Search

CN-122022337-A - Analysis management method for industrial production and assembly of products

CN122022337ACN 122022337 ACN122022337 ACN 122022337ACN-122022337-A

Abstract

The invention relates to the technical field of intelligent production scheduling of industrial products, in particular to an analysis management method for industrial production and assembly of products, which comprises a plurality of components, wherein the production period prediction step of the components comprises S1, the production period of the components on each production line is predicted based on each component and a plurality of production lines matched with each component, S2, the assembly period prediction step of the components is used for predicting the assembly period of at least two components on the assembly line matched with each component, S3, the full production period prediction step of the products is used for predicting the shortest product production scheme used for the full production period based on the optimal combination, so that the production line and the assembly line of the products are scheduled based on the production scheme.

Inventors

  • Nicholas Bickley

Assignees

  • 上海维式商务咨询有限公司

Dates

Publication Date
20260512
Application Date
20260130

Claims (10)

  1. 1. The analysis management method for the industrial production and assembly of the product is characterized by being applied to the scheduling management of a plurality of production lines and a plurality of assembly lines of the product to be produced, wherein the product to be produced comprises a plurality of components, and comprises S1, a component production period prediction step, wherein the production period prediction of the components predicts the production period of the components on each production line based on each component and a plurality of production lines matched with each component; s2, a component assembly period prediction step, wherein the component assembly period prediction step is used for predicting the assembly period of at least two components on an assembly line matched with the component assembly period prediction step; S3, a product full production cycle prediction step, wherein the product full production cycle prediction step is based on the predicted production cycle of each component on each production line, the predicted assembly cycle and the production assembly scheme of the product to be produced, which are characterized by the complete waiting time, and finally screens out the shortest product production scheme used in the full production cycle based on the optimal combination, so that the production scheme is used for scheduling the production line and the assembly line of the product.
  2. 2. The method for analysis and management of industrial production and assembly of products according to claim 1, wherein the step of predicting the production cycle of the S1 component comprises the following steps: s11, component production data feature construction The component production data feature comprises three parts, component task feature Component execution progress feature And real-time status features of a production line ; Component task feature Characterization of the component to be produced The occupation requirement for production resources is as follows: , Wherein, the As a component in the current production line Is a planned production total amount of (1); complexity of processing the assembly on a production line; Component execution progress feature Characterization of the component to be produced The current processing stage comprises the following components: , Wherein, the Is a component Working procedure completed on current production line Is the number of (3); Representing a predicted time span; real-time status feature of production line Characterization production line Middle (f) Assembly of process pairs The actual production capacity of (2) is constituted as follows: , , , , , Wherein, the Characterization line in the first The production and feeding buffer status of the working procedure comprises the length of a feeding queuing sequence Accumulated waiting time of feeding ; Characterization line in the first Machine tool state of the process including machine tool state code Length of time for feeding Real-time failure rate of production machine tool ; Characterization line in the first The production discharge buffer status of the working procedure comprises the length of a discharge queuing sequence Accumulated waiting time with discharging ; Characterization line in the first From one process to the next Intermediate transport state comprising transport sequence length And time consuming transport ; S12, component production data feature sampling Component production data feature By component task features Component execution progress feature And real-time status features of a production line The common components are as follows: , From the group consisting of Raw data set of individual samples In (1) by Sub-randomly sampling with put back to construct a sampling set with the same size as the original data set This process is repeated Secondary, finally generate Independent sampling data sets 。
  3. 3. The method of analytical management of production and assembly of a product industry as claimed in claim 2, wherein the step of S1 component production cycle prediction further comprises: s13, component production data feature extraction Each sample set Each depth feature extractor comprises an encoder and a decoder, wherein the encoder performs data feature conversion from an input layer to a hidden layer, and the decoder performs data feature conversion from the hidden layer to an output layer; In each depth feature extractor, the first Layer hidden layer to The mapping relationship between hidden layers follows: , Wherein, the And The weight matrix and bias vector of the depth feature extractor hidden layer respectively, Is an activation function; First, the The hidden layer features extracted by the depth feature extractor are marked as sub-feature vectors Will be The outputs of the parallel depth feature extractors are subjected to feature stitching to form a composite feature vector for prediction : 。
  4. 4. The method of analytical management of production and assembly of a product industry as claimed in claim 3 wherein the step of S1 component production cycle prediction further comprises: In training the depth feature extractor, a loss function is defined Directing its ability to acquire production data features: , Wherein, the For producing reconstruction error terms of data features by calculating the first of the encoders Individual component production data features And the first reconstructed from the decoder Individual component production data features The sum of the squares of the distances between them, Representing the total characteristic quantity of the production data; For weight decay regularization term, by computing weight matrices for each layer in encoder and decoder The inner elements are squared and summed while introducing weight decay coefficients Preventing the depth feature extractor from overfitting; maximizing terms for feature mutual information for enhancing information correlation between hidden layer features extracted by a depth feature extractor and original input features, wherein Weight coefficients are lost for mutual information.
  5. 5. The analysis management method for industrial production and assembly of products according to claim 4, wherein the characteristic mutual information maximization item The calculation mode of (2) is as follows: , , Wherein, the Representing a hidden layer feature vector set obtained after all the component production data features are encoded by a depth feature extractor; is the first Hidden layer feature vectors for individual components; is the first Transpose of hidden layer feature vectors of individual components; a set of production data feature vectors representing all components; is the first Individual component production data features; for scoring function for measuring hidden layer characteristics And component production data features Degree of matching between, when When representing the matching score of a positive sample pair, when A match score for a negative sample pair; and the scoring weight matrix is used for establishing a mapping relation between the hidden layer feature space and the input feature space.
  6. 6. The method of analytical management of production and assembly of a product industry as claimed in claim 5, wherein the step of S1 component production cycle prediction further comprises: S14, extracting composite characteristic vectors through a residual error network Performing nonlinear mapping to fit the production cycle prediction value; In the residual block, the first Layer residual block to th The mapping relationship between the layer residual blocks follows: , Wherein, the A weight matrix representing the residual block; representing a nonlinear residual mapping operator; activating a function for sigmoid; After the residual error network carries out nonlinear mapping on the composite feature vector, the feature space is further reconstructed and dimension reduced through two full-connection layers, and the calculation mode is as follows: , Wherein, the And The weight matrix and the bias vector of the full connection layer are respectively, Activating a function for sigmoid; s15, predicting production cycle of component Residual network converts component production data characteristics into predicted production progress The predicted production progress is output by the last fully-connected layer, and the output predicted production progress Is used to calculate regression loss to guide parameter updates of the residual network model The calculation is as follows: , Wherein, the The actual assembly production progress is; To predict assembly production progress; to train the total amount of component samples, output-based predictive production progress Achieving assembly production cycle through iteration First, according to the predicted production progress Executing progress features at a component Medium adjustment prediction time span When meeting the following requirements I.e. when the assembly production is completed, the corresponding predicted time span Namely the production cycle of the components under the current production line 。
  7. 7. The method for analytical management of industrial production and assembly of products according to claim 1, wherein the step of predicting the assembly cycle of the S2 component comprises: S21, component assembly data feature construction Component assembly data features Comprising four parts, component-mounting task features Component assembly execution progress feature Real-time status feature of assembly line And component alignment status feature ; , Assembly task feature Characterization of the product to be assembled The occupation requirement for assembly resources is constituted as follows: , Wherein, the For products currently in the assembly line Is a planned assembly amount of (a); assembling a complexity coefficient for the product; The number of the component types is as follows; component assembly execution progress feature Characterization of the product to be assembled The assembly stage currently in use comprises the following components: , Wherein, the Is a product The assembly process being completed on the current assembly line Is the number of (3); Representing an assembly prediction time span, and fitting out the minimum time required for enabling the assembly progress to reach a finished state by dynamically adjusting the time span in the assembly period prediction of the assembly; Real-time status feature of assembly station Characterization assembly line Middle (f) The actual assembly capacity of the assembly process is constituted as follows: , , , , Wherein, the Characterization of Assembly line on the first An assembly component buffer status for an assembly process comprising a queuing sequence length for each component Accumulated waiting time of each component Batch number of material ; Characterization of Assembly line on the first The state of the assembly machine tool in the assembly process comprises the state code of the assembly machine tool The current assembly has long time Real-time failure rate of an assembly machine tool ; Characterization of Assembly line on the first The output status of the assembly process includes the length of the output queuing sequence Accumulated wait time with output ; Alignment status feature of each component Characterization of the product The production of the required components is completed, and the components are as follows: , Wherein, the Is a collection of components; a set of predicted production cycles for each component, the value being obtained by a component production cycle prediction calculation; The ready rate vector for each component is the ratio of the number of finished components to the total number of components planned to be produced.
  8. 8. The analysis management method for industrial production and assembly of a product according to claim 7, wherein the S2 component assembly cycle prediction step further comprises: s22, component assembly data feature extraction based on feature migration In the component assembly period prediction, a depth feature encoder with the same structure as the component production period prediction is adopted, encoder parameters are simultaneously migrated to a depth feature extractor in the assembly period prediction, component assembly data features are input to realize fine adjustment, and a loss function of the fine adjustment is defined The method comprises the following steps: , Wherein, the To assemble the reconstruction error term of the data feature, the encoder is calculated by the first Personal component assembly data features And the first reconstructed from the decoder Individual component production data features The sum of the squares of the distances between them, Representing the total feature of the component mounting data, To migrate regularization terms, depth feature extractor weight matrices in assembly cycles are constrained Depth feature extractor weight matrix in production cycle To achieve migration of similar features; For weight decay regularization term, by computing weight matrices for each layer in encoder and decoder The inner elements are squared and summed while introducing weight decay coefficients The depth feature extractor is prevented from overfitting.
  9. 9. The analysis management method for industrial production and assembly of a product according to claim 8, wherein the S2 component assembly cycle prediction step further comprises: S23, assembling and sleeving constraint Before the assembly period of the components is predicted, the state of the components in the complete set is required to be judged, and the time of each component is defined Alignment determination function The method comprises the following steps: , Wherein, the Taking 1 when the condition is satisfied, otherwise taking 0; Is a product Middle (f) Ready rate of class component(s); th Is the first A minimum nesting threshold for class components; When (when) When, i.e. the product If the complete set condition is met, the assembly period can be predicted, otherwise, the assembly period is waited until the latest assembly is finished, and the assembly period is time Waiting time for each component to be sleeved The calculation is as follows: , Wherein, the Is the first The production cycle prediction value of the class component is obtained by calculation in the component production cycle prediction step; Is an indication function; Is a product Middle (f) Ready rate of class component(s); th Is the first Minimum flush threshold for class components.
  10. 10. The analysis management method for industrial production and assembly of a product according to claim 9, wherein the S2 component assembly cycle prediction step further comprises: s24, predicting assembly cycle of component The assembly period prediction network of the components adopts the residual network structure which is the same as the assembly production period prediction, and finally outputs the predicted assembly progress Regression loss for assembly cycle prediction The calculation is as follows: , Wherein, the The assembly progress of the actual assembly is determined; To predict assembly progress; Assembling a sample total for training; Output-based predicted assembly progress The assembly period of the product is realized through iteration Is to comprehensively consider the nesting waiting time of each component and the product Is a total assembly period of (a) The calculation is as follows: , Wherein, the Nesting waiting time for the assembly; predicting values for assembly cycles for components by dynamically adjusting predicted time spans in assembly execution progress features When meeting the following requirements When the assembly of the component is completed, i.e. when the corresponding predicted time span I.e. the predicted value of the assembly period of the component 。

Description

Analysis management method for industrial production and assembly of products Technical Field The invention relates to the technical field of intelligent production scheduling of products, in particular to an analysis management method for industrial production and assembly of products. Background The product production assembly scheduling means that under the condition of limited resources, the production sequence, the assembly sequence, the production line, the assembly line and the like are arranged for the product, so that the production period is shortened, and the resource utilization rate is maximized. However, in the prior art, the production, assembly and scheduling of products mostly depend on expert experience, and whether each production line or each assembly line may have a certain production variability (such as a certain production line changes equipment, the production capacity is greatly improved) at different periods, so that the judgment depending on manual experience has uncertainty, low intelligent degree and high requirement on personnel expertise, and currently, there are few prediction schemes on collaborative scheduling of products to be produced in units of components on each production line and each assembly line on the market. Disclosure of Invention The invention aims to provide an analysis management method for industrial production and assembly of products, so as to realize accurate and efficient prediction of the whole production period of the products to be produced. The analysis management method for the industrial production and assembly of the product is applied to the scheduling management of a plurality of production lines and a plurality of assembly lines of the product to be produced, wherein the product to be produced comprises a plurality of components, and comprises S1, a component production period prediction step, wherein the production period prediction of the components predicts the production period of the components on each production line based on each component and a plurality of production lines matched with each component; s2, a component assembly period prediction step, wherein the component assembly period prediction step is used for predicting the assembly period of at least two components on an assembly line matched with the component assembly period prediction step; S3, a product full production cycle prediction step, wherein the product full production cycle prediction step is based on the predicted production cycle of each component on each production line, the predicted assembly cycle and the production assembly scheme of the product to be produced, which are characterized by the complete waiting time, and finally screens out the shortest product production scheme used in the full production cycle based on the optimal combination, so that the production scheme is used for scheduling the production line and the assembly line of the product. According to the product industrial production and assembly period prediction method, accurate prediction of the whole production period of the product is achieved through cooperation of the component production period prediction module, the component assembly period prediction module and the whole production period prediction module. Drawings FIG. 1 is a schematic flow chart of an analytical management method for industrial production and assembly of the product of the present invention. Detailed Description The following description of the preferred embodiments of the present invention is provided in connection with the accompanying drawings, and it should be understood that the preferred embodiments described below are for illustration only and are not intended to limit the scope of the present invention. The invention is further described in detail below with reference to the preferred embodiments: as shown in figure 1, the analysis management method for industrial production and assembly of the product is applied to the scheduling management of a plurality of production lines and a plurality of assembly lines of the product to be produced, wherein the product to be produced comprises a plurality of components, and comprises S1, a component production period prediction step, wherein the production period prediction of the components predicts the production period of the components on each production line based on each component and a plurality of production lines matched with each component; s2, a component assembly period prediction step, wherein the component assembly period prediction step is used for predicting the assembly period of at least two components on an assembly line matched with the component assembly period prediction step; S3, a product full production cycle prediction step, wherein the product full production cycle prediction step is based on the predicted production cycle of each component on each production line, the predicted assembly cycle and the production assembly scheme of the produ