Battery production equipment health state evaluation system based on multi-dimensional data

Abstract

The invention discloses a battery production equipment health state evaluation system based on multi-dimensional data, and relates to the technical field of equipment evaluation. The method comprises the following steps: dividing a production process of a battery into different production nodes, respectively constructing a digital twin model of each production node, obtaining fault events of each production node and fault data of different fault events, obtaining fault data characteristics according to the fault data of different fault events, and constructing a fault evaluation model; respectively setting a monitoring data threshold value for each production node, obtaining the equipment state of each production node, and judging whether a fault event occurs in the corresponding production node or not according to the evaluation data characteristics of the production equipment by using the fault evaluation model; an effective prediction mechanism can be provided for whether a potential fault event exists before the battery leaves the factory on the premise of ensuring the health of the equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of equipment evaluation, and in particular to a battery production equipment health status evaluation system based on multidimensional data. Background Art

[0002] During the battery production process, there are often many objective factors that affect the production process. These factors not only affect the health of the production equipment itself, but also cause potential defects in the processed batteries, which may lead to subsequent failures. How to deal with this situation has become a necessary technical issue.

[0003] Battery production equipment often has multiple parameters that can be monitored during its operation, and the operating status of the equipment will be directly reflected in these parameters. If the fault types of different faulty batteries can be combined with the changes in various parameters during their processing, it may be possible to provide an effective prediction mechanism for whether there are potential faults in the batteries processed by the operating production equipment. To this end, the present invention provides a battery production equipment health status assessment system based on multidimensional data. Summary of the Invention

[0004] The purpose of the present invention is to provide a battery production equipment health status assessment system based on multi-dimensional data.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A battery production equipment health status assessment system based on multidimensional data includes the following modules:

[0006] The node division module is used to divide the battery production process into different production nodes, obtain the basic information and monitoring data of the production equipment at each production node, build a digital twin model for each production node, and obtain the digital twin parameters of the production equipment;

[0007] The data acquisition module is used to obtain fault events of each production node, and obtain fault data of different fault events in combination with the production equipment of the production node, and obtain corresponding fault data features based on the fault data of different fault events;

[0008] The model building module is used to build corresponding fault assessment models based on the fault data characteristics of different fault events and the digital twin parameters of production equipment;

[0009] The first evaluation module is used to set corresponding monitoring data thresholds for each production node and obtain the equipment status of each production node based on the monitoring data of the production equipment;

[0010] The second evaluation module is used to obtain the evaluation data characteristics of the production equipment and use the fault evaluation model to determine whether a fault event will occur in the corresponding production node.

[0011] Furthermore, the process of dividing the battery production process into different production nodes and obtaining basic information and monitoring data of the production equipment at each production node includes:

[0012] The production process refers to the entire process from raw material processing to finished product of the battery, including different production equipment, and the processing process of different production equipment is regarded as a production node;

[0013] The basic information refers to the specifications of the production equipment at each production node, and a physical model of the corresponding production node is constructed using a three-dimensional modeling tool based on the specifications of each production equipment;

[0014] The monitoring data includes first-class monitoring data and second-class monitoring data. The first-class monitoring data refers to the equipment operating parameters of each production equipment, and the second-class monitoring data refers to the process quality parameters of each production equipment.

[0015] Furthermore, the process of constructing digital twin models for each production node and obtaining digital twin parameters of each production equipment includes:

[0016] Using simulation software to simulate the working process of the production equipment on the physical model of the production node to obtain a simulation model, uploading a type of monitoring data of the production equipment to the simulation model for synchronization, and adjusting the values ​​of various simulation parameters during the simulation process;

[0017] The simulation parameters refer to adjustable parameters in the production equipment other than the first and second type of monitoring data, and the second type of monitoring data of the production equipment in the simulation model is obtained under different values ​​of the simulation parameters;

[0018] When the second-category monitoring data in the simulation model is the same as the second-category monitoring data at the corresponding moment of the synchronized first-category monitoring data, the simulation model at this time is used as the digital twin model of the corresponding production node, and the simulation parameters of the corresponding values ​​are used as digital twin parameters.

[0019] Furthermore, the process of obtaining the fault events of each production node and obtaining the fault data of different fault events in combination with the production equipment of the production node includes:

[0020] The failure event refers to the type of failure caused by different factors during the battery processing process of the production equipment of each production node;

[0021] The monitoring data of the battery having a failure event during the processing of the battery by the production equipment is used as the failure data of the failure event, including type I failure data and type II failure data.

[0022] Furthermore, the process of obtaining corresponding fault data features based on the fault data of different fault events includes:

[0023] Preprocess the first-class fault data and second-class fault data of each type of fault event separately, including outlier processing, missing value processing, and normalization processing. Feature extraction is performed on the preprocessed first-class fault data and second-class fault data to obtain corresponding data features.

[0024] The data features include mean, standard deviation, peak value, kurtosis, fundamental frequency amplitude, sideband modulation, and harmonic energy ratio. The data features of type I fault data are marked as type I fault data features, and the data features of type II fault data are marked as type II fault data features.

[0025] Furthermore, the process of building a corresponding fault assessment model based on the fault data characteristics of different fault events and the digital twin parameters of the production equipment includes:

[0026] The first-class fault data features and second-class fault data features of different fault events of the same production node and the digital twin parameters of the production equipment of the production node are included in the fault evaluation set, and divided into a training set and a test set;

[0027] Construct a convolutional neural network, using the different fault data features and digital twin parameters in the training set as the input data of the convolutional neural network, and using whether a fault event will occur at the corresponding production node and what kind of fault event will occur as the output data of the convolutional neural network;

[0028] The convolutional neural network is trained to obtain an initial convolutional neural network, and the initial convolutional neural network is verified using a test set. The initial convolutional neural network with a test error threshold that is less than or equal to a preset test error threshold is output as the fault assessment model of the production node, and the fault assessment model of each production node is obtained respectively.

[0029] Furthermore, a monitoring data threshold is set for each production node, and the process of obtaining the equipment status of each production node in combination with the monitoring data of the production equipment includes:

[0030] A monitoring data threshold is set for each production node's Class A monitoring data, and the current Class A monitoring data of the production equipment of each production node is compared with its corresponding monitoring data threshold to obtain the equipment status of the corresponding production equipment, including normal status and abnormal status. The first feedback information is generated for the production equipment in the abnormal status and fed back to the relevant personnel.

[0031] Furthermore, the process of obtaining evaluation data features of production equipment and using the fault assessment model to determine whether a fault event will occur at the corresponding production node includes:

[0032] Set an evaluation cycle, use the monitoring data of production equipment in normal state in the most recent evaluation cycle as evaluation data, including Class I evaluation data and Class II evaluation data, and use the data features of Class I evaluation data and Class II evaluation data as Class I evaluation data features and Class II evaluation data features, respectively;

[0033] The first-class evaluation data features and second-class evaluation data features of a single production equipment and its digital twin parameters are input into the fault assessment model of the corresponding production node to determine whether a fault event will occur at the corresponding production node and what kind of fault event will occur, and generate corresponding second feedback information and feed it back to relevant personnel.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] By dividing the battery production process into different production nodes, the present invention constructs a digital twin model for each production node and obtains its corresponding digital twin parameters. This can take into account the differences between different production equipment in actual application scenarios, which is conducive to reducing errors in subsequent fault assessment of production nodes.

[0036] By obtaining the failure events that may be caused by different production nodes and the failure data and failure data characteristics corresponding to different failure events, it is possible to build a failure assessment model for different production nodes. This is conducive to judging whether each production node will cause a battery failure event based on the changes in various monitoring data during the operation process. It can provide an effective prediction mechanism for whether there will be potential failure events in the battery before leaving the factory under the premise of ensuring the health of the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a module schematic diagram of the present invention. DETAILED DESCRIPTION

[0038] like Figure 1 As shown in the figure, the battery production equipment health status assessment system based on multi-dimensional data includes the following modules:

[0039] The node division module is used to divide the battery production process into different production nodes, obtain the basic information and monitoring data of the production equipment at each production node, build a digital twin model for each production node, and obtain the digital twin parameters of the production equipment;

[0040] The data acquisition module is used to obtain fault events of each production node, and obtain fault data of different fault events in combination with the production equipment of the production node, and obtain corresponding fault data features based on the fault data of different fault events;

[0041] The model building module is used to build corresponding fault assessment models based on the fault data characteristics of different fault events and the digital twin parameters of production equipment;

[0042] The first evaluation module is used to set corresponding monitoring data thresholds for each production node and obtain the equipment status of each production node based on the monitoring data of the production equipment;

[0043] The second evaluation module is used to obtain the evaluation data characteristics of the production equipment and use the fault evaluation model to determine whether a fault event will occur in the corresponding production node.

[0044] It should be further explained that, in the specific implementation process, the process of dividing the battery production process into different production nodes and obtaining the basic information and monitoring data of the production equipment at each production node includes:

[0045] The production process refers to the entire process of processing the battery from raw materials to finished products, including different process links and production equipment, such as mixers, coaters, roller presses, slitters, winders, liquid injection machines, and welding machines;

[0046] The processing of production equipment in different process links is regarded as a production node, and each production node is bound to its production equipment. This method can divide the entire battery production process into several different production nodes.

[0047] The basic information refers to the specifications of the production equipment at each production node, including all relevant data necessary for building a physical model. A 3D modeling tool is used to build a physical model of each production node based on the specifications of each production equipment.

[0048] In actual application scenarios, monitoring data of production equipment at each production node is obtained separately, including first-class monitoring data and second-class monitoring data. The first-class monitoring data refers to the equipment operating parameters of each production equipment, such as current, voltage, speed, pressure, temperature, amplitude, and noise. The second-class monitoring data refers to the process quality parameters of each production equipment.

[0049] Among them, the process quality parameters of the mixer include viscosity, solid content, and particle fineness; the process quality parameters of the coater include coating adhesion, surface density, and thickness; the process quality parameters of the roller press include compaction density, thickness rebound rate, and electrode crack rate;

[0050] The process quality parameters of the slitting machine include burr height and width tolerance; the process quality parameters of the winding machine include winding needle alignment and winding core ovality; the process quality parameters of the filling machine include filling weight deviation and electrolyte infiltration rate; the process quality parameters of the welding machine include air tightness and weld penetration depth.

[0051] It should be further explained that, in the specific implementation process, the process of constructing the digital twin model of each production node and obtaining the digital twin parameters of the production equipment includes:

[0052] Taking the physical model of a single production node as an example, simulation software is used to simulate the working process of the production equipment based on the physical model to obtain the corresponding simulation model. The monitoring data of the production equipment is uploaded to the simulation model for synchronization, and the values ​​of various simulation parameters in the simulation process are adjusted;

[0053] The simulation parameters refer to the inertia parameters, electromagnetic parameters, thermodynamic parameters, environmental parameters, and performance degradation parameters of the production equipment in actual application scenarios. The simulation parameters include all adjustable parameters except the first-class monitoring data and the second-class monitoring data. The second-class monitoring data of the production equipment in the simulation model is obtained under different values ​​of the simulation parameters;

[0054] When the second-category monitoring data in the simulation model is the same as the second-category monitoring data at the corresponding moment of the synchronized first-category monitoring data, the simulation model at this time is used as the digital twin model of the production node, and the simulation parameters of the corresponding values ​​are used as the digital twin parameters of the production equipment. The same method is used to obtain the digital twin model and digital twin parameters of each production node respectively.

[0055] It should be further explained that, in a specific implementation process, the process of obtaining the fault events of each production node and obtaining the fault data of different fault events in combination with the production equipment of the production node includes:

[0056] In an embodiment of the present invention, fault events of each production node are obtained. The fault events refer to the types of faults caused by different factors during the battery processing process of the production equipment of each production node, which can be obtained through historical data of faulty batteries.

[0057] Specifically, failure events of the mixer include cycle capacity drop, rate performance degradation, and self-discharge rate exceeding the standard. Failure events of the coater include local overcharge / overdischarge, gas bulging, and cycle expansion and rupture. Failure events of the roller press include poor capacity consistency, cycle lithium deposition, and difficulty in cell assembly.

[0058] Failure events of the slitting machine include micro-short circuit self-discharge and internal short circuit of the battery cell. Failure events of the winding machine include accelerated cycle life attenuation and low-temperature performance degradation. Failure events of the filling machine include gas generation during high-temperature storage and poor interface infiltration. Failure events of the welding machine include electrolyte leakage and increased connection impedance.

[0059] Taking a single fault event as an example, the monitoring data of the corresponding battery during the processing of the production equipment where the fault event occurred is used as fault data, including Class I fault data and Class II fault data. The Class I fault data includes current, voltage, speed, pressure, temperature, amplitude, and noise. The Class II fault data refers to the process quality parameters of the production equipment of the production node corresponding to the fault event.

[0060] It should be further explained that, in a specific implementation process, the process of obtaining corresponding fault data features based on fault data of different fault events includes:

[0061] Taking the fault data of a single fault event as an example, the first-class fault data and the second-class fault data are preprocessed separately, including outlier processing, missing value processing, and normalization processing. The outlier processing is used to clean up abnormal data, the missing value processing is used to fill missing data, and the normalization processing is used to unify the data format.

[0062] Since the preprocessed Class I fault data and Class II fault data are both quantized time series data, feature extraction can be performed on them to obtain corresponding data features, including mean, standard deviation, peak value, kurtosis, fundamental frequency amplitude, sideband modulation, and harmonic energy ratio. The data features of Class I fault data are marked as Class I fault data features, and the data features of Class II fault data are marked as Class II fault data features.

[0063] It should be further explained that, in the specific implementation process, the process of building a corresponding fault assessment model based on the fault data characteristics of different fault events and the digital twin parameters of production equipment includes:

[0064] The fault data features of different fault events of the same production node (including first-class fault data features and second-class fault data features) and the digital twin parameters of the production equipment of the production node are included in the fault evaluation set, and divided into a training set and a test set;

[0065] Construct a convolutional neural network, using the different fault data features and digital twin parameters in the training set as input data for the convolutional neural network, and using whether a fault event will occur at the corresponding production node and what kind of fault event will occur as output data for the convolutional neural network;

[0066] The convolutional neural network is trained to obtain an initial convolutional neural network, and the test set is used to verify the model of the initial convolutional neural network. The initial convolutional neural network with a test error threshold less than or equal to the preset test error threshold is output as the fault assessment model of the production node. The same method is used to obtain the fault assessment model of each production node.

[0067] It should be further explained that, in the specific implementation process, the monitoring data threshold is set for each production node, and the process of obtaining the equipment status of each production node by combining the monitoring data of the production equipment includes:

[0068] Set corresponding monitoring data thresholds for the first-class monitoring data of each production node, including current threshold, voltage threshold, speed threshold, pressure threshold, temperature threshold, amplitude threshold, and noise threshold, and compare the current first-class monitoring data of the production equipment of each production node with its corresponding monitoring data thresholds;

[0069] If a type of monitoring data is less than or equal to its monitoring data threshold, the corresponding production equipment is marked as normal. If a type of monitoring data is greater than its monitoring data threshold, the corresponding production equipment is marked as abnormal. The equipment status includes normal status and abnormal status. Corresponding first feedback information is generated for the production equipment in the abnormal state and fed back to the relevant personnel.

[0070] It should be further explained that, in the specific implementation process, the process of obtaining the evaluation data characteristics of the production equipment and using the fault assessment model to determine whether a fault event will occur at the corresponding production node includes:

[0071] Set an evaluation cycle, use the monitoring data of production equipment in normal state in the most recent evaluation cycle as evaluation data, including Class I evaluation data and Class II evaluation data, and use the data features of Class I evaluation data and Class II evaluation data as Class I evaluation data features and Class II evaluation data features, respectively;

[0072] The evaluation data characteristics (including Class I evaluation data characteristics and Class II evaluation data characteristics) of a single production equipment and its digital twin parameters are input into the fault assessment model of the corresponding production node. The fault assessment model is used to determine whether a fault event will occur at the corresponding production node. If it is determined that a fault event will occur, the corresponding second feedback information is generated in combination with the determined fault event and fed back to the relevant personnel.

[0073] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A battery production equipment health status assessment system based on multidimensional data, characterized by: Includes the following modules: The node division module is used to divide the battery production process into different production nodes, obtain the basic information and monitoring data of the production equipment at each production node, build a digital twin model for each production node, and obtain the digital twin parameters of the production equipment; The data acquisition module is used to obtain fault events of each production node, and obtain fault data of different fault events in combination with the production equipment of the production node, and obtain corresponding fault data features based on the fault data of different fault events; The model building module is used to build corresponding fault assessment models based on the fault data characteristics of different fault events and the digital twin parameters of production equipment; The first evaluation module is used to set corresponding monitoring data thresholds for each production node and obtain the equipment status of each production node based on the monitoring data of the production equipment; The second evaluation module is used to obtain the evaluation data characteristics of the production equipment and use the fault evaluation model to determine whether a fault event will occur in the corresponding production node.

2. The battery production equipment health status assessment system based on multidimensional data according to claim 1, characterized in that: The process of obtaining basic information and monitoring data of production equipment at each production node includes: The production process refers to the entire process from raw material processing to finished product of the battery, including different production equipment, and the processing process of different production equipment is regarded as a production node; The basic information refers to the specifications of the production equipment at each production node, and a physical model of the corresponding production node is constructed using a three-dimensional modeling tool based on the specifications of each production equipment; The monitoring data includes first-class monitoring data and second-class monitoring data. The first-class monitoring data refers to the equipment operating parameters of each production equipment, and the second-class monitoring data refers to the process quality parameters of each production equipment.

3. The battery production equipment health status assessment system based on multidimensional data according to claim 2, characterized in that: The process of building a digital twin model of a production node and obtaining digital twin parameters includes: Using simulation software to simulate the working process of the production equipment on the physical model of the production node to obtain a simulation model, uploading a type of monitoring data of the production equipment to the simulation model for synchronization, and adjusting the values ​​of various simulation parameters during the simulation process; The simulation parameters refer to adjustable parameters in the production equipment other than the first and second type of monitoring data, and the second type of monitoring data of the production equipment in the simulation model is obtained under different values ​​of the simulation parameters; When the second-category monitoring data in the simulation model is the same as the second-category monitoring data at the corresponding moment of the synchronized first-category monitoring data, the simulation model at this time is used as the digital twin model of the corresponding production node, and the simulation parameters of the corresponding values ​​are used as digital twin parameters.

4. The battery production equipment health status assessment system based on multidimensional data according to claim 3, characterized in that: The process of obtaining fault events and fault data for different fault events includes: The failure event refers to the type of failure caused by different factors during the battery processing process of the production equipment of each production node; The monitoring data of the battery having a failure event during the processing of the battery by the production equipment is used as the failure data of the failure event, including type I failure data and type II failure data.

5. The battery production equipment health status assessment system based on multidimensional data according to claim 4, characterized in that: The process of obtaining fault data characteristics includes: Preprocess the first-class fault data and second-class fault data of each type of fault event separately, including outlier processing, missing value processing, and normalization processing. Feature extraction is performed on the preprocessed first-class fault data and second-class fault data to obtain corresponding data features. The data features include mean, standard deviation, peak value, kurtosis, fundamental frequency amplitude, sideband modulation, and harmonic energy ratio. The data features of type I fault data are marked as type I fault data features, and the data features of type II fault data are marked as type II fault data features.

6. The battery production equipment health status assessment system based on multidimensional data according to claim 5, characterized in that: The process of building a fault assessment model includes: The first-class fault data features and second-class fault data features of different fault events of the same production node and the digital twin parameters of the production equipment of the production node are included in the fault evaluation set, and divided into a training set and a test set; Construct a convolutional neural network, using the different fault data features and digital twin parameters in the training set as the input data of the convolutional neural network, and using whether a fault event will occur at the corresponding production node and what kind of fault event will occur as the output data of the convolutional neural network; The convolutional neural network is trained to obtain an initial convolutional neural network, and the initial convolutional neural network is verified using a test set. The initial convolutional neural network with a test error threshold that is less than or equal to a preset test error threshold is output as the fault assessment model of the production node, and the fault assessment model of each production node is obtained respectively.

7. The battery production equipment health status assessment system based on multidimensional data according to claim 6, characterized in that: The process of setting monitoring data thresholds and obtaining the device status of each production node includes: A monitoring data threshold is set for each production node's Class A monitoring data, and the current Class A monitoring data of the production equipment of each production node is compared with its corresponding monitoring data threshold to obtain the equipment status of the corresponding production equipment, including normal status and abnormal status. The first feedback information is generated for the production equipment in the abnormal status and fed back to the relevant personnel.

8. The battery production equipment health status assessment system based on multidimensional data according to claim 7, characterized in that: The process of obtaining evaluation data features and determining whether a failure event will occur includes: Set an evaluation cycle, use the monitoring data of production equipment in normal state in the most recent evaluation cycle as evaluation data, including Class I evaluation data and Class II evaluation data, and use the data features of Class I evaluation data and Class II evaluation data as Class I evaluation data features and Class II evaluation data features, respectively; The first-class evaluation data features and second-class evaluation data features of a single production equipment and its digital twin parameters are input into the fault assessment model of the corresponding production node to determine whether a fault event will occur at the corresponding production node and what kind of fault event will occur, and generate corresponding second feedback information and feed it back to relevant personnel.

Images