.

3. The method according to claim 1, characterized in that The determining of the target recommended production anomaly based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the production anomaly text to be processed and the topic distribution probability of each abnormal topic in the production anomaly text to be processed comprises: Determine a matching probability difference between a topic distribution probability of each abnormal topic in the to-be-processed production abnormality text and a topic distribution probability of a corresponding abnormal topic in each trained production abnormality; Determining an actual matching probability value between the production exception text to be processed and the corresponding trained production exception based on the matching probability difference value associated with each of the trained production exceptions; The trained production anomaly with the largest actual matching probability value is used as the target recommended production anomaly.

4. The method according to claim 2, characterized in that: The process of constructing the target abnormal topic distribution model includes: Determine a pre-acquired training industrial production word matrix for each production anomaly to be trained; Randomly initialize the abnormal topic of each industrial production word in the training industrial production word matrix associated with each production anomaly to be trained, to obtain an abnormal topic matrix corresponding to each production anomaly to be trained; Determine the number of times each industrial production word in the training industrial production word matrix corresponds to each abnormal topic, and the number of times each production abnormality to be trained corresponds to each abnormal topic based on the abnormal topic matrix of each production abnormality to be trained; Determine the total number of corresponding abnormal topics based on the number of times all the production abnormalities to be trained correspond to each abnormal topic; The total number of times each abnormal topic occurs, the number of times each production abnormality to be trained corresponds to each abnormal topic, the number of times each industrial production word corresponds to each abnormal topic, and the number of abnormal topics are input into the pre-created initial abnormal topic distribution model for iterative training until the output topic distribution probability of each industrial production word tends to be stable, and the corresponding target abnormal topic distribution model is obtained.

5. The method according to claim 4, characterized in that The step of determining the pre-acquired training industrial production word matrix for each production anomaly to be trained comprises: Segment each pre-acquired production anomaly to be trained to obtain a corresponding initial industrial production word set for training; Filtering out invalid words in the initial training industrial production word set to obtain a corresponding training target industrial production word set; The words in the training target industrial production word set are deduplicated and vectorized to obtain a training industrial production word matrix corresponding to each production anomaly to be trained.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: Based on the target recommended production exception, a form is filled in for the production exception text to be processed to obtain a target production exception form corresponding to the production exception text to be processed.

7. The method according to claim 6, characterized in that The step of filling a form with the production exception text to be processed based on the target recommended production exception to obtain a target production exception form corresponding to the production exception text to be processed includes: Identify and extract each unfilled field in the to-be-processed production exception text, and the actual value of the unfilled field in the target recommended production exception; The actual value of each unfilled field is filled into the corresponding field of the form associated with the production exception text to be processed, so as to obtain the target production exception form corresponding to the production exception text to be processed.

8. A recommendation device for abnormal product production quality, characterized in that: include: A distribution probability determination module is used to determine the topic distribution probability of each abnormal topic in the production abnormality text to be processed; wherein the abnormal topic is used to characterize the abnormal type to which the production abnormality text to be processed belongs; the abnormal topic includes but is not limited to one of the following: product quality abnormality; operation process abnormality; production tool abnormality; An anomaly determination module is used to determine a target recommended production anomaly based on the topic distribution probability of each abnormal topic in each trained production anomaly associated with the production anomaly text to be processed and the topic distribution probability of each abnormal topic in the production anomaly text to be processed.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the method for recommending abnormal product production quality according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the recommended method for abnormal product production quality according to any one of claims 1 to 7 when executed.

11. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the computer program implements the method for recommending abnormal product production quality according to any one of claims 1 to 7.

Images